VOLUME 73 1986 ANNALS OF THE MISSOURI BOTANICAL GARDEN The ANNALS, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, t. Louis. Papers originating outside the Garden will also be ac- cepted. Authors should write the Editor for information concerning arrangements for publishing in the ANNALS. EDITORIAL COMMITTEE Morn, Editor CY Missouri Botanical Garden MARSHALL R. CROSBY Missouri Botanical Garden GERRIT DAVIDSE Missouri Botanical Garden JOHN D. ER Missouri Botanical Garden & St. Louis University PETER GOLDBLATT Missouri Botanical Garden Colophon This volume of the ANNALS of the Missouri Botanical Garden has been set in APS Times Roman. The text is set in 9 point type while the figure legends and literature cited sections are set in 8 point type. The volume has been printed on 70# Centura Gloss, an acid-free paper designed to have a shelf-life of over 100 years. Centura Gloss is manufactured by the Consolidated Paper Company. Photographs used in the ANNALS are reproduced using 300 line screen halftones. The binding used in the production of the ANNALS is a proprietary method known as Permanent Binding. The ANNALS is printed and distributed by Allen Press, Inc. of Lawrence, Kansas 66044, U.S.A. © Missouri Botanical Garden 1986 ISSN 0026-6493 ——————— ——"———— SD T ANNALS ISSUUA BOTANICAL GARDEN IME 73 1986 NUMBER 1 CONTENTS Sudanian Elements in the Flora of Israel A. Shmida & J. A. Aronson ... 1 Systematic Foliar Morphology of FERAM (Euphorbiaceae). I. Con- spectus Geoffrey A. Levi | Systematic Foliar Morphology i Phyllanthoideae (Euphorbiaceae). II. Phe- netic Analysis Geoffrey A. Levin 86 Notes on the Floral Biology of Couroupita guianensis Aubl. (солдон) Sonia Yarsick, Nerida Xena ае Enrech, Nelson Ramirez, & Getulio Agostini 99 Convergent Evolution of the ‘Homeria’ Flower Type in Six New Species of Moraea (Iridaceae—Irideae) in Southern Africa Peter Goldblatt ........ 102 The Calyx in Lycianthes and Some Other Genera "mo DUC =. Serotaxonomy of Solanum, Capsicum, Dunalia, and Other Selected Solan- aceae Richard N. Lester & Philip A. Roberts Notes on the Systematics of Hesperantha (Iridaceae) in Tropical Africa Peter Goldblatt 134 117 Notes on Peruvian Palms Alwyn H. Gentry .... Мои BOTANICAL A Guide to Collecting Palms John Dransfield VOLUME 73 SPRING 1986 NUMBER 1 | ANNALS MISSOURI BOTANICAL GARDEN The ANNALs, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be ac- cepted. Authors should write the Editor for information concerning arrangements for publishing in the ANNALs. Instructions to Authors E are printed on the inside back cover of the first issue of this volume. __ EDITORIAL COMMITTEE Y Morin, Editor Missouri Botanical Garden CHERYL R. Bauer, Editorial Assistant Missouri Botanical Garden Mans L- R- CRO Mi issouri Meise Put GERRIT DAVIDSE Missouri Botanical Garden N D. Dw Missouri "vam la & = ous University PETER болт Missouri Botanical Garden _ For subscription informati сойыс Business Office of the Annals, P.O. Box 299, St. Louis, MO 63 LE volume U.S., ANNALS OF THE MISSOURI BOTANICAL GARDEN VOLUME 73 1986 NUMBER 1 SUDANIAN ELEMENTS IN THE FLORA OF ISRAEL! A. SHMIDA2 AND J. A. ARONSON? ABSTRACT n the Dead Sea Rift boi of vii a northern tongue of penetration of Sudanian elements exists da o diat has traditionally bee rde composed of Miocene relicts. We postulate that most of these eleme ve penetrated ^s area since the end of the Pleistocene. The principal habitats harboring Sudanian elements in Israel are described, including the pseudo-savanna in wadi beds, cliffs and rock Mediterranean elements with paleotropical origins and the extant Sudanian elements in Israel are discussed. Within the portion of the Syrian-African Rift lying between the Red Sea to the south and the watershed of the Jordan River in the Golan Heights, a large number of Sudanian elements, plants as well as animals, are found considerably farther north than anywhere else in their distri- bution range and in many cases growing under ecological conditions quite different from their norm. In addition, a handful of Sudanian species also occur in the low-lying coastal plain of Israel, almost entirely cut off from the Dead Sea Rift Valley and the bulk of the Sudanian region prop- er (Fig. 1). Climatic and geological conditions in the Dead Sea Rift Valley explain in part the pres- ! A. Shmida wishes to express his considerable debt of к to the late Prof. М. Zohary who introduced him (and dozens of а о the intricacies and u e attributes of the Israeli flora. We thank G. E. Wickens who reviewed the manuscript with meticulous care, a ae gather information for some of the distri- bution maps, and mer valuable suggestions for improvement of the text. J. B. Gillett and P. Quézel also kindly reviewed the bise and provided stimulating comments and suggestions. Our thanks also go to Sandy reenberg, Seemah Shemesh, and Marion Milner who provided computer and typing assistance, to Nancy R. Morin and Dr. Marjorie Tiefert for patient and professional editorial guidance, and to Miri Shmida who drew the figures. The assistance of J. Dransfield in preparing Figure 5 is gratefully acknowledged. Any errors or inaccuracies therein are entirely the responsibility of the authors. Finally we thank the numerous guides of the who helped us locate some of the rare and little- known populations research from the Bath-sheva de or Fund to A. Shmida and the Clifford M. Hardin Fund of the Fund for Higher Education to J. A. Ar 2 Department of Botany, The Hebrew University of Jerusalem, Jerusalem, Israel. 3 Rudolf a nd Rhoda Boyko Institute for Agriculture and Applied үс е» Institutes for Applied Research, Ben-Gurion University of the Negev, P.O. Box 1025, Beer-Sheva 84110, Isr ANN. MISSOURI Bor. GARD. 73: 1-28. 1986. ANNALS OF THE MISSOURI BOTANICAL GARDEN CSS Dead-Sea Rift Valley d \ меу iR The Coastal-ploin Sa Haifa „д EE ls. YA © м Voy ee > o Ж с Q jz: Es c] sharon © ` TA plain on ө rM Amman oU] ^ ghe. Gaza A, o Port Said jj o El Arish C Beer Sheba Р 2 \ Sodomw O Y Sde, Boger SS Ismailya N м, P S À Q Bir Gafgata \ S JORDAN Suez \ №: Thamad ` JN e \ Ry Eilat ® Sinai Ne SAUDI St.Katarina ARABIA e EQYPT 65 .. о 50km FIGURE 1. General geographical aspect of the study area with the main districts harboring Sudanian ele- ments. ence of Sudanian species so far north, but several questions remain concerning the present ecolog- ical adaptations and distribution patterns of the Sudanian elements in Israel. The most intriguing of these is simply: how and when did these species, whose principal distribution is in subtropical Af- rica and similar regions in Arabia and the Thar Desert of India and West Pakistan reach Israel particularly in the Dead Sea Rift Valley? Only two papers relating directly to the Su- danian floristic elements in Israel can be found in the literature. One is Eig’s work, in which he cites about 170 species of plants from the Dead Sea Rift Valley, of which more than 30% are Sudanian (Eig, 1931-1932). The second paper (Gruenberg-Fertig, 1954) is a renewed and de- tailed phytogeographical analysis of the Sudan- ian species in Palestine with particular emphasis on the affinities of those elements to the Eritreo- [VoL. 73 Arabian and West-Sudanian subregions of En- gler (1879-1882) and others. Since the publication of Eig’s and Gruenberg- Fertig’s works, our knowledge of the vegetation of Israel and its neighboring areas has increased eatly. This is due mainly to the publication of Egypt (Tackholm, 1974), the Sahara (Ozenda, 1977), the Thar Desert (Bhandari, 1978), and others (e.g., Wickens, 1977; Lind & Morrison, 1974; Danin, 1983). We have to date recorded more than 800 species of plants growing in the Dead Sea Rift Valley, of which 14.5% can be considered Sudanian elements. More ecological information concerning the Sudanian elements in Israel outside the Dead Sea Rift Valley has р conclusions of Eig, Zohary, Gruenberg-Fertig, and others on the age and origin of the Sudanian elements in Israel should be reconsidered. The dominant view of this question, as pre- sented by Zohary (1973), Tchernov (1968), and others has held that the Sudanian elements in Israel, both within the Dead Sea Rift Valley and along the coastal plain, are Miocene relicts sur- viving in specialized habitats from an earlier, warmer, and wetter period when tropical vege- tation occurred as far north as central Europe. The opposing view, held by Tristram (1884), Hart (1891), and Bodenheimer (1935), argues that the Sudanian elements in Israel are recent and arrived in Israel in waves during the warm in- tervals that occurred during the Pleistocene, i.e., during the last million years. Galil (1972) revived this theory and even went so far as to suggest that most of the Sudanian plants occurring in Israel today are the result of the most recent penetration that took place in the hyperthermic period about 4,000 to 8,000 years ago. The purpose of this paper is to enlarge on and analyze the findings related to the Sudanian ele- ments in Israel and to attempt to resolve the issue of the relative age of the extant Sudanian vege- tation in Israel. Since fossil records relevant to been analyzed, e.g., the Saharo-North African Region studied by Monod (1957), Ozenda (1977), 1986] Quézel (1958, 1965, 1978), and others as well as recent general studies of the vegetation of Africa, such as those by Knapp (1973) and White (1976, 1983), and of nearby regions including Iran (Hedge & шол 1978) апа Arabia (Мап- daville, 1984). THE ENVIRONMENT — GEOMORPHOLOGY AND CLIMATE The Dead Sea Rift Valley (Fig. 1), which is subdivided into the Jordan Valley and the Arava Valley, is part of the northern section of the Syr- ian-African Rift that extends from northern Syr- ia southwards through the Red Sea, terminating in East Africa (Garfunkel, 1970, Freund & Gar- funkel, 1978). This rift is a relatively new feature that has been slowly widening since Oligocene- Miocene times (Freund 1965, Freund et al., 1970). During this period, the Arabian Peninsula and Jordan have moved at least 40 km northwards with respect to Egypt, the Sinai, and Israel (Freund et al., 1968). Since the Pliocene, this movement has led to the concomitant formation of a deep rift so that the rivers that once drained from Transjordan into the Mediterranean Sea now drain into the Dead Sea Rift from Dan in the north to the central Arava in the south. The Dead Sea Rift is defined by two large fault systems running in a north-south direction that border it on the west and east. These systems have d the f tion oflarge cliffs and steep inclines on either side of a deep depression (Gar- funkel & Horowitz, 1966; Neev & Emory, 1967). Climatically, this depression constitutes a rain- shadow desert, the intensity of which is deter- mined by the abrupt difference in altitude (ca. 100 m) between the valley floor and the moun- tains to s n beyond which most of the rain fronts ori Rain Sell. пел РИЙ the Middle East decreases rom north to south, and the precipitation gra- dient in the area under discussion decreases from 600 mm annually in the Dan area in the north of Israel to 27 mm in the Eilat area in the south (Anonymous, 1970). In keeping with this, dis- tinct changes in vegetation zones are observed, from the extreme xeric vegetation in the Arava (south) through the steppe vegetation in the Jer- icho-Beit Shean area (central) to the Mediterra- nean vegetation in the Hula Valley area (north) (Fig. 2). The more intense the rain-shadow formed in the Dead Sea Rift Valley the further north the xeric vegetation zones penetrate. Accordingly, SHMIDA & ARONSON — SUDANIAN ELEMENTS 3 1 — 450mm — 300mm Irano Turanian RL (steppe) mm FIGURE 2. Dominant phytogeographical regions within the Dead Sea Rift Valley (modified after Zohary, 1973) and mean annual rainfall isohytes. xerophytic desert vegetation in the rain-shadow area of the Judean Desert penetrates as far north as Wadi Auja, and Irano-Turanian steppe vege- tation penetrates up to the Mt. Gilboa area. Because the flora and fauna inhabiting the slopes descending to the Dead Sea Rift Valley have also been influenced by the sharp depres- sion of the Syrian-African Rift, these slopes are considered part of the Dead Sea Rift Valley for this paper. Like other typical geomorphological depressions, the Dead Sea Rift includes wide al- luvial valleys, some draining to the Red Sea and some with underground drainage systems in which salt marshes and saline springs have formed. Large fault and contact springs have led to the formation of extensive oases that are char- acteristic of the fault escarpment area. Thus, four habitat types exist in the Dead Sea Rift that can support Sudanian floristic (and faunal) elements, as will be discussed in detail below. The great majority of Sudanian species occurring in Israel are found in one or more of these habitats. Outside the Dead Sea Rift, a smaller number of Sudanian elements are found in a variety of 4 ANNALS OF THE MISSOURI BOTANICAL GARDEN habitats throughout the low-lying plains and val- leys of the coastal and inland regions. The two important outlying districts that support Sudan- ian elements, i.e., the Dead Sea Rift and the coastal plain, are separated by the mountainous regions of Judea, Samaria, and the Galilee (Fig. 2). However, a sparse transitional zone does exist between the two in southern Israel from the Ara- va Valley through the northern Negev to the coastal plain. Moving north and west along this transitional zone, the Sudanian elements are in- creasingly restricted to distinctly thermic habi- tats such as southern exposures. The abundance of these elements decreases along the zone with the exception of the hydrophilic species, which are most abundant in the coastal plain. THE PALEOGEOGRAPHY AND PALEOECOLOGY OF THE DEAD SEA RIFT VALLEY During the Eocene a tropical climate prevailed in the latitudinal subtropical belt of the worl (Zohary, 1973; Axelrod & Raven, 1978). As a result, tropical vegetation dominated in the Med- iterranean basin and as far north as central Eu- rope and England as indicated by fossilized leaves and fruits of tropical trees found in those areas. These fossils belong to the paleophytogeograph- which elements of this vegetation have survived today (Zohary, pers. comm.; Thorne, pers. comm.). Great changes took place during the Miocene (Axelrod, 1975). Along with the Alpine orogen- the Tethys Sea, which had covered large areas of the Middle East and central Asia, shrank and was replaced by large steppe areas (Tchernov, 1975). The process of desertification continued throughout the Pliocene, which was, however, a much colder period than the Miocene. The prin- cipal development of the steppe flora of central Asia (including many Chenopodiaceae, Arte- misia, and other typical Irano-Turanian repre- sentatives) is believed to have occurred then. This flora penetrated to north Africa through the Mid- dle East and reached the coast of the Spanish Sahara. During the Pliocene a branch of the Med- iterranean Sea penetrated through the Jezreel Valley of Israel and formed a saline lake in the [VoL. 73 central and upper Jordan Valley (Horowitz, 1979). During the Pleistocene, extreme fluctuations between rainy and dry periods occurred in the subtropical belt. During the last cycle it was cold- er and rainier in Israel than at present by some 3-4*C in temperature and an undetermined amount ofless average annual precipitation (Ho- rowitz, 1 у Following the end of the Wuram period in the Middle East about 17,000 years ago (Bar Yosef, 1975), a trend towards warmer, drier conditions began, reaching present levels by about 8000 B.C. Since then, during the Holocene, very few major climatic changes are believed to have taken place (Horowitz, 1979). Around 6000 B.C., a brief hy- perthermic period occurred and approximately 4000 B.C., a colder one (the Atlantic period) (Nir & Bar Yosef, 1976). Until the Pliocene, the Dead Sea Rift Valley was a shallow depression that did not interfere with the flow of rivers from Transjordan to the Mediterranean Sea (Garfunkel & Horowitz, 1966). Since then, the drastic deepening of the Dead Sea Rift by faults on both sides resulted in the formation of the Judean Desert area, a rain- shadow desert, and the thermophilic, arid-trop- ical conditions of the Dead Sea Rift Valley, par- ticularly around the Dead Sea (Neev & Emory, 1967). This historical geographical description has been derived from geological and geomorpho- logical evidence and not from direct botanical evidence. Israel is extremely poor in plant fossils: the only two finds in this field after the Mesozoic era are leaves of tropical species found in Mio- cene formation in the Yemin Plain and Hatzeva (Lorch, 1966). This suggests that the climate in that period in the Negev and Dead Sea Rift was more tropical. Zoological evidence from the Miocene-Pliocene period is also rather poor but increases from the end of the Pliocene to the present (Tchernov, pers. comm .). DEFINITION AND DELINEATION OF THE SUDANIAN PHYTOGEOGRAPHICAL REGION The Sudanian phytogeographical region con- stitutes one of the most problematic in the Old World in terms of classification and delineation. Grisebach (1884) classified all of tropical Africa as a “Sudanian” zone. Subsequently, throughout the last century, there have been several attempts to further delineate the arid and semi-arid re- 1986] Sudanian enclaves «€ Irano- Turanian [5271 Saharo- Arabian Ш Mediterranean Sudanian Ш Prine M. 2? LR 7 > * A LN TKS Er Y : eis AA 4 fol IS Á 4) ) Lge 0 С 1 Ж Wy, у, E MIA, Y, [ М IGURE 3. ytogeographical regi ] t to the study area (modified after Zohary, 1973). Note the striking penetration of the Sudanian elements north- ward along the Dead Sea Rift Valley. gions in Afri d rel tregions ofthe Middle East and the Indian subcontinent (Engler & Drude, 1896; Chevalier, 1938; Eig, 1938; Good, 1964; Meyer-Homji, 1965; Burtt, 1971; Zohary, 1973; Werger, 1973; Wickens, 1977; Werger & Coetzee, 1978; Hedge & Wendlebro, 1978; Wal- ter, 1979; White, 1967, 1983; Mandaville, 1984). Eig (1938), who was the first to study in detail the phytogeographical regions of and relevant to the Middle East, defined a “Sudano-Deccanian region" that included a relatively narrow belt in west Africa and a much broader portion of east and northeast Africa, thence across the Red Sea, around coastal Arabia, and eastward to the Dec- can Plateau in central India. Concomitantly, he combined the region of Sind (western Pakistan), northwest India (principally Rajasthan), and parts of Baluchisthan and southern Iran with the hy- per-arid Sahara Desert that lies roughly in the same latitudes. This broad region he called “За- haro-Sindian.” Zohary (1945, 1963, 1973) revised Eig’s scheme by combining the “Sindian” region with the Af- rican and Arabian parts of Eig’s “Sudano-Dec- canian region,” while including the Deccan Pen- insula in the Saharo-Arabian portion of Eig's SHMIDA & ARONSON—SUDANIAN ELEMENTS 5 E4. Distribution map of Calotropis procera that е a typical Sudanian distribution (sensu Zohary, 1973). “Saharo-Sindian region.” For simplicity, Zohary called the results of this switch the “‘Sudanian” and “Saharo-Arabian” O: re- gions, respectively (Figs. 3 Gruenberg-Fertig (1954) and Monod (1957) also questioned the affinity between the Deccan flora and that of the Saharan and Arabian re- gions. Both of these authors commented on Eig’s own recognition of the relative individuality of the flora of the semi-arid Deccan Peninsula as compared with the rest of his ““Sudano-Deccan- ian region.” In 1965, Meyer-Homji categorically rejected the validity of a “Sudano-Deccanian re- gion” and proposed instead a “Sudano-Raja- sthanian region," because “Ц is the semi-arid Rajasthan which offers closer analogies with the Sudan region, by consideration of the bio-cli- matic Bien and the ombrothermic dia- Ы re significantly from our point of view, Meyer-Homji (1965) showed that “the strength of the so-called Sudano-Deccanian element in the entire southern semi-arid zone [of India], in- cluding the Deccan, is only 2.696," while in the northern semi-arid zone (i.e., Rajasthan and ad- jacent parts of Gujerat and Punjab), “it is two times higher (5.6%).” Unfortunately, in the same paper Meyer-Homji (1965) supported the use of Eig’s term “Saharo- Sindian" to combine the Saharo-Arabia region with Sind sensu stricto, that is, the Indus River Valley to the west of Rajasthan. Thus, additional confusion was created concerning the division of the ү in question Eig’s term “Saharo- Sindian" wa vived in a а paper by Hedge and Wendlebro (1978) dealing with southern Iran. These authors argued moreover, that this term should also be FIGURE 5. The genus Hyphaene, which represents = H. guineensis, 6 = . the , 7 ^ H. dichotoma, 8 = H. sinaitica i only record: the distinction of this taxon from he- baica is not recognized by Feinbrun-Dothan, 1985) used to describe the portion of southern Iran described by Zohary (1973) as “Sudanian.” Based on their observations in southern Iran, Hedge and Wendlebro cast doubt on the validity of de- scribing any of the species in southern Iran as “Sudanian elements.” Instead, they maintained that most E. the prominent species grouped by and others under the rubric of “Su- (Hedge & Wendlebro, 1978: 456). “The species of this element," they wrote, *are very widely dispersed with no clear pattern other than being widely dispersed in subtropical and tropical parts of the Old World. In Iran, they grow quite in- termixed with the species of the north African/ Arabian element," which is the name they pro- posed in place of Zohary's “Saharo-Arabian” nes e riking contrast, Mandaville (1984) re- ie m strong support for Zohary's phy- togeographic E including the concept ofa “Sudanian” group ofelements. He even went so far as to propose an RE enlargement of the territory delineated by Zohary as *'Sudanian" in Arabia. According to Mandaville (1984), not just the coastal regions, but nearly two-thirds of the Arabian Peninsula should be classified as *Sudanian," at the expense of the “Saharo-Ara- bian" territory as defined by Zohary (1973). We believe that outside of Africa, and most certainly on the temperate fringes of the distri- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 bution of the basically tropical elements under discussion, it is not at all surprising to find “‘Su- danian" elements intermingling with the nu- merically more significant eremaic Saharo-Ara- bian and/or other local elements. In Israel the same situation is found, as will be described in detail below. Within Africa itself, White (1983) has con- firmed the existence of a “Sudanian regional cen- ter of endemism," one of only seven such centers that he recognizes for the entire continent (he points out that it is the “least unambiguously defined" of the centers of endemism by his cri- teria). By contrast, the Sahelian region is treated by White as one of a number of regional tran- sitional zones, “‘showing gradual replacement of one flora with another that is little complicated by endemism" (White, 1983: 43). We will follow White in this matter, rather than Chevalier (1938) and Greenway (1970) who recognized a ‘‘Sahe- lien zone" per se, or Le Houerou and Popov (1981) among others who speak of a “Sudano- Sahelien zone." Burtt (1971) has E the strong floris- tic similarity between t oras of the Sudanian region (sensu du. 1973) and those of the southern half of Africa. Accordingly, Burtt recognized an amalgamated “Sudano-Zambe- sian phytogeographical region,” which was sup- ported and discussed in more detail by Werger (1973) and Werger and Coetzee (1978) (see Fig. 5). It is worth noting that this concept corre- sponds with Engler's original scheme of 1879— 1882. We realize that strong affinities exist be- tween the semidesert and savanna floras of the southern and northern halves of Africa and therefore acknowledge the validity of Burtt’s ter- minology. We argue, however, that given the cur- rent sketchy state of knowledge, phytogeographic ious regional floras under discussion, no hard and fast nomenclature can be agreed upon. In the meantime, we choose to follow Zohary’s and White’s general concept (if not all specific details) of the Sudanian phytogeographical region. We choose this course not only for practical reasons but also to provide historical continuity. In contrast to the lack of agreement on the delineation of the phytogeographical regions un- der discussion, there is overall agreement con- cerning the zoogeographic regions (Bodenhei- 1986] mer, 1935; Udvardy, 1969). Two separate zoogeographical regions are consistently distin- guished in the Old World: the Ethiopian region in Africa and the Oriental region in Asia. It is important to emphasize here the striking difference between the phyto- and the zoogeo- graphic regions as defined by biogeographers concerned with the Old World, for it reflects a major distinction between the macro-distribu- tion of animals and plants i in the area. „екан a relatively t exists between the respective faunas of the Af- rican and Asian continents, no such division ex- ists between the two floras. Above all, we wish to emphasize that our ob- jective here is not to delineate a Sudanian phy- togeographic region in Israel, in our view, such an effort is unjustified. Instead we are concerned with the occurrence, distribution, and ecological behavior of the relatively limited number of Su- danian elements that reach our study area. By the same token, we have no desire to enter into taxonomic disputes. With very few exceptions, м] б 5 g Е d J © [md 5 £e P — © a © than, 1978, 1985) even where it is clear that fur- ther taxonomic revision is required. TYPICAL AND KEY DETERMINANTS OF THE SUDANIAN ELEMENTS The Sudanian zone as a whole is typified by plants of tropical origin that have acquired one ture than true desert species. sd characteristic vegetation type ist v, open forests о bius trees and large spots with perennial the open spaces between them rA 1982). The trees are typically multi- stemmed with sparse, open canopies. Many o them are spiny, with small compound, xero- morphic leaves. Some are evergreen (Balanites, Maerua, Boscia, Salvadora, and some Ziziphus species) and some are drought-deciduous (e.g., Acacia, Albizia, and Commiphora (species). Many are characterized by a reduced leaf area, with green stems assuming some of the photo- synthetic function. In retamoid (broom-like) shrubs this adaptation is most prominent—the plants being leafless throughout most of the year while the stems become the primary photosyn- thetic organs. Some examples of this group are SHMIDA & ARONSON — SUDANIAN ELEMENTS 7 Moringa peregrina, Capparis decidua, Lepta- denia pyrotechnica, Ochradenus baccatus, and Periploca aphylla. Many of the trees have an umbrella shape, which is most striking i in Acacia—the dominant genus of tl Xerophytic vines are occasionally found climbing on the savanna trees (e.g., Cocculus pendulus, Commicarpus and Boerhavia spp., asclepiad vines such as Oxystel- ma esculenta as well as various perennial cucur- its). The Sudanian flora in the Dead Sea Rift is typical à in its arboreal elements but unusual in between the trees. Some Sudanian grasses do occur there — species of Pennisetum, Cenchrus, Cymbopogon, Ennea- pogon, and others. But the overall aspect of the vegetation type in the Dead Sea Rift is best de- scribed as a pseudo-savanna, owing to the scar- city of perennial grasses. Moreover, a wide range of non-Sudanian, desert chamaephytes belong- ing to the Saharo-Arabian phytogeographical re- gion intermingle with the Sudanian elements, creating a kind of bi-regional patchwork or mo- cavannac same pattern is apparently also found in west and north Africa (Quézel, 1965, 1978). STATISTICAL ANALYSIS OF THE SUDANIAN FLORA IN ISRAEL IN RELATION TO GROWTH-FORMS, TAXONOMY, AND GLOBAL DISTRIBUTION A statistical analysis of the flora of Israel (ac- Zohary & Feinbrun-Dothan 1966- , 1978, 1985, pers. mm.) shows that 157 plant species, out of a var of 2,173 (i.e., 7.2%), belong at least partially to the Sudanian phytogeographic region. Of these, 116 species, or 68.8% of the subtotal, are con- sidered to be predominantly Sudanian in distri- bution, sensu Zohary (1973) (Table 1). This rep- resents about 5.3% of the total Israeli flora; it is on this group of 116 species that all further sta- tistical analysis will be based. Most of the species accepted here as Sudanian show either a straightforward Sudanian distri- bution or chorotype (55.1%) or a bi-regional Su- danian-Saharo-Arabian chorotype [10.3% (SU- SA) + 18.1% (SA-SU) = 28.4%]. In both of these 8 ANNALS OF THE MISSOURI BOTANICAL GARDEN (VoL. 73 TABLE 1. Phytogeographical analysis of the Sudanian elements in the flora of Israel and their growth forms [data from the “Нога Palestina” (Zohary & Feinbrun-Dothan, 1966-1974; Feinbrun-Dothan, 1978, 1985) with corrections]. An- nua Facul- Cham- Hemi- tative Num- Per- ae- to- An- Peren- Para- Chorotype* ber cent Trees Shrubs phytes phyte nual nial Vine site Sudanian 64 55.1 17.22 18.7 31.3 6.2 14.1 3.1 7.8 1.6 Sudanian-Saharo-Arabian 12 10.3 11.1 11.1 33.3 22.2 11.1 — 11.1 — Sudanian-Tropical 9 7.7 — 25.0 4.0 4.0 160 25.0 — Sudanian-Mediterranean 3 2.6 33.3 — — — 33.3 33.3 — — ace Irano-Turanian 3 2.6 33.3 66.6 — — — — — — Pluriregional 4 3.5 — — — 25.0 25.0 25.0 — 25.0 Saharo- qur Sudanian 21 18.1 19.1 48 33.3 19.1 9.5 14.3 — — Total 116 100.0 15.5 16.4 26.7 11.2 15.5 7.7 5.2 1.7 * Chorotype = distribution, sensu Zohary (19 * 17.2 here refers to the percent of species of a bM growth form within the total number of species within the chorotype group, e.g., 11 trees/64 Sudanian = 17.2%. groups, the high percentage of trees, shrubs, and monotypic. Overall, the mean number of Su- chamaephytes is especially notable. danian species per genus among Sudanian ele- Of the eight families most abundant in Su- ments in Israel is 1.30, which is quite low relative danian elements in the flora of Israel, five fam- to the mean number (3.05) of species per genus ilies show significantly higher importance than within the entire flora of Israel. the reference (meaning a greater percentage con- tributed by the family in the total number of species in the Israeli flora; see Table 2). These e Chenopodiaceae (5.290 versus 3.296), Eu- phorbiaceae (5.2% versus 1.6%), Zygophyllaceae The environmental conditions that support (4.3% versus 0.9%), Asclepiadaceae (6.9% versus Sudanian vegetation in the heart of the Sudanian 0.796), and Capparaceae (696 versus 0.496). The region in east Africa and the Thar Desert are last four of these py have tropical diversity high ambient temperatures during the growing centers (Ozenda, 1977). The higher Tepre resenta- season, a relatively warm, frost-free winter, and tion of the acid nee in the S flora two rainy seasons. The rains come in spring and than in the reference is perhaps related to the autumn, with total annual rainfall averaging be- large number of pluriregional chorotypes in this tween 250 and 500 mm (Walter, 1979; Zohary, family. 1973). In the west African region alternately called The lower portion of Table 2 reveals the neg- Sudanian or Sahelian, a single rainy season oc- ligible representation ofa few typically temperate curs in summer, with precipitation averaging 75 families in the Sudanian flora of Israel. While to 400 (to 500) mm per year (Wickens, pers. the Rosaceae, Ranunculaceae, and Apiaceae have comm.). no Sudanian representatives in Israel whatso- The first mentioned determinant of the distri- ever, the Brassicaceae contribute three Sudanian bution of Sudanian elements—high year-round species, which account for 2.6% of the total Su- temperatures—is found in Israel primarily in the d DISTRIBUTION AND ECOLOGY OF SUDANIAN ELEMENTS IN THE DEAD SEA RIFT VALLEY danian flora in Israel. Dead Sea Rift Valley. Accordingly, assuming Although a total of 92 genera in 40 families successful dispersal of diaspores from the Su- are represented, v showlargenum- danian region to the Dead Sea Rift, or their pre- rs of species а оош; Sudanian dis- existence there in the past, Sudanian elements tribution. Only Acacia has four species in the should be found in those habitats in the Dead group. Five genera have three species each: Com- Sea area with moisture regimes similar to those micarpus, Capparis, Cleome, Moretia, and of the main Sudanian region. However, such Tamarix. Ten genera are bitypic and 76 are habitats are relatively very small and scattered. 1986] Moreover, they are specialized habitats that dif- fer ecologically from the true savanna environ- ments of the Sudanian region in Africa. PRINCIPAL HABITATS OF SUDANIAN ELEMENTS THE DEAD SEA RIFT (1) Pseudo-savanna in wadi beds and flood plains. In the large wadi beds of the Arava Val- ley, Acacia species (A. tortilis and A. raddiana— sometimes described as A. tortilis subsp. raddi- ana) dominate the landscape, and a variety of xeric desert sp tersp d g them We define pseudo-savanna as an association dominated by large savanna trees that can tap the deep year-round ground water available in the wadi beds, but which is virtually lacking the understory of drought-resistant perennial grasses typical of a true savanna. Instead most of the associated species are xeric desert shrubs and annuals such as Zilla spinosa, Moricandia ni- tens, Anabasis articulata, and Achillea fragran- tissima, all of which belong to the Saharo-Ara- bian phytogeographic group. The non-arboreal Sudanian elements, as well as most of the shrubs and vines, have migrated up to the cliffs and rocky sites on either side of the Rift Valley. It is important to emphasize that these pseudo-sa- vannas do not cover large areas as they do in Africa but are confined to relatively narrow water courses in flood plains. Zohary (1945) suggested the term “gallery forest" to describe this asso- ciation; since this term has been extensively used to describe riparian vegetation on the border of the humid tropics (see Wickens, 1977: 30-31), we propose a new term — *Arterial Desert Sa- vanna" for bis pseudo-savanna described here. s term corresponds to the “‘savane and east, characterized by warm a relatively stable water regime, support numer- ous Sudanian elements, including small shrubs (e.g., Ephedra foliata), dwarf shrubs (Abutilon frutescens), and perennial grasses (e.g., Tetra- pogon villosus). In addition, large cliff faces in side of the Dead Sea Rift and scattered throughout the Negev and Sinai provide moist, thermic conditions that e cartilaginea, pau aphylla, and Cocculus pendulus SHMIDA & ARONSON—SUDANIAN ELEMENTS TABLE 2. Relative importance of species numbers within the important families of the Israeli flora and the Sudanian elements within them B in rael with doo Sudanian Total Species Distribution? in Israel Num- Per- Num- Per- Family ber cent be cen Papilionaceae 9 dul 264 12.1 Compositae 4 3.4 246 11.3 ramineae 9 7.7 208 9.6 Chenopodiaceae 6 5.2 70 3.2 uphorbiaceae 6 5.2 34 1.6 Zygophyllaceae 5 4.3 20 0.9 Asclepiadaceae 8 6.9 15 0.7 arid 7 6.0 8 0.4 Brassicaceae (Temperate) 3 2.6 115 5.3 Apiaceae (Temperate 0 0 93 4.3 Rununculaceae (Temperate) 0 0 45 2.1 Rosaceae (Temperate) 0 0 21 1.0 a The two columns of the table do not total 100%, because not all the families are inclu * The total number of species in Israel with a pre- dominantly Sudanian distribution © The total number of species in p entire flora of Israel is 2,173. Some of these species are obligate chasmo- phytes and they (or vicarious species) grow only on cliffs even in the Sudanian region proper. Ex- amples of these are Sonchus suberosus and Car- alluma spp. Some species that occur only on rocky cliffs in the Dead Sea Rift, however, grow as scrambling climbers supported by Acacia and other trees on non-rocky ground in the east Af- rican savannas. The outstanding example is Coc- plants has shifted in the more arid environment of the Dead Sea Rift so that their growth con- ditions, in terms of temperature and water, re- main relatively unchanged. However, for at least a few of the scandent species (e.g., Oxystelma alpini), the shift to rocky habitats also represents a shift in growth con- wn 200- logical phenomenon (MacArthur, 1972) but one 10 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 O А = € о E D £ o д = Mean Temp. ох с 254 i -500 © O е л a 9 247 E Раа +400 _ — = Е а Е Е Е 23 5 [42 300 ~ = Е = > E = 5 22 = [30 200 5 e a N Ф Ф So EM o 2! о LLL Е А о 538 [100 o Rainfall Jm > > Daiana Dead-Sea area Sea of Galilee ч 4 e 1 = la Уч La e Le. t 100 + | 20 ! 300 ! Of Eilat Yothveta Hatzeva EinJGedi Jerico ! Degonio Dafna Sedom Tirat- Zvi South 4«&—— Geographical Distance (km) э» North URE 6. Mean annual rainfall and average daily temperature and average maximum temperature along the Dead Sea Rift axis so far almost unnoticed in botany (Cody & Moo- ney, 1978). (3) Oases. The largest concentration of Su- danian species is found in the moist habitats of oases. тарен existing in oles conditions enjoy both of see Here can be found arboreal Sudanian species generally uncommon in the Dead Sea Rift due to the rarity of oases. Around the Dead Sea, the major large oases are Ein Gedi, Safi, and Kallirhoi. Sudanian species non-existent in other parts of the Levant (Israel, Lebanon, Syria, and Jordan) grow in these oases. Worthy of mention are Maerua crassifolia, Capparis decidua, Acacia laeta, and Cordia sinensis. 4) Warm salines. The geomorphological structure of the Dead Sea Rift gives rise to un- drained depressions (“Ка” — Evenari et al., 1982) with permanently high ground water where sa- line conditions occur. In addition, saline habitats occur around the Dead Sea near springs and al- luvial fans of large wadis (Zohary, 1945; Zohary & Orshan, 1956). The majority of the species occurring in these saline depressions are Saharo-Arabian elements; only a few Sudanian taxa are present, yet they are prominent. The species worth noting are Suaeda monoica, S. fruticosa, and Tamarix spp. These species belong to a peculiar phyto-geo- graphical group (Zohary, pers. comm.) distrib- uted from central Asia all the way to South Af- rica. They occur throughout the Sudanian (or Sudano-Zambesian) region but are not restricted to it. It is unclear whether their center of origin is Irano-Turanian, Sudanian, or South African. They might even represent part of the “paleo- Welwitchia flora" that connected the ancient deserts of Asia and Africa (Zohary, 1973). Two additional ancient elements of tropical origin that are concentrated primarily in these saline habitats are the sole two palm species that occur in Israel. The first is the doum palm Hy- phaene thebaica that, in Israel, grows exclusively in the high-groundwater salt marshes of the southern Arava and iis the Gulf of Eilat. The of Sudanian origin is the origin has been a subject of controversy for cen- 1986] SHMIDA & ARONSON—SUDANIAN ELEMENTS 11 4 130 P o Diversity Ф 2 оо Ф a o» c 90F O © 5 (09) sun fOr © © = B5OF 30r О Dead -Sea area Sea of Galilee ^ „ ME loot ] 400 Eilat Jethveta Hatzeva Ein* бед! Jerico Tirat-Zvi Degania Dafna South *— Geographical Distance (km) — > North FIGURE 7. Sudanian species diversity along the Dead Sea Rift axis. turies. As a result of human introduction, cul- tivated or feral P. dactylifera is now found in virtually every desert oasis in Africa, the Middle East, and adjacent regions (Chevalier, 1938; Zo- hary & Spiegel-Roy, 1975). However, through- out the Saharo-Arabian and Sudanian regions of its present range, it is almost impossible to dif- ferentiate between natural and introduced pop- ulations. In some of the salines where the species occurs in the Dead Sea Rift Valley, the popula- tions may well be subspontaneous, i.e., escape from cultivation, but we assume it is indigenous to the area and not an introduction from far away (Zohary, 1982). The salt marshes at Neot HaKikkar and in the Beit Shean Valley might be typical of the kind of natural habitat from which the date palm was originally dispersed to other saline and freshwater springs in nearby desert areas. [^7] DISTRIBUTION PATTERNS AS FUNCTIONS OF ENVIRONMENTAL CONDITIONS ALONG E DEAD SEA RIFT Figure 6 illustrates the opposing gradients of mean annual temperatures and precipitation along the length ofthe Dead Sea Rift. Progressing south from the Dan Valley, the mean annual precipitation decreases from 600 to mm at Eilat (Anonymous, 1970). In contrast, the mean temperatures of the warmest and coldest months increase from 36.4 to 40.2?C in the north and from 7.8 to 10.1°C in the south (Anonymous, 70) — No As described in the preceding section, in areas with low temperatures such as are found north of the Jerusalem-Jericho line, by contracting to southern exposures with increased insolation. Thus, from the standpoint of temperature ad- aptation, the number of Sudanian species should increase to the south; however, the opposite is ing that the overall num along the Dead Sea Rift decreases from south to north, both absolutely and in relation to the total 12 ANNALS OF THE MISSOURI BOTANICAL GARDEN | Hyphaena thebaica poo Ziziphus at hrist 2 Calotropis procera H is" ` RU “Phoenix dactylifera Balanites aegyptiaco 4 Hyphaena sinaitica FiGURE 8. Northern distribution borders of typical L зс 1 = . Rift ¡ft RA Arava region, RD— Dead Sea region, RL— Lower Jor dan Valley, RU—Upper Jordan Valley, RH— Haleh Valley. Numbers indicate solitary distribution poin of rare Sudanian trees. vegetation cover of шег агеа (Fig. т is parallel to the d g р gr from south to north. Figures 8 and 9 show several typical po species according to their northernmost dist bution limit in the Dead Sea Rift. For the pur- pose of presentation and due to the existence of relatively sharp geomorphological climatic bor- ders along the Rift, we have grouped the plants according to distribution (Table 3). Most of the topographical “hole” of the Dead Sea (399 m below sea level) and also to the existence of ap- propriate habitats such as rocky cliffs and oases. Additional Sudanian species end gradually be- tween the Dead Sea and the Sea of Galilee. The last topographical barrier, the Corrasim Saddle, which rises from the Sea of Galilee (— 212 m) to the Hula Valley (30-180 m above sea level), [Vor. 73 t Ipomoea cairica hontus acaciae p> RU poses a Maerua crassifolia Cassia italica Capparis decidua GURE 9. Northern distribution borders of small shrubs and woody vines in the Dead Sea Rift Valley. probably presents a drastic climatic change through which none of the Sudanian species have yet broken. North of the Sea of Galilee there are almost no true Sudanian elements. Accordingly, e Rift is the world’s northern limit (locus ter- minus) for most Sudanian species. Theres is, HOWEVER, one phenomenon that seems to dient of Sudanian species from south to north: - the population den- sities of certain Sudanian species increase to the north from the Dead Sea to Mt. Gilboa, where the annual precipitation reached 150-300 mm (Fig. 2). These species, while rare or completely gs, m e.g. Salvadora persica, Balanites aegyptiaca, (2) cliff vegetation such as Periploca aphylla, Sonchus tuberosus, and Iphiona maris-mortui, and (3) plants growing in rock crevices that absorb large amounts of runoff water, such as Moringa pere- grina, Grewia villosa, and Abutilon indicum. around the northern end of the Dead Sea (due to its 1986] special geomorphology) than in other regions of the Rift. Thus, the frequency of the habitats here can explain the “abnormal” abundance of the Sudanian species around the Dead Sea itself. TAXA ENDEMIC TO THE DEAD SEA RIFT VALLEY Species in small enclaves outside their main area of distribution have a higher probability of speciation (Stebbins, 1952, 1971) so that an an- cient population cut off from its principal area of distribution for thousands or even millions of years may differentiate to the species or even can be expected to differentiate only to the level of forms or varieties, if at all. Thus one method for determining whether the Sudanian vegetation in the Dead Sea Rift is a Miocene relict or the result of recent invasion is to compare the degree of speciation of Sudanian species in the Dead Sea Rift Valley with that of (1) related taxa in the Sudanian region of east Africa and (2) species from other phytogeograph- ical regions growing in the Dead Sea Rift Valley, such as true desert elements. Another rule of thumb used by many bio- geographers is that the more ancient and isolated the vegetation in an enclave, the greater the chances of finding systematic relicts in it (Shmi- da, 1985). Thus, if Sudanian vegetation has been present in the Dead Sea Rift since the Miocene, at least some relict taxa with no close systematic relatives and whose distribution areas do not ex- tend outside the Dead Sea Rift, should be found. This is a phenomenon known from other simi- larly isolated regions, such as the Inyo region in southern California, where many relicts have been recorded (Reveal, 1976; Raven, 1977). For a de- tailed probability analysis of these two biogeo- graphic rules : see Shmida (17835 otally absent from the area. This fact is stikins d in view of the seemingly ideal conditions for speciation in which the various Sudanian species occur in the Rift Valley (small, remote, and disjunct populations subject to extreme desert stress, see Stebbins, 1952). A possible exception is Trichodesma bois- sieri, a chasmophyte endemic to the region be- tween Ein Gedi and Mt. Gilboa. This species is unrelated to others in the genus found in the Dead Sea Rift and Sinai (T. africana and T. eh- renbergii) and is probably both a systematic and geographic relict. Trichodesma boissieri may SHMIDA & ARONSON— SUDANIAN ELEMENTS 13 continue southwards to Yemen through the Ara- bian Peninsula, but this area is one of the least investigated in the Old World, and no data are available. Table 4 presents the list of species endemic to the Dead Sea Rift. Only four of them are of cer- tain Sudanian origin and they are limited to the Sdom-Wadi Kelt area. Thirteen species are en- demic to the Dead Sea Rift and the Judean Des- ert and are not of Sudanian origin but derived from other, nearby phy Most of the endemic species (except for Trichodesma boissieri) have allopatric sibling relatives with n, 1978) may be another pa- leoendemic but, Owing to its uncertain pre-taxo- nomi (compare Boulos, 1962), its taxo- nomic- eae iat. position is unclear. This pattern hints at the relative youth of the specia- tion process in the Dead Sea Rift area. If the populations in this area had been isolated from each other for many thousands of years, clear morphological disjunctions would be expected within pairs of main species. If that were the case, the gradual geographical-morphological transi- tion between them wou ot be discernible at present. We might also expect to see, in at least some of the species, the additional evolutionary step of sympatric expansion and differentiation to different niches. (There are many such ex- amples recorded for other regions, see Mayr, 1970; MacArthur, 1972). Yet none of the species endemic to the Dead Sea Rift or Sudanian species in Israel follows this pattern. mong Sudanian trees and shrubs, no vari- eties and certainly no species endemic to the Dead Sea Rift have been found, with the sole exception of Maerua crassifolia var. maris-mortui (Zo- hary, 1955). However, it is worth remembering that the speciation process of arboreal species is slower than that of herbaceous and annual ones (Stebbins, 1971; Davis & Heywood, 1963). Most of the arboreal Sudanian species in Israel have very large geographic ranges, from the southern Sahara through southern Arabia to southern Iran and India (Fig. 2). Examples are Calotropis pro- cera, Capparis decidua, Cassia italica, Grewia villosa, Maerua crassifolia, and Ochradenus bac- catus. Some of these taxa spread even further south along the east African savannas to South Africa (Werger, 1978). The distribution of the genus Moringa (Fig. 10) exemplifies this distri- bution pattern. The distribution maps of the ge- 14 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 TABLE 3. Geographical distribution of some Sudanian elements in Israel.* Northern Limit of Distribution in the Dead Sea Rift Eilat District Hatzeva District Dead Sea (Ein Gedi) Jericho — Wadi Kelt District District вилино пеи Bothriochloa ischae- Ast се iin na a гузаба Cometes abyssinica Crotalaria aegyptiaca Laisiurus scindicus oe pyrotechni- ee leyseroides Pentatropis spiralis Lindenbergia sinaica Oxystelma alpini Tephrosia apollina Otostegia Pterogaillonia calycoptera Pycnocycia tomentosa Seetzenia orientalis Tephrosia nubica Abutilon hirtum Acacia tortilis Capparis decidua Cleome droserifolia Abutilon fruticosum A. hirtum Acacia raddiana A | Convolvulus glom- Aizoon canari eratus C d lone Cordia sinensis Cleome a с F DRE tenacissima Iphonia жге ы subsp. maris-m Lavendula UP Moringa peregrina Ochradenus baccatus Grewia villosa Hibiscus micran- thus Maerua crassifolia Pergularia tomento- Pulicaria inuloides Stipagrostis hirti- gluma Zygophyllum sim- plex nus Balanites (Fig. 11) and Salvadora (Fig. 12) illustrate different kinds of distribution patterns within the Sudanian group. In conclusion, the extant environmental and genecological conditions in the Dead Sea Rift Valley are potentially ideal for speciation (Steb- bins, 1952; 1972). The Sudanian species there are represented by small isolated populations, separated by thousands of kilometers from their center of distribution in east Africa. In compar- ison to prevailing conditions in the east African savannas, many of the habitats of these species have changed о and the species survive under extremely a climatic conditions. All these factors are ae of exceptionally strong selective pressures that should have led to rapid speciation within the isolated populations (Steb- bins, 1972). However, just the opposite has oc- curred in the Е Sea p ‘Valley. Very few species 1 n, and when they have, it is biy to thé Manil. or subspecies level. In contrast, a relatively high rate of endemism occurs in the Judean Desert, which borders the Dead зел. Rift Valley, consisting mainly of Sa- aro-A i d Irano-Turanian species. These distribution and speciation patterns seem to sup- port the argument that the Sudanian elements in Dead Sea Rift Valley are of recent origin. DISTRIBUTION AND ECOLOGY OF THE SUDANIAN ELEMENTS IN ISRAEL OUTSIDE THE DEAD SEA RIFT The distribution of a few Sudanian species in Israel is not limited to the Dead Sea Rift and the thermic wadis of the Negev but also extends into the Mediterranean region (see Fig. 2). These more widespread species represent only about 10% of the Sudanian elements in Israel, but their eco- logical “behavior” and distribution patterns shed additional light on the question of the relative antiquity of all the Sudanian elements in Israel. The majority of the Sudanian elements occur- ring outside the Dead Sea Rift are found pri- marily in disturbed secondary habitats created, or at least heavily influenced, by human activity. This fact has led to the formulation of the widely accepted theory, expressed by Zohary (1973), that these species, once limited to the Arava Valley in the southern portion of the Dead Sea Rift and to the few scattered oases and saline marshes in the Rift, gradually spread north and west over large areas of the country; this expansion was made possible by the greatly intensified agricul- tural and settlement activities of the last century. However, some clear and some apparent excep- tions to this rule are recognizable, namely in the group of Sudanian aquatic elements and the few 1986] TABLE 3. Continued. SHMIDA & ARONSON—SUDANIAN ELEMENTS 15 Northern Limit of Distribution in the Dead Sea Rift Species Penetrating into the Mediterranean Region Beit Shean Sea of Galilee Fully Integrated Aquatics and District District lements Ruderals Hygrophiles Balanites aegyptia- Восора monnieri Aristida sieberiana Chrozophora Bacopa monni- ca Calotropis procera Commicarpus afri- tinctoria? | Ci pos prophetar- | Commicarpus verti- canus Withania somnif- Cyperus jemini- cillatus Desmostachya bipin- era? cus th ym acaciae nata Nymphaea ceru- pois aphylla Tricholaena teneriffae Epipactis consimilis la Salvadora persica Hyparrhenia hirta Vigna lutea Trichodesma boissi- Ipomoea cairica Pennisetum setaceum Ziziphus spina-christi E: Ошу well-documented unambiguous cases are included. hese species are рын эры = the fact that Zohary (1973) did not record its chorotype as Sudanian, since its Sudanian vicariads clos —both кой and systematically. Therefore we assume that it is an instance of a ae derived awed of Sudanian origin species that appear to be fully integrated today with the predominant Mediterranean vegetation in the coastal plain. The three groups of the Su- danian elements— Ruderals, Aquatics and Hy- drophiles, and Fully Integrated — will be dis- cussed in turn, along with an Anomalous group. RUDERALS he most prominent of the Sudanian elements that has rapidly expanded and continues to ex- pand into disturbed sites outside the Dead Sea Rift is the so-called Syrian Christ-Thorn, Zizi- phus spina-christi. The present distribution of this tree, whose edible fruits are dispersed by birds and mammals, includes the valleys of the central plain and Galilee and even some low- lying regions of the central Negev. Within the Rift, it extends all the way to the Dan Valley, i.e., farther north than any other Sudanian ar- boreal element in Israel. However, a striking di- chotomy occurs within the different portions of this extended distribution area. In the Rift Valley up to the Sea of Galilee and in the southern part of the coastal plain, Z. spina-christi is evergreen. By contrast, in areas north of the Sea of Galilee and in the northern portion of the coastal plain, it is winter deciduous (Danin, pers. comm.). We hesitate to call the corresponding areas primary and secondary habitats for Z. spina-christi, but there can be little doubt that the colder areas to the north represent areas into which Ziziphus has recently spread, and is continuing to spread, par- ticularly along roadsides and in abandoned fields. D e again "Below in the section on fully integrated elements. Several additional ruderal species among the Sudanian elements are found in the Gramineae. The Sudanian grasses Hyperrhenia hirta and Pennisetum setaceum, to take two major ex- amples, were formerly limited to southern ex- posures in the Rift Valley. As a result of inten- sified human activity, they have recently spread plain. The northerly penetrations of these grasses 16 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE 4. List of the endemic plant species restricted to the Dead Sea Rift Valley. Vicarious Species from Chorotype Which it of the Probably Parent Type of Distribu- Endemic Species Originated Taxa Habitat tion and Speciation 1. Anthemis maris-mortui 2. Blepharis attenuata ы . Centaurea lanata 4. Fagonia grandifolia* 5. Galium hierochuntinum 6. Hammada eigii ч oo . Iphiona maris- mortui? o . Kickxia judaica o . Matthiola aspera -— — . Podonosma syriaca юм Sonchus suberosus* uy . Reseda maris- mortui* 14. Silene oxydonta* — CA . Suaeda asphaltica 16. Tamarix palaestina > . Trichodesma boissieri Heliotropium maris-mortui A. hebronica B. ciliaris C. aegyptiaca F. mollis G. judaicum H. scoparia H. rotundifolium I. mucronata K. acerbiana M. livida P. syriaca var. typica S. suberosus ssp. suberosus R. boissieri T. jordanis ? Irano-Turani- an Sudanian Saharo-Ara- bian Saharo-Ara- bian Mediterranean Irano-Turani- an Irano-Turani- Sudanian E. Saharo-Ara- bian Saharo-Ara- bian Mediterranean Sudanian Saharo-Ara- bian Irano-Turani- an Saharo-Ara- bian-Irano- Turanian Irano-Turani- an Sudanian Thermic slopes Thermic slopes Rocks & wadis of J mE Des Slopes ip Ju- dean Desert Rocks & wadi beds in can- yons Slopes & collu- vium Slope of Judean Desert Cliffs & rocks Cliffs & rocks Slopes of Ju- dean & Jer- icho Desert valley Rocks Cliffs & canyons Thermic slopes In wadi beds Chalky slopes Wet habitat Rock outcrop- ping Allopatric distribu- tion with gradual transition Parapatetic distribu- ol with transi- Ala distribu- , no transition кө rved Allopatric distribu- wi dí ual sitio Allopatric E Allopatric distribu- tion with disjunc- tio hie distribu- tio ins = adual Allopatric distribu- transition Allopatric distribu- i ith gradual transition Allopatric distribu- tion with gradual tion of hundreds of km Allopatric distribu- tion with gradual transition 9 No transition Allopatric distribu- tion No transition or al- lied * The ae rank of these taxa has recently been lowered to varieties. > Very ra SHMIDA & ARONSON—SUDANIAN ELEMENTS 17 e Moringa oleifera El Г) M. peregrina A other species FIGURE 10. Distribution map of some Moringa spp. (after Kerauden, 1965). are probably accomplished with the aid of phys- iological-biochemical adaptations of new popu- lations to lower and lower temperatures. Other examples of savanna grasses turned ru- deral include Desmostachya bipinnata and Ar- istida sieberiana. As was mentioned with respect to Ziziphus spina-christi, so again in the grasses, a certain transitional character must be exam- ined. Within the Rift Valley most of these grasses occur strictly in moist habitats, salt marshes, or rocky outcrops— or more recently, as weeds in plowed fields and along roadsides. In the warm, moist parts of the coastal plain, where they occur in far fewer numbers, they are found in disturbed de S 5 г. 1 B.roxburgii < Bolanites | aegyptiaca 15°} @ B.maughamii A В. pedicillaris 30°} FiGURE 11. Distribution map of some Balanites spp. (with the assistance of M. J. S. Sands). D €2 Salvadora persica A 5 angustifolia e 6 5. oleoides FIGURE 12. Distribution map of some Salvadora spp. sites. However, in some cases they also appear to be well integrated in the oak woodland asso- ciation, as will be discussed below. AQUATICS AND HYDROPHILES A second group of Sudanian elements found mainly outside the Sudanian enclave of the Rift are hydrophilic species. Some of these grow in +1 2 18 4 SWallps OI OUICT регешпапу 4 V g , Cyperus jeminicus, Nymphaea alba, and Vigna р à i 2 . luteo. a). UICIS which grow in migrated to cultivated fields (e.g., Chrozophora tinctoria, C. plicata, and Commi- carpus africanus). Those areas of primary wet habitats in which the water level and channels can change rapidly and unpredictably are pop- ulated in part by species pre-adapted to second- ary habitats such as roadsides and cultivated s. FULLY INTEGRATED SUDANIAN ELEMENTS Desmostachya bipinnata is an important Su- danian grass that grows in a variety of soils in the east African savannas (Wickens, pers. comm.) and in the Thar Desert (Bhandari, 1978). In the Arava it is dominant in salt marsh and saline sand communities. Along the coastal plain it oc- curs in the natural oak woodland community of Quercus ithaburensis on red sandy loam (Eig, 6; Zohary, 195 his thermophilic plant community also includes additional Sudanian m grostis obtusifolia, and Ziziphus spina-christi. The Quercus ithaburensis-Desmostachya bipinnata association is part of a broad transitional trend 18 ANNALS OF THE MISSOURI BOTANICAL GARDEN A Acacia gerrardi ssp. negevensis РА <> subsp. gerrardii ) т FIGURE 13. Distribution of Acacia gerrardii subsp. negevensis and subsp. ger Aided (after Halevy, 1971; Ross, 1966; Wickens, pers. comm.). from the holarctic to the tropical regions. A similar transition gradient occurs in North America, where northern oaks, along with trop- ical Prosopis and Poaceae constitute the “ѕауап- na” vegetation in parts of southern Texas and southern Arizona (Whittaker et al., 1979). In this oak woodland community of the coastal plain is found the only recorded endemic Sudanian species in Ben occurring entirely outside the Dead Sea — Aristida sieberiana. The genetic- systematic ык of this grass species аге пої clear, and considerable controversy exists on the endemic status of this taxon (Feinbrun, pers. comm.). Further studies may alter its current sta- tus. As mentioned above, both Ziziphus spina- christi and some of the grasses belong both to the ruderal group and to the group of fully-in- tegrated species. In fact a Ziziphus spina-christi- Hyparrhenia hirta community is recognized along the coastal plain. Zohary (pers. comm.) assumed that the climax area of this community occurred in the southern portion of the coastal plain. Due to intensive cultivation and woodcutting, how- ever, not many locations remain that can still testify to this presumed climax vegetation. In the northern portion of the coastal plain, it is re- placed by the Ceratonia siliqua—Pistacia lentis- cus community that dominates the coastal foot- hills. Even in this community where most of the components are typical Mediterranean, the dom- inant species show a strong Sudanian affinity. For example, Ceratonia siliqua grows wild in the mountains of Yemen and clearly belongs to a tropical section within its family. Only recently, [Vor. 73 a new species of Ceratonia — C. oreothauma was discovered in isolated, disjunct regions of Arabia and Somalia providing further indication of a tropical origin for this genus (Hillcoat et al., 1980). Pistacia lentiscus is also reported from the col- line regions of east Africa (Lind & Morrison, 1974; Zohary, pers. comm.). At present, the prin- cipal distribution of both these species is in the colline belt around the Mediterranean Sea. Yet they exhibit both systematic and ecological (they are both evergreen sclerophyllous and thermo- philic) links to the Sudanian tropical vegetation. Because the links are usually evident only at the generic level, Zohary (1955, 1973) assumed that they were very ancient links going back as far as the Miocene, when tropical vegetation dominat- ed the entire Mediterranean region and central Europe. ANOMALOUS GROUP In addition to the three groups described above, two additional species should be mentioned that conform to none of the above patterns and the- ories. The first of these is Acacia gerrardii subsp. negevensis. The map in Figure 13 shows the dis- tribution of Acacia gerrardii with geographical delineation of the two recognized subspecies. The generic distribution pattern reveals a remarkable degree of disjunction between subspecies negev- ensis in southern Israel and a few sites in south- western Arabia and the more widespread sub- species gerrardii in eastern and southern Africa. Moreover, А. gerrardii subsp. negevensis does not occur, in Israel at least, in typical Sudanian conditions (Halevy, 1970). Instead of occupying wadi beds in the Arava Valley, such as are pre- ferred by Acacia tortilis and A. raddiana, this subspecies grows in only a few adjacent wa- tersheds in the highlands of the southern Negev and one or two sites in eastern Sinai. It appears to be adapted to cool habitats, which represents an ecological shift from its primary warm hab- itats, as seen in Africa. Its restricted geographical area and lack of penetration to adjacent cool, open areas (e.g., the Mt. Ramon Range and the Sante Katerina Highlands in Sinai) are indicative of an extreme relict position in the area. Further evidence for this view is provided by the anom- ttern (southern Negev, eastern a ously disj unct p a subsp. gerrardii is also a watershed species throughout its range in Africa (see Fig. 13, Wick- 1986] ens, pers. comm.). Finally, it is worth noting that the Jordan Sparrow, Passer moabiticus, exhibits a similar disjunct pattern, occurring only in Is- rael, Jordan, and the Persian Gulf region (Boros & Horvath, 1954; Enero, 1971). The second Acacia albida (= Faidherbia albida), which occurs in a few iso- lated sites of the Mediterranean coastal plain, a few sites in the Jordan Valley north of the Dead Sea, and one or two isolated spots elsewhere in the country (Halevy, 1971). It does not occur in the Negev or Arava Valley (Aaronsohn, 1913; Karschon, 1961; Aloni & Orshan, 1972; see Fig. 14). Unlike its typical ecological patterns in Af- rica (Radwanski & Wickens, 1967; Wickens, 1969; Ross, 1966, 1979, 1981), itis not restricted to riparian or high ground-water sites in Israel. It also differs in that flowering and fruiting are relatively rare in all the sites at which it occurs in Israel, while vegetative propagation by suckers is far more common (Halevy & Orshan, 1972). Zohary (1962) has argued for the relict position of this species in Israel; indeed, the anomalous position of A. albida within the genus Acacia, and its occurrence with typical Mediterranean associations such as Artemisia monosperma and Cyperus mucronatus along the coastal plain, sup- port that conclusion. However, the overall dis- tribution pattern of the species in Israel (Fig. 14) and its atypical ecological characters here, could instead indicate a recent arrival and penetration to this area from the Nile River Basin rather than along the Syrio-Africa Rift. Throughout Africa Acacia albida exhibits a phenological feature unusual to but not unique among savanna trees in that it sheds its leaves at the beginning of both rainy seasons (Wickens, 1969; Pelissier, 1977). In Israel, bi-annual shed- ding of the leaves occurs although there is only one rainy season per year. In the case of an an- cient relict, we would expect this character to have been modified rather than preserved. More- over, recent investigations on the northernmost branch of the Nile River indicate that as recently as the sixth century A.D. this so-called Pelusiac branch was active and extended a considerable distance into the northern Sinai Peninsula (Sneh & Weissbrod, 1973). It does not seem impossi- ble, therefore, that Acacia albida should have been gradually dispersed by one or more vectors along the Mediterranean coast and finally to the Jezreel Valley and Beit Shean on the one side and as far north as southern Lebanon on the other (Halewy. 1971). SHMIDA & ARONSON—SUDANIAN ELEMENTS 19 35° [#2 35° > 433° TEL AVIV, 32°} a PZ JERUSALEM AMMAN + OBEER-SHEBA a Я 431° b \ ) : \ ) \ ) 30°} pl : 730° 5:0 \ O 10 20 : EILAT amem 35* FIGURE 14. Distribution of Acacia albida in Israel. The arrows indicate the presumed penetration route (after Halevy, 1971). ADAPTATION OF SUDANIAN SPECIES TO ARIDITY AND LONG-RANGE DISPERSAL Regardless of the age of the Sudanian elements Sea Rift and the other regions in Israel in which Sudanian elements are found do not enjoy the same amounts of rain that fall in the Sudanian region proper, and that what rain does occur is concentrated in only one rainy season, it might be supposed that various adaptations or pre-ad- aptations to drought would be found in the Su- danian elements in Israel. Obviously, it is diff- cult or impossible to analyze such factors 20 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE 5. Dispersal analysis of the Sudanian arboreal floras of Sudan, Egypt, and Israel (see text for details). Long-distance га bd Тагна Disper Sudan Egypt Vector Number Percent Number Percent Number Percent Wind 56 14.0 11 8.5 9 18.4 Endozoochory 189 47.25 73 56.1 22 44.9 Exozoochory 2 0.5 8 6.2 4 8.1 one apparent 153 38.25 38 29.2 14 28.6 Total 400 100.0 130 100.0 49 100.0 a Species list drawn from Andrews (1950-1956) and modified after Wickens (1968). > Species list drawn from Táckholm (1974) with modifications. © Species list drawn from Zohary and Feinbrun-Dothan (1966-1974), оогоо (1978, 1985), апа personal observations of the authors. Full species list with information о unit, and presumed dispersal vector is available from A Shm chorotype, dis form, flowering period, da. 4 Full species list т details of growth forms and presumed dispersal systems is available from the authors. quantitatively, short of in situ field comparisons of the same species occurring in Sudan, for ex- ample, and in Israel. On the other hand, a partial analysis of ad- aptations for long-distance dispersal of seeds is possible, on the basis of personal observations and study of the published floras of Israel, Egypt, and Sudan. Anemochory and zoochory would clearly facilitate more efficient dispersal of Su- danian elements to the Dead Sea Rift Valley and enable dispersal units or diaspores to reach the rare and disjunct habitats within the Rift that are suitable to their requirements. Members of the Asclepiadaceae (e.g., Calotropis, Periploca, Lep- tadenia, and Solenostemma) have a pseudopap- pus attached to each seed, which is clearly im- portant in long-distance dispersal. Certain grasses (e.g., Pennisetum, Tetrapogon, and Tricho- laena), Geraniaceae, Monsonia spp., and Abu- tilon spp. have dispersal mechanisms very sim- ilar to sui of the ы eames Yet another e distance Scd mechanisms is the endozo- ochorous species, whose dispersal units are juicy berries eaten by birds (e.g., Maerua, Ephedra, Lycium, and Salvadora) or la O eaten by mammals and other large animals (e.g., Aca- cia, Balanites, Cordia, and Ziziphus). However, many Sudanian species in Israel lack any appar- ent long-distance dispersal mechanism. Table 5 shows a comparison of the relative importance of the major long-distance dispersal vectors of seeds, anemochory and zoochory of trees, shrubs, and perennial vines in the Sudan- 1an elements of Israel, Egypt, and Sudan exclud- ing Equatoria Province. This analysis unavoid- ably relies to a certain extent on speculation; cient long- however, in all cases where no readily apparent morphological adaptation for long-distance dis- persal was discernible, none was assumed. The flora of Sudan was chosen as the basis for com- parison to Israel, since a fairly reliable flora was published rather recently (Andrews, 1950-1956) and considerable systematic work has been con- ducted in Sudan since then (cf. Wickens, 1968, 1977). All species occurring in the mountainous Red Sea Hills—were considered representative of the Sudanian phytogeographic region. Cer- tainly, a west African region should have been compared as well, but we were unable to find a suitable flora. In addition, the Sudanian elements of the ar- boreal flora of Egypt (Táckholm, 1974) were ana- lyzed as a kind of mid-way station between Su- dan and Israel. Finally, the Sudanian arboreal elements in Israel were analyzed in somewhat more detail; in almost all cases, the geographical ranges (chorotypes) were accepted as given by Zohary and Feinbrun-Dothan (1966-1974) and Feinbrun-Dothan (1978, 1985). We concentrate only on the arboreal elements in each of the three countries for two main rea- sons: 1) the dispersal spectra and taxonomy of the non-arboreal floras of Egypt and Sudan are less well-known; and 2) the arboreal floras of Egypt and Israel are presumed to give the best other non-Sudanian chorotypes e.g., Pistacia, 1986] SHMIDA & ARONSON— SUDANIAN ELEMENTS TABLE 6. Habitat shifts of some Sudanian elements in different geographical areas. Name Sudan ead Sea Rift Valley Outside Dead Sea Rift Valley, esp. Coastal Plain Trees Acacia albida Acacia gerrardii ssp. gerrardii Balanites aegyptiaca Moringa peregrina Ziziphus spina-christi Halophytes Desmostachya bipin- nata Hyphaene thebaica Phoenix dactylifera Vines Cocculus pendulus Ephedra foliata Pentatropis spiralis Pergularia tomentosa Grasses Pennisetum setaceum Trichlolaena tenerif- Aristida sieberiana Tetrapogon villosus high ground water or flood plains (habitat) flood plains scattered trees in sa- vanna Red Sea Hills; savanna scattered trees in sa- vanna savanna grass sparse in savanna introduced; naturalized climbers in savanna climbers or shrubs slender climber in sa- vanna small shrub savanna sparse in savanna dry savanna dry savanna wadibeds in extreme cold desert oases and rocky cliffs oases and rocky cliffs oases and moist wadi beds saline sinks saline marshes oases and saline marsh- cliffs shrubs fleshy climber in saline sink small shrub or climber rocky cliffs rocky cliffs rocky cliffs crevices in rocky cliffs a variety of diversified habi- tats—sands, basaltic slopes, and cds wadi ruderal, segetal oak woodland aggressive climber on culti- vated trees small shrub ruderal recently invading roadsides in patches south-facing rocks Juniperus, Colutea, Calligonum, and Haloxylon. Introduced and cultivated species were excluded as well, for example, Ficus sycamorus in Israel, and Moringa oleifera, Dalbergia sissoo, and is throughout. Suc Sudanian elements in Israel show a higher pro- portion of species with adaptations for long-dis- tance dispersal than those in Egypt or Sudan, although a и г percentage of Su- srae wind dispersal and соня The most striking re- sult is that the arboreal Sudanian floras in all three areas show a very high percentage of long- distance dispersal adaptations (61.75, 70.8, and 71.4% in Sudan, Egypt, and Israel, respectively). On the basis of this analysis we conclude that adaptations for long-distance dispersal do not provide a clue as to the question of the age of the Sudanian flora in Israel. However, they cer- tainly play an important role i in n tae dispersas pad survival of th in their geographical ranges. DISCUSSION The Sudanian assemblage of plants in Israel exhibits a typical example of the evolutionary and ecological processes that can take place in a locus terminus, where populations are small, dis- junct, and vulnerable to extinction. In short, such populations face ecological and environmental conditions quite different from those in their cen- tral distribution area in Africa (Fig. 1). In the following paragraphs, we summarize the distinc- tive features that characterize the Sudanian ele- ments in Israel. 22 ANNALS OF THE MISSOURI BOTANICAL GARDEN ENVIRONMENTAL CONDITIONS The environmental conditions in which the Sudanian elements occur in Israel are generally similar to those in the main Sudanian dominion in Africa. However, there are some critical dif- ferences: in Israel there is a greater contrast be- tween winter and summer conditions, and the overall yearly precipitation is less predictable. Moreover, the rainfall patterns are completely different. In the Sudanian region udi east iin there are two principle rainy seasons—autu (October-November) and spring (M AL ri ub (Lind & Morrison, 1974), but in Israel virtually all of the rain comes in the winter (December- March) (see Fig. 10). GEOGRAPHICAL AND ECOLOGICAL SPLITTING OF THE SUDANIAN ELEMENTS The Sudanian elements penetrating Israel from the south split and inhabited two distinct habi- tats when they reached the Dead Sea Rift Valley: arboreal elements that grow in large wadis and grasses and other smaller lifeforms that tend to grow in the thermic micro-environments occur- ring among open rock formations. Both a splitting of the natural Sudanian as- sociation and an intermingling of the Sudanian elements with the Saharo-Arabian and Mediter- ranean floras have occurred. In most places, even in the most typical “Sudanian habitats" in Israel, 1973, 1982) devoted a special geographical dis- trict to the Sudanian elements in the Arava Val- ley and Dead Sea area, but phytogeographical analysis of the community shows that the Su- danian elements account for less than 1596 ofthe total species list (see section on Distribution Pat- terns below). Accordingly, Feinbrun-Dothan (1985) does not recognize a Sudanian phytogeo- graphic territory in the “Flora Palaestina." As mentioned earlier, we heartily concur with this view HABITAT SHIFT AMONG THE SUDANIAN ELEMENTS Habitat shift is a phenomenon consisting of space in another portion ofthe species' geograph- ical range (MacArthur, 1968, 1972; Pianka, 1978; [Vor. 73 Diamond, 1975). Such patterns have been fre- quently found in zoology (Diamond, 1975; Pian- ka, 1975; Krebs & Davies, 1978) but are almost completely unnoticed in botany (see Cody & Mooney, Our study РУ cases of habitat shift on two geographical scales: a) the east African savannas and the Arterial Desert savannas of the Dead Sea ины and 2) the Dead Sea Rift and the coastal . Table 6 summarizes the most obvious tst shifts displayed by the Sudanian ele- ments under study. Most of the habitat shifts can be at least partially explained by the Eco-Geo- graphical Rule of Boyko (1947). Since the plants in Africa are accustomed to a fairly high regime of rainfall, where they penetrate into the arid parts of Israel, they tend to be confined to hab- itats with improved water regimes relative to the surrounding area. Nevertheless, these specialized habitats have components that differ markedly from the African savannas even if the amount of water is approximately the same. In saline habitats, for example, the plants must contend with a high concentration of salts, whereas in rocky habitats they face a ium degree of unpre- dictability in their water su nother major habitat shift occurs in the coastal plain, where most of the Sudanian species occupy ruderal and especially segetal sites. As the cultivation of summer crops has expanded in the area over Mie por 100 | уган, certain Su- d a remarkable pre- adaptation to disturbed sites with artificially sup- water in the summertime (Dafni, danian elemente weeds throughout the coastal plain (e.g., Chro- zophora tinctoria, see Table 5). Yet another case of habitat shift is displayed by those Sudanian vines that are typically climbers but in the Dead Sea Rift are restricted to cliffs and only seldomly occur on trees (e.g., Cocculus pendulus). We pro- pose that some of the cucurbits have also under- gone a habitat shift, wherein typical savanna climbers became prostrate spreading vines in the arterial wadi beds (e.g., Cucumis prophetarum, Citrullus colycynthis), but in this case the shift apparently took place at the genus level, rela- tively early in the evolution of the relevant sec- tions of the family. A similar trend occurs in the Commicarpus/Boerhavia complex, in which c sively prostrate wadi-bed vines (e.g., Commi- carpus boissieri). This trend seems to represent 1986] a general pattern of shift of growth form con- comitant with a migration from semi-desert sa- vanna environments to extreme desert habitats. ASSOCIATION AND RE-ARRANGEMENT In general terms, in comparing the association of Sudanian elements in Israel with the entire association of the Sudanian flora in Africa, we find a complete lack of agreement between them. The main differences are that in Israel the Su- danian elements occur together with many species of other chorotypes and that even the details of their co-occurrence are different from those found in Africa. This pattern of non-consistent asso- ciation corresponds to Gleason’s (1926) individ- ualist concept of plant RT and phytoso- ciology, an approach that s quantitatively developed and substantiated bs wai ttaker (1962). Nevertheless, a small *core group" of taxa can be distinguished that are, in Israel, restricted to the Dead Sea Rift Valley. We shall call this the Arava Group (a). They are all species universally recognized as belonging to the basic Sudanian flora, which occurs continuously, if disjunctively, Sal procera, and Acacia tortilis are good examples of this group (see Figs. 4, 11). A second group is the Fully Integrated Group (b) These are paleo-Mediterranean elements (sensu Zohary, 1973) that do not have phylo- genetic allies among temperate floras but instead replaced there by a congeneric relative (e.g., Olea europaea replaced by O. chrysophylla, = O. eu- ropaea subsp. africana) or by related genera within the same familial section or tribe (e.g., Laurus, Nerium, Phillyrea, and Tamus vis-à-vis their tropical counterparts). Zohary (1973) pos- tulated that this group has survived since the Miocene around the Mediterranean Basin while most of the tropical elements g with them at that time his m to the warmer regions of Africa (and Asia). The кы Group (с) includes those species that grow in both the Dead Sea Rift and in the Mediterranean region of Israel, e.g., Ziz- iphus spina-christi, Demostachy bipinnata, and Hyparrhenia hirta. These species are character- ized by differences in habitat in the two areas, SHMIDA & ARONSON— SUDANIAN ELEMENTS 23 as described earlier. In the Dead Sea Rift, they are restricted to moist habitats, i.e., salt marshes or rocky outcrops; and in the Mediterranean re- gion, most of them are relatively widespread, especially in warm and at least slightly disturbed habitats. Indirect and published literary evidence suggests extensive invasion and expansion of these elements throughout the Mediterranean re- gion within the last century. Yet it seems clear that the bine distribution of these species, cup now weedy grasses, is Sudanian 1938; Zohary, 1973). For this reason we ue them as a separate group. In addition to the an Anomalous Group (d) са can be recognized, whieh includes Acacia albida and A. gerrardii subsp. negevensis. As described earlier, these two species show complex distributional and ecological pat- terns in Israel, the elucidation of which will re- quire further study of the geological and paleo- ecological past of the area. = RARITY, DISJUNCT DISTRIBUTION PATTERNS, AND THEIR ECOLOGICAL CONSEQUENCES Habitats i in Israel where Sudanian plants dom- desert slopes that do not permit the existence of non-xeric plants. Consequently, most popula- tions of these Sudanian species in Israel are small and disjunct. In many rocky habitats in the Dead Sea area and along the Arava Valley, common desert species (Saharo-Arabian elements) are found in- d e to high rates o cf. "Selene, 1952) and rare recolonization nts THE AGE OF THE SUDANIAN ELEMENTS IN THE FLORA OF ISRAEL Points in favor of the antiquity of Sudanian vegetation in the study area are the following: 1. Strong paleontological evidence exists for the gradual decrease (attenuation) of tropical an- imals in the area from the Miocene to the Holocene (Tchernov, 1968, 1975). 2. The “paleomediterranean elements" (sensu Zohary, 1973) of Israel show a distinctly relict о раце with а a apta- tionstot r ы ANNALS OF THE MISSOURI BOTANICAL GARDEN Olea europaea and Ceratonia siliqua, both of which are present-day components of the ev- а еи that have decidedly tropical affiniti . The Od and disjunct character of many Sudanian species growing in oases and by springs in cliffs and gorges might suggest that the vegetation is relict in character (Zohary, 1962, 1973). According to this theory, the vegetation that once covered the whole area contracted to moister habitats as the climate grew dryer Points in favor of the recency of most of the Sudanian vegetation in the study area are: — N Ww + сл . Cold conditions in the Pliocene and the glacial periods of the Pleistocene were unfavorable to Sudanian vegetation existing in the stud area, which was е destroyed after the Miocene (Galil, 1972). Up to the Pleistocene the Dead Sea Rift was apparently only a shallow depression, in which a pseudo-Sudanian climate appropriate for thermophilic Sudanian species could not ex- ist. During the last glacial period the central Emory, 1967; Horowitz, 1979) and a not appropriate for the existence of the xe- rophytic Sudanian species growing there to- ay. . Most Sudanian species in Israel have under- gone virtually no speciation relative to their larger populations in east Africa. Those taxa that have speciated have done so to a very low degree (variety or subspecies level). In most species with endemic taxa, gradual mor- phological transitions can be found between the endemic taxa and the corresponding Su- danian sibling species. . Many Sudanian species have efficient long- distance dispersal mechanisms and could have reached the disjunct habitats suitable for their growth by this method (see previous section on Long-Range Dispersal). In light of recent research on the seed dispersal ability of many plants occurring on oceanic islands (Carl- quist, 1974), this evidence seems significant. . The disjunct and apparently random distri- bution of Sudanian vegetation in the Dead Sea Rift can be attributed not only to their being relicts from a different climatic period, but also to long- distance dispersal and the scarcity of appropriate habitats. According to [VoL. 73 this view, many habitats suitable for the ex- istence of Sudanian vegetation can be ex- pected to remain empty or populated by only one species (cf. Ellner & Shmida, 1981). What seems to be a random pattern at the single species level can be explained by consistent statistical evidence at the community and to- tal Sudanian flora level (see Ellner & Shmida, 1981). . That the indisputably relict “paleomediter- ranean elements" occur entirely in the Med- patterns in the study area. The unique distri- bution of Acacia gerrardi subsp. negevensis also stands in marked contrast to that of the remainder of the Sudanian elements in the area. The former shows a clearly relict pattern while the latter do not. . The existence of a transitional group and a wide range of marked habitat shifts (see pre- vious sections) in the Sudanian elements oc- curring in Israel outside of the Dead Sea Rift also argues for the recency of the Sudanian flora as a ing place at a rapid rate in localities charac- terized by human disturbance . Very few endemic species are found in the Dead Sea Rift where the great majority of Sudanian floral and faunal elements in Israel occur. The only clear exceptions are the bo- raginaceous chamaephyte Trichodesma bois- sieri and the sunbird, Cinnyris osea, both of y contrast, in both the Irano- Turanian and the Mediterranean elements of the Israeli flora, a large number of endemics or sub-endemics exist. Many of these occupy rocky or otherwise locally isolated habitats. Similar habitats supporting Sudanian ele- ments, although many hundreds of kilome- ters away from their nearest congeneric neighbors, are nevertheless devoid of pa- leoendemics. CONCLUSIONS n our opinion, the large majority of the Su- danian plant species occurring in the Negev and 1986] Dead Sea Rift Valley today are of recent origin and have reached the area only after the last glacial period. That is, most of the Sudanian species presently occurring in Israel, with the ex- ception of Trichodesma boissieri and Acacia ger- rardi subsp. negevensis, started to penetrate from the south about 12,000 to 17,000 years ago. This process continued throughout the Holocene. This conclusion does not exclude the possibility that udanian elements, including some of the same species found here today, existed in Israel in more ancient times during the Pleistocene and late Tertiary. Nevertheless, the existence of Tertiary tropical flora in Israel has yet to be proved by fossils. By contrast, evidence for tropical fauna during the early Pleistocene is plentiful, mainly from the fossils found at Ubediyah of the ele- phant, giraffe, and hippopotamus (Tchernov, 1975; Bar Yosef & Tchernov, 1972). It is interesting to note that analogous ques- were investigated by Sealy (1949). The occur- rence of Arbutus unedo in habitats edaphically unsuitable for woodland in a few sites in Brittany and western Ireland, surrounded by plants of very different overall chorotypes, was investigated geogr ical region, Sealy concluded that Arbutus unedo had invaded tł y recently, for reasons very similar to those favored by us concerning the Sudanian elements in Israel. Kosswing (1955) analyzed the biogeography of tropical elements in Turkey and concluded that some of the Sudanian elements are ancient (20 million years or more) while others are recent (hundreds « oF thousands of years); The same is i an Israel, according o the opinion held by most biogeographers in а today (Galil, nie Heller, pers. comm.; Lipkin, pers. comm.; ernov, pers. comm.). The question now is be size and importance of the patterns, speciation level, and phylogenetic relationships of Sudanian vegetation. We con- clude that in the Negev and Dead Sea Rift Valley the overwhelming majority of Sudanian ele- ments are very recent. They penetrated the area primarily along the Syrian-African Rift during the various inter-glacial periods. Most of the ex- tant Sudanian vegetation probably reached the area only at the end of the upper Pleistocene and during the Holocene, after the Lisan Lake dried SHMIDA £ ARONSON— SUDANIAN ELEMENTS £5 up and the rapid deepening ofthe Rift took place. On the other hand, a significant portion of the species with tropical affinities occurring in the Mediterranean area of Israel are probably an- cient and can be considered Miocene relicts, both from their geographical disjunction and system- atic isolation (Zohary, 1973). LITERATURE CITED AARONSOHN, A. 1913. Notules de slate ha et palestinne (I). Une station peu connue de Г.А albida Del. Bull. Soc. Bot. France 60: 495-502. ALONI, R. & G. ORSHAN. 1972. The vegetation of the Lower Galilee, Israel. Israel J. Bot. 21: 209-227 ANDREWS, F. W. of the Anglo-Egyptian Sudan, 3 rnment by T. Bunck, Arbroath, Scotland. ANONYMOUS. 1970, Atlas of Israel. 2nd edition. De- ment of Surveys, Ministry of Labor, Jerusa- em. AXELROD, D. 1975. Evolution and biogeography of the Madrean-Tethyan sclerophyll vegetation. Ann. Missouri Bot. Gard. 62: 280-334, P. H. Raven. 1978. Late Cretaceous and Tertiary vegetation history of Africa. Pp. 99-130 in M. J. A. Werger коке Biogeography and Ecology of Southern Africa. W. Junk, The Hague. BAR YoseF, O. 1975. The Epi- Paleolithic in Palaes- tine and Sinai. Pp. 363-378 in F. Wendorf & A. Marks (editors), Problems in Pre-history: North Africa and the Levant. Southern Methodist Univ., Dallas. E. TCHERNOV. 1972. On the Paleo-ecolog- ical History of the Site of Ubeidiya. = Acad- emy of Science and Humanities, Jerusal BHANDARI, M. M. 1978. Flora of the Indian y; Scientific Publishers, Jodhpur BODENHEIMER, F. = 35. Animal Life in Palestine. L. Mays, Jerusalem. Bonos, I. & L. кага 1954. Тһе PT of Passer moabiticus. Acta Zool. 1(1-2): 43-51. Bouros, L. 1962. Cytotaxonomic studies in ies genus Sonchus. 4. The generic status of some species earlier treated as Sonchus. Bot. Not. 115: 58—60. Boyko, Н. 1947. On the role of plants as quantitative climatic indicators and the geo-ecological law of distribution. J. Ecol. 35: 138-157. Burtt, В. L. 1971. From the south: an African view of the floras of western Asia. Pp. 135-154 in P. H. Davis, P. C. Harper & I. C. Hedge (editors), Plant-Life in South-West Asia. Botanical Society of Edinburgh, Edinbur ia S. 1974. Island Biology. Columbia Univ. s, New York. aa А. 1934. а observations sur quelques cue de l'Afrique. Rev. Bot. Appl. Agric. Trop. 14: Sahara, : Cento d'Origine des Plantes et al., La Vie dans la Mond Copy, M. L. & H. A. Mooney. 1978. Convergence 26 ANNALS OF THE MISSOURI BOTANICAL GARDEN versus nonconvergence in Mediterranean-climate ecosystems. Annual Rev. Ecol. Syst. 9: 265-321. DarNi, A. 1975. Comparative biology of Prosopis farcta in four different habitats in Israel. Ph.D. thesis. Hebrew Univ., Jerusalem. [In Hebrew with English summary. id A. 1983. n Vegetation of Israel and Si- ai. Cana, Jerusa Das P. H. & V.H. He ywoop. 1963. mn of Angiosperm Taxonomy. Oliver & Boyd, Edin burgh. DIAMOND, J. M. 1975. VU SUA A of esr com- munities. Pp. 342-444 in M. Cody & J. Diamond (editors), Ecology = Evolution of шш Belknap, Cambridge, Massachu- se EIG, A. togéographiques auxiliares dans la nienne. Feddes Repert. Spec. Nov. Regni Veg. 63(1): 1-201 [text, 1931]; 63(2): 1-120 la! tables, 1932 “193 1-1932. Les éléments et les groupes phy- flore iim n the phytogeographical i ES of Palestine. Palestine d. Bot., Jerusalem Ser. 12. 946. Synopsis ofthe соон es of Palestine. Palestine J. Bot., Jerusalem 3 183-246. ELLNER, S. & A. SHMIDA. 1981. Why are adaptations for long-range seed dispersal rare in desert plants? 4. Oeco ENGLER, A. 1879-1882. Versuch einer Entwicklungs- geschichte der Pflanzenwelt i Flo- rengebiete seit der Tertiárperiode, 2 volumes. En- gelmann, Apal ig. & UDE ару 1896. Die Vegetation der Erde. Engelm nn, Leipzig. EvENARI, M., L. SHAN ONAN. TADMOR. 1982. N egev: The Challenge of a reds 2nd x Harvard н Ргеѕѕ, Сат FEINBRUN- Don Flora Palestina. Part 3. Israel reins of Sciences and Humanities, Je- Flora Suse Part 4. жард Academy nities, Jeru rene of the sonia devel- ince U Cretaceous times. Geol. . GARFUNKEL. 1978. Guidebook to Ex- cursions Along the Dead Sea Rift. rea of Geology, Hebrew Univ., Jerusalem , 1. ZAK & Z Z. GARFUNKEL. 1968. ` Аве and rate of sinistral pre along the Dead Sea Rift. Nature 220: 253-2 , Z. с L ZAK . GOLDBERG, T. WEISSBROD & B. PERIN. i. The shear along the Dead Sea Rift. Philos. Trans., Ser. B, 267A: S J. 1972. Theorigin ofthe Sudanian vegetation n Israel. Tev' Va'Arets 14(4): 139-145. [In He- uina GARFUNKEL, Z. 1970. The tectonics of the western margins of the Southern Arava. Unpubl. Ph.D. thesis. Hebrew Univ., Jerusalem. [In Hebrew.] Horowitz. 1966. The upper Tertiary and Quaternary morphology of the Negev. Israel J. Earth Sci. 15: 101-117 [Vor. 73 GLEASON, H. 1926. The individualistic concept E the plant association. Bull. Torrey Bot. Club 5 26 7-26. Goop, R. 1964. The Geography of the Flowering Plants. Longman, London. GREENWAY, P. J. 1970. A classification of the vege- tation of East Africa. Kirkia 9: 1-68. GRISEBACH, A. Die Vegetation der Erde nach ihrer klimatische Anordung, 2 volumes. W. En- mann, Leipzig. GRUENBERG- FERTIG, I. 1954. On the “Sudano-Dec- canian" elements in ls flora of Palaestine. Pal- estine J. Bot., Jerusalem Ser. 6: 234-240. HM p: 197 0 Distribution a ne Ladies p. negevensis and the Tev’ Va Ашы ш 209-212. [In M En "English summa | 1971. A study of Acacia albida in Israel. La- Ya'aran 21: 52-63. [In Hebrew with English sum- mary.] & G. ORSHAN. 1972. Ecological studies on Acacia species in the Negev and Sinai. I. Distri- bution of Acacia raddiana, A. tortilis and A. ger- rardii subsp. negevensis. Israel J. Bot. 22: 120- 138. HAMILTON, C. A. 1982. Environmental History of East Africa. Academic Press, London . 1891. Some Accounts of the Fauna and ra of Sinai, Petra and Wády Arabah. Com mittee of the а d Fund, Lon dii. HEDGE, I. C. & P. WEN 1978. Patterns of distribution and icis mism in Iran. Notes Roy. : 441-464 RDCOURT. 80. A minosae-Cesalpi- noideae) from Arabia and the Somali Republic. Kew Bull. 35: 260-271. HoRowirrz, А. ns The Quaternary in Israel. John Wiley, New Yor и о В. De Acacia albida Del. in Israel and = EI East. La- Bie aran 11(2): 4—6. [In Hebrew English sum Калон М. 1965. ere genre Moringa en Afrique a Madagascar. Webbia 19: 815- a R. 1973. Die Vegetation von Africa. Fischer, tuttga d С. 1955. uus of the Near East. t. Zool. 4: 49-1 VIE ( A; h v- y: An о Арргоасћ. Black- ford. ] Ecolog well Scientific Publ., Ox 71. e Dead Sea sparrow: a sec- ceding place on Turkish and first known шош place on Syrian territory. Ibis 3: 617— 618 LARSEN, T. R. 1974. Butterflies of grin. National Council for Scientific Research, Beiru ГЕ Houerou, Н. & С. Е. Popov. 1981. da eco-cli- matic classification of intertropical Africa. FAO Pl. Protect. jeu 31: 1-40. имр, E. M. & M. E. S. M 1974. East African Vegetation. Longman on LirAv, M. 1967 Micro-environmental factors and species interrelationships in three Batha associa- tions in the foothill region of the Judean Hills. Israel J. Bot. 16: 79-99 1986] LorcH, J. 1966. A Pleistocene aes from the Cen- tral Jordan Valley. Israel J. B 5: 31-34. MACARTHUR, R. H. 68. The ms of the niche. Pp. 159-186 in R. C. Lewontin (editor), Popula- tion Biology and Evolution. Syracuse Univ., Syr- acuse. 1972. Geographical Ecology. Harper & Row, TT J.P 1984. Studiesin the flora of Arabia XI:s e historical and geographical aspects of a а floristic frontier. Notes Roy. Bot. Gard. Edinburgh 42: 1-15. Mayr, E. 1970. Animal Species and Evolution. Har- vard Univ. Press, Cambridge. МЕуУЕк-Номл, V. M. 1965. On the ‘Sudano-Dec- canian' floral element. J. Bombay Nat. Hist. Soc. Monon, T. 1957. Les grandes divisions chorolo- giques de l'Afrique. Publ. Cons. Sci. Afrique Sud Sahara 24: 1-146. МЕЕХ, D. & D. O. EMory. 1967. The Dead Sea de- positional processes and ee of evapo- rites. Bull. Geol. Surv. Israel 41: 1-147. Nir, D. & O. BAR Yoser. 1976. ie Envi- ronment and Man in ма Israel Nature Protec- Flore du Sahara, 2nd edition. PELISSIER, P. 1977. Competition and the integration rid Regions. Univ. of California, Riv- PIANKA, E. R. 1975. Niche relations of desert lizards. -3 M. dy & (editors), о and Evolution of Communities. Ha d Uni ress, ES Ecology, 2nd edition. E du Sahara. Fischer, Stuttga —-. тү 8. Analysis of the flora of Mediterranean and Saharan Africa. Ann. Missouri Bot. Gard. 65: RADWANSKI, s. А. & G. E. Wickens. 1967. The ecol- ogy of Acacia albida on mantle soils in ro m Jebel Marra, Sudan. J. Appl. Ecol. 4: 569-5 Raven, P. H. 77. The California flora. Pp. E 137 in M. Barbour & J. Major (editors), Terrestrial egetation of California. Wiley & Sons, New diis n REVEAL, J. L. 1976. Dedeckera (Pol genus from California. Brittonia 28: 245-251. Ross, J. H. 1966. Acacia albida in Africa. Bol. Soc. Brot., Ser. 2, 40: 187-205. A conspectus of the African Acacia species. Mem. Bot. Surv. South Africa 44: 1-155. 1981. A biogeographical analysis of the Af- rican Acacia species. Bothalia 13: 389—402. SEALY, J. В. 1949. Arbutus unedo. J. Ecol. 37: 365- 388 SHMIDA, A. 1982. Mediterranean vegetation in Cal- ifornia and Israel: similarities and differences. Is- rael J. Bot. 30: 105-123 SHMIDA & ARONSON— SUDANIAN ELEMENTS 27 . 1985. D of the desert floras of the world. Jn M. Evenari & I. Noy-Meir (editors), Hot Desert Ecosystems of the World, Volume 12. El- sevier, New Yor & S. ELLNER. T Coexistence of plant species with similar niches. Vegetatio 54: 29-55. SNEH, A. & T. WEISSBROD. 1973. Nile Delta: the de- funct Pelusic Branch identified. Science 180: 5962. STEBBINS, G. L. 1952. Aridity as a stimulus to plant evolution. Amer. Naturalist 86: 33-44. —. 1971. Relationships between adaptive radia- tion, speciation and major evolutionary trends. Taxon : 3- 1 72. Ecological distribution of — of I. Seeds and seedlings. Pp. 7-34 in D. . Valentine (ede) Taxonomy, Phytogeography and Evolutio ademic Press, New York. TACKHOLM, V. 974. Students’ Flora of Egypt, 2nd edition. Cairo Univ., Beirut. TAKHTAJAN, A. 1969. Flowering Plants: ын апа Dis persal. Oliver & Boyd, Edinbur TCHERNOV, E. 1968. Succession of Ro Fauna During the Upper ге of Israel: sy petals lia epicta. Paul Pan e Early Pleistocene Molluscs of Erq r. Israel dis; ademy of Sciences and Hu- . B. A he p of Western Pal- estine: The Fauna oa a of Palestine. Com mittee of the Palestine Explorati on Fund, London. Upvarpy, M.D. 19 ynamic Zoogeography. Van Nostrand Reinhold, New York. WALTER, H. 1979. Vegetation of the Earth, 3rd edi- tion. Springer, РЕ н rk. WERGER, M. J. A. 1973. Notes on the phytogeograph- ical affinities of the southern Kalahari. Bothalia 11: 177- x 197 пое а division id South Africa. P | 45-1 169 i . Wer, (editor, Gee af and Ecology ‘of жо Айка Junk, The & . 1978. The Sudano-Zam- 193-224 in M. erger т , Biogcosraphy I 8 of Southern he Ha The savanna nee of the Zam- udanian domains: an ecological and a phytogeographical comparison. Webbia 19: 651-671. . 1983. The Vegetation of Africa. Unesco, Paris. WHITTAKER, R. H. 1962. Classification of natural communities. Bot. Rev. (Lancaster) 28: 1-239. 77. Evolution of peus diversity in land communities. Evol. Biol. —65. , L. E. GILBERT & J. H. Ede. s 1979. Anal- ysis of two-phase pattern in a mesquite grassland, Texas. J. Ecol. 67: 935—952. киш С. E. 1968. Some additions and correc tion . W. Andrew’s Flowering Plants of the ieu Forests Bull. Forests Dept., Sudan n.s. 1-64. 1969. A study of na albida Del. (Mi- mosoideae). Kew Bull. 23: 1976. The flora of n s (Sudan Re- 28 ANNALS OF THE MISSOURI BOTANICAL GARDEN d and its geographical affinities. Kew Bull. Add. Ser. 5: 1-368. ZOHARY, D. &Р. SPIEGEL-Roy. 19 Beginnings of fruit-growing in the Old World. Science 187: 319- ZOHARY, M. 1945. Outline of the vegetation in Wadi Araba. J. Ecol. 32: 204-213. . 1955. Geobotany. Sifriat Poalim, Merhavia. [In Hebrew 1962. Plant Life of Palestine: Israel and Jor- dan. Ronald Press, New York. 1963. On the geo-botanical structure of Iran. Bull. Res. Council Israel, Sect. D, Bot. 11 (Suppl.): 1-113. [Vor. 73 . 1973. Geobotanical oo of the Mid- dle East. 2 parts. Springe rlin. . 1982. Plants of the Bible. Cambridge Univ. Press, a & NBRUN-DOTHAN. 1966-1974. Flora Palestina. an 1 and 2. Israel Academy of Sci- ences and Humanities, Jerusalem 1956. Ecological studies in the II. Wadi Ara- . ORSHAN. vegetation of the Near East Deserts. ba. Vegetatio 7: 15-37. & a SHANSKY. 1949. Structure and ecol- g of the egetation in the Dead Sea Region of P Palestine T, Bot., Jerusalem Ser. A 177- 206. SYSTEMATIC FOLIAR MORPHOLOGY OF PHYLLANTHOIDEAE (EUPHORBIACEAE). I. CONSPECTUS! GEOFFREY A. LEVIN? ABSTRACT Leaf architecture and epidermal morphology of 259 species representing 51 genera in the Phyllan- thoideae (Euphorbiaceae) vary in a taxonomically significant way. Most tribes are epa homo- 1 geneous and well defined, риу tic era. Tribes that are generally considered to be closely related have similar ‚ең "Bienes in y. Foliar m support placement of Euphorbiaceae near the Violales in the Dilleniidae, rather than in the Rosidae. Paleobotanists, faced with the task of identi- fying fossil angiosperms in the absence of mega- fossils of reproductive parts, have instead de- mendous increase n the sophistication of the techniques for in- а and description of foliar morphology (review by Dilcher, 1974; see also Fryns-Claes- sens & van Cotthem, 1973; Hickey, 1973; Hill, 1980; Payne, 1979; Stace, 1965). Particularly im- portant in this progress was Hickey’s (1973) de- velopment of a precise and non-overlapping ter- minology for the description of shag hae leaf architecture, defined by Hickey as “the placement and form of those elements en tuting the outward expression of leaf structure, including venation patterns, marginal configu- ration, leaf shape, and gland position." Taxonomists studying higher taxa of extant angiosperms have generally ignored foliar char- acters, particularly aspects of architecture. Yet Hickey and Wolfe (1975), in their pioneering survey of dicotyledon leaves, found “coherent patterns of morphology of apparent value in de- termining taxonomic and phylogenetic relation- ships" at the ordinal and subclass levels. A few le mamelididae (Wolfe, 1973), Onagraceae (Hick- ey, 1980), Marcgraviaceae (Bedell, 1981), Rubiaceae (Pray, 1953), and Rutaceae (Dede, 1962). Clearly, however, the leaf architecture of extant а remains largely unexplored by system ата pa have had a longer history of use in taxonomic investigations (see review by Stace, 1965), although here, too, our knowl- edge of fossil cuticle may exceed what we know of extant angiosperms. Epidermal features have art phorbiaceae (Raju & Rao, 1977), and Magnoli- aceae and related families (Baranova, 1972). e Euphorbiaceae are an ideal family with which to explore the potential contribution of leaf architecture and epidermal morphology to higher level taxonomy. This large family, which contains about 300 genera and 7,000 species (Webster, 1967), has long attracted systematists' attention, resulting in an impressive series of in- frafamilial classifications. The first treatment, by Jussieu (1824), has been succeeded by further treatments prepared by Baillon (1858), ” Mueller (1866), Bentham (1880), Pax (1890), Pax and Hoffmann (1931), Hurusawa (1954), and Hutch- inson (1969). In addition to the information on floral morphology on which these systems are largely based, evidence has accumulated on anat- ! This study с part of a Ph.D. dissertation submitted to the University of California, Davis. My his committee, es A. Doyle, anuscript and a жаы and guidance. Com onymous reviewer also imp many cleared lea is gratefully acknowle Wolfe, made helpful comments on the ordon McPherson, Nancy R. Morin, and ce oved the manuscript. 1 шаш. hank Dr. Wolfe for the generous loan of aves. Financial on from the University of California, Davis, and Ripon College, Wisconsin, ? Natural History Museum, P.O. Box 1390, San Diego, California 92112. ANN. MISSOURI Bor. GARD. 73: 29-85. 1986. 30 ANNALS OF THE MISSOURI BOTANICAL GARDEN omy (Rothdauscher, 1896; Gaucher, 1902; Pax & Hoffman, 1931; Metcalfe & Chalk, 1950: 1207- 1235; Hayden, 1980), palynology (Erdtman, 1952: 165-175; Punt, 1962; Kóhler, 1965), and cytology (Webster & Ellis, 1962; Miller & Web- ster, 1966). Webster (1975) produced a treatment taking into account data from all these fields. Yet, in spite of all this attention, the foliar morphol- ogy of the Euphorbiaceae remains little known (but see Sehgal & Paliwal, 1974; Hayden, 1980). The phylogenetic relationships of the Euphor- biaceae are controversial, which also makes the study of euphorbiaceous leaf architecture poten- tially valuable. There are two schools of thought regarding the position of the family. The first has accepted rosid affinities. Based on the pluriloc- ular gynoecium containing two (or one) ovules in each locule and the presence of a nectary disk, Cronquist (1981) argued for a close relationship to the Celastrales. Webster (1967) was much more cautious, suggesting that Euphorbiaceae evolved from a primitive rosalean plexus. Among living orders, he considered the Euphorbiaceae closest to the Geraniales, with which they share a similar gynoecium and epitropous ovules. The second school has placed the Euphorbiaceae in the Dil- leniidae near the Violales and Malvales (Croizat, 1940; Hickey & Wolfe, 1975; Takhtajan, 1980; Thorne, 1976). These authors emphasized the similarities between the wood, leaves, gynoeci- um, and pollen of the Euphorbiaceae and the primitive members of these orders, particularly the Flacourtiaceae and Sterculiaceae. A few oth- ers, including Hutchinson (1969) and Airy Shaw (1965, 1973) have proposed that the family is polyphyletic, with relatives among both the ros- ids and dilleniids. In order to examine the systematic value of foliar morphology in general and to try to shed some light on the relationships of the Euphor- biaceae, I chose to work with the Phyllanthoi- deae, which is generally recognized to be the most primitive subfamily because many of its mem- bers have petaliferous flowers that, although uni- sexual, contain large androecial or gynoecial ru- diments. This subfamily of 57 genera and about 2,000 species can be recognized by its lack of latex; alternate, stipulate leaves; tricolporate to rate, non-spinose pollen; biovulate locules; and ecarunculate seeds. In this paper I will describe phyllanthoid foliar morphology and discuss its bearing on the ancestry of the Euphorbiaceae. In later papers (Levin, 1986 and in press) I will present results of phenetic and cladistic analyses [VoL. 73 of the leaf characters that demonstrate that they do vary in taxonomically and phylogenetically meaningful ways. MATERIALS AND METHODS I examined leaves of 259 species belonging to 49 genera that Webster (1975) included in the Phyllanthoideae. I was unable to obtain leaves of Apodiscus Hutch., Chascotheca Urb., Chori- sandrachne Airy Shaw, Danguyodr) Leandr., oe Muell. d^ Oreoporanthera d ricto 1 also studied a ate Martretia a Par- adrypetes, genera that are usually included in the Phyllanthoideae but were excluded by Webster. For all but the very largest genera (e.g., Glochid- ion and Phyllanthus), my sample includes a min- imum of 1096 of the species. Leaves were removed from herbarium speci- mens and cleared and stained according to the method of Wolfe (cited in Dilcher, 1974), except that some were mounted in Eukitt rather than Permount. Most were photographed with Kodak Technical Pan Film SO-115, a black-and-white, panchromatic negative film with extended red sensitivity, which was then developed and print- for maximum contrast. The resulting prints frequently showed details of venation more clearly than the original cleared leaves. Addi- tional photomicrographs were taken with Koda Panatomic-X, often using a green filter to en- hance contrast. Cleared leaves and negatives are stored in the collections of Dr. Jack Wolfe, United piles Geological Survey, Denver, Colorado, and d their vouchers are listed below with the descriptions of their e same or closely related species. Whenever possible, herbarium specimens that had been an- notated by authorities v were chosen. In a few cases, I checked suspected d using the ap- propriate regional manuals or monographs. Initially, савд species y Was scored individually. Ith eng Most of these groups correspond | to genera, sub- genera, or sections, following Pax and Hoffmann (1922) for most genera, Jablonsky (1915) for Bri- delia and Cleistanthus, and more recent mono- graphs when these were available. In a few cases (e.g., Bridelia, Drypetes) the species within a sec- 1986] tion formed two or more homogeneous groups, one or more of which was highly similar in ar- chitectural and epidermal characters to groups belonging to other sections. In these cases I dis- regarded the taxonomy and lumped similar species into a single group. Following the ter- minology of numerical taxonomists, I will refer to the homogeneous groups as operational taxo- nomic groups, or OTUs, in this and future papers (Levin, 1986 and in press). Table 1 lists the OTUs included in this study. CHARACTERS Leaf architectural terminology generally fol- lows Hickey (1971, 1973). Leaf size classes are those of Raunkiaer (1934); nothing about ve- nation e implie d by the terms microphyll or mega . Terminology for cuticular characters al follows ‘Pilcher 1974), who systema- tized Stace’s (1965) terms. Specific trichome types follow Payne (1977) and stomatal terminology includes terms introduced by Payne (1970, 1979). I also recorded a few features of the leaf's internal anatomy, including type and distribution of crys- tals, degree and extent of sclerification of the bundle sheath, size of the terminal tracheids rel- ative to the other tracheary elements in the free- ly-ending veinlets, and prominent idioblasts. few characters that I scored exhibited con- siderable variation within groups that were oth- erwise homogeneous. These characters include size and shape of the leaf, shape of the apex and base (with the exception of base balance), relative size ofthe terminal tracheids, sclerification of the bundle sheath, ultimate marginal venation, epi- dermal cell size, and trichome density. The last three of these have been shown in other groups to be sensitive to the environmental conditions & Bjorkman, 1978); such phenotypic Жарк, would be particularly troublesome with small sample sizes such as mine. Marginal aia and sclerification of the bundle sheath can also show evolutionary instability, at least in the Magnoliaceae (Tucker, 1977) and Hibbertia (Rury & Dickison, 1977). Leaf shape characters are fre- quently used in keys to species in the Phyllan- thoideae, suggesting that shape varies consider- ably at the species level, possibly in response to interspecific ecological differences, and thus will not be useful at higher taxonomic aiios When I eliminated t t varied LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 31 too much or that were monomorphic within the Phyllanthoideae, 43 characters remained. The character states for these ae and their codes are listed in Table 2. The order of character states within a character does not жалы ир: im- ply any evolutionary sequence. The following characters require more explanation: 3. Margin. In most members of the Aporu- seae, the margins are entire to crenulate with small glands in the sinuses (Figs. 64—67). See the descriptions of this tribe for more details of this condition, which I recorded as state G. Three genera, Bischofia, Drypetes, and Putranjiva, bear theoid teeth. See the descriptions of these genera for a more thorough discussion of the tooth mor- phology. 4. Venation. I recognized two expressions of by well differentiated secondary, tertiary, and quaternary veins (Figs. 32, 34). In the second, state W, the secondary loop is accompanied by additional secondary loops that gradually de- crease in size and merge into tertiary loops (Figs. 31, 36, 97), a pattern I will refer to as weakly brochidodromous. This pattern differs from eu- camptodromy (state U) in that in the latter the connections between one secondary and its superadjacent secondary are not formed by prominent secondary loops but are simply higher order crossveins (Figs. 41, 44). In the Phyllan- thoideae, weak brochidodromy may form the evolutionary link between eucamptodromy and simple brochidodromy. 7. Angle of origin of basal secondary. If the angle formed between the lowermost pair of sec- ondary veins and the midrib was consistent with the angles formed by the other lower secondaries (Figs. 14, 36, 97), I recorded state S; if the basal angle was either markedly more acute (Figs. 55, 64) or more obtuse (Figs. 3, 62), I recorded states A or O, respectively. 12. Size of outer loops. The sizes of the loops, if any, accompanying a secondary loop may be uniform (Fig. 90), decrease upward from the sec- ondary loop (Fig. 55), or vary irregularly (Figs. 90, 119) 17. Composite intersecondary frequen- cy. Thethree classes I recognized appear to have well marked boundaries in the Phyllanthoideae. 18. Intramarginal vein. In some sections of Bridelia the outer portions ofthe secondary loops orm a vein at the leaf margin (Figs. 37, 38). Ww N TABLE 1. Character states of taxa studied. Sequence of genera follows Webster (1975). Two genera that were examined, Poranthera and Reverchonia, are hliched elcewhere r 30, 39) for explanation of character states. Numbers in parentheses refer to operational taxonomic units (OTUs) used in numerical (Levin, 1986 and in press). ANNALS OF THE MISSOURI BOTANICAL GARDEN (1E) рарирајә г $ Y (0£) с dnoi8 y (67) І dnosd ршәрпир (87) snyjuvipuods (LZ) sisdoyjunyjAyd $ Y (97) І dnoi8 əuyovapuy (SZ) роо] (pc) sixádo197 $ 9 (€Z) sajouisny $ 9 (ZZ) 111142105 7) (Ic) snyjunisiajD $ 2 (07) 12001404) $ Э (61) € dnoi8 ‘Э (8D nomdns $ 2 (11) 57ш155ђрипипэр SNYIUDISIO]O (91) € dnoJ3 D1]2p14g (S1) с dnos3 руармЯ (pI) 1 dnoj3 руарыЯ (£6) 211221049) (СП) z dnoi$ p1ydoJ oy (11) 1 dnos3 pjn(dojop (01) Роиршү (1) 2147$ $ "S (9) DIADSOL9J2J $ DIADS (8) snosipojviad (£I) tnij2r4qpiuod (p) sipGsouuov'] (€) vipoowtag (©) sndupvooosiq (6) 01108 (T) DISVDIOLIS Y (1) DIpudjol Organization 2. Basal balance 3. Margin 4. Venation l. un N N N N N N N Nn N N N Nn N N un un un ua un Nn п п п un N Y un un п N Nn umuoz«uwwaczmrrom«co ш ЕБ nnn2 ESnnNnDAOALX4MO Ott Dams nett Deine m e nen n Zu Nan a nan B WWUB U WB W WB B nan nan U фо ES и Ean nen Ean Ean Ean Ean Е5 n2n ЕЁ < Ean ES EZO Ean р Wp р р Pp Ww ш ш m w ui m ea) n са щ га шщ m щ а) un га ш га щ W ш са ш ea) [e| щ щ B B BB BB UB B В В В WB У S S AS S S AAS S S S 5. Primary size 4. 3? angle, exmedial ---- —– А А А А А А А А А А А А А А А А А А 1 16. Simple intersecond. 17. Composite intersecond. 18. Intramarginal vein [Vor. 73 о 5 м о Zw ош м Xu eX о Zw 02" м 5 к 5 ОЁ оё о Zw оё ОЁ ОЁ оё о Highest order present 19. High order pattern 20. High order, size 21. 1986] Continued. TABLE l. LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. (TE) DipudadaL $ “Y (ОЕ) c dnox3 y (6c) | dnoJ3 вибарпиу (87) snyjuvipuods (LZ) Sisdoyjunjyjdyd $ Y (9c) | dnoJ3 auyovapup (sz) рцәола (pz) 51х440127 $ “9 (£z) sapmauisny $ `D (ZZ) 11414215 Э (12) snuiuvisi2]) $ 2 (07) 12001404) $ `Э (61) $ dnoj3 5 (81) ивтан$ $ 2 (11) smurissipununop SNYIUDISIAD (91) $ dnoj8 руария (S1) с dnoi$ vyapug (p1) 1 dnoiz вуариЯ (€6) 211021049 (CD с 1п013 руду (11) 1 dnos3 ругуаәру (01) роиршу (L) 2142$ $ "S (9) DIADSOL919J $ 01405 (8) snosipojpiad (€ 1) 1mmnjovaqpiuad (f) sitsouuor'T (€) vIpoomda yy (с) sndavooosiq (6) 27019 (с) DISDIOLISY (1) призами m MeN а 4 5 сч - MEO Я. ер ч 24 uu и 5 - om >с! m. 22. Areole development 23. Areole arrangement | | | | | | | | + | | | | 27. Prismatic, mesophyll 28. Prismatic, veins 29. Druses, mesophyll 30. Druses, veins S 31. Epidermal wall, adaxial Ег тл г Liki b it ib ELL LE ълът т Р Р Р Р Р Р Р Р Р Р Р А А Р Р Р Р Р Р Р Р Р Р Р Р Р A A P P P P 36. Stomatal type А А А А А А 5 А А А А А 5 S AAAS $ 5 5 А А А А $ $ $ А $ $ $ 38. Unicellular trichomes 40. Peltate trichomes 1. Filiform sclereids 4 + + + + 43. Sclerified epidermal cells Continued. TABLE 1. ANNALS OF THE MISSOURI BOTANICAL GARDEN (09) рширшләши17 (65) 51/Аабоицоворпаз4 | (85) ютузәшәрү (p6) 010119ц5]ә8и17 | (LS) uospuapodvay (95) ртуѕиојдог ($$) vaifuvamd (06) DADAMDMOAR | (ps) 9 dnoi3 (7 (ES) ç dnos3 q (TS) у dnoi3 q (15) € dnos3 q (0$) c dnol3 q (6b) | dnoi8 sozad4ug (8p) 01424213 (Lp) 0140q032w01044 (9p) 2441040522! (Sp) snisioowpiq (pp) иомриро$] $ `q (£p) ovina ‘a (cv) мримаа $ “E (Ip) | dnoij8 vounvoopg (0p) Diuolysy (6€) DSnsody (8€) si402p221 I (LE) ршаиола2Т (9€) nuu u0424H (сє) вИзивиао (ре) в/и4шазавцо $ ^V (££) »souimjaA $ y (TE) DUDIUON $ Y SS S S S S S S S S S S S S S S S S S S S S S S S S S S S SS 1. Organization 2. Basal balance 3. Margin S S S S S S S S S S S S S S S SS А А А А А А А $ S S S SS E T TT E TE E E E E E T T B ЕСЕ G UB B B WB B B B B B B W W WWW 4. Venation Е5 Ean EZn Е5 ЕЁ Ean >< mmm M M M A AS >< >< gz« mS S m m M MMMM SASSA оё < nan S SS MM M A OA о 5 < nan m N O n2n ZZ n20 ЕЁ < Е 5 < 7. Angle, basal 2° n о n Nn n « N S S SS S S S S S S S S S S S S S S S S S S S S 8. Angle, lower 2? OLAZ n camas р<«ибы<& ntMxO nia к< ©о<<ДО< а б <<[До=< с о‹<%Г[ДС ш < а и<иа к< 2 z a А А А А А A А І АЕ А А А А A A ATI I А А А 1 [VoL. 73 о 5 м OZ m Ex о 5 GM 5 о Zw oZ OZ © 02 Lact о Ё OZ о 5 о м о Ё м о 5 Ф о Ё м ОЁ о Ф о 5 м eX о Ё м о Zw о 5 м о Ё м о м о 5 м wast о Ё м о Ё м о 5 ш 1986] Continued. TABLE 1. LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. (09) Ре,иршләшш:7. (65) sijfisouuopjopnasQq (85) ваша! (p6) vnui2usposurT (LS) uodpuapodvay (95) viysuojqor (Сс) vatfupamq (06) рәрлмрмоәд (ps) 9 4пол3 q (єс) 6 п018 q (сс) y dnos3 q (15) € dno q (05) z dnoi3 q (6p) | dnoi8 sajadtiq (8p) мау (Lp) v140q0321401044 (9$) »«410q0820]y (Sp) snisi20144piq (pp) моыриро$5] $ `Я (ЄР) 2514па `Ч (гу) примета $ `9 (Ір) 1 00018 2241022049 (0$) viuolysp (6€) vsnaody (8€) si40272214 L (LE) vw2uoid21 (0€) nuuo424]t (SE) ByjauntjaD (РЄ) 0//19шәѕәрцо) $ V (£g) »souunjaA $ "V (TE) DuDJUOW $ Y M uu >с! Дд Омиш — — — — 22. Areole development 23. Areole arrangement м >с | an м 5 сч | >с - Be >с 27. Prismatic, mesophyll 28. Prismatic, veins + + + + + + + + 29. Druses, mesophyll + + + + , adaxial 31. Epidermal wall 34. Stomatal location m m m m m em am m m ea m m ea m ea m e « A ea em em ea e e O e O O ea m P Р Р Р Р Р Р А А А А А А Р A А А Р Р Р Р Р Р Р Р Р Р Р Р Р Р S SS AAS S S S TAS S TS SS А А А А А А А А А $ АА $ А 38. Unicellular trichomes 39. Multicellular trichomes 40. Peltate trichomes + + + + + + + ++ + 42. Tannin. epidermal cells 43. Sclerified epidermal cells Continued. TABLE 1. ANNALS OF THE MISSOURI BOTANICAL GARDEN (T6) 279110 (16) vyouosig (68) Р1р4рэоиәш&Н (88) гора (99) uowajsouds (L8) sndounvg (98) nj[21421/2131 ($8) vwspdaog $ d (p8) uoipigoojBouotsq $ ‘а (£8) uo1pij20]304purá]oq. $ "d (c8) u01puspoyiuo¡J4Yd $ ‘а (18) wnipiydwospivd $ ‘d (08) snyiuvyduiáN $ d (61) vw2]20110N $ ‘а (8L) uoipi20]304212H. $ `4 (LL) v2211142H $ ‘а (91) с dno18 ‘wnipiyduoy “3qns y (SL) 1 dnoi8 ‘ширщашог "qns q (pL) 1punquoj4 $ d (£L) vot]qui3 $ d (TL) шти‹8оләцтәд $ `4 (14) 22215 $ d (OL) Djjasosody $ ‘4 (69) ршаиозтиу $ d | (89) u01p1/20/80U9py $ SNYIUD]IAYA (L9) 0140140840уү (59) € dnoi8 1) (#9) z dnos3 :9 (£9) wuoipi20]2) $ ио1р1Ц20] 1) (c9) VIUÁDAZ (91) 2288эп] 1 un Q Nn un п un un Nn un un un Nn N N N N N N Nn un N N N N N N un N YN N N Organization 1. 2. Basal balance о N Nn N N N N < N N < un N N N N un un uo N YN un п Y п N < YN N N un nan2nAntO0_OXO “Ка ЕЁ ми<<<0ожш< ж nin <ии<иа к? шоши 5 ш ш8 ЕБ и we ЕБ и N E B m mm Sm m m m m M MM M M M M M M S S OO $ $ $ $ $5 Ean о 2 n SS SSSSSSSOSSSSSSSSSSSSS S mm mm m mS S S M MMM MM NN MM S SS S S OO OS О S WW WWWBB W WWBWBB BWB WwWB WB В В S M S A E E E E E EE E E EE E E E EE E E E E E E E E 3. 3? angle, admedial 7. Angle, basal 2? 8. Angle, lower 2? 6. Simple intersecond. == [VoL. € 5 € 5 „ оё о о Zw мш + о Zo о Ё MH о ч о 5 у Lat oZ f сё х X о х 5 х 5 as o Zw “> > wan о 5 о Zo о 5 n о 5 о о Zu о 5 n о ш ош о 5 м о Ё м 19. High order pattern 21. Highest order present 20. High order, size 73 1986] Continued. TABLE 1. LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. (T6) DDN] (16) vyouosig (68) Pip4v20u2iuAH (88) 222421 (99) uowajsouds (L8) sndounpg (98) D]]914942 y (68) vuusodaos $ d (p8) uompryrojsoyodsg $ "d (£8) u01p14y90]801puvajOg $ ‘а (c8) uospuapoylunyjAyd $ d (18) wmnipigdwuosvavq $ d (08) nyuDYAWÁN $ d (6L) ршәЈоціом $ ‘а (81) u01p1/90/804919H $ `4 (LL) v2210u2H $ d (91) z dnoi8 читрщашоо '3qns ‘а (SL) 1 п018 ‘wunipiydwoy ‘8405 ‘4 (pL) ipunquoj4 $ d (£L) DO1quiz $ d (cL) штиволацта]4 $ d (11) 222012 $ d (0L) vjjoso4ody $ ‘а (69) ршәиоѕиу $ d (89) ио1рицзо8оиару $ snyjunydyg (L9) 0140)14084риг (¢9) € dnoi3 ‘0 (#9) с dnoid `0 (£9) иотртуэоо) $ иотртцэоо) (c9) vrut24g (91) vaaSonj4 ер а Sean Симона = 2с - м - 24 R — I RR RRRRRRRRRRRRRRRRRRRRRR 22. Areole development 23. Areole arrangement — — — S MM MMS MS SS MSS MMMMS MMMMMS MS 22222 22 222 262722 8 2222273 8 252 27. Prismatic, mesophyll 28. Prismatic, veins 29. Druses, mesophyll 5555 55 SS SS UUS 5$ 5 S US $ $ US S S S S S S S SUUS S S S S S US + 31. Epidermal wall, adaxial 32. Epidermal wall, abaxial bel oL bd L L LL Lob hohe bL LL LE bb hb LLLBLLELSo.6 34. Stomatal location P P P P P P P Р P PP P PP P PP P PP P PP P PP P P Р P P P P A P P P А А А А А А А А А А А А А А А А А А А А А А А А А А А 5 5 А А 38. Unicellular trichomes . Filiform sclereids 43. Sclerified epidermal cells 38 ANNALS OF THE MISSOURI BOTANICAL GARDEN TABLE 2. Characters and character states. See Hick- ey (1973), Dilcher (1974), and text (pp. 30, 39) for more explanation. — . Organization: Simple = $; Compound = С. 2. Base balance: Symmetrical = S; Asymmetrical = ы . Margin: Entire = Е; Entire or crenate, but with glands = G; Toothed = T. 4. Venation: Brochidodromous = B; Weakly broch- idodromous = W; Eucamptodromous = U 5. Primary size: Moderate = m; Stout = S; Massive = М. . Secondary angle: Narrow (<45°) = N; Moderate = M; Wide (> 65°) = W. Angle of basal secondaries, relative to adjacent secondaries: More acute = A; Similar = S; More obtuse = . Angle of lovis secondaries, relative to middle sec- ondaries: More acute = A; Similar = S; More ob- tuse — . Angle oft upper secondaries, relative to middle sec- Е More acute = A; Similar = S; Моге ob- = О. an ~ со © m : Secondary course: Curved uniformly = U; Curved abru =A. ; Anel ee secondary loops: Acute = A; Right = Obtuse = O. . Size " outer loops: Irregular — : Uniform = U; Decreasing upwards = D; Abse Е TUM iio of origin, и oral Right Obtuse = O. — N — w A А dens bud of origin, exmedial: Acute — A; Right = R; Obtuse = O. — un . Tertiary pattern: Ramified = r; Random reticulate = p; Strongly У angle to midrib predom- inantly right : = cee Вы: Absent = —; Present = an — N . р БИ ИЕ Absent = A; шйе- quent (in fewer than 20% of intercostal areas) = I; Frequent (in more ad 2096 of intercostal areas) =F 18. Intramarginal vein: Absent = —; Present = +. 19. Higher order vein pattern: Ramified = r; All ran- dom = R; 4° orthogonal, a orders random = 0; 4° and 5? orthogonal = 20. Higher order vein size: All PENA — M; 4? mod- erate, 5? fine = m; 4? and 5? fine = F; 4° moderate, 5? heavy — h. 21. Highest order present: 4? = 4; 5° = 5; 6° = 6. 22. a development: Incomplete = i; Imperfect = I; Well-developed = D. 23. Aishe arrangement: Random = R; Ordered = O. 24. Areole shape: Irregular = I; Regular = R. 25. Areole size: Large = L; Medium = M; Small = S. [VoL. 73 TABLE 2. Continued. 26. Veinlets: Absent = A; Spk = S; Branched 1-2 x = 2; Branched 2-3 x 27. ee mesophyll: Absent = —; Pres- 28. ЕК crystals with veins: Absent = —; Present = + 29. Druses in mesophyll: Absent = —; dies - 30. Druses with veins: Absent = —; Present = +. 31. oo anticlinal walls, айа назр Ма = $; Undulate = U. 32. Epidermal anticlinal walls, abaxial: Straight = S; Undulate = U. 33. senasa papillae: Absent = —; Present = +. 34. Stomatal location: Abaxial only = L; Primarily abaxial, but a few de Approximately equal n both surfaces 35. al index ce 09 <10% = A; 10-2096 = 36. Stomatal ipe т = P; т =A. 37. Water stomata: Absent = sent = +. 38. Unicellular Se pim Antas A; esas = $; O = Т. 39. Uniseriate, multicellular trichomes: Absent = — Present = + 40. Peltate trichomes: Abse —; Pre = +. 41. Filiform sclereids in E Hund = —; Pres- > N = +. Я Tanniniferous epidermal cells: Absent = —; Pres- nt= +. A [9v] i Slee epidermal cells: Absent = —; Present = Although technically marginal in position be- cause no leaf tissue lies outside of it, this vein appears to be homologous to Hickey’s (1973) intramarginal vein and I will use his term. See the description of Bridelia for more details. 26. Veinlets. The degree of branching of the freely-ending veinlets did not break down cleanly into simple, branched once, branched twice, etc. UE branched 1-2 times, and branched 2- 3 times appeared to reflect the variation more accurately. 35. Stomatal index. These classes appear to represent natural ranges that are distinct from each other in the Phyllanthoideae. 6. Stomatal type. Because both paracytic and brachyparacytic (Dilcher, 1974) stomata oc- cur in many taxa, or even on the same leaf in 1986] many species in the Flueggeinae, I did not dif- ferentiate between these basically paracytic pat- terns in Tables 1 and 2. For the specific stomatal types, see the descriptions of the taxa. 37. Water stomata. Some Andrachne O bear large, apparently n ad- dition to their Sd: numerous, smaller aniso- cytic stomata (Fig. 38. Unicellular m In the Phyllan- thoideae, unicellular trichomes are usually sim- ple with a peg-like, unmodified base and radial basal cells (Figs. 48, 57). Tanniniferous epidermal cells. In several tribes, some of the epidermal cells are enlarged and spherical, with the adjacent cells forming a radial pattern (Figs. 58, 110). These contain tan- nins, according to Rothdauscher (1896) and шо ia I have not attempted to confirm their 43. pon cells in the lower epidermis. See the descriptions of 4manoa and Discocarpus for more on these cells. In addition to these characters, one character not among those listed in Table 2 but included in the descriptions below requires further expla- nation. Hickey (1971, 1977), based on his ex- amination of cleared and fossil dicotyledon leaves, categorized the degree to which venation is organized into four levels or ranks. Rank is based primarily on the highest vein order in which the veins follow relatively regular courses and delimit areas of relatively consistent size and shape. If the secondary and higher order veins have irregular courses and the intercostal areas, therefore, are not uniform in size and shape, the organization is said to be first rank: only the primary vein has a regular course. If the second- ary veins follow regular courses and the inter- costal areas have uniform size and shape, but the tertiary and higher vein orders are disorganized, the level of organization is second rank. As the degree of organization increases further, the ranks increase to third and finally fourth, in which even the areoles show at least rudimentary orientation within domains. Leaves with low organizational rank appear first in the fossil record and, among living dicotyledons, the rank of foliar venation organization correlates well with the degree of advancement of the whole plant as determined by other evidence (Hickey, 1971, 1977; Hickey & Doyle, 1972; Doyle & Hickey, 1976). In general, rank is fairly easy to assign by fol- lowing Hickey's criteria. Problems may arise, however, with leaves from arid, arctic, or alpine LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 39 habitats. In plants from these habitats, the or- ganization of leaf venation frequently shows a regression below the level typical of their more temperate relatives, particularly in the lower vein orders (Hickey, 1977; Rury & Dickison, 1977). As a result, leaves can be found in which the higher order veins are more organized than are the lower order veins. It is not clear what ranks Hickey would assign in such cases. I recommend basing rank on the highest vein order that is reg- ular in course, regardless of the behavior of lower order veins. In this way, the degree of evolu- tionary advancement achieved by the leaves will be better reflected by the rank assigned. RESULTS Table 1 summarizes the distribution of char- acter states for the characters in Table 2. Two genera that I examined, Poranthera and Rever- chonia, are not included in Table 1 because the extremely xeromorphic leaves of these genera contain too few characters. In the discussion that follows I will first de- scribe the leaf architecture and cuticular features of the subfamily. I will then describe each genus in turn, following the sequence in Webster's (1975) classification, including only those char- acters that are not adequately summarized in Tables 1 and 2. At the end of the generic de- scriptions for each tribe, I will discuss the taxo- nomic implications of the foliar morphology. DESCRIPTIONS AND DISCUSSION PHYLLANTHOIDEAE ASCHERSON Leaves simple (compound in Bischofia), mi- crophylls to mesophylls, some nanophylls; blades typically symmetrical, shape various; apex and b vari 1 gina venation typically second to third rank, but first or fourth rank in some; venation pinnate, bro- chidodromous to eucamptodromous, a few species kladodromous or reticulodromous; pri- mary vein moderate to stout, rarely weak or mas- sive, typically straight; secondary vein angle of divergence acute, uniform or varying, thickness moderate, curved uniformly or abruptly, loop- forming branches, if any, joining superadjacent secondary at acute to obtuse angle, only in Bri- delia forming an intramarginal vein; intersecon- 40 ANNALS OF THE MISSOURI BOTANICAL GARDEN dary veins simple, composite, or absent; tertiary veins Originating at various angles, reticulate or percurrent, in a few species ramified admedially, if percurrent straight or sinuous, branched or un- branched, oblique to primary vein with angle constant or decreasing upward or outward, typ- ically alternate; higher order venation random to orthogonal, typically resolved to fifth or sixth order; marginal ultimate venation various; ar- eoles imperfect to well developed, incomplete in me, ngement random or oriented, sha typically irregular, size small to large; veinlets simple to branched three times, rarely absent; terminal tracheids normal or swollen; bundle sheath of primary vein almost always sclerified, higher order veins parenchymatous or sclerified; epidermal cells isodiametric, arrangement ran- dom or tetragonal to polygonal, anticlinal cell walls straight or undulate and typically unor- namented, surface smooth or papillate; stomata typically only abaxial, on both surfaces in a few, stomatal index about 8-25%, paracytic, brachyparacytic, parallelocytic, aniso- cytic, or, rarely, anomocytic, development me- sogenous or mesoperigynous; trichomes typical- ly unicellular with peg-like base and radial basal cells, some uniseriate with an unmodified base, rarely multiseriate or with a multicellular or en- larged head or branched; domatia absent except in Bischofia with marsupiform domatia; crystals typically present within leaf, single and prismatic or druses O WIELANDIEAE BAILL. EX HURUSAWA Wielandia Baill. (OTU 1) W. elegans Baill., Seychelles: Bernardi 14639 (BM). Figures 1, 2. Leaves large microphylls, symmetrical, ovate- elliptic; apex acute; base acute; organization of ultimate marginal $ higher order veins iones stomata paracytic. Astrocasia Robn. & Millsp. (OTU 2) A. neurocarpa (Muell. Arg.) I. M. Johnst., Mex- Tamaulipas, Stanford et al. 2264 , Mexico: Tres Marias Island, Howell 10405 (CAS). [Vor. 73 A. phyllanthoides Robn. & Millsp., Mexico: Yu- catan, Gaumer 475 (UC). Leaves microphylls to small mesophylls, sym- metrical, very wide ovate to orbiculate; apex rounded or emarginate; base rounded or peltate; organization of venation mid third rank; mar- ginal ultimate venation fimbriate; terminal tra- m normal (A. peltata, A. phyllanthoides) or swollen (A. neurocarpa), bundle sheath paren- a bra tic; lower epidermis papillate (4. ph и or smooth. Blotia Leandri (OTU 9) B. oblongifolia (Baill.) Leandri, Madagascar: Serv. For. 6157 (P). Figures 5, 6. I ll trical, narrow ellip- tic; apex acuminate; base obtuse; organization of bundle sheath of higher order veins lightly scle- rified; stomata paracytic. Discocarpus Klotzsch (OTU 5) D. spruceanus Benth., Guiana: Spruce 3527 (MO). Figures 8, 9. Leaves microphylls, symmetrical, elliptic; apex attenuate; base obtuse; organization of venation mid fourth rank (but see below); marginal ulti- mate venation fimbriate; terminal tracheids nor- mal; bundle sheath of higher order veins par- enchymatous, but extensions sclerenchymatous; stomata brachyparacytic. My anatomical observations agree with de- scriptions by Rothdauscher (1896) and Gaucher (1902). The leaves of Discocarpus are remarkable in many respects. Although the higher order ve- nation is strikingly orthogonal, forming ex- tremely well developed areoles, the lower order venation is much less organized, with uncom- mon simple intersecondaries and weakly per- current tertiaries that are quite irregular and sin- ower epidermis appearing to be subsidiary cells 1986] LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 41 LX 0) Рр ox 4 Р Gy a8 Я aut Sv. A E : Жу IRSA Lae m, wey ONSE WO КУБЫ СЕ ома E 9 1 P "71 X FIGURES 1-6. Leaves of Wielandieae.— 1, 2. Wielandia elegans. — 3, 4. Astrocasia neurocarpa. — 5, 6. Blotia oblongifolia. Bars equal 1 cm in Figures 1, 3, 5 and 1 mm in Figures 2, 4, 6. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 о» г 0,60 ea LP 4 ee, b ә 24 са aj% FIGURES 7-13. Leaves of Wielandieae.—7. Astrocasia neurocarpa, swollen terminal tracheids.—8, 9. Dis- cocarpus spruceanus. — 10, 11. Heywoodia lucens. — 12, 13. Lachnostylis hirta. Bars equal 1 cm in Figures 8, 10, 12; 1 mm in Figures 9, 11, 13; and 100 um in Figure 7. 1986] or guard cells. The lower epidermis is also strong- ly sclerified, with the outer wall of most cells thickened (cf. Amanoa, below). Heywoodia Sim. (OTU 3) H. lucens Sim., Tanganyika: Watkins 453 (EAH). Figures 10, 11. Leaves microphylls, symmetrical, elliptic; apex acuminate; base obtuse-cuneate; tracheids nor- mal; bundle sheath of higher order veins par- enchymatous; stomata paracytic. Lachnostylis Turcz. (OTU 4) L. hirta (L. f.) Muell. Arg., South Africa: Burchell 5213 (G). Figures 12, 13. Leaves microphylls, symmetrical, oblanceo- late; apex obtuse; base acute; organization of ve- nation high second rank; marginal ultimate ve- nation looped; terminal tracheids normal; bundle sheath of higher order veins lightly sclerified; sto- mata brachyparacytic, rarely each subsidiary cell divided again by a wall perpendicular to the guard cells. My anatomical observations agree with the re- ports of Rothdauscher (1896) and Gaucher (1902). Pentabrachium Muell. Arg. (OTU 13) P. reticulatum Muell. Arg., Cameroons: Zenker 3869 (MO). Figures 14, 15 Leaves mesophylls, symmetrical, ovate-ellip- tic; apex long acuminate; base obtuse; organi- zation of venation mid third rank; marginal ul- timate venation fimbriate; terminal tracheids normal; bundle sheath of higher order veins par- enchymatous; stomata paracytic; trichomes scat- tered on abaxial epidermis. Petalodiscus Baill. (OTU 8) P. danguyana Leandri, Madagascar: Schlieben 8120 (UC). Figures 16, 17 Leaves microphylls, symmetrical, narrow ovate-elliptic; apex acuminate; base obtuse; or- ganization of venation low fourth rank (but see below); marginal ultimate venation incomplete to weakly looped; terminal tracheids normal; bundle sheath of higher order veins lightly scle- rified; stomata paracytic. The leaves of Petalodiscus, as do those of Dis- cocarpus, exhibit an unusual combination of rel- atively disorganized lower order venation with highly differentiated orthogonal higher order ve- LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 43 nation forming oriented, but still imperfect ar- eoles Savia Willd. Sect. Heterosavia Urb. (OTU 6) S. bahamensis Britt., well 8782 (UC). Figures 18, 2 S. clementis Alain, Cuba: Pondus et al. 4036 о Cald- S. clusiifolia p Cuba: Alain & Clemente 1485 (DAV). S. pri pe Urb., Cuba: Alain & Clemente 919 (DAV). S. a Griseb., Haiti: Leonard 8334 C (UC). Sect. Savia (OTU 7 S. pee Willd., Haiti: Holdridge 1415 (UC). Figures 20, Leaves microphylls, symmetrical, obovate or elliptic; apex acute, obtuse, rounded, or emar- ginate; base acute-decurrent or cuneate or ob- tuse; organization of venation low to mid third rank; marginal ultimate venation looped or fim- briate; terminal tracheids normal; bundle sheath of higher order veins parenchymatous or lightly sclerified; stomata paracytic or occasionally par- allelocytic (Fig. 22). My observations agree with earlier reports on S. bahamensis, S. clusiifolia, S. erythroxyloides, and 5. sessiliflora (Rothdauscher, 1896; Gau- cher, 1902; Hayden, 1980) The leaves of the genera Webster placed in the Wielandieae generally form a homogeneous group. Two genera appear anomalous; however. Astrocasia, with its weakly percurrent tertiaries and narrow intercostal areas, appears more sim- ilar to some members of the Phyllantheae, a re- lationship suggested by both Punt (1962) and Kohler (1965) on the basis of pollen and even by Webster in an earlier paper (Webster, 1956). The leaves of Pentrabrachium are also of higher presence of unicellular trichomes, and the un- dulate anticlinal walls of the epidermal celis In these and other characters, Pentabrachium is much like Dicoelia, with which it shares the un- usual condition of having the lower and medial secondaries eucamptodromous and the apical secondaries brochidodromous. Both phenetic and cladistic analyses of leaf characters suggest a close relationship between these genera (Levin, 1986 and in press), a possibility that has not been pro- 44 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 age p y ОО BIE SES ES 3: SOLS SZ | { ^ | 2 A its a Vern ЖАЧ ЕЭ, үө, RS COS D] (t bene ў '" С 3 x^ ZA E Q M DAS a $ ‘> =e Y Сеа УМ g "- | ^ A GV EERI гр doe 9264 FIGURES 14-17. Leaves of Pentabrachium and Petalodiscus. — 14, 15. Pentabrachium reticulatum. — 16, 17. Petalodiscus danguyana. Bars equal 1 cm in Figures 14, 16 and 1 mm in Figures 15, 17. LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 1986] У NO a te aed ъъ „^^“ M RIT „© £2 aS ^ E EL) 25 À Y a) 95, UL 4 A [7 КО? ea » W' H m p cr \ DU ANS Lv. ә f V MW <) 93375479 м] ху » у ДАТ 2T. NA «= } 21. S. sessiliflora. — 22. S. 19. Savia bahamensis. — 20, 3 FIGURES 18-24. Leaves of Savia and Amanoa. — 18 bahamensis, abaxial epidermis, showing paracytic and parallelocytic (arrow) stomata.—23. Amanoa oblongi- folia. — 24. A. guianensis. Darkening of margin is due to sclerified epidermal cells. Bars equal 1 cm in Figures and 50 um in Figure 22. , 24; 21 , 1 mm in Figures 19 18, 20, 23; ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Scl erified Cells Adaxial Abaxial 46 TABLE 3. Epidermal characters in species of Amanoa. Cell Size Patterns Species Adaxial Abaxial glaucophylla small cells in small and large groups of 5-15 cells randomly with matrix of mixed large cells grandifolia small and large small and large cells randomly cells randomly mixed i guianensis small and large two size classes cells randomly not well differ- mixed entiated oblongifolia only small cells only small cells present dense near margin none or scattered near margin none or scattered near margin scattered isolated ls in intercos- tal areas; dense over low order scattered isolated cells and m of cells in intercostal areas; dense beneath low order veins and near margin scattered isolated cells and small groups of cells in intercostal areas; dense beneath low order veins and near margin few scattered isolated cells in intercostal areas; beneath low order almost all cells sclerified, except scattered groups in intercostal areas and stomatal cells veins and near margin posed previously but that deserves further con- sideration. Blotia, Petalodiscus, Savia, and Discocarpus appear to form a closely related group. A trend toward disorganization of the lower order ve- nation coupled with increasing organization of the higher order venation, a very unusual pat- tern, can be traced from Savia sect. Heterosavia through sect. Savia to Petalodiscus and Disco- carpus. Blotia is quite similar to Savia, differing mainly in the large size of its leaves. AMANOEAE (PAX & HOFFM.) WEBST. Amanoa Aubl. (OTU 10) A. о Muell. Arg., Guyana: Amazonas, Krukoff 6370 (A). Figure 25. A. E Muell. Arg., Venezuela: Pittier 14799 (UC). Figures 26-28. A. Penn Aubl., Venezuela: (UC). Figure 24. A. oblongifolia Muell. Art., Guyana: Amazonas, Krukoff 7015 (A). Figures 23, 29, Cardona 401 Leaves microphylls to mesophylls, symmet- rical, oblong or elliptic; apex acuminate; base acute, obtuse, or rounded; organization of ve- nation low third rank; marginal ultimate vena- tion looped or weakly fimbriate; terminal tra- cheids normal to slightly enlarged; bundle sheath of higher order veins parenchymatous; stomata paracytic or rarely parallelocytic. Of the seven to nine species in this genus, the leaves of only A. glaucophylla and A. oblongifolia have previously been described (Rothdauscher, 1896; Gaucher, 1902; Hayden, 1980). My ob- servations agree with the earlier descriptions ex- cept for Rothdauscher's report, repeated by Met- calfe and Chalk (1950), of papillae and corkworts in the lower epidermis of A. glaucophylla. Re- ports of enlarged terminal tracheids vary in the literature. My observations suggest that this is a variable character within the genus, perhaps even within some of the species. As observed by Rothdauscher, Gaucher, and Hayden, cells of the lower epidermis of Amanoa exhibit different patterns of sclerification. In ad- dition, rather than being of relatively uniform size as in most phyllanthoids, the epidermal cells of most Amanoa species are of two sizes: larger cells with diameters of about 35 um and smaller cells with diameters of about 16 um. The distri- butions of cells of the two sizes vary from species to species. Table 3 details the distribution of large and small cells and sclerified cells in the four species I examined; some of the patterns are il- lustrated in Figures 25-30. The variation in these 1986] LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 47 FIGURES 25-30. Epidermal cell patterns of oT Compare with Table 8. beg adaxial epiderm —27, spectively. Sclerified cells are a with polarized light. — 3.—25. A. pr die a pae A. grandifolia, abaxial epidermis, photographed w — 29, 30. A. ds adaxial and abaxial ее respectively. Bar equals 100 u epidermal characters in the different species and the apparent constancy of the patterns within species suggest that the patterns may be useful in classification and identification. Actephila B1. Group 1 (OTU 11) A. anthelminthica Gagnep., Annam: Poilane 9216 (UC) A. nitidula Gagnep., Cochinchina: Pierre (1907) (UC). Figures 32, 33. Leaves microphylls, symmetrical, narrow el- liptic; apex acute; base acute-cuneate; organiza- tion of venation mid second rank; marginal ul- timate venation incomplete (4. anthelminthica) or fimbriate (A. nitidula); terminal tracheids nor- mal or slightly swollen; bundle sheath of higher rder veins parenchymatous; stomata paracytic. FIGURES 31-33. Leaves of Actephila.—31. A. excelsa. 33 32 and 1 mm in Figure 33. Group 2 (OTU 12) A. excelsa (Datz) Muell. Arg., Philippines: Mindanao, Ramos & Edano 49158 (UC). Figure 31. I hall к. 1 phylls, sy narrow ovate; apex acuminate; base rounded; organization of ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 YS —32, 33. A. nitidula. Bars equal 1 cm in Figures 31, venation low third rank; marginal ultimate ve- nation looped; terminal tracheids normal; bun- dle sheath of higher order veins p hy t ; stomata paracytic; trichomes scattered on abax- ial epidermis. As Table 2 and the descriptions indicate, the leaves of Actephila form a very heterogeneous 1986] group. Comparison with uncleared leaves of these and other species indicated that this variation is not an artifact of my small sample size; assessing the taxonomic significance of this variation re- quires study of cleared leaves from more species. Croizatia Steyerm. (OTU 93) C. preci Steyerm., Venezuela: Tillett et al. 82 (DAV). Figures 34, 35. Leaves mesophylls, symmetrical, elliptic; apex acute; base acute-decurrent; organization of ve- nation low third rank; marginal ultimate vena- tion looped; terminal tracheids normal; bundle sheath of higher order veins parenchymatous; stomata paracytic; trichomes scattered on abax- ial epidermis and beneath midri Although Actephila and Amanoa have tradi- tionally been placed together, pollen of the two genera appears to be quite different (Punt, 1962; Köhler, 1965), as does the wood (Metcalfe & Chalk, 1950; Hayden, 1980). Leaf morphology does little to resolve the controversey. Cladistic and phenetic analyses of leaf characters place Amanoa near the Wielandieae, particularly Wie- landia, Blotia, and Savia sect. Heterosavia, a po- sition implied by Webster's classification and suggested by pollen and wood anatomy (Levin, 1986 and in press). The position of Actephila is less clear, largely due to the heterogeneity of its leaves. Leaves of Actephila anthelminthica and A. nitidula suggest a fairly close relationship to Amanoa and the Wielandieae, whereas the leaves of A. excelsa share more with the leaves of An- drachne, a relationship first proposed by Baillon (1858) and later supported by both Punt (1962) and Kóhler (1965) on the basis of pollen. A re- lationship between Actephila and Andrachne is also more consistent with wood anatomy than is the association of Actephila and Amanoa (Met- calfe & Chalk, 1950; Hayden, 1980). Further study of leaves and other organs of Actephila will be necessary to clarify the relationships of this genus. When Steyermark (1952) described Croizatia, he weh it with Actephila on the basis of strik- ingly similar fruits. Leaf morphology suggests a pads “ыйл to Savia апа its relatives, раг- ticularly Blotia. More sampling within Actephila could conceivably turn up species more similar to Croizatia than the three I examined. Because male flowers of Croizatia remain unknown, the pollen has never been examined. We are simi- LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 49 larly ignorant of most other aspects ofthe biology of this little known genus. BRIDELIEAE MUELL. ARG. Bridelia Willd. B. atroviridis Muell. Ag. Cameroons: Bre- teler 2866 (UC). Sect. Scleroneurae Gehrm. B. cathartica Bertol. f., Tanganyika: Tanner 3541 (UC). B. exaltata F. v. M., ссы ee Kajewski 72 (UC). Figur B. mollis Hutch., Rhodesia: AT (1951) (UC). Group 2 (OTU 15) Sect. Cleistanthoideae Gehrm. B. balansae Tutch., China: Kwangtung, Chun 5982 (UC). B. glauca Bl., Sumatra: Boeea 7657 (UC). Figure 40. B. о Hook. f., Philippines: Ramos (1925) (UC). Group 3 aa 16) Sect. Micranthae Gehrm. B. glabrifolia (Muell. Arg.) Merr., Philip- pines: Ramos & Edano (1924) (UC). Figure 37. B. micrantha (Hochst.) Baill., Ivory Coast: de Wilde 452 (UC). Figure 39b. Sect. Scleroneurae Gehrm B. monoica (Lour.) Merr., Sumatra: Boeea 8440 (UC). Figures 38, 39a. B. scleroneura Muell. Arg., Tanganyika: Tanner 4160 (UC). Sect. Stipulares Gehrm. B. stipularis (L.) Bl., Sumatra: Boeea 8275 (UC) Leaves microphylls or mesophylls, symmet- rical, ovate, elliptic, or obovate; apex acuminate, acute, obtuse, or rounded; b high third rank; marginal ultimate venation fim- briate or consisting of an intramarginal vein formed by the stiffening of the brochidodromous loops (see below); terminal tracheids normal; bundle sheath of higher order veins parenchy- matous; stomata brachyparacytic, rarely one of the subsidiary cells divided by an additional wall perpendicular to the guard cells; trichomes typ- ically at least beneath low order veins, may also 50 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ute q > оуб» Y z i^e, oda ASR epp, б ARN их «У ср. ЧУ y ^ TAS хо E оао Ru За ae Ue D» Boe ese: LU » р, z x 2, ae? PRONE av M Ne Se ^ УТ M DE G} | < ES SS «m S SS т S А. y ЖОО Ss Sane "TTD «m ES fe, FIGURES 34-39. Leaves of Croizatia and Bridelia. — 34, 35. Croizatia naiguatensis. — 36. Bridelia exaltata. — 37. B. glabrifolia. — 38, 39a. B. monoica. — 39b. B. micrantha. Bars equal 1 cm in Figures 34, 36-38 and 1 mm in Figures 35, 39a, b. 1986] be scattered on abaxial or less on adaxial surfaces, rarely the leaves glab Gaucher (1902) reported o bundle sheaths in B. micrantha, B. obtusa, and B. zen- keri. Although all the species of Bridelia I examined are identified easily by their numerous second- aries and closely spaced, regularly and strongly percurrent tertiaries, three groups of species can be recognized on the basis of the patterns formed by the outer portion of the secondaries and the marginal ultimate venation. The first group, which includes some members of sects. Micran- thae and Scleroneurae, has weakly brochidod- romous secondaries and fimbriate marginal ul- timate venation (Fig. 36). The second group, all members of sect. Cleistanthoideae, also has a fimbrial vein, but the secondaries are eucamp- todromous (Fig. 40). In the third group, the sec- ondary veins are brochidodromous with the out- er portions ofthe loops forming an intramarginal vein. Close examination reveals that in some species, members of sects. Stipulares and Scle- roneurae, the loops join without any abrupt change of direction (Figs. 38, 39a), whereas in other species, members of sect. Micranthae, each secondary vein bends sharply exmedially within ] mm or less of joining the intramarginal vein (Figs. 37, 39b). As both Micranthae and Scle- roneurae also contain species with the first leaf type, this difference may reflect independent or- igin ofthe intramarginal vein in the two sections. Resolving this and the question of the relation- ship of the eucamptodromous Cleistanthoideae awaits study of more species in this genus of over 60 species. Cleistanthus Hook. f. ex Planch. Group 1 (OTU 17 C. acuminatissimus Merr., British N Bor- neo: Elmer 21153 (UC). Figures 41, 42. Group 2 (OTU 18) Sect. Stipulati Jabl. C. bridelifolius C. B. Rob., Sarawak: Chew Wee-Lek 558 (UC). C. siamensis Craib, Annam: Poilane 5186 Group 3 (OTU 19) Sect. Ferruginosi Jabl. C. barrosii Merr., Philippines: 16870 (UC). Figure 43. C. ferrugineus Muell. Arg., Ceylon: Galston C Clemens 223 : Sect. Pedicellati Jabl. LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 51 FIGURE 40. Leaf of Bridelia glauca. Bar equals 1 C. integer C. B. Rob., Philippines: Loger 12372 (UC), Ramos & Edano (1926) (UC). Group 4 (OTU 20) Sect. Chartacei Jabl. C. cunninghamii Muell. Arg., Australia: Queensland, Clemens (1944) (UC). Group 5 (OTU 21) Sect. Cleistanthus C. polystachyus Hook. f. ex Planch., Ivory Coast: Leeuwenberg 3736 (UC). Group 6 (OTU 22) C. saichikii Merr., China: Hainan, Lau 15 (UC). Group 7 (OTU 23) Sect. Australes Jabl. C. micranthus Croizat, Fiji: Smith 9531 (UC). Group 8 (OTU 24 Sect. Leiopyxis (Miq.) Jabl. C. apodus Bth., Australia: Queensland, /r- vine 670 (DAV). C. blancoi Rolfe, Philippines: Ramos & Edano 1928 (UC). 52 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 о Leaves of и Dicoelia, and Poranthera. —41, 42. Cleistanthus acuminatissimus. — . Dolabrate trichome of C. bar —44. Dicoelia affinis. —45. Filiform sclereids of D. affinis. —46. Poranthera Sire Bars equal | cm in ы 41, 44; ] mm in Figures 42, 46; and 100 um in Figures 43,4 1986] C. ripicolus Leonard, Nigeria: Onochie FHI 33213 (DAV). Leaves microphylls or mesophylls, symmet- rical, typically narrow elliptic or lanceolate; apex acute or acuminate; base acute, obtuse, or round- ed; organization of venation typically low to mid third rank, rarely (C. micranthus) second rank; paracytic or brachyparacytic; leaves glabrous or with trichomes scattered abaxially. Two features unusual in the Phyllanthoideae appear in some Cleistanthus species. Members of sects. Chartacei, Cleistanthus, and Leiopyxus have a network of filiform fibers in the mesophyll giving the veins a frayed appearance. Only Di- coelia has similar fibers. Unique in the Phyllan- thoideae are the two-armed dolabrate trichomes (Fig. 43) borne by species of sects. Chartacei and Ferruginosi. Overall, most species of Cleistanthus included i thus) that I examined. Both have strongly broch- idodromous secondaries arising at higher angles than typical for the genus and less well organized higher order venation than most of the other species. The latter is particularly true of C. mi- cranthus in which the tertiary veins are random reticulate rather than percurrent. The leaves of C. cunninghamii are not typical of Chartacei (Ja- blonsky, 1915); it is unfortunate that this is the bibtiaky (1915) felt that Australes were rather isolated in Cleistanthus, which is corroborated by my evaluation of the leaves. Typical leaves of the two genera in the Bri- delieae do not seem to resemble each other very closely, but a connection between them can be seen through some of the higher ranked species of Cleistanthus and the fimbrial-veined groups of Bridelia, in agreement with the general inter- pretation, based on floral characters, that Bri- delia is derived from Cleistanthus-like ancestry. After studying pollen exine sculpturing of four phyletically from Cleistanthus. Examination of the pollen and leaves of more species in both LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 53 genera might help clarify the relationship be- tween them DICOELIEAE HURUSAWA Dicoelia Benth. (OTU 25) D. affinis J. J. Sm., Borneo: Hallier 1255 (UC). Figures 44, 45. T h 1 а 1 ‚ narrow ellip- tic; apex acuminate; “base acute-cuneate; orga- nization of venation third rank; marginal ulti- mate venation looped; terminal tracheids normal; bundle sheath of higher order veins parenchy- matous; stomata © id trichomes scattered on abaxial epiderm The leaves of pe om share with some Cleis- anthus species having filiform fibers in the me- Mies (Fig. 45). In general, however, the resem- blance between these genera is not strong, which is consistent with the evidence from pollen mor- phology (Punt, 1962; Kóhler, 1965) and wood anatomy (Metcalfe & Chalk, 1950). Indeed, as discussed above, foliar morphology suggests a relationship between Dicoelia and Pentabrach- ium, a relationship not before considered. PORANTHEREAE (MUELL. ARG.) GRUNIG Poranthera Rudge. P. corymbosa Brongn., Australia: New South Wales, Boorman (1903) (UC). Figure 46. Leaves simple, sessile, nanophylls, symmet- mous; primary vein massive, straight; secondary vein angle of origin very narrow acute and more or less uniform, thickness moderate to fine, course zig-zag and branched; higher vein orders not well resolved, forming an irregular reticulum of grad- ually thinning veins; marginal ultimate venation looped; areoles large and extremely irregular in size and shape; veinlets irregularly branched; ter- minal tracheids normal; bundle sheath of higher order veins parenchymatous; adaxial epidermal cells rectangular, abaxial cells isodiametric, both arranged randomly, anticlinal walls undulate, abaxial cells with thickened ridges on the un- canis surfaces T stomata abaxial, ori- entation i sets thd guard cells sunken and al- most obscured by the subsidiary cells; trichomes and crystals none. Me] d ES Б ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 m Mx cutus 59 ЕРА | +. к n 47-53. Leaves of Andrachne and Spondianthus. —47. ос chinense. ae Abaxial aga of A nse, oe baie о richom phyllantholdes — A e and anisocytic stomata (arr in Figures 49, 51, EO 2 100 um in Figure . A. australis. — 50, da ia preussii. Bars equal 1 cm in Figures. 47. 52: 5 mm in Figure 50; | m mm 1986] LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 55 FIGURE 54. Abaxial epidermis of Andrachne chinense, showing anisocytic (A) and large anomocytic (B) stomata. Bar equals 100 um. Andrachne L. Group | (OTU 26) Sect. Arachne (Neck.) Endl. A. australis Zoll., Annam: Poilane 4755 C). Fi 4 . Figure 49. A. chinense Bunge, India: Punjab, Koelz 4423 (UC). Figures 47, 48, 54. A. cordifolia Muell. Arg., Pakistan: Rodin 5378 (UC). Sect. Phyllanthidia (Didrichs.) Muell. Arg. A. microphylla (Lam.) Baill., Peru: Saga- stegni & Cabanillas 8738 (DAV). Leaves microphylls, symmetrical, ovate; apex attenuate or rounded to emarginate; base acute, obtuse, or rounded; organization of venation high second rank; marginal ultimate venation looped; stomata" also present (Fig. 54); trichomes scat- tered on abaxial epidermis, denser beneath lower order veins, in some species also scattered on adaxial epidermis. Group 2 (OTU 27) Sect. Phyllanthopsis (Scheele) Muell. Arg. A. arida (Warnock & M. C. Johnst.) Web- ster, U.S.A.: Texas, Butterwick € Lott 3582 (DAV). A. phyllanthoides (Nutt.) Muell. Arg., U.S.A.: Missouri, Palmer 45338 (UC). Figures > Leaves nanophylls, symmetrical, elliptic; apex obtuse; base acute or rounded; organization of venation low second rank; marginal ultimate ve- nation looped (4. phyllanthoides) or incomplete (A. arida); terminal tracheids normal; bundle sheath of higher order veins parenchymatous; stomata anisocytic; trichomes scattered on both surfaces, densest beneath midrib and secondar- ies. My anatomical observations generally agree with those of Rothdauscher (1896), who included A. chinensis, A. cordifolia, and A. phyllanthoides among the seven species he studied. He did not, however, report the large anomocytic stomata I found in sects. Arachne and Phyllanthidia. Га" 56 ANNALS OF THE MISSOURI BOTANICAL GARDEN The members of this tribe are perennial herbs or subshrubs, many extending into drier and more temperate habitats than most other phyllan- thoids (Pax & Hoffmann, 1922; Webster, 1967). Surprisingly, the rank of the venation of the two species in Andrachne sect. Phyllanthopsis re- mains comparatively high for such xeromorphic leaves, which generally show a regression in or- ganization below the leaves of their more me- sophytic relatives (Hickey, 1971). The strikingly low rank of the leaves of Poranthera reflects the extremely xerophytic nature of this Australian endemic and reveals nothing about its relation- ships Section Phyllanthopsis has been repeatedly transferred between Savia and Andrachne (cf. Mueller, 1866; Pax & Hoffmann, 1922; Webster, 1967). The anisocytic stomata and undulate an- ticlinal walls on the epidermal cells suggest that Phyllanthopsis is better included in Andrachne, the conclusion both Punt (1962) and Kohler (1965) came to when studying the pollen. The venation of Phyllanthopsis also appears some- what more organized than would be expected from a xeric Savia derivative, being more com- parable to other Andrachne sections. SPONDIANTHEAE WEBST. Spondianthus Engl. (OTU 28) S. preussii Engl., Cameroons: Zenker 563 (CAS), Ghana: Oldeman 818 (MO), Congo: Louis 13728 (UC). Figures 52, 53. Leaves mesophylls, symmetrical, wide elliptic; apex rounded; base rounded; organization of ve- nation second rank; marginal ultimate venation fimbriate; terminal tracheids normal; bundle sheath of higher order veins sclerified; stomata paracytic. The paracytic stomata and tanniniferous epi- dermal cells would associate Spondianthus with the Antidesmeae, in which the second rank ve- nation, few secondaries enclosing large intercos- tal areas, and random-reticulate tertiaries with a tendency to ramify admedially would be anom- alously primitive. I am inclined to agree with Webster’s (1975) placement of this genus in its own tribe near the Antidesmeae, of which it may be the primitive sister group (Levin, in press). ANTIDESMEAE (ENDL.) HURUSAWA Antidesma L. Group | (OTU 29) Sect. Roxburghiana Pax & K. Hoffm. [Vor. 73 A. vogelianum Muell. Arg., Tanganyika: Tanner 2564 (UC). Sect. Venosa Pax & K. Hoffm. A. henryi Hemsl., China: Hainan, Tsang et al. 23 (UC) A. japonicum Sieb. & Zucc., Taiwan: Ta- naaka & Shimada 13570 (UC) A. venosum Tul., Tanganyika: Tanner 3424 (UC). Figure 55. Group 2 (OTU 30) Sect. Laciniata Pax & K. Hoffm. A. laciniatum Muell. Arg., Congo: Louis 11467 (MO). Sect. Venosa Pax & K. Hoffm. A. cuspidatum Muell. Arg., Sumatra: Toroes 5296 (UC). A. delicatulum Hutch., China: Chekiang, Ching 2038 (CU). Figures 57, 58. A. gracile Hemsl., China: Kwangtung, Tsang 21272 (UC). A. maclurei Merr., China: Hainan, Fung 20210 (UC). A. membranaceum Muell. Arg., Belgian Congo: Louis 6088 (UC), Ghana: Hall & Abbiw 43391 (GC). Group 3 (OTU 31) Sect. Tetrandra Pax & K. Hoffm A. angusanense Elm. Philippines: Ramos « Pasgasio (1919) (UC) A. digitaliforme Tul., 15967 (UC). A. fructifera Elm., Philippines: Ramos (1915) (UC). Figure 56. Group 4 (OTU 32) Sect. Montana Pax & K. Hoffm. A. bunius Spreng., China: кз иш Tsiang Ying 1013 (UC). A. insulare Gillespie, Fiji: Smith 656 (UC). A. platyphyllum Mann, U.S.A.: Hawaii, Meinecke (1933) (UC). Group 5 (OTU 33) Sect. Velutinosa Pax & K. Hoffm A. fordii Hemsl., Hong Kong: Chun 5089 (UC) Philippines: McClure Group 6 (OTU 34) Sect. Ghaesembilla Pax & K. Hoffm A. ghaesembilla Gaertn., China: Ha ainan, Tsang 15564 (UC), Sumatra: Toroes 4828 (UC). Leaves microphylls to mesophylls, symmet- rical, elliptic, narrow elliptic, or narrow obovate; apex typically acuminate or acute, less typically obtuse or rounded; base acute, obtuse, or round- ed; organization of venation low to mid third LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 1986] FIGURES 55-60. Leaves of Antidesma and Celianella.—55. Antidesma venosum.—56. A. fructifera. —57. Abaxial epidermis of A. delicatulum, showing unicellular trichome.—58. Adaxial epidermis of A. delicatulum , 60. Celianella montana. Bars equal 1 cm in Figures 55, 59; and 50 um in Figure 58. 59, showing tanniniferous epidermal cell (arrow). — 2 100 um in Figure 57; 60; , 1 mm in Figures 56 58 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 FIGURES 61-63. nioides rank; marginal ultimate venation looped; ter- minal tracheids normal; bundle sheath of higher order veins parenchymatous; stomata paracytic; trichomes typically present at least beneath low order veins and in many species also scattered on abaxial intercostal areas, less typically over low order veins or the leaves glabrous. My anatomical observations agree with Roth- dauscher (1896) and Gaucher (1902) except in one respect: they reported that stomata were re- stricted to the abaxial epidermis in all 11 species they examined, including A. ghaesembilla. In the two specimens of this species that I examined, I found a few (stomatal index 1-296) stomata in the adaxial epidermis. Celianella Jabl. (OTU 35) C. montana Jabl., Venezuela: Amazonas, Ma- guire & Maguire 3507 1 (US). Figures 59, 60. Leaves microphylls, symmetrical, oblanceo- late; apex acute with a small apiculum; base acute- decurrent; organization of venation low third rank; secondary veins quite numerous and closely spaced, alternating with intersecondaries; ter- tiary veins percurrent but sinuous; marginal ul- timate venation formed by basal secondaries in lower portion ofleaf, looped above; terminal tra- cheids normal; bundle sheath of higher order veins parenchymatous; stomata paracytic. Hyeronima Fr. Allem. (OTU 36) H. alchorneoides Fr. Allem., Venezuela: Ama- zonas, Krukoff 5032 (UC). Figure 61. H. guatemalensis D. Sm., Costa Rica: Stork 4213 (UC) Leaves of Hyeronima and Leptonema.—61. Peltate trichomes of Hyeronima alchor- — 62. Leptonema venosa. — 63. Multicellular trichome on margin of leaf of L. venosa. Bars equal 5 mm in Figure 62 and 100 um in Figures 61, 63. H. poasana Standl, Costa Rica: Smith 2012 (UC) Leaves microphylls to mesophylls, symmet- rical, elliptic or obovate; apex acuminate; base acute or rounded; organization of venation low | of higher order veins lightly sclerified; stomata paracytic; unicellular trichomes beneath primary and secondaries; peltate trichomes scattered on adaxial epidermis, densely covering abaxial epi- My omical observations agree with those of E (1896). Leptonema A. Juss. (OTU 37) L. venosum (Poir.) Juss., Madagascar: Decary 13624 (P). Figures 62, 63. Leaves nanophyile to | graphy is, symmetri- d cuspidate; base ounded; organization of venation third rank; 2d ultimate venation looped; terminal tra- cheids swollen; bundle sheath of higher order veins parenchymatous; stomata paracytic; uni- cellular trichomes scattered on adaxial epider- mis, especially beneath low order veins; multi- cellular, uniseriate trichomes with single-celled glandular heads scattered beneath lower order veins and dense on margins of blade. са ат E. ti ;apex 7 Thecacoris A. Juss. (ОТО 38) T. gymnogyne Pax, Cameroons: Zenker 23 (CAS 1986] T. leptobotrya (Muell. Arg.) Brennan, Camer- s: Olorunfemi 30638 (DAV). T. trichogyne Muell. Arg., Rhodesia: Brenan & Greenway 8031 (EAH). Leaves microphylls to mesophylls, symmet- rical, obovate or elliptic; apex acute to acumi- nate; base acute; organization of venation low third rank; marginal ultimate venation looped; terminal tracheids normal; bundle sheath of higher order veins parenchymatous; stomata paracytic or occasionally parallelocytic; tri- chomes dense beneath low order veins and scat- tered on adaxial epidermis. The central genera of the Antidesmeae, Anti- desma, Hyeronima, and Thecacoris, have ex- tremely similar leaves, which can be recognized architecturally by their regularly spaced second- aries enclosing more or less parallel-sided inter- costal areas and very we ries. The additional combination of paracytic stomata and tanniniferous epidermal cells con- firms the identification of members of this tribe. Airy Shaw’s (1973) resurrection of the family Stilaginaceae for Antidesma receives no support from foliar morphology; his suggestion of a re- lationship to the Icacinaceae directly contradicts the evidence, for the latter have anomocytic or anisocytic stomata (Metcalfe & Chalk, 1950), fewer, more curving secondaries, more strongly percurrent tertiaries, and generally more orthog- onal and organized higher order venation (pers. observ.). In fact, the leaves of some sections of Antidesma and Hyeronima are so similar that if it were not for the peltate trichomes of Hyero- nima, one could not reliably distinguish leaves of the two genera. The other two genera that I examined, Celi- anella and Leptonema, have leaves that depart considerably from the typical architecture of the tribe. Both, however, share the tanniniferous epi- dermal cells and paracytic stomata characteristic of the Antidesmeae, suggesting that they do, in fact, belong here. Pollen evidence supports this placement of Leptonema (Kohler, 1965). The al- ternative treatment, followed by most authors, places Leptonema among the genera included in Webster’s Phyllantheae. Only the uniseriate, multicellular trichomes with unmodified bases argue strongly for this relationship. However the spherical glandular cell at the apex contrasts markedly with the blunt or tapering, non-glan- dular terminal cell in the Phyllantheae (cf. Figs. 63, 100, 102). Both phenetic and cladistic anal- LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 59 yses of leaf characters ep Leptonema with the Antidesmeae (Levin, 1986 and in press). Celianella, Vie: dh by Jablonsky in 1965, remains poorly known, and its wood and pollen have never been described. Even more so than Leptonema, its leaves contrast with the other members of the Antidesmeae. No other phyllan- thoids exhibit similar architecture, with numer- ous, closely spaced secondaries typically alter- nating with intersecondaries, sinuous, percurrent tertiaries, and few vein orders. However, these features шау result Hom modification of - ba- sic A fthe coriaceous, sub- succulent leaves of this genus, a possibility supported by cladistic analysis (Lev- in, in press). APORUSEAE (LINDL. EX MIQ.) AIRY SHAW Aporusa Bl. (OTU 39) A. aurita (Tul.) Miq., Philippines: Salvoza (1924) (UC). Figure 67. A. chinensis (Champ.) Merr., China: Hainan, Gressitt 876 (UC). A. frutescens Blume, British N Borneo: Ramos 1364 (UC). Figures 64, 65. A. lanceolata Hance, China: Kwangtung, Chun 6391 (UC). Figure 66. A. microcalyx Hassk., Indonesia, cult. Hort. Bog. (UC 234706). A. сеч Merr., Philippines: Cenabre (1924) (U A. оды ме: Philippines: Loher 14360 (UC). Figure 68. A. villosa AL Baill., Siam: Rock 1673 (UC). Leaves microphylls to mesophylls, symmet- rical, lanceolate, ovate, or elliptic; apex acumi- nate to caudate or obtuse; base acute, obtuse, or rounded; organization of venation low to mid third rank; marginal ultimate venation looped or c stomata anisocytic; trichomes typically present at least beneath midrib and secondaries, may also be scattered on abaxial epidermis and over midrib and secondaries. My anatomical observations generally agree with those of Rothdauscher (1896) and Gaucher (1902), as Rothdauscher remarked, the tanni- niferous cells in the adaxial epidermis are re- markably large in Aporusa and the other genera in the Aporuseae. 60 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ES wks Шу, & P EO чы; Y e, e PEO vs ty, > y XJ FicunES 64-68. Leaves of Aporusa. —64, 65. A. frutescens. Arrow indicates marginal gland. – 66. Marginal gland of A. lanceolata. —67. Marginal gland of A. aurita. —68. Adaxial epidermis of A. symplocifolia, with tanniniferous epidermal cells. Bars equal 2 cm in Figure 64; 1 mm in Figure 65; 200 um in Figures 66, 67; and 100 um in Figure 68. 1986] LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 61 FIGURES 69-74. Leaves of Ashtonia, Baccaurea, and Didymocistus. — 69, 70. Ashtonia praeterita. — 71. A. whitmorei. — 72. Abaxial epidermis of Baccaurea deflexa, with solitary and tufted trichomes. — 73, 74. Didy- mocistus chrysadenius. Bars equal 2 cm in Figure 73; 1 cm in Figure 69; 1 mm in Figure 74; 200 um in Figures 70, 71; and 100 um in Figure 72. 62 ANNALS OF THE MISSOURI BOTANICAL GARDEN Ashtonia Airy Shaw (OTU 40) A. praeterita Airy Shaw, Malaya: Everett KEP 104934 (A). Figures 69, 7 A. whitmorei Airy Shaw, Malaya: Stone 8654 (A). Figure 71. Leaves microphylls, symmetrical, elliptic; apex obtuse; base obtuse; organization of venation mid third rank; marginal ultimate venation mostly incomplete; terminal tracheids more or less nor- mal (A. whitmorei) or swollen (A. praeterita); bundle sheath of higher order veins lightly (A. whitmorei) or heavily (A. praeterita) sclerified; stomata anisocytic; trichomes scattered beneath midrib. If the specimens I examined are representa- tive, the degree of enlargement of the terminal tracheids and of the sclerification of the bundle sheath allow one to distinguish easily between leaves of the two species (Figs. 70, 71). Baccaurea Mp Group 1 (OTU Sect. н (Міа.) Muell. Arg B. wilkesiana Muell. Arg., Fiji: Smith 9692 (UC). Sect. Everettiodendron (Merr.) Pax & K. Hoffm. B. bracteata Muell. Arg., Singapore: Ridley 1893 (UC). B. deflexa Muell. Arg., British N Borneo: Elmer 21118 (UC). Figure 72. B. philippinensis Merr., Philippines: Wenzel 3007 (UC) Group 2 (OTU 42) Sect. Pierardia (Roxb.) Muell. Arg. B. obtusa A. C. Smith, Fiji: Smith 1780 (UC), Webster & Hildreth 14140 (DAV). B. pendula Merr., British N Borneo: E/mer 21290 (UC). B. racemosa (Reinw.) Muell. Arg., British N Borneo: Elmer 21195 (UC). Group 3 (Dubiae) (OTU 43) B. cauliflora Lour., China: Hainan, Fung 20024 i Group 4 (OTU Sect. ee (Baill.) Muell. Arg. B. stylaris Muell. Arg., Fiji: Smith 8060 UC), Webster & Hildreth 14046 (DAV) Leaves microphylls to mesophylls, symmet- rical, ovate, elliptic, or obovate; apex acuminate, acute, or obtuse; base acute-decurrent, acute, ob- tuse, or rounded; organization of venation third rank; marginal ultimate venation looped or par- [VoL. 73 tially incomplete; terminal tracheids normal; bundle sheath of higher order veins lightly or heavily sclerified; stomata anisocytic, rarely the smallest cell divided again parallel to the guard cells; trichomes typically of two lengths, the lon- ger solitary, the shorter solitary or tufted in groups of 2-5(-8), both scattered beneath lower order veins or generally on abaxial epidermis (Fig. 72), or rarely the leaves glabrous. Rothdauscher's (1896) and Gaucher's (1902) anatomical observations agree with mine. Didymocistus Kuhlm. (OTU 45) D. chrysadenius Kuhlm., Peru: Asplund 14760 (CAS). Figures 73, 74. Leaves mesophylls, symmetrical, ovate-ellip- tic; apex acuminate; base rounded; organization of venation fourth rank; marginal ultimate ve- nation looped; terminal tracheids normal; bun- dle sheath stomata | f higher OI der ftwo types, the longer ones 5 solitary, scattered beneath midrib and dense abaxially in junction of midrib and secondaries, the shorter ones tufted in groups of 3-15, scattered on both surfaces but denser abax- ially and beneath veins; lower epidermis densely covered with sessile, disk-shaped glands. Maesobotrya Benth. (OTU 46) M. barteri (Baill.) Hutch., Liberia: Leeuwenberg M. bipindensis (Pax) Hutch., Cameroons: Zen- ker 2598 (MO M. dusenii (Pax) Hutch., Cameroons: Zenker (1906) (CAS), Olorunfemi s.n. (DAV 47305). M. floribunda Benth., Congo: Leonard 251 (UC). M. pynaertii (De Wild.) Pax & Hoffm., Congo: Leonard 334 (MO). M. staudtii (Pax) Hutch., Congo: Donie 2113 (MO), Cameroons: Zenker 568 (UC). Leaves microphylls to mesophylls, symmet- rical, elliptic or oblong; apex long acuminate; base acute or obtuse; organization of venation mid third rank; marginal ultimate venation in- complete or partially looped; terminal tracheids normal or somewhat swollen (M. dusenii and M. floribunda); bundle sheath of higher order veins parenchymatous; stomata anisocytic; trichomes scattered or dense beneath midrib and second- aries and on margin, may also be scattered on abaxial epidermis. Among the Aporuseae, the closest to being toothed, with species such as M. AL L Ф +h 1986] floribunda and M. staudtii having denticulate margins. Protomegabaria Hutch. (OTU 47) P. stapfiana Hutch., Ghana: Enti FHI 8167 (G). Leaves mesophylls, symmetrical, elliptic; apex acute; base acute; organization of venation mid third rank; marginal ultimate venation looped to partially incomplete; terminal tracheids normal; bundle sheath of higher order veins parenchy- matous; stomata anisocytic; trichomes dense be- neath midrib and secondaries. Richeria Vahl. (OTU 48) R. australis Muell. Arg., Santa Catarina: Reitz & in 4044 el R. grandis Vahl., Venezuela: Wurdack & Mona- chino 41 O). R. obovata (Muell. Arg.) Pax & Hoffm., Matto Grosso: Irwin € Soderstrom 6445 (NY). R. racemosa (Poepp. & Endl.) Pax & Hoffm., Venezuela: Maquire & Politi 28440 (DAV). Leaves mesophylls, symmetrical, obovate; apex acute or obtuse; base acute or obtuse; organiza- tion of venation mid third rank; marginal ulti- mate venation looped to partially incomplete; terminal tracheids normal or swollen; bundle sheath of higher order veins parenchymatous; stomata basically anisocytic, but with a tendency toward being helicocytic; trichomes scattered or dense beneath midrib and proximal portions of secondaries, or the leaves glabrous. y observations generally agree with those of Rothdauscher (1896), Gaucher (1902), and Hay- den (1980), except that Hayden considered the stomata to be anomocytic. Indeed, as Roth- dauscher noted, the number of subsidiary cells surrounding each pair of guard cells varies be- tween three and four; this variation, with the concomitant variation in cell size and shape, may suggest the anomocytic pattern. I, however, in- terpret the stomatal complex as developing in a spiral pattern, described by Payne (1970) as giv- ing rise to either anisocytic or helicocytic sto- mata. The Aporuseae were first recognized as a dis- tinct group by Köhler (1965), who considered the group closely related to the Antidesmeae, in which the genera had been scattered by earlier authors. Foliar morphology certainly supports this treatment. The architecture of the two tribes is quite similar, but the Aporuseae can be sep- LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 63 arated on the basis of having much larger tan- niniferous cells in the epidermis, anisocytic rath- er than paracytic stomata, and marginal glands. Also the level of venational organization tends o be higher in the Aporuseae. Cladistic analysis of the leaf characters suggests that the Aporuseae are specialized derivatives of the Antidesmeae (Levin, in press). The marginal glands, located either on the ends of small teeth or in the sinuses of small crenu- lations, may represent remnants of the theoid teeth found in Bischofia, Drypetes, and Putran- jiva. However, as the Aporuseae appear not to be particularly closely related to these genera but to be derived from entire-leaved ancestors, the alternative hypothesis that the glands have evolved de novo may be more parsimonious (Levin, in press). Two genera placed in the Aporuseae by Web- ster (1975) do not conform completely with the typical foliar morphology of the tribe. Of these, Protomegabaria departs only in lacking marginal glands. Kóhler (1965) commented that pollen characters placed P. stapfiana in the Aporuseae, b mediate between these tribes. Alternatively, the glands may be secondarily absent; some other genera have glandless species (e.g., Richeria obo- vata). It would be worthwhile to examine leaves of P. macrophylla to determine the stomatal type and size of the tanniniferous epidermal cells. ular, the leaves of Didymocistus have more high- ly organized (fourth rank), have brachy- paracytic stomata, and lack tanniniferous cells and marginal glands. The high rank and sessile, flat-topped glands suggest a possible relationship with Hymenocardia. Certainly the relationships of this little known, monotypic genus deserve far more consideration than Kuhlmann (1940) gave in his original description. DRYPETEAE (GRISEB.) HURUSAWA Drypetes Vahl. (Figs. 75-82 See Table 4 for species examined. Leaves microphylls to mesophylls, symmet- rical, ovate, elliptic, oblong, or obovate; apex attenuate, acuminate, acute, or obtuse; base broadly acute, obtuse, or rounded, almost in- variably oblique; organization of venation mid second to mid third rank; marginal ultimate ve- 64 ANNALS OF THE MISSOURI BOTANICAL GARDEN TABLE 4. Species of Drypetes examined. [Vor. 73 Species Collector Location Herbarium Group 1 (OTU 49) Sect. Drypetes oit. Ekman 12160 Hispanolia DAV diversifolia Krug & Urban Hackett 188 Bahamas UC Sect. Sphragidia (Thwait.) Pax & Hoffm. aframensis Hutch. Mensah FH 6732 Ghana GC floribunda (Muell. ied Hutch. Hall & Enti 40228 Ghana GC natalensis (Harv.) H Я Wood (1891) Natal MO rhakodiscos (Hassk.) ps Shaw Elmer 20768 British N Borneo UC principum (Muell. Arg.) Hutch. Hall & Abbiw 44753 Ghan GC Sect. unknown ivorensis Hutch. & Dalz. Hall & те 43178 Сһапа ОС lisolinoli Leonard Leonar Congo UC littoralis (C. B. Rob.) Merr. Keinholz (1 2 Philippines UC Group 2 (OTU 50) Sect. Drypetes ilicifolia Kr. & Urb. Proctor 31554 Jamaica DAV pellegrini Leandri Hall & Abbiw 44118 GC perreticulata Gagnep. Lau 304 China, Hainan UC standleyi Webster Foster & Croat 2308 Panama DAV Sect. Sphragidia (Thwait.) Pax & Hoffm. assamicus Pax & Hoffm. Soejarto 102 Indonesia, cult. DAV Hort. Bo; aylmeri Hutch. & Dalz. Hall & Abbiw 44147 Ghana GC battiscombei Hutch. Tanner 1326 Tanganyika UC Sect. unknown gosseweileri S. Moore Louis 11651 UC iwahigensis Merr. Elmer 17290 Philippines UC Group 3 (OTU 51) Sect. Sphragidia (Thwait.) Pax & Hoffm. laevis (Miq.) Pax & Hoffm. Soejarto 95 Indonesia, cult. DAV maquilingensis Merr. ramiflora Pax & Hoffm. Group 4 (OTU 5352) Sect. eint Pax & Hoffm. gerrardii Hut lateriflora cun Kr. & Urb. Sect. Stemonodiscus Pierre in sched. gilgiana Pax & Hoffm. leonensis Pax parvifolia (Muell. m Pax & Hoffm. staudtii (Pax) Hut Sect. Stipulares Pax pa Hoffm. singroboensis Ake Assi unknown Ramos & Edano (1928) Brass 17837 unknown Schlieben 3953 Louis 3479 Hall 43837 K. Da— (ill.) 215 Hall & Abbiw 43781 Hall & Abbiw 44732 Hall 1192 Zenker 3398 Hall & Swaine 43266 Philippines Philippines Nyasaland Mexico, Sonora Tanganyika Gold Coast Ghana Ghana Ghana Cameroons Ghana UC 272672 UC UC UC 44009 1986] TABLE 4. Continued. LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 65 Species Collector Location Herbarium Group 5 (OTU 53) Sect. Oligandrae Pax & Hoffm. aubrevillei Leandri Hall & Swaine 43239 Ghana GC Sect. Stipulares Pax & Hoffm. bipindensis (Pax) Hutch. Zenker 3788 Cameroons MO Group 6 (OTU 54) Sect. Drypetes australasica Muell. Arg. Clemens (1944) Australia, UC Queensland deplanchei Brongn. & Gris McKee 1996 New Caledonia UC glauca Vahl Webster et al. 8790 Puerto Rico DAV vitiensis Croizat Parks 20691 Fiji UC nation typically looped or fimbriate, less typi- cally incomplete; terminal tracheids normal; bundle sheath of higher order veins parenchy- matous or rarely sclerified; stomata brachypara- cytic, the subsidiary cells typically almost com- T. covered by the pcm cells. Observations generally agree with those of снег 6), Gaucher (1902) Smith and Ayensu (1964), and Hayden (1980). Smith and Ayensu, however, reported anomocytic stomata from some species. Having examined many of the same species myself, I suspect that Smith and Ayensu missed the subsidiary cells, which in many species are completely sunken beneath the guard cells and therefore difficult to see in clear- ings. The teeth of Drypetes are traversed by a single vein that terminates below a clear, deciduous seta (Figs. 80-82), and thus are of the theoid tooth type described by Hickey and Wolfe (1975). The teeth are concave to straight along the upper side, convex along the lower side, with an obtuse apex and rounded sinus. The axis of the tooth forms an angle of about 40° to the tangent of the margin. The teeth are typically regularly distrib- uted along the complete margin, or are absent on the apex if this is attenuate or long acuminate. In many species the teeth are little more than н н crenations (Fig. 77), whereas in a few (e.g., D. natalense, juvenile foliage of D. diver- dono is teeth are spinose the description and Table 1 indicate, leaf architecture in Drypetes is quite variable. Inter- estingly, this variation does not seem to follow Pax and Hoffmann’s (1922) classification of the genus (see Table 4), which is based primarily on characters of the gynoecium and disk. Careful study of more of the approximately 160 species will be required in order to evaluate the taxo- nomic significance of the architectural variation. Neowawraea Rock (OTU 90) N. phyllanthoides Rock, U.S.A.: Hawaii, ster 13889 (DAV). Figures 83-85. Web- Leaves mesophylls, symmetrical, ovate; apex broadly acute; base rounded to subcordate; or- ganization of venation fourth rank; marginal ul- timate venation looped; terminal tracheids nor- mal; bundle sheath of higher order veins stomata were more or less anomocytic and that crystals were absent in the specimen of N. pAyl- lanthoides that they examined. Putranjiva Wall. (OTU 55) P. roxburghii Wall., Thailand: collector illeg. (UC 237213). Figures 86-89. | ж» |, РЕА. 1 ‚ narrow ellip- tic; apex acute: base acute; organization of ve- nation high second rank; marginal ultimate ve- nation incomplete; terminal tracheids normal; bundle sheath of higher order veins parenchy- matous; stomata brachyparacytic. The teeth of Putranjiva, like those of Drypetes, are of the theoid type. The upper side is concave to straight, the basal side convex, the apex acute, and the sinus rounded. The axis of the tooth forms an angle of about 20—25? to the tangent. The teeth are regularly distributed along the complete margin. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ГА ay NN D 14.229 = x а Qut Vot e > n а a s | Т8 ae 6 2 RL й FAS LE ? - TR g. Ss 259, aS 5 М SA Si 8 з E^ i se D o See DA 272 = r 4 SM $ Paz g di $ T» rx ` est А к РЕ YY O FIGURES 75-82. Leaves of Drypetes.—75, 76. D. australasica.—77. D. battiscombei. — 78—80. D. chevalieri. The glandular seta abscises at the base of the darkly stained region. —81, 82. Teeth of D. arguta, with and without the deciduous seta. Bars equal 1 cm in Figures 75, 77, 78; 1 mm in Figures 76, 79; and 200 um in Figures 80-82. 1986] LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 67 FiGURES 83-89. Leaves of Neowawraea and Putranjiva. —83-85. Neowawraea phyllanthoides. Note the pa- pillae and brachyparacytic stomata in the abaxial epidermis. —86-89. Putranjiva roxburghii. Figure 87 shows i 200 abaxial epidermis. Bars equal 1 cm in Figures 83, 86; 1 mm in Figures 85-89; in Figures 84, 87. um in Figure 88; and 50 um 68 ANNALS OF THE MISSOURI BOTANICAL GARDEN My anatomical observations agree with Gaucher (1902). Although Pax and Hoffmann (1922) placed Putranjiva in the vicinity of Glochidion and Breynia on the basis of the lack of a disk and ovary rudiment in the flowers, foliar characters support Bentham’s (1880) treatment, followed by Webster (1975) and most other recent au- thors, in which Putranjiva is viewed as a close relative of Drypetes. Evidence from wood anat- omy (Metcalfe & Chalk, 1950) and pollen mor- phology (Punt, 1962; Kóhler, 1965) also points to this relationship. Some taxonomists, e.g., Hu- rusawa (1954), have even submerged Putranjiva in Drypetes. The numerous secondaries, frequent intersecondaries, and closely spaced, acute teeth that arise at a very low angle do make the leaves of Putranjiva distinctive, but until more is known about the architectural variation within Dry- petes, the taxonomic significance of these char- acteristics cannot be determined. Neowawraea, however, appears quite distant from Drypetes. Rock (1913), when he originally described the genus, likened it to Phyllanthus. Indeed, the wide angle of origin of the second- aries, the strongly and regularly percurrent ter- tiaries that form a more or less uniform oblique angle to the midrib, and the brachyparacytic sto- mata (Fig. 84) suggest a relationship to the more primitive genera in the Flueggeinae, such as Mar- garitaria and Flueggea, a relationship proposed by Stone (1967) on the basis of floral morphol- ogy. The wood, which is of the Glochidion-type (Hayden & Brandt, 1984), and pollen (Selling, 1947) would be more at home in this subtribe than in the Drypeteae. The putative relationship beween Neowawraea and Drypetes is particularly suspect in that D. forbesii Sherff, which Sherff (1939) used as the basis for reducing Neowa- wraea to synonymy with Drypetes, was later de- termined to be a previously described species of Xylosma in the Flacourteaceae (Sherff, 1942). PHYLLANTHEAE DUMORT./SECURINEGINAE MUELL. ARG Jablonskia Webst. (OTU 56) J. congesta (Muell. Arg.) Webst., Venezuela: Amazonas, Ducke 1631 (UC). Figures 90, Leaves mesophylls, symmetrical, ovate to lan- ceolate; apex acute; base acute-cuneate; organi- zation of venation second rank; marginal ulti- [VoL. 73 mate venation looped; terminal tracheids normal; bundle sheath of higher order veins parenchy- matous; stomata paracytic or parallelocytic (with three or more lateral subsidiary cells parallel to the guard cells; Payne, 1970). My anatomical observations agree with Roth- dauscher’s (1896) description of this species un- der its earlier name, Securinega congesta. Keayodendron Leandri (OTU 57) K. bridelioides (Mildbr.) Leandri, Gold Coast: Vigne 3122 (A). Figure 95. Leaves mesophylls, symmetrical, wide-ellip- tic; apex rounded; base obtuse-decurrent; orga- nization of venation mid to high second rank; marginal ultimate venation fimbriate; terminal tracheids normal; bundle sheath of higher order veins parenchymatous; stomata paracytic; tri- chomes scattered beneath midrib and second- aries and on margins. Lingelsheimia Pax (OTU 94) L. frutescens Pax, Congo: Leonard 1759 (UC). I hyll trical, elliptic: apex ^ d » к acuminate; base acute; organization of venation low third rank; marginal ultimate venation looped; terminal tracheids normal; bundle sheath of higher order veins parenchymatous; stomata paracytic, rarely the subsidiary cells extending beyond the guard cells at one end (Fig. 96). Meineckia Baill. (OTU 58) M. capillipes (Blake) Webst., Guatemala: Pittier M. neogranatensis (Muell. Arg.) Webst., Brazil: Glaziou 13193 (DAV). M. parvifolia (Wight) Webst., Ceylon: Wheeler 12227 (DAV). Leaves nanophylls to microphylls, symmetri- cal, elliptic or ovate; apex acute or obtuse; base acute, obtuse, or rounded; organization of ve- nation high second to low third rank, the veins very fine; marginal ultimate venation looped; terminal tracheids normal; bundle sheath of higher order veins parenchymatous; stomata paracytic. Raju and Rao (1977) reported that the stomata of M. parvifolia are predominantly anisocytic. The specimens of all three species I examined possess paracytic stomata almost exclusively. 69 LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 1986] FRE p... EUER OND их О SSA pe as S bie КЕЕ UES de a + > 69 са? €, d Fá adi w e >. ^N 2076753 № FIGURES 90-95. Leaves of Securineginae. — 90, 91. Jablonskia congesta. —92. Pseudolachnostylis maprou- nifolia. —93. P. glauca. —94. Zimmermannia acuminata. —95. Keayodendron bridelioides. Bars equal 1 cm in Figures 90, 92, 94, 95 and 1 mm in Figures 91, 93. 70 ANNALS OF THE MISSOURI BOTANICAL GARDEN Мрт FiGURE 96. Developmental sequence typical of Flueggeinae, giving rise to paracytic (A), intermediate (B), and brachyparacytic (C) stomata. All three types appear on most leaves Pseudolachnostylis Pax (OTU 59) P. dekindtii Pax, Rhodesia: Coxe 187 (MO). P. glauca (Hiern) Hutch., Rhodesia: Rodin 4409 (UC). Figure 93. P. maprouneifolia Pax, Rhodesia: Farrell 213 (MO). Figure 92. P. polygyna Pax & Hoffm., Rhodesia: Stolz 1754 (CAS) Leaves microphylls, symmetrical, elliptic or Ovate; apex acute or obtuse; base obtuse to rounded or subcordate; organization of venation low fourth rank; marginal ultimate venation looped; terminal tracheids normal; bundle sheath of higher order veins sclerified; stomata para- cytic; trichomes typically absent, but in P. de- kindtii dense beneath midrib and secondaries. My observations agree with those of Gaucher (1902) and "Hayden (1980) Zimmermannia Pax (OTU 60) Z. acuminata Verdc., Tanganyika: Vaughn 2665 (EAH). Figure 94. Z. capillipes Pax, Tanganyika: Greenway 5862 Leaves microphylls to mesophylls, symmet- rical, narrow elliptic or lanceolate; apex acute or acuminate; base obtuse to rounded; organization of venation low third rank; marginal ultimate venation looped; terminal tracheids normal; bundle sheath of higher order veins parenchy- matous; stomata brachyparacytic. Webster's (1975) Securineginae consist of a collection of genera not found together in any other system. Pollen evidence (Punt, 1962; Kóh- ler, 1965) suggests two groups of genera, one con- [Vor. 73 sisting of Danguyodrypetes, Lingelsheimia, Pleiostemon, and Zimmermannia, and the other consisting of Securinega (except S. congesta; see below) and Chascotheca. Both Punt and Kóhler related Pseudolachnostylis to Amanoa. The po- sition of Meineckia is unclear and Keayodendron pollen has not been described. Leaf morphology does little to clarify the re- lationships in this group, largely because the gen- era with aberrant pollen frequently have aberrant leaf morphology. Lingelsheimia and Zimmer- imilar leaves. Pseudolach- specialized than Атапоа; there is little to suggest a close relationship between the two genera. Un- fortunately, the leaves of Pseudolachnostylis are also more specialized than the other Securine- ginae, although there are trends toward increas- ing rank in this subtribe, making Pseudolach- nostylis more at home here than elsewhere. Meineckia, in contrast, exhibits lower rank, and appears most similar to Andrachne. This could be convergence, as both genera have herbaceous tendencies and small, thin leaves in which re- version to lower rank would be expected. Web- ster (1965) suggested that Zimmermannia is the closest relative of Meineckia; the leaves of the latter could easily be derived from the former by reduction in rank associated with the change in abit. In his review of the pollen of the Euphorbi- aceae, Punt (1962) pointed out that the pollen of Securinega congesta differed radically from the pollen of other species of Securinega he studied. Webster (1984b) described the new genus Ja- blonskia for this species. The leaves suggest a relationship to the Wielandieae, which also have u amined leaves of the species that Webster main- tained in Securinega. FLUEGGEINAE MUELL. ARG. Flueggea Willd. (OTU 61) Е. flexuosa Muell. Arg., Philippines: Luzon, El- mer 15662 (UC). F. suffruticosa (Pallas) Baillon, Hondo: Ohwi (1951) (UC). Figures 97, 98. F. virosa Roxb. ex Willd., Ivory Coast: Leeu- wenberg 3317 (UC), Tanganyika: Tanner 4266 (UC) Leaves microphylls to mesophylls, symmet- rical, elliptic; apex acute; base obtuse; organi- 1986] zation of venation mid third rank; marginal ul- timate venation incomplete or looped; terminal tracheids normal; bundle sheath of higher order veins parenchymatous; stomata brachyparacytic or with the subsidiary cells extending beyond the guard cells at one end only (Fig. 96). Gaucher (1902) air sclerified bundle sheaths in F. suffrutico I am following MES (1984a), who recog- nized Flueggea as a genus distinct from Securi- Breynia Forst. (OTU 62) B. acuminata Muell. Arg., Philippines: Luzon, Clemens 16742 (UC). B. angustifolia Hook. Е, Sumatra: Boeea 8969 (UC). B. cernua Muell. Arg., Culion: Herre 1024 (UC). B. discigera ed Arg., Sumatra: Yates 1171 (UC). Figure 100. B. disticha с Tuamotu Island: Wilder 1148 (UC), Fiji: Smith 9623 (UC). Figure 99. B. fruticosa (L.) Hook. f., Hong Kong: Taam 1771 (UC). B. officinalis Hemsl., Hong Kong: Chun 5072 (UC) Leaves microphylls, symmetrical, ovate, elip tic, or oblong; apex acute, obtuse, or rounde base acute, obtuse, or rounded, in many donee decurrent; organization of venation high third rank; marginal ultimate venation incomplete or partially looped; terminal tracheids normal or swollen (B. acuminata, B. angustifolia, B. cer- nua); bundle sheath of higher order veins par- enchymatous; stomata paracytic to brachypara- cytic, in many the subsidiary cells extending beyond the guard cells at one end only (Fig. 96); gera with trichomes covering abaxial e mi (Fig. 100); а. Ms не у papillate, in some the papillae quite elongate y observations of the species examined by а (1902), Raju and Као (1977), апа Hay- den (1980) agree with their reports. Glochidion Forst. See Table 5 for species examined. (Figs. 101, 102 Leaves microphylls or small mesophylls, sym- metrical or asymmetrical (sect. Hemiglochidion obtuse, rounded, or cordate, oblique in species LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 71 with asymmetrical blades; organization of ve- nation high third rank; m nation looped; terminal tracheids normal; bun- dle sheath of higher order veins parenchymatous or sclerified; stomata brachyparacytic, in some the subsidiary cells extending beyond the guard cells at one end (Fig. 96); trichomes sparse to dense on abaxial epidermis, particularly dense beneath veins, in some species di on adaxial epidermis, or the leaves glabrou y anatomical observations do not conflict with those of Gaucher (1902) Three groups of species can be recognized on the basis of leaf architecture. Section Glochidion (OTU 63) has symmetrical leaves wach tertiaries that I rib. Section Hemiglochidion can be divided into two groups, the first (ОТО 64) with symmetrical leaves and tertiaries that vary from oblique to perpendicular to the midrib with the angle de- creasing upwards in the leaf, and the second (OTU 65) with asymmetrical leaves with oblique bases and tertiaries that are predominantly perpendic- ular to the midrib. the mid- Margaritaria L. f. (OTU 67) M. nobilis L., d à Luteyn & Foster 949 (G). Figures 103, M. obovata (Baill. е Tanganyika: Tanner M. scandens (Wr. ex Griseb.) Webst., Bahamas: Andros, Brace 4867 (NY). M. tetracocca (Baill.) Webst., Cuba: Leon & Cle- mente 23339 (NY). Leaves microphylls, symmetrical, obovate or u stomata brachyparacytic; trichomes none or scattered beneath veins. Phyllanthus L. See Table 6 for species examined. Leaves nanophylls, microphylls, or meso- phylls, symmetrical or some asymmetrical, ovate, elliptic, or obovate; apex acuminate, acute, venation incomplete or looped; a cheids normal or swollen; o sheath of higher 72 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 97-104. Leaves of Flueggeinae. —97, 98. Flueggea suffruticosa. —99. Breynia disticha. — 100. Mul- ticellular trichomes of B. discigera. — 101. Glochidion philippicum. — 102. Multicellular trichomes of G. eriocar- pum. —103, 104. Margaritaria nobilis. Bars equal 1 cm in Figures 97, 99, 101, 103; 1 mm in Figures 98, 104; and 100 um in Figures 100, 102. 1986] LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 73 TABLE 5. Species of Glochidion examined. Species Collector Location Herbarium Group 1 (OTU 63) Sect. Glochidion atalotrichum A. C. Smith Smith 8747 Fiji UC cordatum (Muell. Arg.) Seem. Smith 9403 Fiji UC dasyphyllum K. Koch Merrill 10810 China, Kwangtung UC dasyphyllum K. Koch Chun 6405 China, Kwangtung UC eriocarpum Champ. Lei 899 China, Hainan UC hirsutum Wight Boeea 6700 Sumatra UC hongkongense Muell. Arg. Chun 4783 China, Lantoa UC Island lancifolium C. B. Rob. mos & Edano (1924) Philippines UC littorale B a. (1915) British N Borneo UC Group 2 (OTU 64) Sect. Hemiglochidion Muell. Arg. daltoni Kurcz Maire 3865 China, Yunnan UC ferdinandi Muell. Arg. White (1926) Australia, UC Queensland fortunei Hance Merrill 11369 China, Kaingsu UC merrillii C. B. R Clemens 18750 Philippines UC obovatum Sieb. " 2% Hiroe 1403 Japan UC rum Bl. Skukor AS 95 Malaya UC seemannii Muell. Arg. Smith 8758 Fiji UC Group 3 (OTU 65) album (Blco.) Boerl. Ramos & Edano (1927) Philippines UC benguetense Elm Ramos & Edano (1926) Philippines UC borneense Boerl. Yates 2481 Sumatra UC bracteatum Gill. Smith 7366 Fiji UC fagifolium Miq Liang 62257 China, Hainan UC hypoleucum (Miq.) Boerl Cenabre (1924) Philippines UC lanceolatum Ha huang & Kao 5541 iwan UC leiostylum Kurz Toroes 383 mat UC philippicum "v ) C. B. Rob. Ramos (1925) Philip UC puberum (D.) Hutch. Gressitt 1407 China, Kwangtung UC wrightii Benth. Gressitt 1111 China, Hainan UC order veins parenchymatous or sclerified; sto- mata brachyparacytic or the subsidiary cells ex- tending beyond the guard cells at one end (Fig. 96); trichomes present beneath veins or covering the abaxial epidermis, or none; abaxial epidermis typically papillate, the papillae quite long in many species. The leaves of most Phyllanthus species are borne on determinate plagiotropic branches that leaves vary considerably, the diversity within this one huge genus equalling or exceeding the entire subfamily. Many more of the 600-700 species must be examined before any evaluation of the overall taxonomic value ofthe foliar morphology can be made. Early steps in this direction were made by Webster (1956, 1957, 1958) in his monograph of the West Indian species, but far more remains to be done. Reverchonia A. Gray R. arenaria A. Gray, U.S.A.: Texas, Rowell 11531 (DAV). Figure 105. Leaves nanophylls, symmetrical, narrow-el- liptic to oblanceolate; apex obtuse; base acute; margin entire; organization of venation high first rank; venation pinnate, brochidodromous, a the vein orders beyond the primary very fine; pri- mary vein massive, straight; ау veins 74 ANNALS OF THE MISSOURI BOTANICAL GARDEN TABLE 6. Species of Phyllanthus examined. (VoL. 73 Species Collector Location Herbarium Sect. Adenoglochidion Muell. Arg. (OTU 68) aeneus Baill. Baumann 6295 New Caledonia DAV aeneus Baill. Baumann 6286 New Caledonia DAV Sect. Anisonema (A. Juss.) Griseb. (OTU 69) microcarpus (Benth.) Muell. Arg. Bartlett 14708 Philippines UC reticulatus Poir. Tanner 3720 Tanganyika UC Sect. Aporosella (Chodat) Webster (OTU 70) chacoensis Morong. Eyerdam & Beetle Argentina UC Sect. Cicca (L.) Muell. Arg. (OTU 71) acidus (L.) Skeels Mexia 1005 Nayarit UC acidus (L.) Skeels Berngardt (1923) Philippines (cult.) UC distichus H. & A. Webster 1631 Hawaii UC Sect. кн чен Muell. Arg. (OTU 72) induratus S. Moore Baumann 12504 New Caledonia DAV Панта Baill. Webster 14780 New Caledonia DAV Sect. Emblica (Gartn.) Baill. (OTU 73) emblica L Lei 446 China, Hainan UC Sect. Floribundi Pax & Hoffm. (OTU 74) floribundus Muell. Arg. Tanner 3022 Tanganyika UC Subg. Gomphidium Baill. Group 1 (OTU 75) comptonii (S. Moore) Guill. о 9520 New Caledonia US comptonii (S. Moore) Guill. 4829 New Caledonia DAV ripicolus Guill. и 7232 New Caledonia A Group 2 (OTU 76) balanseanus Guill. Webster 14735 New Caledonia DAV bourgeosii Baill. Guillaumin & Baumann New Caledonia DAV buxoides Guill. McMillan 5103 New Caledonia UC cornutus Baill. Baumann 13486 New Caledonia DAV pancherianus Baill. Guillaumin & Baumann Мем Caledonia DAV 6577 poumensis Guill. Webster 14642 New Caledonia DAV poumensis Guill. McKee 4620 New Caledonia UC Sect. Hemicicca (Baill.) Muell. Arg. (OTU 77) flexuosus (Sieb. & Zucc.) Muell. Arg. Kasapligil 3640 China, Honshu UC Sect. Heteroglochidion Muell. Arg. (OTU 78) baladensis Baill. Webster 14875 New Caledonia DAV francti Guill. Guillaumin 8408 New Caledonia DAV peltatus Guill. Webster 14645 New Caledonia DAV ronyensis Guill. Webster 14414 New Caledonia DAV salacioides S. Moore Webster 14871 New Caledonia DAV serpentinus S. Moore McKee 4568 New Caledonia UC Sect. Nothoclema Webster (OTU 79) acuminatus Vahl Dwyer & Duke 7916 Panama UC acuminatus Vahl. Hinton et al. 15965 Mexico UC 1986] TABLE 6. Continued. LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 75 Species Collector Location Herbarium Sect. Nymphanthus (Lour.) Muell. Arg. (OTU 80) ruber (Lour.) Spreng. Gressitt s.n. China, Hainan UC 1352893 Sect. Paragomphidium Muell. Arg. (OTU 81) vieillardii Baill. Webster 14711 New Caledonia DAV Sect. Phyllanthodendron (Hemsl.) Beille (OTU 82) mirabilis Muell. Arg Wolfe (1964) England, cult. — Kew Sect. Polyandroglochidion S. Moore (OTU 83) ngoyensis Schltr. McMillan 5184 New Caledonia UC Sect. Psychoglochidion Muell. Arg. (OTU 84) koghiensis Guill. Webster 14589 New Caledonia DAV Sect. Scepasma (Bl.) Muell. Arg. (OTU 85) buxifolius (Bl.) Muell. Arg. Ramos (1923) Philippines UC arising acutely at a more or less uniform angle of about 50°, course curved abruptly, loops join- ing at an obtuse angle; tertiary veins originating at random angles, transverse ramified; highest vein order tertiary; marginal ultimate venation incomplete; areoles extremely irregular in size and shape; veinlets branched once or twice; ter- minal tracheids swollen; bundle sheath of all veins parenchymatous; epidermal cells isodiam arranged randomly, with straight anticlinal E stomata equally abundant on both epidermises (stomatal frequency about 12%), with anomo- cytic stomata; trichomes and crystals absent. My observations agree with Hayden (1980). Richeriella Pax & Hoffm. (OTU 86) R. gracilis (Merr.) Pax & Hoffm., China: Hainan, Liang 65438 (UC). Leaves microphylls, symmetrical, elliptic; apex acuminate; base obtuse; organization of venation of higher order veins parenchymatous; stomata brachyparacytic. Sauropus Bl. (OTU 87) S. androgynus (L.) Merr., Malaya: Yapp 198 UC (UC). S. grandifolius Pax & Hoffm., Cochinchina: Pierre s.n. (UC). S. hirsutus Beille, Cochinchina: Pierre 564 (UC). S. rhamnoides Bl., British N Borneo: Elmer 20803 (UC) S. spectabilis Miq., Indonesia: cult. Hort. Bog. (UC 301610) Leaves microphylls to mesophylls, symmet- rical, lanceolate, ovate, or elliptic; apex acumi- nate, acute, or obtuse; base obtuse; organization of venation mid to high third rank; marginal ultimate venation incomplete; terminal tra- cheids normal; bundle sheath of higher order yond the guard cells at one end only (Fig. 96). My anatomical observations on these five species agree with. Rothdauscher (1896), who studied four others. Synostemon F. Muell. (OTU 66) S. bacciforme (L.) Webst., Ceylon: Wheeler 12034 (DAV). Figures 106, 107. S. thesioides Hj. Eichl., Australia: South Austra- lia, Lothian 3234 (UC). Leaves nanophylls, symmetrical, obovate; apex cr Б mplete; terminal tracheids normal (S. thesioides) or swollen (S. bacciforme); bundle sheath of higher order veins parenchy- matous; stomata brachyparacytic or the subsid- iary cells extending beyond the guard cells at one end only (Fig. 96); epidermis of S. thesioides con- taining groups of sclerified, papillate cells. My observations of the epidermis of S. bac- ciforme agree with those of Raju and Rao (1977). 76 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 105-1 107. Synostemon bacciforme. — 108. 10. Leaves of Reverchonia, Synostemon, and Uapaca. —105. Reverchonia arenaria. — 106, Uapaca guineensis. — 109. U. heudelotii. — 110. Adaxial epidermis of U. , with tanniniferous epidermal cells adi Bars equal 1 cm in Figure 108; 5 cm in Figure 107; 1 mm in Figures 105, 106, 109; and 100 um in Figure 1 In contrast to the Securineginae, the Flueg- geinae appear to form a closely knit group. Flueg- gea, Margaritaria, and Richeriella, the more primitive genera in the subtribe, have very sim- ilar pes architecture and epidermal morphology. ore primitive sections of Phyllanthus e I und e.g., Anisonema, Cicca, and Flori- bundi, share many features with these three gen- era. Apparently the shift within Phyllanthus to phyllanthoid branching (Webster, 1956, 1967), in which determinate, plagiotropic lateral branchlets mimic pinnate leaves, has led to an adaptive radiation into a variety of habitats with an attendant radiation in foliar morphology. The leaf architectural diversity within Phyllanthus and its closely related segregates, Breynia, Glochid- ion, Sauropus, Synostemon, and (probably) Rev- 1986] erchonia (Webster, 1967; Webster & Miller, 1963) equals that of the rest of the subfamily, reflecting this radiation. Still, certain patterns can be de- tected. All these genera, with the exception of the herbaceous Reverchonia, share a similar sto- matal development pattern, in which the divi- sions giving rise to the subsidiary cells and the guard mother cell exhibit a certain flexibility, giving rise to paracytic, brachyparacytic, and in- termediate stomata on the same leaf (Fig. Rao (1977) demonstrated, the herbaceous taxa within this plexus, which are not closely related to each other, generally bear aniso- or anomo- cytic stomata, suggesting that the latter pattern in Reverchonia results from convergence. Also, the similarity of Breynia, Glochidion, and Sau- ropus suggests that they may belong to a single line derived from the more central Phyllanthus subgenera. The taxonomist interested in study- ing this group in detail should find leaf mor- phology a rich source of information. HYMENOCARDIEAE (MUELL. ARG.) HUTCH. Hymenocardia Wall. ex Lindl. (OTU 89) H. acida Tul., Ghana: Enti R775 (RSA). H. laotica Gagnep., Indochina: Thorel 1989 (UC). H. lyrata Tul., Ivory Coast: Leeuwenberg 2987 (UC). H. mollis Pax, Tanganyika: Tanner 5258 (UC). Figure 111. H. wallichii Tul., Burma: Dickason 6879 (RSA). Figure 112. Leaves microphylls, symmetrical, ovate, ellip- tic, or obovate; apex acuminate, acute, or obtuse; base acute, obtuse, or rounded; organization of venation low fourth rank; marginal ultimate ve- nation looped; terminal tracheids normal; bun- dle sheath of higher order stomata paracytic; unicellular trichomes typi- cally above and beneath midrib and secondaries, in some species also beneath other veins, abax- ially in junctions of midrib and secondaries, scat- tered on abaxial epidermis, or none; sessile, disk- shaped glands somewhat sunken in pits on the abaxial epidermis. y anatomical observations agree with those of Rothdauscher (1896), Gaucher (1902), and Pax and Hoffmann (1922) The leaf morphology of Hymenocardia reveals little about its relationships within the Phyllan- thoideae, although the high rank of the leaves LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 77 suggests proximity to the Phyllantheae. This is in accord with its wood, which is of the Glochid- ion type (Metcalfe & Chalk, 1950). Only Didy- mocistus bears similar sunken disk-shaped glands, suggesting that these genera are related. Airy Shaw (1965) recognized for Hymenocar- dia a distinct family related to the Ulmaceae on the basis of the winged fruits and pollen that closely resembles Celtis. The leaves of Hymeno- cardia are of much lower rank than is typical of the Ulmaceae and have symmetrical bases in contrast to Celtis and many other members of the Ulmaceae. Nothing in its foliar morphology suggests that Hymenocardia is out of place in the Euphorbiaceae, and both wood anatomy (Met- calfe & Chalk, 1950) and chromosome numbers (Mangenot & Mangenot, 1958; Webster & Ellis, 1962) support its retention. UAPACEAE (MUELL. ARG.) HUTCH. Uapaca Baill. (OTU 88) U. guineensis Muell. Arg., Nigeria: Ujor FHI 26153 (DAV), Liberia oca 771(UC). Figure 108. U. heudelotii Baill., Nigeria: Taylor FHI 31132 (DAV), Cameroons: Bereteler et al. 2541 (UC), Zenker 59 (UC). Figure 109. U. nitida Muell. Arg., Rhodesia: Rodin 4397 (UC). Figure 110. U. sansibarica Pax, Nyasaland: Brass 17466 ( U. togoensis Pax, Nigeria: Onochie FHI 32206 (DAV) Leaves microphylls to mesophylls, symmet- rical, obovate to oblanceolate; apex obtuse to rounded; base acute-cuneate; organization of ve- nation mid third rank; marginal ultimate vena- tion incomplete or partially looped; terminal tra- cheids normal; bundle sheath of higher order veins parenchymatous; stomata anisocytic; tri- chomes beneath veins (U. sansibarica, U. to- goensis) or none; elongate tanniniferous cells originating in the epidermis and running through the mesophyll; tanniniferous cells in epidermis. My anatomical observations generally agree with those of Gaucher (1902) and Comyn (1947), except that neither of them reported the large several species show the guard cells surrounded by a single subsidiary cell, a pattern that is known only from a few ferns (Fryns-Claessens & van Cotthem, 1973). 78 ANNALS OF THE MISSOURI BOTANICAL GARDEN " жле ST „түү; ~ io 62, FIGURES 111- Leaves of оо апа Bischofia.- —1 11, (VoL. 73 112. Hymenocardia mollis.—113, 114. 5, Terminal leaflet of Bischofia javani Marsupiform domatium —11 of В. javanica. Bars equal 1 cm in Figures AL, 113; 1 mm in Figures 112, 114; and 500 um in Figure 115. The leaf morphology of Uapaca is particlarly intriguing because the genus has unique invo- lucrate heads of disk-less flowers and its affinities us cannot be determined using traditional characters. Architecturally Uapaca appears most like the Antidesmeae/Aporuseae; the large tan- = = niniferous epidermal cells and anisocytic sto- mata strongly suggest a relationship to the Apo- ruseae. Pollen (Punt, 1962; Kóhler, 1965) and wood anatomy (Metcalfe & Chalk, 1950) con- tribute little toward resolving the question Airy Shaw's (1965) suggestion of an afinity 1986] between Uapaca and the Anacardiaceae gains no support from foliar morphology. Anacardiaceae leaves typically are pinnately compound and ex- hibit distinct tendencies for the secondary veins to thin markedly at the margin and for the ter- tiary veins to thin markedly between the sec- ondaries (J. Wolfe, pers. comm.). Uapaca shares none of these characteristics and also has higher venation that branches less than is typical of the Anacardiaceae (pers. observ.). Furthermore, An- acardiaceae generally have anomocytic stomata (Metcalfe & Chalk, 1950). BISCHOFIEAE (MUELL. ARG.) HURUSAWA Bischofia Bl. (OTU 91) , Philippines: бару Elmer UO), Luzon: Elmer 15228 (UC), её cult. California, Reid s.n. (DAV 88589). Figures 113-11 Leaves trifoliolate, mesophylls; leaflets sym- metrical above the base, obovate; apex obtuse to rounded; base obtuse, symmetrical on terminal leaflet, asymmetrical on lateral leaflets; organi- zation of venation second rank; marginal ulti- orm domatia abaxially in junctions of midrib and secondaries and secondaries and tertiaries; irregularly shaped secretory cells in epidermis. My observations agree with those of Roth- dauscher (1896), Gaucher (1902), Penzig and Chiabrera (1903), and Raju and Rao (1977). The teeth of Bischofia are traversed by a single vein that is shifted toward the sinus and ends he as modified theoid teeth (Hickey & Wolfe, 1975). The teeth are concave on the upper side, convex on the lower side, with a rounded apex and rounded sinus. The axis of the tooth forms an angle of about 15° to the tangent. The teeth are regularly distributed along the upper two-thirds of the margin. Within the Phyllanthoideae, the compound leaves and marsupiform domatia of Bischofia are unique. Airy Shaw (1967) pointed out that the terminal petiolule is longer than the lateral ones, suggesting that the leafis basically pinnately rath- er than palmately compound, having become tri- foliolate through loss of leaflets. In further sup- LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 79 port of his hypothesis, he cited a specimen bearing a pinnately 5-foliolate leaf and previous reports of young shoots with 5-foliolate leaves. Other genera of the Euphorbiaceae have compound posal of the family Bischofiaceae, to be near the Staphyleaceae because of supposed similarity in habit and foliage between Bischofia and Tapiscia (Airy Shaw, 1965). The bulk of the evidence suggests otherwise. The floral morphology of Bischofia and the Staphyleaceae is completely different (Webster, 1967), as is the wood anatomy (W. J. Hayden, pers. comm.). Significantly, Bischofia has pen- dent, anatropous, bitegmic, crassinucellate ovules, with the micropyle roofed by a placental obturator (Bhatnagar & Kapil, 1974), a suite of characteristics that is diagnostic of the Euphor- biaceae (Webster, 1967) and not found in the Staphyleaceae (Davis, 1966). Foliar similarity is only superficial. Rather than theoid teeth, Ta- piscia and other staphyleaceous genera have cu- nonioid teeth, in which the vein leading into the tooth forks, with one branch ending in a knot in the sinus and the other continuing into the tooth (Hickey & Wolfe, 1975 and pers. observ.). Fur- ther, in Tapiscia leaves, the venational organi- zation is of higher rank thani in ше Bischofia, апа the former have and anisocytic stomata (pers. observ.). Based on this evidence, Bischofia is probably best placed in a separate tribe in the Phyllan- thoideae, as originally proposed by Hurusawa (1954) and accepted by Webster (1975). Both phenetic and cladistic analyses of leaf characters emphasize the isolation of Bischofia (Levin, 1986 and in press). venation GENERA OFTEN INCLUDED IN THE EUPHORBIACEAE BUT EXCLUDED Y WEBSTER (1975) Martretia Beille (OTU 92) M. quadricornis Beille, Congo: Corbisier- Boland 1391 (MO), Ivory Coast: Leeuwenberg 2697 (UC). Figures 119, 120. Leaves large microphylls or small mesophylls, symmetrical, narrow elliptic; apex acute; base acute; organization of venation mid second rank; marginal ultimate venation looped; terminal tra- cheids normal; bundle sheath of higher order veins parenchymatous; stomata paracytic. 80 [Vor. 73 ANNALS OF THE MISSOURI BOTANICAL GARDEN "X 1 di m ALEA 49 € FIGURES 116-122. Leaves of lora Martretia, and раман — 116, 117. Teeth of PERLE мм ithout deciduous seta. — 118. Tooth of В. javanica.—119, 120. Martretia quadricornis. — 121. drypetes ilicifolia. — 122. Raphides in mesophyll of P. ilicifolia. Bars equal 1 cm in Figures 119, 121; 1 mm in ith and wi Figure 120; 500 um in Figures 116, 117; and 150 um in Figures 118, 122. 1986] The foliar morphology of Martretia fits well within the Phyllanthoideae, and is most similar to Actephila (Levin, 1986 and in press), which share the straight-sided int tal with fre- quent intersecondaries and reticulate tertiaries. However, the false partition in the fruit (Pax & Hoffmann, 1922) and atriate apertures in the pol- len (Punt, 1962; Köhler, 1965) would be unique within the Phyllanthoideae. Clearly more infor- mation on this monotypic genus is required be- fore its relationships will be understood. Paradrypetes Kuhlm. P. ilicifolia Kuhlm., Brazil: Tombos, Barreto 1663 (F). Figures 121-123. Leaves simple, mesophylls, symmetrical, ob- ovate; apex obtuse; base acute; margin with spi- mary vein stout, straight; secondary veins arising at a wide acute angle, the angle more or less uniform, their thickness moderate, course curved abruptly, loop-forming branches joining acutely and accompanied by uniform tertiary loops; in- tersecondary veins composite, zig-zag, one pres- ent in each intercostal area and running its length; tertiaries originating from secondaries at acute angles, weakly percurrent, branching to form in- tersecondaries, essentially parallel to the midrib; ourth order veins orthogonal, their size mod- erate; fifth and sixth order veins random, their sizes moderate; highest vein order sixth; mar- ginal ultimate venation fimbriate; areole devel- opment imperfect, arrangement random, shape triangular to polygonal, size moderate; veinlets simple or branched once; terminal tracheids nor- mal; bundle sheath of higher order veins scleri- fied; epidermal cells isodiametric, arrangement random, anticlinal walls curved; stomata abax- ial, orientation random, stomatal index about 1296, basically paracytic, but each subsidiary cell divided again by one or more wall at right angles to the guard cells, thus similar to the laterocytic type of den Hartog and Baas (1978); trichomes none; crystals present as raphides in abundant idioblasts just below the epidermis. The teeth of Paradrypetes are spinose, tra- versed by a single vein terminating in a point at the tooth apex. Both sides ofthe tooth are straight, the apex acute, and the sinus rounded. The axis of the tooth forms an angle of about 25? to the tangent. The teeth are distributed very irregularly along the upper two-thirds of the margin. LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 81 FIGURE 123. Abaxial epidermis of Paradrypetes ili- cifolia. Bar equals 100 um. Paradrypetes, as the name suggests, was con- sidered by Kuhlmann (1935) to be closely related to Drypetes. However the foliar morphology of Paradrypetes, of which my observations agree with those of Milanez (1935), would make this genus quite anomalous in the Phyllanthoideae and even in the Euphorbiaceae. The most ob- vious discrepancies are the long, zig-zag inter- secondaries, unlike any that I have seen in the subfamily, the tertiaries parallel to the midrib, and the raphides. I have never observed raphides in any phyllanthoides; Gaucher (1902) firmly stated that raphides are never found in the Eu- phorbiaceae. Milanez (1935) reported that raph- ides are abundant in every part of the plant, in- cluding the embryo. I agree with Webster (1975) that the resemblance between Paradrypetes and Drypetes is superficial, and that the former should be excluded from the Euphorbiaceae. There are few families with both stipules and raphides (Metcalfe & Chalk, 1950); further study may re- veal the proper relationships of this monotypic genus. RELATIONSHIPS OF THE EUPHORBIACEAE The foliar morphology of the Phyllanthoideae helps narrow the choice of hypotheses regarding the relationships of the Euphorbiaceae. Parti ularly significant are the basically theoid teeth of Bischofia, Drypetes, and Putranjiva. Hayden (1980) reported teeth intermediate between the theoid and violoid types in Choriceras (Oldfield- ioideae), and Hickey and Wolfe (1975), who pre- sumably examined the uniovulate subfamilies = [e] 82 ANNALS OF THE MISSOURI BOTANICAL GARDEN almost exclusively, characterized the Euphorbi- aceae as having violoid teeth. This series can be interpreted as a parallel to the shift from theoid to violoid teeth that Hickey and Wolfe (1975) inferred to have occurred in the Violales, sup- porting the suggestions of Takhtajan (1980) and Thorne (1976) that the two groups share com- mon ancestry. The Rosidae typically have teeth assigned by Hickey and Wolfe (1975) either to the cunonioid type or its presumed derivative, the rosid type. These types differ from the theoid type in their venation and non-deciduous glandular apex. In contrast, teeth of some Celastrales, particular] members of the Aquifoliaceae and Celastraceae, bear deciduous setae, although these teeth differ from typical theoid teeth in that the former have the apex strongly turned inward toward the sinus whereas the latter are essentially symmetrical. Hickey and Wolfe (1975) originally interpreted the teeth of the Celastrales as modifications of the theoid tooth type, and largely on that basis transferred this order from the Rosidae, in which it is included in the systems of Cronquist (1981) and Takhtajan (1980), to the Dilleniidae. More recently, teeth of the celastroid type have been found in Brexia and related genera, which are placed in or near the Saxifragaceae by most au- thors. Hickey and Wolfe now view the type of tooth found in Brexia and the Celastrales as a modification of the similarly asymmetrical cu- nonioid tooth, and agree that the Celastrales be- long in the Rosidae (L. Hickey and J. Wolfe, pers. c omm.). In light of the proposed relationship between the Euphorbiales and the Celastrales, the asym- metrical teeth of Bischofia deserve particular at- tention. Although they, like the teeth of Brexia and the Celastrales, have the vein and seta shifted toward the sinus, the degree of shifting is much less than is typical of brexioid/celastroid teeth (Levin, pers. observ.). Tooth morphology, there- fore, supports the proposed dilleniid relation- ships of the Euphorbiaceae somewhat more strongly than the relationship to the Celastrales, but the latter possibility remains. A relationship between Euphorbiaceae and the Geraniales ap- pears much less probable than either of these other possibilities. Venati 445+; 1 ++ d levidence on in this regard. In Drypetes and many entire-leaved phyllanthoids, the basal secondary veins arise at a significantly different angle from the other sec- ondaries. This pattern suggests either incipient [VoL. 73 or vestigial actinodromy (Hickey & Wolfe, 1975; Hayden, 1980) and is also found in various prim- itive Violaceae and Flacourtiaceae (Hickey & Wolfe, 1975). Better developed actinodromy is common in other subfamilies of the Euphorbi- aceae, and characterizes a group that Hickey and Wolfe (1975) referred to as the “palmate Dille- niidae," consisting of the Violales, Malvales, and related orders. Although palmate venation oc- curs sparingly in some rosid orders, including the Geraniales, leaves of the Rosidae typically have phorbiaceae) the only genus with any suggestion of pinnately compound leaves is Bischofia. Thus neither venation nor tooth morphology alone allows the possibility of rosid relationships to be eliminated for the Euphorbiaceae. Taken together, however, these aspects of leaf architec- ture suggest that the dilleniid relationship is more tenable. The relationship to the Geraniales is un- likely given the theoid teeth of the Phyllanthoi- deae; although the Celastrales have somewhat similar teeth, differences in the teeth apparent upon close comparison and the suggestion of in- cipiently actinodromous venation in the Phyl- lanthoideae make a relationship to this order doubtful. However, the Euphorbiaceae share with the Violales and Malvales both basically similar teeth and venation patterns. CONCLUSIONS Leaf architecture and epidermal morphology vary in a taxonomically useful way in the Phyl- lanthoideae. Most tribes recognized by Webster (1975) appear to be relatively homogeneous and well defined, with most exceptions being tribes that contain genera that have become either her- baceous or highly xerophytic. Within the subfamily (but excluding herbaceous or highly xerophytic genera), venational organization is of lowest rank in those tribes generally agreed to have the most primitive flowers, i.e., the Wie- landieae and Amanoeae, and of highest rank in putatively more derived tribes, e.g., Phyllantheae and Hymenocardieae, a pattern that conforms to Hickey’s (1971, 1977) generalization that the level of organization increases evolutionarily. Cladis- tic and phenetic analyses of the foliar morphol- ogy presented elsewhere (Levin, 1986 and in press) clarify additional evolutionary patterns, such as the closeness of the Antidesmeae, Apo- 1986] ruseae, and Spondiantheae, and the proximity of Astrocasia, Andrachne, and the Phyllantheae. In addition to confirming the general aspects of Webster’s (1975) classification indicated above, foliar morphology indicates some genera whose Prominent among these are о Croizatia, Didy- mocistus, Jablonskia (= Securinega congesta), and a a of which are poorly known and/or recently described. The relationships be- tween Amanoa and Actephila may also deserve examinaton, because the leaves of these genera differ considerably from each other. Finally, at- tention should be paid to the suggestion, based primarily on unusual epidermal characteristics but supported by architecture, = a ы between Uapaca and the Aporu Leaf architecture also proves аай valuable at higher levels. The presence of theoid teeth and suggestion of vestigial or incipient ac- tinodromy in the most primitive subfamily o the Euphorbiaceae argue for placement of the family in the vicinity of the Violales in the Dil- leniidae. The most popular alternative, that of rosid affinities, gains little support from leaf ar- chitecture. LITERATURE CITED AIRY SHAW, Н. ja 1965. Diagnoses « or new families, new names, etc., for th n of Willis's “Dictionary.” " Kew Bull. 18: 249-21 3. . 1967. Notes on the genus еш Bl. (Bisch- ofiaceae). Kew Bull. 21: 327- 973. Willis's A Dictionary oí the Flowering Plants and Ferns, 8th edition. Cambridge Univ. Press, Cambridge BAILLON, Н. 1858. Etude Générale d Groupe des Euphorbiacées. Victor Masson, Par BARANOVA, M. 1972. Systematic bred of the leaf epidermis in the Magnoliaceae and some related families. E 21: 447—484 BEDELL, H. G. . [Abstract:] Leaf architecture and foliar dre in Marcgraviaceae oo ). Publ. Bot. Soc. Amer., Misc. Ser. 160: 6 BENTHAM, G. 1880. Euphorbiaceae. In G. Bentham D. Hooker, Genera Plantarum 3: 239-340. . Reeve & Co., London. BHATNAGAR, A. K. & К. М. КАРИ. 1974. Bischofia javanica —its relationship with d Euphorbi- aceae. Phytomorphology 23: 264- pom euille du genre a Baill. (Euphorbiactes) Lille Imp. Morel & Corduant, ari E L. 1940. On the phylogeny of the Euphor res aceae a and some of od Presumed allies. Revista iacet 25:2 eid A. An pec System of Clas- LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 83 sification ofthe аы Plants. Columbia Univ. Press, New Yor Davis, G. L. 1966. ` Systematic Embryology of the Angiosperms. John Wiley & Sons, New Yor Depe, R. A. 1962. Foliar venation patterns in the Rutaceae. Amer. J. Bot. 49: 490-497. Den HARTOG, R. M. 1978. Epidermal characters of the Celastraceae : sensu lato. Acta Bot. Neerl. 27: 355-388. DiLcHER, D. L. 1974. Approaches to the identifica- tion of angiosperm leaf remains. Bot. Rev. (Lan- caster) 40: 1-157. DoYLE, J. A. & L. J. Hickey. 1976. Pollen and leaves lution of Angiosperms. Columbia Univ. Press, New York. EHLERINGER, J. & O. BJORKMAN. 1978. Pubescence and leaf spectral characteristics in a desert shrub, Encelia farinosa. Oecologia 36: 137—148. ERDTMAN, G. . Pollen Morphology and Plant о I. Angiosperms. Chronica Botanica, Stockh FRYNS- в Е. & W. van COTTHEM. 1973. А new classification of the n P ue of sto- mata. Bot. Rev. (Lancaster) 39: 7 8. GAUCHER, L. 1902. Recherches AN sur les Euphorbiacées. Ann. Sci. Nat. Bot., Sér. 8, 15: —309. HAvDEN, W. J. 1980. Systematic anatomy of Old- fieldioideae (Euphorbiaceae). Ph.D. dissertation. Univ. of Maryland, College Park .S. BRANDT. 1984. Wood anatomy and г I 1 2. s X Sy t Bot. 9: 458-466. Hickey, L. J. 1971. [Abstract:] Evolutionary signif- icance of leaf rd dy features in the woody dicots. Amer. J. Bot. 58: 469. . 1973. Crib don of the architecture of di- cotyledonous leaves. Amer. J. Bot. 60: 17-33. Stratigraphy and paleobotany of the Golden Valley formation pad Tertiary) of west- North Dakota. Mem. Geol. Soc. Amer. 150: i —. 1980. [Abstract:] Leaf architecture of Ona- graceae. 2nd Int. Cong. Syst. Evol. Biol., Abstracts, p. 69. J. A. DovrE. 1972. [Abstract:] Fossil evi- dence on evolution e angiosperm leaf venation. Amer. J. B 9: ‚ А. WOLFE. de The bases of angio- sperm phylogeny: vegetative morphology. Ann. Missouri Bot. Gard. 62: 538-589. Нил, В. S. 1980. A numerical taxonomic approach o the study of angiosperm leaves. Bot. Gaz (Crawfordsville) 141: 213-229. Hurusawa, I. 1954. Eine nochmalig Durchsicht des herkómmlichen Systems der Euphorbiaceen im weiteren Sinne. J. Fac. Sci. Univ. Tokyo, Sect. 3, Bot. 6: 209-342. HUTCHINSON, J. 1969. Tribalism in the family Eu- phorbiaceae. Amer. J. Bot. 56: 738-758. JABLONSKY, ОКОВ Farino de ae-Bridelieae. Pflanzenreich IV, 147-VIII(65): —98. 84 . 1965. Euphorbiaceae. Pp. 150-178 in B. Ma- guire (editor), The Botany of the Guyana High- land, Part IV. Mem. New York Bot. Gard. 12(3): 1-285. Jussieu, A. DE. 1824. De Euphorbiacearum generi- bus a. earumdem viribus tentamen. Di- 65. Die Pollenmorphologie der bio- vulaten Euphorbiaceae und ihre Bedeutung fiir die Taxonomy. Grana Palynol. 6: 26-120. KUHLMANN, J. C. 1935. Novas especies botanicas da Hyléa. Arq. Inst. Biol. Veg. 2: 83-89. 1940. Especies novas equatoriais. Anais Reu- niáo Sul-Amer. Bot. 3: 78-86. Levin, G. A. 86. Systematic foliar morphology of Phyllanthoideae (Euphorbiaceae). II. Phenetic analysis. Ann. Missouri Bot. Gard. 73: 86-98. In press. Systematic foliar morpholog Phyllanthoideae as III. Cladistic analysis. Syst. Lucic, P. C. 1970. олы leaf venation studies of selected species of Acer. M.A. thesis. Univ. of Cal- ifornia, Berkeley. MANGENOT, S. & G. MANGENOT. 1958. Deuxiéme chez diverses dicotyledones et monocotyledones d'Af- rique occidentale. Bull. Jard. Bot. État 28: 315- 329. METCALFE, C. R. & L. CHALK. 1950. Anatomy of the Dicotyledons, 2 volumes. Oxford Univ. Press, Ox- ord. MiLANEZ, F. В. 1935. Anatomia de Res ili- cifolia. Arq. Inst. Biol. Мы 2: 133-1 MILLER, K. I. & G. L. WEB 1966. CN numbers in the Euphorbiaceae. Brittonia 18: 372- 379. Шан», J. 1866. Euphorbiaceae [except Euphor- bieae]. Jn A. de Candolle (editor), Prodomus Sys- peaa Naturalis de Vegetabilis 15(2): 1 89-1 286. ortin Masson, Pari PAYNE, W. W. 1970. Helicocytic and allelocytic sto- Amer. J. Bot. 57: 140-1 A glossary of oi hair terminology. Brittonia 30: 239-255. tomatal patterns in embryophytes: Е evolution ontogeny and interpretation. Tax- n 28: -132. Pax, F a perc T PR In A. Engler & K. Prantl (editors), Die Natürlichen Pflanzenfamilien, 1st edition. 3(5): 1-119. Wilhelm Engelmann, Leip- zig. K. HOFFMANN. Euphorbiaceae- Phyllanthoideae- Phyllantheae. Pflanzenreich IV, 147-XV(81 349. . Euphorbiaceae. Jn A. Engler & K. Prantl (editors), Die Natürlichen Pflanzen- familien, 2nd edition. 19c: 11-233. Wilehlm En- gelmann, Leipzi PENZIG, O. & C. CHIABRERA. 1903. Contributo alla conoscenza delle piante acarofile. Malpighia 17: 429-487. , T. R. 1953. Morphological ries е studies on the foliar venation of c angio- sperms. Ph.D. dissertation. Univ. 2 California, Berkeley. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Punt, W. 1962. Pollen morphology of the Euphor- biaceae with special reference to taxonomy. Wen- tia 7: 1-116. Raju, V. S. & P. N. Rao. 1977. Variation in the structure and development of foliar stomata in the Euphorbiaceae. Bot. J. Linn. Soc. 75: 69-97. RAUNKIAER, C. The Life Forms of Plants and e m Plant Geography. Clarendon Press, Ox- : F 1913. Indigenous Trees ed ios Hawaiian nds. Privately publis и Нопо e r die EN ond. Axe der dde or m Auschluss iet Euphylla antheen). Bot. Cen : 65-79, 97-108, 129-136, 161-169, 193- 203. 248-253, 280-285, 305-315, 338-346, 385- ‚Р.М. & W. С. Dickison. 1977. Leaf venation patterns of the genus Em (Dilleniaceae). J. Arnold Arbor. 58: 209-24 SEHGAL, O. & G. S. PALIWAL. due Studies on the leaf anatomy of up rod II. Venation patterns. Bot. J. Linn. Soc. 68: —208. SELLING, О. Н. 1947. ы in Hawaiian pollen sta- tistics. II. The pollen of Hawaiian phanaerogams. Special Publ. Bernice Pauahi Bishop Mus. 38: 197- 206. RURY SHARMA, G. K. & B. D. DUNN. 1969. Environmental ао ie leaf surface traits in Datura stra- monium. . J. Bot. 47: 1211-1216. SHERFF, Е. E. 1939. Additional studies of the Hawai- ian Euphorbiaceae. Publ. Field Mus. Nat. Hist., Bot. Ser. 17: 547-576. 942. Some new or otherwise noteworthy Seem. Publ. Field Mus. Nat. Hist., Bot. Ser. 22 570-572. SMITH, А. C. & E. S. AvENsU. 1964. The identity of the genus Calyptosepalum S. Moore. Brittonia 16: 220-227. diles ise A. 1965. Cuticular studies as an aid to plant xonomy. Bull. Bri 78. t. Mus. (Nat. Hist.) Bot. 4: 1- STEYERMARK, J. A. 1952. Botanical exploration in Venezuela —II. Fieldiana, Bot. т A, 43-447 STONE, B. 1967. A review of the endemic genera of Hawaiian plants. Bot. Rev. cen] 33: 216- 2 TAKHTAJAN, А. L. 1980. Outline of the classification of flowering plants ene phyta). Bot. Rev (Lancaster) 46: 225-3 THORNE, R. F. 1976. mo NS classification of the Angiospermae. Evol. Biol. 9: 106. Tucker, S. C. 1977. Foliar sclereids in the Magno- liaceae. Bot. J. Linn. Soc. 75: 325-356 WEBSTER, G. L. 1956. monographic study of the West Indian species of Phyllanthus dv 1-3]. J. Arn е bor. 37: 91-122, о 40-359. 57. А monographic study of the West In- dian pales of Phyllanthus [Parts 4 —6]. J. Arnold Arbor. 38: 51-80, 170-198, eu —. 1958. A monographic study of the West In- dian species of désert (Pars 7, 8]. J. Arnold Arbor. 39: 49-100, 1986] A revision of the genus Meineckia (Eu- phorbiaceae). Acta Bot. Neerl. 14: 323-365. 1967. The genera of Euphorbiaceae in the southeastern United States. J. Arnold Arbor. 48: 303-430. 1975. Conspectus of a new classification of the Euphorbiaceae. Taxon 24: 593-601. 1984a. A revision of IUe (Euphorbi- aceae) Allertonia 3: 259-312 984b. Jablonskia, a new genus of Euphor- LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. I. 85 biaceae from South America. Syst. Bot. 9: 129- 132 & J. R. Еша$. 1962. Cytotaxonomic studies in the poi e subtribe Phyllanthinae. Amer. J. Bot. 4 18. I. all 1963. The genus Reverchonia T Rhodora 65: 192-207. Wo re, J. А Fossil forms of Amentiferae. Brit- tonia 25: 334-355 SYSTEMATIC FOLIAR MORPHOLOGY OF PHYLLANTHOIDEAE (EUPHORBIACEAE). Il. PHENETIC ANALYSIS! GEOFFREY A. LEvIN? ABSTRACT Leaf architectural and cuticular characters in the Phyllanthoideae (Euphorbiaceae) were analyzed using Similarity Graph Clustering. The resulting groups and their similarity relationships correspond Ullal kably appear to A either cases in which other evidence also suggests that previous classifications are invalid or in which total similarity could be expected to clear results in be a poor indicator of relationship. Obta a group not known for having diagnostic leaf characteristics indicates the кас: ing such potential foliar morphology has for classification of modern and fossil angiosperms. This paper reports the first numerical analysis of leaf architectural and cuticular features of a related group of extant flowering plants. This is part of a project undertaken both to test the sys- 980) have tested foliar character sets by at- tempting to correctly identify leaves taken from randomly selected, usually unrelated plants. For example, Hill collected five leaves each from 20 species of woody plants growing in a botanic garden. His sample included 19 genera from 12 families. peque both he and, to a lesser extent, olph achieved their stated goal of taximetri- cally grouping together leaves taken from the same species, neither was able to recognize any higher taxa using their methods and character sets. I instead chose to test the systematic usefulness of a leaf character set in a natural group by com- paring the results of numerical analysis of leaf characters with recent classifications based on more traditional sources of systematic data. I selected for study the Phyllanthoideae, putative- ly the most primitive subfamily of the Euphor- biaceae, because the long history of systematic interest in the family has resulted in a series of infrafamilial classifications based on character- icance of the subfamily in clarifying the rela- tionships of m e That the leaves PI lati vely nondescri ipt makes analysis of this group of additional inter- est. The recent treatment by Webster (1975; see Table 1), who integrated evidence from floral morphology, palynology, cytology, and wood anatomy, serves as the principal classification with which to compare the foliar analysis. I derived leaf architectural characters from the system proposed by Hickey (1973). Although the use of this character set for numerical analysis has been criticized because many characters do not have equidistant arre states (Hill, 1980), the characters have been s o be of systematic importance at higher taxo ee levels Hickey & Wolfe, 1975). баат characters were selected from the list prepared by Stace (1965; see also Dilcher, 1974). In another paper (Levin, 1985), I discussed the characters and their states and described the leaves of the 51 genera in the Phyllanthoideae that I examined. In a fu- ture paper (Levin, in press), I will present the results of a cladistic analysis of the same data set. — MATERIALS AND METHODS DATA GATHERING PROCEDURES The taxa included in this study are listed in Table 1, along with the numbers I assigned to them for convenience of representation in Fig- ures 1-6. For reasons that I explain below, I have deleted from the list all the taxa with phyllan- ! This study — ҮЕ of a Ph.D. dissertation submitted to the University of California, Davis. J. A. er, Doyle, G. L. We guidance. Comments particularly thank J. A. time available gratefully acknow ? Natura ANN. MISSOURI Bor. GARD. 73: 86-98. 1986. jn cPherson, C. A. Meacham, and N. R. W uius for the Hoc den loan of many cleared leaves and T. reip a support from the University of California, Davis, and Ripon College, Wisconsin, is d J. Au oa made Eros comments on the manuscript and provided support and Morin also improved the manuscript. I Duncan for making com puting l Hi iens Museum, P.O. Box 1390, San Diego, California 92112. 1986] LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY, II. 87 TABLE 1. Taxa includes in study and numbers as- signed to OTUs. Classification follows Webster (1975). TABLE 1. Continued. Number Taxon Number Taxon WIELANDIEAE — Savia sect. Heterosavia . sect. 1 AMANOEAE OV 00 ө р ө tA SD м м D & > 8 З E = ta anoa 11 Actephila group 1: A. anthelminthica, A. nitida 12 А. group 2: A. excelsa 93 Croizatia BRIDELIEAE 14 Bridelia group 1: sects. Micrantheae pro parte, Scleroneurae pro parte 15 B. group 2: sect. Cleistanthoideae B. group 3: sects. Micranthae pro parte, Scleroneurae pro parte, Stipulares 17 Cleistanthus acuminatissimus 18 C. көн Stipulat 19 Cs pin i ud 20 e sect. Charta 21 . sect. Cleistanthus 22 C.saichikii 23 C. sect. Australes 24 C. sect. Leiopyxis DICOELIEAE 25 Dicoelia PORANTHEREAE 26 Andrachne sects. Arachne, Phyllanthidia 27 A. sect. Phyllanthopsis SPONDIANTHEAE 28 Spondianthus ANTIDESMEAE Antidesma sects. Roxburghiana, Venosa pro parte . sects. Laciniata, Venosa pro parte 30 A 31 A. sect. Tetrandra 32 A. sect. Montana 33 А. sect. Velutinosa 34 A. в Ghaesembilla 35 Celianella 36 | Hyeronima 37 Leptonema 38 Thecacoris APORUSEAE Aporusa 40 Ashtonia 41 Baccaurea sects. Everettiodendron, Calyp- troon 42 В. sect. Pierardia 43 B. Dubiae 44 В. sect. еи 45 Didymoc 46 Мае Bias 47 Коло im 48 | Richeria DRYPETEAE 49 ороо sects. Drypetes pro parte, Sphra- gidia 50 D. sects. je ee pro parte, Sphragidia 51 D. sect. Sphragidia pro parte 52 D.sects. Oligandrae pro parte, Stemono- discus, Stenogynium, Stipulares pro parte 53 D. sects. Oligandrae pro parte, Stipulares pro parte 54 Р. зем. ii im pro parte 90 Neowawra 55 Pune PHYLLANTHEAE Securineginae 56 | Jablonskia 57 Keayodendron 58 | Meineckia 59 Pseudolachnostylis 60 Zimmermannia Flueggeinae 61 Flueggea 67 Margaritaria UAPACEAE 88 | Uapaca HYMENOCARDIEAE 89 Hymenocardia BISCHOFIEAE 9] Bischofia INCERTAE SEDIS 92 Martretia thoid branching, in which plagiotropic branch systems resemble pinnately compound leaves (OTUs 62-66 and 68-87). In all but a few cases, I studied cleared leaves of a minimum of 1096 of the species of each genus. I then grouped the 88 ANNALS OF THE MISSOURI BOTANICAL GARDEN TABLE 2. Characters and character states. See Hick- ey (1973), Dilcher (1974), and Levin (1985) for more explanation. The type of each character, and for or- dered characters the value chosen for j, is indicated in parentheses. — . Organization: Simple or Compound. (Simple) . Base кише Symmetrical or Asymmetrical. тшге кә . Ма ntire, Entire ог as but with glands, or Too vien (Ordered . Venation: LN Weakly brochidod- romous, or Eucamptodromous. (Ordered, 1) . Primary size: Moderate, Stout, or Massive. (Or- dered, . Secondary angle: Narrow («45?), Moderate, or Wide (>65°). (Ordered, 1) 7. Angle of basal secondaries, relative to adjacent secondaries: More acute, Similar, or More obtuse. (Ordered, 1 . Angle of lower secondaries, relative to middle sec- ondaries: More acute, Similar, or More obtuse. (Ordered, 1) Angle of upper secondaries, relative to middle sec- ondaries: More acute, Similar, or More obtuse. (Ordered, 1) . Secondary course: Curved uniformly or Curved abruptly. (Simple) . Angle of secondary loops: Acute, Right, or Obtuse. (Ordered, 1) . Size of outer loops: Irregular, Uniform, Decreasing upwards, or Absent. (Ordered, 1) . Tertiary angle of origin, admedial: Acute, Right, or Obtuse. (Ordered, 1) . Teritary angle of origin, exmedial: Acute, Right, or Obtuse. aquo. 1) . Tertiary pat : Ramified, Random reticulate, Orthogonal e Asi percurrent, Strong- ly percurrent with angle to midrib oblique, or Strongly percurrent ith peek to midrib predom- inantly right. (Ordered, 3 . Simple intersecondaries: Absent or Present. (Sim- le A сл е оо 2 — o — — — N — uy A — wa — an — N . Composite intersecondaries: Absent, Infrequent (in fewer than 20% of intercostal areas), or Fre- quent (in more than 20% of intercostal areas). (Or- dered, 18. Intramarginal vein: Absent or Present. (Simple) 19. Higher order vein pattern: Ramified, All random, 4° orthogonal and higher orders random, or 4° and 5° orthogonal. (Ordered, 2 20. Higher order vein size: 4° moderate and 5° heavy, All moderate, 4° moderate and 5° fine, or 4° and 5° fine. (Ordered, 2) 21. Highest order present: 4°, 5°, or 6°. (Ordered, 1) 22. Areole development: Incomplete, Imperfect, or Well- dd (Ordered, 1) . Areole arrangement: Random or Ordered. (Sim- ple) кә uy [VoL. 73 TABLE 2. Continued. 24. Areole shape: Irregular or Regular. (Simple) 25. Areole size: Large, Medium, or Small. (Ordered, 1) . Veinlets: Absent, Simple, Branched 1-2x, or Branched 2-3x. (Ordered, 2) 27. Pri ti tals i phyll: Absent or Present. N an (Simple) 28. Prismatic crystals with veins: Absent or Present. (Simple) 29. Druses in mesophyll: Absent or Present. (Simple) 30. Druses with veins: Absent or Present. (Simple) 31. Epidermal anticlinal walls, adaxial: Straight or Undulate. (Simple) 32. Ep idermal anticlinal walls, abaxial: Straight or Undulate. (Simple) 33. Epidermal papillae: Absent or Present. (Simple) 34. Stomatal location: Abaxial only, Primarily abaxial but a few к or Approximately equal on both surfaces. (Orde 1) 35. Stomatal index neil < 10%, 10-20%, or > 20%. (Ordered, 1) 36. Stomatal type: Paracytic or Anisocytic. (Simple) 37. Water stomata: Absent or Present. (Simple) 38. Unicellular жарт даш Absent, Solitary, or Some tufted. (Ordere 39. Uniseriate, deci trichomes: Absent or Present. (Simple 40. PEINE trichomes: Absent ог ны (Simple) 41. F nt or Present. (Sim 42. Sa MA epidermal cells: Absent or Present. (Simple) 43. Sclerified epidermal cells: Absent or Present. (Sim- ple) species into homogeneous groups, which gener- ally correspond to sections or undivided genera. I treated each group as an operational taxonomic unit, or OTU, for this study. Details of the pro- cedures I used for collecting and clearing the leaves and for selecting the OTUs have been pub- lished in another paper (Levin, 1986). Table 2 lists the 43 characters I scored for each leaf. These characters and their states are de- scribed by Stace (1965), Hickey (1973), and Lev- in (1986). All are binary or multistate qualitative characters. The basic data matrix has been pub- lished elsewhere (Levin, 1986: table 1) and will not be reproduced here. DATA-ANALYTIC METHOD Rather than using a clustering method like UPGMA that merely indicates the level of sim- ilarity at which a taxon or group of taxa first joins 1986] another taxon or group of taxa and expresses the results in the form of a dendrogram, I chose to use Similarity-Graph Clustering (SIMGRA). This method, because it does not force the results into a hierarchical form but instead shows all rela- tionships at a given similarity level, summarizes methods (Prance et al., 1969; Legendre & Rogers, 1972). The theoretical and mathematical frame- works of SIMGRA can be found in Estabrook (1966), Estabrook and Rogers (1966), Wirth et al. (1966), and Legendre and Rogers (1972), and other examples of its use in Prance et al. (1969), Rogers and Fleming (1973), and Duncan (1980). The explanation presented here is derived from these references The SIMGRA algorithm consists of two steps. First, a similarity measure is calculated for all pairs of objects (the OTUs in Table 1). The sim- ilarity measure used here is the generalized Sim- ple Matching Coefficient of Estabrook and Rog- ers (1966). This is defined as: 5 S,(a,b) S(a,b) = where S(a,b) is the similarity between any two acter k, then S,(a,b) = 1. If the two OTUs differ for character k, then one of three rules may be chosen by the taxonomist to calculate S,(a,b): 1. S(a,b) = 0 (simple character); 2g + 1 — d) 23+2+dj 0 2. Si(a,b) = whenever d < j when d > j, where j is the maximum number of character states by which two OTUs may be separated and still be considered at all similar, and d is the distance apart that the two states are ina pre-specified ordering of the states (ordered character); or S,(a,b) = arbitrary values assigned by the tax- onomist in advance, by providng a matrix of values between O and 1 (matrix character). 195] Thus the partial similarities for both ordered and matrix characters involve subjective decisions on the part of the taxonomist. For the ordered LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. II. 89 characters in this study, I set the value of j for character k at the number of character states for that character less two or three; these values cause only the most extreme distances between char- acter states to yield a partial similarity of 0. There were n matrix characters in this study. The type and, for ordered characters, the value of j, are died in Table 2 After an overall similarity value S(a,b) has been calculated for each pair of OTUs, the similarity values are examined in order of decreasing mag- nitude. At each level of similarity, a connection is formed between two OTUs if they are at least as similar as the specified level. A cluster is a group of OTUs for which there exists at least one continuous pathway of connections joining the OTUs. The procedure continues until all the OTUs belong to one cluster. In this study, I used the program SIMGRA by G. F. Estabrook (Univ. of Michigan) on an IBM 4341 at the University of California at Berkeley. RESULTS During the SIMGRA analysis, clusters were formed and enlarged at 45 levels of similarity. I grouped these levels and summarized them in Figures 1-6. On each drawing the lowest simi- larity value at which connections were made (S), the levels summarized (L), and the number of OTUS that have not yet formed a connection with any other OTU (single member clusters or SMCs) are listed on the left side. Each OTU that has formed a connection with another is indi- lowest-numbered OTU belonging to the cluster. Within a cluster, tightly connected subsets of OTUS that are less tightly connected to other subsets are termed subclusters and referred to by the prefix *SC-," with the lowest-numbered OTU as the suffix. For highly connected groups, either clusters or subclusters, it is often inconvenient to show all the connections. Such a group of OTUs, therefore, is shown diagrammatically as a circle, with the OTUs forming the cluster rep- resenting by placing their numbers at the edge of the circle. The fraction in the center, in which the numerator equals the number of connections actually formed within the cluster and the de- 90 ANNALS OF THE MISSOURI BOTANICAL GARDEN Aporus 39-40 Richeria Ashtonia 1817 ү MERI $>.92000 11-11 49 SMC Astrocasia ve dicia [Vor. 73 © „©. <. TE Y P Thecacoris Antidesma 49-50 Drypetes 61 67 Phyllanthinae FIGURE 1. Clusters present in SIMGRA analysis at level II or similarity value 0.92000. In this and all following oe OTUs are represented by numbers, as listed in Table 1. Dashed lines indicate connections S rmed between OTUs at any rmed at previous levels, and thick solid lines indicate connections between circular subclusters. The lengths of lir lines reflect the constraints of two-dimensional representation and in no way indicate the degree of connect- edness between the groups. See text for further explanation nominator equals the number of connections that are possible, indicates the connectedness of the cluster's members. Connections between circular subclusters, designated by the prefix '*CSC-," are represented by a heavy solid line accompanied by a fraction that represents the degree of con- nectedness between the subclusters; again, the numerator of the fraction indicates the number of actual connections and the denominator equals line c: no = to the connectedness between the grou initially, I ша all the OTUs for which I had data (Levin, 1985: table 1) except for Lin- gelsheimia, which I received after I had com- pleted all the SIMGRA analysis. It immediately became clear that members of the Flueggeineae exhibited overwhelming convergence with other OTUs, completely obscuring relationships both within the subtribe and between the other OTUs. The foliar morphology of this species-rich group has apparently undergone an adaptive radiation, associated with the evolution of phyllanthoid ranching, in which determinate lateral branch- lets mimic pinnately compound leaves (Webster, 1956, 1967; Levin, 1985). Fortunately for the systematist, the more primitive members of the subtribe, F/ueggea, Margaritaria, Richeriella, and a few species ОР Phyllanthus, retain the normal habit. I therefore removed the genera with phyl- lanthoid branching (Breynia, Glochidion, Phyl- lanthus, Sauropus, and Synostemon) from the analysis, leaving the 68 OTUs in Table 1. After the first 11 similarity levels (Fig. 1), 19 OTUs have joined to form six clusters, and 49 OTUs remain as SMCs. The only large cluster, C-29, consists of Thecacoris (OTU 38) and a completely 1 d subcluster containing all but one section of Antidesma. The other clus- ters have only two or three members and are minimally connected. As Figure 1 indicates, the first-formed clusters generally consist of OTUs from a single genus or of closely related E which might be expected to be quite similar. Lingelsheimia been included in the a. it would have joined C-2 through a connection with Zimmermannia (60) at level 6. The next seven levels, to S = 0.90000 (Fig. 2), see the formation of five new two-membered clusters and the enlargement of the earlier-formed clusters. Two new internal connections form in C-29, very tightly connecting Thecacoris to the other OTUs in the cluster. C-2 and C-61 join to 1986] LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. II. 91 48—39— 40 ar д Baccaurea 21 asses bii 17 22 p» 25 Dicoelia 13 Pentabrachium 16 ------. 15 S>.90000 112-18 34 SMC e 49—50----54-...5 2 is ; 7 ..*60 1 Phyllantheae ч. 4 7, “в FIGURE 2. Clusters present in SIMGRA analysis at level 18 or similarity value 0.90000. form a single cluster, with OTU 56, Jablonskia, connected to only one OTU ofan otherwise com- pletely interconnected cluster of members of the Phyllantheae. The clusters representing Cleistan- thus and Drypetes each gain one or two new OTUs belonging to their respective genera, but each cluster remains minimally interconnected. Two sections of Baccaurea rather loosely join C-39. Thirty-four SMCs remain. By similarity level 0.88222 (Fig. 3), only one new cluster forms and C-1 remains unchanged. The most marked change is the joining of C-13, C-17, C-22, and C-29 into a single cluster. CSC- 29 is joined by other OTUs in the Antidesmeae, Antidesma sect. Ghaesembilla (34) and Hyero- nima (36). C-2, C-49, and C-39 become more tightly connected internally, and the latter is joined by OTU 46, Maesobotrya. OTUs 6 and 9 connect to C-7, which now consists of Savia and its segregates Blotia and Petalodiscus. OTU 14 joins the other OTUs in Zridelia. Twenty-five SMCs remain. By level 31 (Fig. 4), only 16 SMCs remain. C-1 and C-6 join by a connection between Wielandia (1) and Savia (6) to form most of the Wielan- dieae. No new internal connections form within C-2, but it is joined by three genera, Heywoodia (3), Keayodendron (57), and Bischofia (91). C-39 joins CSC-29 by a single connection between OTUs 36 and 40, and also forms connections with a section of Baccaurea (43), Protomega- baria (47), and Uapaca (88). In other portions of C-13, additional connections form between the subclusters, with two sections of Cleistanthus (22 and 24) beginning to become connected to the other sections of the genus. A connection between OTUSs 11 and 12 joins the two OTUs in Actephila; the latter OTU had earlier con- nected with a section of Andrachne (27). Neo- wawraea (90) joins Bridelia (C-14), and Putran- jiva (55) forms a single connection with its close relative Drypetes (C-49), whose members are be- coming increasingly tightly interconnected. Levels 32 through 37 (Fig. 5) see connections forming to seven SMCs and the coalescence of the earlier clusters into two large clusters. Con- nections between OTUS 1 and 55, 6 and 56, and 55 and 56 loosely join CSC-1 (Wielandieae pro parte), CSC-2 (roughly the Phyllantheae), and CSC-49 (Drypeteae). These subclusters remain 92 ANNALS OF THE MISSOURI BOTANICAL GARDEN 14 16—15 $2.88222 Bridelia L19-26 25 SMC [Vor. 73 36 Hyeronima 6 Savia : 1 Blotia 9 8 Petalodiscus FiGURE 3. Clusters present in SIMGRA analysis at level 26 or similarity value 0.88222. relatively unchanged at these levels: CSC-1 gains two internal connections, CSC-2 gains OTU 20, a section of Cleistanthus, by a connection with OTU 60, and CSC-49 gains one internal con- nection and another group in Drypetes, OTU 51. OTUs 27 and 38 connect to 26, thereby joining CSC-11 to CSC-29, the core genera of the An- tidesmeae. No new connections form between CSC-13 and CSC-29, and only one between CSC- 13 and CSC-17. CSC-17 gains SC-22, thus fur- ther consolidating Cleistanthus, and Meineckia (58) connects to OTU 24. Hymenocardia (89) simultaneously joins both Neowawraea (90) and OTU 18 to connect SC-14 to CSC-17. CSC-29, in addition to becoming increasingly highly in- ternally connected, increases its association with CSC-17 through four new connections and with CSC-39 through three. CSC-39 gains only OTU 42, but increases both its internal connectedness and its connectedness to OTU 47. This subclus- ter now corresponds to the Aporuseae of Webster (1975), less Didymocistus (45), which remains among the nine SMCs. At level 45 (Fig. 6), all OTUs belong to one cluster. This large cluster is composed of eight circular subclusters and six fairly isolated OTUs. ielandieae pro parte, CSC-1, in addition to becoming more tightly connected internally, gain Discocarpus (5) through a connection to Pet- alodiscus (8), form a new connection to CSC-11, and connect further to CSC-2 both directly and through newly added OTUs 10 (Amanoa) and 93 (Croizatia). The connection between CSC-1 and CSC-2 is the strongest between any two sub- clusters. CSC-2, which with the addition of Pseu- dolachnostylis (59) consists of most of the Phyl- lantheae, Astrocasia (2) and Heywoodia (3) of the Wielandieae, and a section of Cleistanthus, con- nects with six other subclusters, albeit generally weakly. Amanoa (10), Celianella (35), and Croi- zatia (93) all form connections with CSC-2 late in SIMGRA analysis; in fact Amanoa is the last OTU to join the cluster, connecting simulta- neously to OTUSs 1 and 56 at level 45. CSC-11 (Actephila and Andrachne), CSC-14 (Bridelia), and CSC-45 (Didymocistus, Hymenocardia, and Neowawraea) become very highly or completely tremain fair- ail ter nally 1986] 88 Uapaca 46 , Protomega- "47 baria S>.87111 127-31 16 SMC Keayodendron 57.... э” Bischofia LEVIN — PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. II. 93 11 UO Actephila 12—27 Апагасһпе Putranjiva Drypetes а 4 | Lachnostylis 7 of N FiGure 4. Clusters present in SIMGRA analysis at level 31 or similarity value 0.87111. ly isolated from other subclusters. CSC-49, the Drypeteae less Neowawraea, is less tightly inter- nally connected than these, but also is quite iso- lated. Two OTUs connect to CSC-11: a section of Cleistanthus (23), which connects at level 38, and Martretia (92), which connects at level 42. Leptonema (37) forms six connections to mem- bers of SC-29, which, together with CSC-17, si- multaneously joins to CSC-13. CSC-13 thus con- sists of most of Cleistanthus (17-19, 21, 22, and 24), Dicoelia (25), Meineckia (58), Pentabrach- ium (13), and the Antidesmeae (29-34, 36-38) less Celianella (35). Spondianthus (28) forms six connections to SC-29. The final subcluster, CSC- 39, consisting of the relatively tightly connected Aporus seae PA 46- A mos ia 201080 by with CSC- 13, but is otherwise каша isolat ed. In addition to the SI analysis using both architectural and cuticular characters, I also per- formed an analysis using only the 26 architec- tural characters in order to examine their power to group the OTUs. The results were similar in oth cases. Inclusion of the genera with phyllan- thoid branching introduced convergence that muddled the clusters, so I again deleted them. The same major clusters formed in approxi- mately the same order, but tended to be some- what less tightly connected internally and to have more connections to other clusters. This was par- ticularly true of CSC-2, which consists of genera Webster (1975) placed in the Phyllantheae and the Wielandieae. CSC-39, the Aporuseae and Uapaceae, appeared more similar to the Anti- desmeae when only architectural characters rath- er than all the foliar characters were used. Cleis- tanthus, with Dicoelia and Pentabrachium, remained more distinct from the Antidesmeae, so that at the conclusion of the clustering, CSC- 13 of Figure 6 would have been represented bet- ter by two circular subclusters. Thus the inclu- sion of cuticular characters generally improved the resolution of the groups but may have caused a few to appear more similar than they otherwise 94 ANNALS OF THE MISSOURI BOTANICAL GARDEN Aporuseae S 2.86222 L32-37 9 SMC *Cleistanthus p.p. FIGURE 5. might. I attempted no analysis using only cutic- ular characters because too few characters were involved DISCUSSION COMPARISON WITH PREVIOUS CLASSIFICATIONS The groups that formed during SIMGRA anal- ysis of leaf characters alone correspond remark- ably well to taxa proposed by other systematists, particularly Webster (1975), whose use of differ- ent sources of systematic information was the most comprehensive. The large subclusters in Figures 5 and 6 are almost equivalent to ei ma- ably the Antidesm meae, Aporu e, nd Spon thus, connect late in the analysis, corroborating their distinctiveness. Furthermore, putative re- lationships between taxa are frequently reflected [VoL. 73 Drypeteae Wielandieae Clusters present in SIMGRA analysis at level 37 or similarity value 0.86222. by the degree of connectedness between the cor- in the cases of the Antidesmeae and the Aporuseae, Spondianthus and the Antidesmeae, and between the Phyllan- theae and the Wielandieae, and the relatively isolated position of the Drypeteae. A one-to-one correspondence does not exist between the SIMGRA results and any previous classification, however, and there are several no- table discrepancies. One of the most obvious of and two more joined the subcluster soon there- ае. (Fig. 5). The two remaining OTUs, sects. Chartacei (20) and Australes (23), remained SMCs until late in the analysis, and even then did not form connections with the Cleistanthus subclus- ter. Chartacei formed connections with eight OTUs, five of which are in CSC-2 and none of 1986] Aporuseae & Uapaca Didymo- 34 cistus 2c " S>.81333 L38-45 OSMC Celianella 35--A. 91 Bischofia Phyllantheae & Wielandieae *Cleistanthus p.p. 32 31 56 .........Атапоа LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. II. 95 Meineckia & 22| Pentabrachium ^ 10 `... Drypeteae ә Wielandieae %e *e е ..5 Discocarpus FIGURE 6. Clusters present in SIMGRA analysis at level 45 or similarity value 0.81333. which are in Cleistanthus. Australes are even more distant, forming a single connection to Actephila at level 38. Jablonsky (1915) felt that Australes, although clearly in Cleistanthus, are rather iso- lated within the genus and also noted that the leaves of C. cunninghamii, the only species in Chartacei I examined, are not typical of the sec- tion. Apparently leaves of these two taxa exhibit sufficient non-divergent change that overall sim- ilarity poorly reflects relationship. In most systems Cleistanthus and Bridelia (14— 16) are considered to be closely related because they have very similar flowers with valvate se- pals (Jablonsky, 1915) and similar pollen (Punt, 1962; Kohler, 1965). Webster (1975), on the ba- sis of the pollen evidence, placed the two genera in their own tribe near the Wielandieae. Phenetic analysis of the foliar characters considerably sep- arates the two genera. Most members of Cleis- tanthus were closely allied with Dicoelia and the Antidesmeae, whereas Bridelia remained fairly isolated and formed only a single connection with Cleistanthus at the final level of similarity. Bri- delia leaves are extremely specialized within the Phyllanthoideae (Levin, 1986), and therefore ap- pear isolated in a phenetic analysis. Cladistic analysis, which de-emphasizes the unique char- acter states (autapomorphies) of Bridelia, sup- ports the relationship between the genera (Levin, in press). Another difference from the classifications of Webster and others lies in the assignment of gen- era to the Wielandieae and the Phyllantheae. In particular, leaves of Astrocasia (2) and Hey- woodia (3), genera usually included in the Wie- landieae, are more similar to leaves of the Phyl- 6 ler (1965) remarked on the close resemblance 96 ANNALS OF THE MISSOURI BOTANICAL GARDEN among the pollen of these taxa. Pollen and other reproductive structures of Heywoodia are less similar to the Phyllantheae. The main reason for retaining Astrocasia and, to a lesser extent, Hey- woodia in the Wielandieae is their petaliferous flowers, a characteristic that, because it is prim- itive ha the subfamily (Webster, 1967), might not necessarily be expected to reflect phylogenetic os Cladistic analysis of the foliar characters in fact clearly associates Astrocasia with the Phyllantheae 2 retains Heywoodia in the Wielandieae (Levin, i Meineckia (58), е, їп de lesa by Webster (1965, 1975), connected with the bulk of Cleistanthus late in the SIMGRA analysis (Figs. 5, 6). Associated with the herbaceous tendencies of Meineckia is a reduction in the organization of its leaves (Levin, 1986), which obscures its relationships. A similar phenomenon probably explains the unexpected connection at level 43 between Celianella (35), a member of the An- tidesmeae (Jablonsky, 1965; Webster, 1975), and Zimmermannia (60) of the Phyllantheae (Fig. 6). The leaves of Celianella are quite thick and have unusual reduced venation (Levin, 1986). In both of these cases, cladistic analysis of leaf characters yields results that are much closer to the classi- fications of Webster (1975) and others (Levin, in press). Another difference between the SIMGRA re- sults and Webster’s (1975) classification involves the treatment of Amanoa (10), Actephila (11,12), and Croizatia (93), which Webster included in their own tribe, the Amanoeae. At similarity level 45, the final level shown in the SIMGRA analysis (Fig. 6), Amanoa simultaneously formed con- nections with Wielandia (1) and Jablonskia (56), and at only slightly lower similarity levels ad- ditional connections would form between 4ma- noa and other OTUs in CSC-1, containing genera in the Wielandieae, but not with either Actephila or Croizatia. Thus Amanoa appears most closely related to the Wielandieae, a relationship also suggested by pollen morphology (Punt, 1962; Kohler, 1965) The leaves of Actephila (11, 12), in contrast, more closely resemble those of Andrachne (26, 27). The two genera first connected at level 24 (Fig. 3), and, by the end of SIMGRA analysis, they formed a tightly connected and rather iso- lated subcluster (Fig. 6). Both Punt (1962) and Köhler (1965) emphasized the similarity be- tween the pollen of Andrachne and Actephila. Again, although SIMGRA analysis of the foliar [Vor. 73 morphology contradicts previous classifications, the concordance with pollen implies that leaf re- sults may better reflect actual relationships. Croizatia (93) has similar fruits to Actephila (Steyermark, 1952) but has very different leaves (cf. Fig. 6). Until Croizatia is better known, its relationships will remain uncertain In addition to its implications regarding the relationships of Actephila, the constitution of CSC-11 is significant also in that Andrachne sect. di poda (27) has been placed in Savia (6 7; CSC-1) by many authors (e.g., Pax & Hoff- man, 1922), although palynology (Punt, 1962; Kóhler, 1965) and some aspects of floral mor- phology (Webster, 1967) suggest that this section is in fact better placed in Andrachne, in accord with the leaves. The most striking feature uniting the leaves of different Andrachne sections is their anisocytic stomata, which is a derived condition 39, Fig. 6). As I discussed previously (Levin, 1985), floral morphology, palynology, and wood anatomy, which have been the principal sources of taxonomic information in the Phyllanthoi- deae, have little to offer toward clarifying the relationships of Uapaca because it has so many unique specializations. However, in SIMGRA analysis, as in cladistic analysis (Levin, in press), both the totality of foliar characters and archi- suggest that the other organs of the genus be re- examined with this possible relationship in mind. Equally novel is the suggestion (Fig. 6, CSC- 45) of relationships between Didymocistus (45), Hymenocardia (89), and Neowawraea (90). Webster (1975), following other authors, placed these genera in three different tribes: the Apo- ruseae, the monotypic Hymenocardieae, and the Drypeteae, respectively. Both Didymocistus and Neowawraea have leaves that would be quite anomalous in the tribes to which Webster re- ferred them (Levin, 1986). The high degree of organization of the venation makes all three gen- 1984). The palynological literature has little to offer, because Hymenocardia has very unique pollen and the other two genera have not been examined. 1986] Finally, Webster (1975) excluded Martretia кеш the Phyllanthoideae, and indeed from t , although previous au- thors had placed iti in the Antidesmeae. Its fruit and pollen characters are indeed unique in the family (Levin, 1986), and because it is one of the last OTUs to join a cluster in SIMGRA analysis, Webster may have been correct to remove Mar- tretia from the Phyllanthoideae, if not also from the family POTENTIAL FOR USE IN IDENTIFICATION Both Dolph (1976) and Hill (1980) had as their primary goal identification of fossil leaves, not classification of an extant group. Both found that phenetic analysis of their character sets generally associated different leaves of the same species but led to the recognition of no higher taxa. Their character sets tend to emphasize aspects of size nd shape, most of which I found to be relatively invariable at the species level within the Phyl- lanthoideae, but highly variable within genera and higher taxa (Levin, 1986). However, given an unknown leaf from a member of the Phyllan- thoideae, the characters I used would probably associate it with the correct genus or higher tax- Two problems arise using these characters for the identification of fossil leaves. The first is the obvious one: How do we know whether a par- ticular fossil leaf belongs to the Phyllanthoideae? As I discussed in another paper (Levin, 1985), characters that neither Dolph, nor Hill, nor I included in our character sets may help narrow the choice of higher taxa. An example of such a character is the type of marginal toothing that Hickey and Wolfe (1975) found to be quite char- acteristic of families and higher taxa. Dolph (1976) and Hill (1980) excluded tooth type from their character sets to avoid statistically weight- ing the analysis in favor of toothed leaves at the expense of entire-margined leaves. I did not in- clude tooth type in my s because it is in- variable within the Phyllanthoideae; for the same reason I did not include other characters, e.g., pinnate versus palmate venation, that may help separate the leaves of the Phyllanthoideae from those of other taxa. Thus the placement of leaves in higher taxa requires not phenetic analysis us- ing equally weighted characters, but the recog- nition of diagnostic characters that delimit large groups. Hill (1980) noted the same problem. Clearly, detailed examination of many more ex- LEVIN —PHYLLANTHOIDEAE FOLIAR MORPHOLOGY. II. 97 ant groups, coupled with cladistic analysis to identify diagnostic characters in those groups like the Phyllanthoideae that have relatively non- descript leaves, will be necessary before we can accurately relate many fossil leaves to leaves of their extant relatives. The way SIMGRA analysis handled Martretia offers some indication of the way in which phe- netic analysis might treat taxa not initially as- signed to the correct higher taxon. Because of its substantial differences from other members of the Phyllanthoideae (although more in the com- bination of character states than in any particular character), Martretia joined late in the analysis. However, it still connected earlier than such aberrant but bona fide Phyllanthoideae as Ama- noa, m Discocarpus, Leptonema, and Spon he sec nd problem, which was extensively addressed E both Dolph (1976) and Hill (1980), involves missing data. Many fossil floras include leaves that lack well-preserved higher order ve- nation and/or cuticle. That the SIMGRA anal- ysis of architectural characters alone yielded clusters that were less well resolved but substan- tially similar to clusters that formed using the could be expected to decrease considerably. CONCLUSIONS Phenetic analysis using leaf architectural and cuticular characters in the Phyllanthoideae pro- duces groups similar to those in classifications based on characters more widely recognized as being of systematic value, such as floral mor- phology, palynology, and wood anatomy. Most exceptions appear not to be cases in which the e classifications are invalid. A similar but less well resolved clustering results from analysis of ar- chitectural characters alone. That numerical analysis yields such clear results ina group p pat known for having diagn the widely held idea ‘that angiosperm leaves are so evolutionary and environmentally plastic that they are of little systematic value. At least in the Phyllanthoideae, the amount of parallelism and convergence in leaf architectural and cuticular characters is small enough that overall similarity 98 ANNALS OF THE MISSOURI BOTANICAL GARDEN usually reflects presumed evolutionary relation- ship. In a few cases, for example Meineckia, Celi- anella, some sections of Cleistanthus, and the phyllanthoid-branching genera in the Flueggei- nae, excessive convergence does cloud relation- ship. In these cases, I feel that it is not so much a fault of the characters as it is the inherent weak- ness of systematic methods based on total sim- ilarity. The results of cladistic analysis of the same data even more closely resemble previous classifications (Levin, in press). The greater abil- ity of cladistic methods to discriminate эга апа to factor out autapomorphies коре the effects of these problem Because this is the a report of numerical analysis of the foliar morphology of a related group of flowering plants, the need for more stud- ies remains. The characters employed were de- rived from systems developed for the description of fossil and modern angiosperms (Stace, 1965; Hickey, 1973) and are therefore widely appli- cable. Further examination and analysis of ex- tant angiosperms should help us to understand better the patterns of leaf evolution in the flow- ering plants and to classify more confidently the leaves of fossil plants. LITERATURE CITED DiLcHER, D. L. 1974. Approaches to the identifica- tion of angiosperm leaf remains. Bot. Rev. (Lan- T. 76. Taximetric partitioning of leaf collections Palaeontographica, Abt. B, Paláophy- tol. —86. Bo T. 1980. A taxonomic study of the Ra- nunculus hispidus Michaux complex in the West- ern Hemisphere. Univ. Calif. Publ. Bot. 77: 1- 125. EsTABROOK, G. F A mathematical model in graph theory for biological classification. J. Theor. Biol. 12: 297-310. . ROGERS. 1966. A general method of taxonomic description for a computed similarity measure. BioScience 16: 789-793. HAYDEN, W. J. & D. S. BRANDT. 1984. Wood anat- omy and relationships of Neowawraea (Euphor- Bot. 9: 458-466. 73. Classification of the architecture of dicotyledonous leaves. Amer. J. Bot. 60: 17 33. & J. A. WoLFE. 1975. The bases id angio- sperm phylogeny: vegetative morphology. Ann. Missouri Bot. Gard. 62: 538- ides Y » ta ae taxonomic a tudy osperm leaves. Bot. Gaz и 141: gti 229. [VoL. 73 HUTCHINSON, J. 1969. ee in the family Eu- phorbiaceae. Amer. t. 56: 738-758. JABLONSKY, тк тей ceae-Phyllanthoi- de ac- Bridelieae. шо, IV, 147-VIII(65): ES 965. Euphorbiaceae. Pp. 150-178 in B. Ma- ers (editor), The Botany of the Guyana High- land. Part IV. Mem. New York Bot. Gard. 12(3): „1965. 1-285. "m E Die Pollenmorphologie der bio- aic as. Grana Palynol. 6: 26-120. & D. J. ROGERS. Da P. Вос 1972. pe esos же clustering in ta a synthesis metric arcu a : 567 Levin, G. A. matic foliar morphology o of Phyllanthoideae стари ае). І. Conspectus. ан гі Bot. Ga ae ———. In press. System : ae morpholog of Phyllanthoideae (Euphorbiaceae) III. Cladistic analysis MENNEGA, Wood structure of Ja- M. 4. blonskia congesta (Euphorbiaceae). Syst. Bot. 9: 236-23 о R. & Г. CHALK. 1950. Anatomy of the Dicotyledons, 2 volumes. Oxford Univ. Press, Ox- ord. Pax, Е. & K. HOFFM 1922. Euphorbiaceae-Phyl- lanthoideac-Phyllantheae. Pflanzenreich IV, 147- J. ROGE RS & F. WHITE. 1969. А giosperm family: generic delimitation in the Chry sri aime New Phy- tol. 68: 1203-1234. Punt, W. 1962. Pollen morphology of the c biaceae with special reference to taxonomy. W tia 7: 1-116. ROGERS, D. J. & Н. S. FLEMING. 1973. A monograph of Manihot esculenta Crantz with an EEE the taximetric methods used. Econ. Bot. 27: 1- Sace C. A. 1965. Cuticular studies as an aid to н axonomy. Bull. Brit. Mus. (Nat. Hist.) Bot. 4: 1- STEYERMARK, J. A. 1952. Botanical exploration in Venezuela —II. Fieldiana, Bot. 28: m 447. WEBSTER, "б. L. 1956. phi see of the West Indian n of Phyllanthus [Parts 1 -3]. J. 7: 91-122, 217-268, 340-359. 57. ү monographic study of the West In- dian species of Phyllanthus [Parts 4-6]. J. Arnold Arbor. 38: 51-80, 170-198, 295-373. A monographic study of the West In- dian species of Phyllanthus [Parts 7, 8]. J. Arnold Arbor. 39: 49-100, 111-212. A revision ofthe genus Meineckia (Eu- phorbiaceae). Acta Bot. Neerl. 14: 323-365. 1967. The genera of Euphorbiaceae in the southeastern United States. J. Arnold Arbor. 48: 0 75. Conspectus of a new classification of the arcae Taxon 24: 593-601. WIRTH, M., BROOK & D. J. ROGERS. pia A graph theory ode for systematic biology, wi an example for the Oncidiinae (Orchidaceae). iiim Zool. 15: 59-69. NOTES ON THE FLORAL BIOLOGY OF COUROUPITA GUIANENSIS AUBL. (LECYTHIDACEAE)' SONIA YARSICK, NERIDA XENA DE ENRECH, NELSON RAMIREZ, AND GETULIO AGOSTINI? Floral biology has been studied in relatively few Lecythidaceae, despite their ecological im- portance in the Neotropics. Only a few authors have dealt with the subject, notably Jackson and Salas (1965) on Lecythis ellipitica H.B.K., Dias (1967) on Bertholletia excelsa Humb. & Bonpl., Mori and Kallunki (1976) on Gustavia superba (Kunth) Berg., Prance (1983) on Eschweilera garagarae Pittier, and Prance (1976) and Mori et al. (1978) on pollination and androecial struc- ture in the family as a whole. Ormond et al. (1981) published a contribution to the floral bi- ology of Couroupita guianensis after all of our field observations were completed and a draft of this paper was submitted for publication. Prance (1976) and Mori et al. (1978) pointed out that the androecial zygomorphy and the position of the hood (adpressed to the ring) in Couroupita subsessilis Pilg. suggest that the genus Couroupita may be intermediate in the family with respect to androecial evolution. Few studies of anthesis behavior in tropical plants, including Lecythi- daceae, have been carried out. Notably, Mori et al. (1978) reported diurnal anthesis and shedding of stamens during the late afternoon for Lecythis amara Aublet (= L. alba Mori, nom. nud.), and asynchronous anthesis during the morning for Eschweilera longipes (Poit.) Miers.; the latter of which sheds its androecia the following day. Dif- ferential behavior between pollen produced in ring anthers and that found in hood anthers is documented by Mori and Orchard (1979), Mori et al. (1980a), and Ormond et al. (1981). Floral visits by potential pollinators, such as wasps and small bees, have been reported by Prance (1976), Mori et al. (1978), and Ormond et al. (1981). METHODS AND RESULTS Cultivated individuals of Couroupita guianen- sis Aubl. were studied at the Jardin Botanico in Caracas, Venezuela, located near the limit of nat- ural distribution of the species. Observations were recorded during two days in November 1979, and again for four days in June 1980, although the observed individuals flower throughout the year. The flowers of Couroupita guianensis are very fragrant and attract many insect visitors. The petals are yellow both on the exterior and the interior, with rose to red-rose margins. The an- droecial column and filaments of the hood sta- mens are lilac, with the tip of the anthers yellow and the ring stamens white. A small amount of sticky secretion is present on the stigma. Anthesis is diurnal and asynchronic, the flowers beginning to open gradually between 7:00 and 8:30 A.M., with a peak around 8:00 A.M. (Fig. 1). The ring anthers open simultaneously with those on the hood, shortly after anthesis, but they retain pol- len longer during the day (Fig. 2). The number of open flowers presenting pollen reaches a peak around 9:00 A.M Pollen in Couroupita is dimorphic; grains pro- duced by anthers in the hood are released in tetrads and as such are clearly larger than the simple grains produced in the ring anthers. Fur- thermore, the hood anthers themselves are larg- er, containing an average of 2,850 tetrads, as compared with approximately 450 grains per ring anther. Tests for pollen viability show no ger- mination of hood pollen in sacharose, as previ- ously reported by Mori et al. (1980b) and by Or- mond et al. (1981). However, Thompson (1921) reported that hood pollen was as fertile as ring pollen. Hood pollen stained well using cotton blue in lactophenol and it is reported by Ormond et al. (1981) as having significant protoplasm in 88% of pollen grains tested using aceto carmine These reports raise doubts as to the reliability al both staining methods for assessing pollen via- bility. ! We appreciate the use of facilities at the Jardin Botanico and Instituto Botanico (MARNR) in Caracas. We wish to mention the valuable collaboration of the 1980 course in ra es the help э Diaz (Technician), Mr. Sixto Garcia (Artist), and Lic. Celia Gil. We thank G. T. Pra Me iret adl York Botanical Garden, and Stephen S. Tillett ofthe о V. nies Ovalle Facultad de Же жеу Universidad т de Venezuela, for reading and correcting the riginal m 2 Universidad Central de Venezuela, Facultad de Се. Escuela « de Biologia, Departamento de Botánica, Apartado 47850, Caracas 1041A, Venezuela. ANN. Missouni Bor. GARD. 73: 99-101. 1986. 100€ Non cumulative 90] O Cumulative 80- 70- 60- % OPENED FLOWERS C P 7777] ОТОО 0800 09:00 TIME (h) Anthesis behavior in Couroupita gui- ТА 10:00 FIGURE 1 Insects captured visiting flowers were bees be- longing to Apis mellifera, Bombus sp., Trygona sp., and Xylocopa frontalis (Oliver), the wasp Po- lybia, and the flower fly Ornidia obesa F. (Dip- tera, Syrphidae). More frequent visits were made by the larger insects, such as individuals of the bee genera Bombus and Xylocopa, which have been reported as visitors to Lecythidaceae, in- cluding species of Gustavia, Eschweilera, Cou- ratari, and Couroupita guianensis. Less frequent visits were made by the smaller insects, such as —— Hood pollen dcus Ring pollen % FLOWERS WITH POLLEN ош o 2 "ER 10 a `0 Y Li LI T T 7:00 9:00 1:00 13:00 15:00 17:00 TIME (h) F Comparison of behavior of anther de- hiscence and pollen extinction between the hood and the ring in flowers of Couroupita guianensis ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Madina] } > : г A in Couroupita guianensis. species of Apis, Trigona, and the wasp genus Po- lybia, which have been observed in numerous Lecythidaceae, including Couroupita subsessilis, a species which has much smaller flowers (Prance, 1976). Thus, Couroupita guianensis differs from C. subsessilis, as well as other Lecythidaceae, in atracting both small and larger insects, including species of the mentioned genera. The hood anthers of C. guianensis provide a convenient landing platform for these floral vis- 8. cium facilitates access to insects of various sizes. While collecting hood pollen, larger bees rub their dorsal areas against the ring anthers and the stig- ma, detaching several ring anthers in the process. This behavior assures transfer of ring pollen to the stigma. Apparently hood pollen may also reach the e however, because insects were observed to turn over completely within the flower. E of Trigona forage for pollen e A E AN > / \ 2 a \ a \ " \ 4 \ о Я й 07:30 08:30 0930 1030 Изо 12:30 1330 1430 19:30 TIME (h) FIGURE 4. Bombus and Xilocopa visits recorded, as an average of five days, in flowers of Couroupita guianensis 1986] on ring anthers during long periods, probably affecting pollination, whereas other small floral visitors wander throughout the flower in such a manner that only their ventral surfaces contact the anthers, and pollination thus occurs only ac- cidentally. Data on frequency of floral visits by individ- uals of Bombus sp. and Xylocopa frontalis were recorded (Fig. 4). Visits began at anthesis and continued throughout the day, with peaks oc- о . В species tended to visit predominantly in the morning, while Xylocopa were more common in the early afternoon. DISCUSSION Our findings on pollen fertility and pollen be- havior confirm that pollination is achieved only with ring pollen. However, hood pollen, which should be rich in nutrients, as suggested by its living protoplasm, provides only food for the pollinators, and plays an important role in the process of pollination, as was previously stated by Ormond et al. (1981 Our observations for Couroupita guianensis as compared to reported data for C. subsessilis (Prance, 1976) suggest that differential use of pol- linators may be important in the evolution and maintenance of species within the genus. It is interesting to note that individuals of Bombus and Xylocopa have been seen on putatively prim- itive Lecythidaceae such as Gustavia, as well as on more advanced taxa such as species of Cou- ratari. It is therefore not surprising to find these larger floral visitors on Couroupita, whose an- droecial structure is apparently intermediate within the family. The asynchronous behavior of anthesis ob- served in C. guianensis may tend to compensate for the short life of the flowers of this species and could be viewed as a way to promote cross pol- lination by means of prolonging pollen presen- tation for a period of nearly two hours. This hy- pothesis is supported by the distribution of visits by insects: most visits were recorded during YARSICK ET AL.—FLORAL BIOLOGY OF COUROUPITA 101 morning hours, at a time when a maximum num- ber of anthers is open. The recorded frequency of floral visits, which shows that Bombus species tend to visit predom- inantly in the morning whereas Xylocopa were more common in the early afternoon, may doc- ument a case of temporal resource partitioning by the insects to reduce competition for pollen. LITERATURE CITED Dias, D. P. DES. 1967. Polinacào de Castanheira de Para por agentes naturais. Contr. do IPEAN a I . SALAS. | Insect visitors of os elliptica H. В. К. J. Agric. Univ. Puerto Rico 49: 133-1 Мон, S. ‚ А. KALL UNKI. 1976. Phenology and floral biology of Gustavia Т (Lecythidaceae) in Central Panama. Biotr 8 & J. E. ORCHARD. 1979. уе biologia floral e evidencia sobre dimorfismo fisiologico do pollen de Lecythis pisonis Canbess су ceae). Anais So 16. os. 1980a. Observacoes sobre a fenologia e biologia floral de Lecythis pisonis Y dien (Lecythidaceae). Theo- broma 10: 103-11 LE: iue. M T. PRANCE. o In- trafloral pollen differentiation in the New World Lecythidaceae, subfamily Е ial Science 209: 400-403. G А. В. BoLTEN. 1978. Ad- ditional Notes on the floral biology of Neotropical Lecythidaceae. Brittonia 30: 113-130. ORMOND, W. T., M. C. B. PINHEIRO & A. R. C. DE CasrELLS. 1981. A contribution to the floral bi- ology and reproductive system of Couroupita gui- anensis by Hee man Ann. Missouri Bot. Gard. 68: 5 PRANCE, G. T. oa The pollination and androphore structure of some T ian Lecythidaceae. Bio- tropica 8: 235-24 1983. qo AR de polinización de Esch- weilera garagae Pittier en el Chocó, Colombia. Mutisia - ‚ 1977. What is Lecythis? Taxon 1979. Lecythidaceae. Jn Flora ге 21: 1-270. THOMPSON, J. M. 1921. Studies in floral morphology II. The staminal zygomorphy of Couroupita gui- anensis Aubl. Trans. Roy. Soc. Edinburgh 53: 1- 15. CONVERGENT EVOLUTION OF THE ‘HOMERIA’ FLOWER TYPE IN SIX NEW SPECIES OF MORAEA (IRIDACEAE-IRIDEAE) IN SOUTHERN AFRICA! PETER GOLDBLATT? ABSTRACT Six new species of Moraea, all with similar small, pale to deep blue-purple flowers and reduced style branches instead of the broad petaloid style branches typical of the genus, are described from the Southwest and interior west coast of southern Africa pical of the sou only M. pseudospicata appears related to M. crispa. The others differ either in vegetative morphology, chromosome number, or in details of the flower, and the unusual flower seems to have evolved independently in at least three of them. Moraea graniticola, M. hexaglottis, and M. rigidifolia are each radi from single populations in the southern Namib a. Moraea worcesterensis, known from one site near Vidas in the southwest Cape, is probably piene related to M. algoensis of sect. Vieusseuxia sect. Mora Desert of Namibia and are referred to aea deserticola, restricted to the Knersvlakte in southern Namaqualand, is allied to M. speciosa of E Polyanthes Moraea is a widespread African genus of some 115 species of seasonal, corm-bearing perennials of the Old World tribe Irideae of the large and nearly worldwide subfamily Iridoideae. It is the major genus of subtribe Homeriinae, an alliance centered in southern Africa but extending through tropical Africa into the Mediterranean and Mid- dle East. The alliance is d by having an astelic corm of a single internode and dis- tinctive, secondarily bifacial us Mig iso- bilateral leaves are basic in Iridaceae). Moraea is the largest of the eight ei recognized genera of the subtribe and occurs almost throughout sub-Saharan Africa, but species are concentrated in highland areas of southeast trop- ical and southern Africa and in the winter rainfall mically, (Goldblatt, a о Ee Despite inten- sive study, new taxa continue to be discovered, mainly in the winter rainfall areas of South Africa and recently in southern Namibia. Some of these were described in 1982 together with a synopsis of the genus in which 105 species in five sub- genera and 12 sections were recognized (Gold- blatt, 1982). Several new species have been discovered since 1982, notably along the arid west coast and in- terior of southern Africa. The low and variable rainfall of the region is insufficient for species to flower every year, and this combined with the rugged landscape and general inaccessibility leaves this area relatively poorly known botan- ically. Six new species are described in this paper. Moraea graniticola, M. hexaglottis, and M. ri- gidifolia are each known from single populations in the southern Namib Desert of Namibia and are referred to sect. Moraea. Moraea worcester- ensis, known from one site near Worcester in the southwest Cape, is probably closely related to M. algoensis of sect. Vieusseuxia. Moraea deserti- cola, restricted to the Knersvlakte in southern Namaqualand, is allied to M. speciosa of sect. Polyanthes, while M. pseudospicata, from the Nieuwoudtville escarpment, is closely related to the widespread Karoo species, M. crispa, also sect. Polyanthos. П six share an unusual feature, a type of flow- characteristic of the related genus Homeria (Goldblatt. 1980, 1981b) in which the style arms are narrow and inconspicuous and the tepals are similar in color, shape, and disposition. The ! Supported by grants DEB 78- idi and 81-19292 from the United States National Science Foundation. I d fo thank Neil MacGregor, from Glenlyon, Nieuwou Jo tville, and assistance while in the field; and provided living and preserved Or in hospitality mib species d Dee Snijman, from Compton Herbarium, Kirstenbosch, who rediscovered and drew my attention an to the жын overlooked ана pseudospicata and assisted me іп recollecting it. I also acknowledge with thanks the drawings of Margo 2B. A. Kruk off dise of African Botany, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. ANN. MISSOURI Bor. GARD. 73: 102-116. 1986. 1986] flowers of the new species are similar also in being blue-purple, and, at least in four of them, small and extremely fugacious, opening in the late afternoon and fading at or shortly after dusk. This type of flower was previously known in a few taxonomically isolated species of Moraea. Despite their similarities, the flower types prob- ably evolved independently in at least three of them. The adaptive significance of this flower is discussed after an outline of the morphology of the Moraea oe ne flower and a review of the new spec The as of the three new species from Namibia more than doubles the number of Mo- raea species recorded in this country from the two admitted by Sólch (1969), M. polystachya and M. namibensis (treated as M. edulis by Sólch). Moraea carsonii has also now been found in the northeast of Namibia bringing the total species of Moraea recorded there to six, of which three genus in Namibia, both from the extreme south of the country, at present awaiting description when adequate type material can be assembled. MATERIALS AND METHODS CYTOLOGY The chromosomes of five of the six new species were studied. Live material of Moraea deserti- cola was not available. Methods used are the same as those employed in similar studies in Iridaceae (Goldblatt, 1979, 1980). Root tips of sprouting corms were routinely harvested in mid- morning, pre-treated in 0.002 M hydroxyquin- oline for seven to eight hours, then fixed in 3:1 absolute ethanol:glacial acetic acid for a few minutes and stored in 70% ethanol. They were hydrolyzed for six minutes in 10% HCl warmed to 60°C, then rinsed in water and squashed in FLP orcein (Jackson, 1975). TAXONOMY Live material of all species except Moraea de- serticola was examined carefully, and the illus- trations were made from living plants. Only cul- tivated plants of M. graniticolaand M. hexaglottis were seen. Herbarium material from important southern African collections was also examined. Measurements were made from live plants ex- cept for M. deserticola. Delicate parts of the flow- ers probably shrink some 10-1596 and the mea- GOLDBLATT—NEW SPECIES OF MORAEA 103 surements for M. deserticola apply only to dry plants. Flower color fades progressively in dry specimens, eventually changing completely, earing. dE and are frequently mentioned by col- lectors. Although only two sheets each of M. graniti- cola, M. hexaglottis, and M. deserticola were available, it seems better to describe these species now because they are relatively rare and flower irregularly, so the chance of obtaining more ma- terial is small. Also describing these species now is likely to stimulate further plant exploration in the dry interior Namib Desert where M. grani- ticola and M. hexaglottis occur. Material examined is cited according to the grid reference system based on geographical de- gree coordinates of latitude and longitude in cur- rent use in southern Africa (Edwards & Leistner, 71) CvTOLOGY The three Namib Desert species have 2n = 20 and a similar karyotype (Fig. 1) with four long acrocentric chromosome pairs, one of which has oraea and is believed to be basic М И 1971, 1976a). Moraea pseu- dospicata and M. worcesterensis have 2n = 12. The karyotype of Moraea pseudospicata cor- responds well to those of other members of sect. Polyanthes in consisting of relatively large ac- rocentric to submetacentric chromosomes (Fig. 1D, E). Satellites are evident on the distal arms of a long and a short acrocentric chromosome pair. Satellites have been recorded in this posi- tion in some populations of M. crispa and also on the end ofthe long arm ofa long chromosome pair (Goldblatt, 1980; Fig. 1F) and in M. poly- anthos of this section. The satellite position in other species of the section such as M. bipartita corresponds with that of M. pseudospicata. Moraea worcesterensis has a long and a short pair of metacentric chromosomes and a long pair of strongly acrocentric cl with a smal satellite at P end of the short arm (Fig. 1F). This karyotype is typical of many species of sect. Vieusseuxia (Goldblatt, 19768, and in prep.), and ds well with the placement of this species in sect. Vieusseuxia. 104 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Sez, >» = = “e, CF, <>) : SA S {т } CTS Ф FIGUR for com M. worcesterensis. (All to same scale. E]. Mitotic metaphase figures. MORPHOLOGY THE BASIC MORAEA FLOWER The typical and apparently basic Moraea flow- er has petal-like, flattened and broad style branches (Fig. 2A) that diverge from the style just above the apex of the filament column. Each style branch has a broad, transverse stigmatic lobe on the upper abaxial surface, and above the stigma it continues as a pair of flat erect ap- pendages, the style crests. These elaborate style branches are like those found in the related /ris and Dietes, the latter believed to be ancestral to both Moraea and Iris (Goldblatt, 1981a). The Moraea flower is Iris-like also in having each stamen appressed to the opposed style branch and concealed by the claw of the outer tepal. The outer tepal is somewhat to much larger than the inner and always has a nectar guide at the base of the limb and a well-developed nectary at the base of the claw. The outer and inner tepals may be similarly oriented or the inner may be erect, or somewhat to very reduced, and then trilobed, aristate, ciliate, or completely lacking. In most species of Moraea the filaments are united in the lower half into a column around the style, but at least in subg. Visciramosa and in M. ramosis- parison (count but not figure reported in Goldblatt, 1976c). — А. Moraea hexaglottis. — B. M. rigidifolia. —C. Barnardiella Ar —D. M. crispa.—E. M. pseudospicata. — A1 4 11 sima (subg. Moraea) tl but entirely e THE HOMERIA FLOWER In Homeria (31 spp.) the flower is an appar- ently simple structure (Fig. 2B, D, E). It com- bines the derived features of united or nearly united filaments and broad tepal claws, which form a cup-like structure including or enclosing all or part of the stamens, as contrasted with apparently unspecialized subequal, similarly dis- posed inner and outer tepals, and style branches that either lack or have weakly developed paired crests. In addition both whorls of tepals have nectar guides at the base of the limbs and both have nectaries at the base of the claws. In Home- ria the flower is known to be adapted for fly pollination (Goldblatt, 1981b). Several members of the genus are visited by species of Diptera and not by bees, which pollinate the flowers of most species of Moraea. However, some species of Homeria, in which the tepal claws are very short and the anthers are held well above the tepals, "1 adapted to bee p (Gold- blatt, 1981 Until кесе (Goldblatt, 1976a, 1980) all species of Homeriinae with the type of flower 1986] GOLDBLATT—NEW SPECIES OF MORAEA 105 FIGURE 2. Floral features of species of Moraea and Homeria. — A. Moraea hic as ки ое flower typical of Moraea, the style branches and stamens ages separately and enlarged. — B. H atens, a typical Homeria. — C. M. crispa, with a flower modified in a manner similar to cae the sole a and stamens much enl d. —D. Style branches and енй only of Homeria elegans.—E. Н. bi —F. the panied genus кош especially for comparison with М. hexaglottis. Whole flowers life size; аван style branches and stamens, x2. 106 described above were assigned to Homeria de- spite a lack of uniformity in details of stem, leaf, and corm tunic morphology. Then as a result of a study of chromosome cytology and crossing relationships of the Homeriinae (Goldblatt, 1976a, 1980), it became clear that Homeria as defined by its characteristic flower alone was ar- tificially constituted and polyphyletic. Informa- tion on cytology, interspecific hybridization, and vegetative structure resulted in the transfer of species with n = 6 and karyotypes matching those found in Moraea sect. Polyanthes to that section; the species with n = 10 were assigned to a new genus Rheome that was assumed to be derived from some extinct or at least not yet identified ancestor in Moraea. THE NEW SPECIES The Namib species and Moraea pseudospicata resemble the widespread Karoo M. crispa (Fig. 2C) in their small, pale to deep blue-purple flow- ers with short style branches, partly free fila- ments, and short tepal claws that form a cup including the base of the filament column. The flowers of all these species are extremely fuga- cious and open after 3 P.M. (even up to 5 P.M.) and fade soon after dark. Only M. pseudospicata, with n = 6, corresponds with M. crispa in chro- mosome number and karyotype, and it seems likely that the two are closely allied. It can be distinguished from M. crispa by its sessile lateral inflorescences, large corms with accumulated tunics, and small, globose, irregularly dehiscing capsules. The three Namib species, all with л = 10, are regarded as unrelated to M. crispa because of the difference in karyotype and are presumed to be allied to species of sect. Moraea (or possibly the closely related sect. Deserticola), which also have and a similar karyotype of four to five long acrocentric pairs and five to six much smaller pairs. Moraea rigidifolia has a thick, relatively short and rigid leaf, sessile lateral inflorescences, 3); and M. hexaglottis has short deeply divided style branches, the arms of which extend hori- zontally below the anthers. The morphological differences between the three Namib species sug- gest that they are not immediately related to one another. Moraea rigidifolia shares its terete leaf and sessile lateral inflorescences with the mono- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 typic Barnardiella (Goldblatt, 1976c), and it is a possible link between this isolated genus and sect. Moraea. The immediate relationships of M. graniticola and M. hexaglottis are not apparent and their possible relationships and origins are discussed in the systematic treatment. Moraea deserticola resembles the western Ka- roo M. speciosa in its flower and general ap- pearance. Its cytology is not known but it seems certain that it is closely related to M. speciosa with which it probably shares a common origin from the widespread southern African M. poly- stachya or its immediate ancestors. Moraea worcesterensis has n — 6 and a karyo- type that corresponds well with species of sect. Vieusseuxia. It is perhaps most closely related to M. algoensis of this alliance, with which it shares a long-lasting, deep blue-purple flower, a slender few-branched stem, and heavily clawed, light brown corm tunics. Its affinity with sect. Vieus- seuxia has been further established by successful crosses with M. tripetala, allied to M. algoensis. Control crosses with M. bipartita (sect. Polyan- thos) did not succeed. Thus, among these six new species of Moraea, there appear to be three separate assemblages with a broadly similar Homeria type flower. Ad- ditional groups with this type of flower are subg. Visciramosa (M. elsiae Goldbl.), subg. Vieus- seuxia (including M. thomsonii Baker of sect. Polyanthes), and some isolated species of sect. Vieusseuxia (e.g., M. lurida Ker, M. neopavonia R. Foster, and M. insolens Goldbl.) in which the flower is less obviously Homeria-like. ORIGIN AND ADAPTIVE VALUE OF THE FLOWER There seems little doubt that the Homeria flower-type arose several times in the history of the Moraea—Homeria alliance. It is interesting and probably significant that there does not ap- pear to be a progression of intermediate forms leading to the apparently complex and integrated set of features that comprise the flower. Instead, it seems to have arisen abruptly and it seems likely that one or a very few genic mutations have ced a phenotype so different that the result with similar flowering phenology in several species, mostly in the arid interior of western southern Africa suggests that there is some direct adaptive value to this flower. I can only speculate GOLDBLATT—NEW SPECIES OF MORAEA 107 E 3. Morphology of Moraea graniticola. Whole plant, x0.5; flower, separated tepals, and leaf tran- E. life size; gynoecium and androecium, x 2; separated style branches, much enlarged. that the color is attractive to small bees, the most likely pollinators. Flowering late in the day may strengthen the chances for pollination at a time when other nectar and pollen sources have been exhausted. Fugaciousness probably кисен — flower by limiting its exposure predation from herbivores for the least а time, still allowing for cross pollination. Li nearly all members of Homeriinae, the new species are strongly self-incompatible (the con- dition in M. deserticola is not known). ed © SYSTEMATIC TREATMENT SECTION MORAEA і . Moraea graniticola Goldbl., sp. nov. TYPE: Namibia. Aus (26.16): southern Namib Desert, Aus townlands, in sand around gran- ite domes, ca. 1,300 m (cult. Kirstenbosch Botanic Gardens in 1983) (CB), Lavranos & Pehlemann 20007 (holotype, WIND; iso- type, MO). FIGURE 3. Plantae parvae plus minusve acaulescentes, foliis duobus secundo parvo vel vestigiale, caule ramis prope 1 т fe rem colu m includentibus, lim- bis horizontalibus extensis, filamentis 5-6 mm longis i lumn nnatis s mm liberis et diver- gentibus, antheri 3.5 longis divergentibus, ovario fusiforme rostrato 7-8 mm longo, ramis styli n mm longis bifurcatis quam antheris brevioribus. Plants, low, acaulescent or nearly so, to 5 cm high. Corm 2.5-3.5 mm diam., with dark brown, coarsely fibrous tunics, live corm to 8 mm diam. Leaves usually 2, the second small and vestigial, grey-green, channeled, spreading, white in the midline, somewhat twisted, margins undulate, to 20 cm long and 6 mm wide. Stern subterranean or produced just above the ground, usually with 1-3 branches clustered near the base. /nflores- cence spathes herbaceous, 2.5—3 cm long, the in- ner somewhat membranous and slightly shorter. Flower with a perianth tube, blue-violet with yel- low nectar guides, tepal claws ascending and en- closing the lower part of the filament column, eS spreading horizontally; perianth tube 5-6 m long, white; outer tepals 18-20 mm long, RR 9-11 mm wide, claws to 3 mm long, the inner to 17 mm long and 9 mm wide. Fila- ments 5-6 mm long, united in a white cylindrical column, free and diverging in the upper 1 mm; anthers to 3.5 mm long, diverging, cream. Ovary 7-8 mm long, fusiform, included in the spathes, with a short sterile beak to 2 mm long; style dividing near the base of the anthers, branches diverging, 2 mm long, deeply bilobed, the lobes diverging, to 1 mm long, stigmatic towards api- ces. Capsule unknown. Chromosome number 2n — 20 (Lavranos & Pehlemann 20007) Flowering time. September to early October (in cultivation). Distribution. Southern Namib Desert, in sand around granite domes, known only from around Aus. Figure 5. 108 Relationships. The flower of Moraea grani- ticola resembles closely those of M. crispa, M. pseudospicata, and M. rigidifolia, but as outlined in the introduction, the vegetative morphology and chromosome cytology indicate that the re- semblance is due to convergence. It seems most nd that M. graniticola is most closely related t. Moraea in which all species except cla and M. hexaglottis have a typical Mo- raea flower with well-developed style branches and crests. However, it differs from all members of this section in having the perianth and fila- ments united below into a short tube. The tube is completely closed and apparently functions to raise the flower above the spathes for better dis- play to pollinators. The acaulescent habit is un- known elsewhere in the section and is rare in Moraea, being restricted to the four species of sect. Acaules. This is an apparently monophy- letic alliance distinguished by contractile pedi- cels that raise the tubeless flowers above the spathes and later draw them back for the com- pletion of fruit development. It seems unlikely on morphological grounds that M. graniticola is related directly to M. rigidifolia despite the sim- ilarity of their subequal, short-clawed spreading tepals and short, narrow crestless style branches. History. Moraea graniticola was discovered in 1982 by John Lavranos and Inge Pehlemann in a Sterile state, in dry ground around granite boulders on the townlands at Aus, in southern Namibia. The town lies inland from Luderitz on the rail line to the interior, and towards the i inner parently rare in the wild, even at the single site from which it is known. It is likely that the species occurs elsewhere in rocky parts of the southern Namib where there is normally some winter pre- cipitation. 2. Moraea hexaglottis Goldbl., sp. nov. TYPE: Namibia. Witpütz (27.16): southern Namib, farm Aub, Huib Plateau, ca. 80 km N o Rosh Pinah, 1,200-1,400 m (BB), Lavranos & Pehlemann 21704 (holotype, MO; iso- type, WIND). FiGURE 4. Ж; Plantae parvae, 8—10 cm altae, folio solitario gracili, caule ramoso, floribus caeruleis, tepalis ca. 10 mm lon- gis, unguibus partem inferiorem columnae filamento- rum includentibus, limbis horizontalibus extensis, fi- lamentis 3 mm longis in columna connatis supra ad 1 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 mm liberis et divergentibus, antheris ad 2.5 zm longis ,ca. 3mm lon- go, ramis styli bifurcatis a basi, ramulibus ad 2 mm longis, inter apices filamentorum extensibus. Plants small, 8-10 cm high. Corm 10-15 mm diam., with coarse brown fibrous tunics. Leaf solitary, basal, slender, and apparently terete but linear and channeled, with tightly inrolled mar- gins, sometimes somewhat twisted above, about twice as long as the plant. Stem somewhat flex- uose, 1-2-branched, stem bracts subtending the branches 12-18 mm long, herbaceous, becoming dry above, axes flexed slightly below the spathes. Inflorescence spathes green, becoming dry and pale straw colored above, 16-19 mm long, the outer about half to nearly as long as the inner. Flowers blue-violet with pale yellow nectar guides on outer tepals; outer tepals 12-14 mm long, claw vestigial, limb spreading almost from the base, 6-7 mm wide, obovate; inner smaller than the outer, 10-12 mm long, lanceolate. Filaments ca. 4 mm long, united in the lower half in a smooth cylindrical column, diverging above; an- thers 3-4 mm long, ascending, yellow. Ovary narrow, cylindric, ca. 3 mm exserted from the spathes; style dividing at the apex of the filament column, branches divided almost to the base into paired slender tapering arms extending upwards between the filaments, the arms 2.5-3 mm long, apically stigmatic, crests lacking. Capsules globose, 4—6 mm diam.; seeds angular. Chromosome number 2n = 20 (Lavra- nos & Pehlemann 21704). Flowering time. September to October, flow- ers opening in mid-afternoon, at about 4 P.M., fading in early evening. Distribution. Known from only one site on the Huib Plateau at about 1,300 m, in southern Namibia in silt on black limestone. Figure 5. Relationships. The small, blue-violet flowers with spreading, subequal tepals of Moraea hex- aglottis are very like those of M. rigidifolia and the Karoo species M. crispa. However, the short style branches, each divided almost to the base into a pair of slender, ascending arms are unlike those of any other species of Moraea and resem- ble those of tk lat glottis (5 spp.). In Hexaglottis the filaments are united only to- wards the base and the style arms are more slen- der than in M. hexaglottis, and the flowers are always yellow. In addition, Hexaglottis has x = 6 (Goldblatt, 1971, and in prep.) in contrast to n= 10 in M. hexaglottis. It is clear that M. 1986] RE 4. Morphology of Moraea hexaglott is. Whole plant, x0.5; flower, life size; gynoecium and androecium, x2. hexaglottis and Hexaglottis acquired similar style Huib Plateau in southern Namibia. It is known from a few sites on the farm Aub and is, ac- cording to Lavranos, widespread there. It prob- ably occurs elsewhere on the western part of the Huib Plateau that receives limited precipitation in the winter. The habitat is described as silt on black limestone, quite different from the rocky granite where the other two southern Namibian species grow. The specific epithet refers to the six slender style arms or tongues that are char- acteristic for the species. 3. Moraea rigidifolia Goldbl., sp. nov. TYPE: Namibia. Witpütz (27.16): southern Namib Desert, rocky granite flats ca. 40 km N of Rosh Pinah, farm Süd Witpútz, ca. 1,100 m (DA), Goldblatt 7016 (holotype, WIND; isotypes, MO, NBG, PRE, S, US, WAG). FIGURE 6. GOLDBLATT—NEW SPECIES OF MORAEA ме 7- 12 cm altae, foli ad recu it raro infra ramoso, floribus caeruleo-malvinis, pali 13-18 mm longis, unguibus partem inferiore o 5-7 mm longo, ramis styli 2-2.5 mm longis pericias quam antheris bre- vioribus. Plants slender, 7-12 cm high. Corm 2-3 cm diam., with dark brown, coarsely fibrous tunics accumulating to form a thick layer and extending upwards in a neck, live corm ca. 8 mm diam. Leaf solitary, basal, terete, to 3 mm diam., fis- tulose, pithy inside, falcate to recurved. Stem brown, to 2.5 cm lon cept the terminal, spathes herbaceous, becoming dry and brown above at anthesis, 2.2 cm long, the outer slightly shorter than the inner. Flowers blue-mauve with cream nectar guides on the outer tepals, the claws forming a shallow cup round the base of the filament column, limbs spreading horizontally; tepals lanceolate, 14-18 mm long, claw 3-4 mm, limb 6-7 mm wide, widest in the upper third, the inner 13-15 mm ong, 5-6 mm wide. Filaments 4—5.5 mm long, united below, free in the upper 1.3-2 mm and diverging, forming a slender smooth cylindric column; anthers 2.2-3 mm long, diverging, white. Ovary linear-fusiform, included in the spathes, 5-7 mm long, with a sterile beak 1-2 mm long; style dividing at the apex of the filament column, branches diverging, 2-2.5 mm long, bilobed above and stigmatic apically, reaching to mid anther level, crests lacking. Capsules not known Chromosome number 2n = 20 (Goldblatt 7016). — Flowering time. Late September to October, flowers opening in late afternoon, at about 4:30 P.M. and fading in early evening at about 7 P.M. Distribution. Rocky granite flats, in the southern Namib Desert north of Rosh Pinah. Figure 5 Relationships. The small, pale blue-mauve flowers of Moraea rigidifolia have subequal te- to short narrow style branches. They clo semble those of the Karoo species M. crispa. My initial presumption was that the two were closely related. However, M. rigidifolia has a base num- 110 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Cal Ф м. pseudospicata $ O м. graniticola W M.hexaglottis nw Ay y Y, KA ; ) Y YN WZ “у ЗГ 4 14 б 12 Wee | m У ДИ SAGA KE | 0 AN К "7. ID RI E OEE PRA || MORI dd SEER ed EE AP EP. FiGURE 5. Geographical distribution of Moraea pseudospicata, M. graniticola, M. hexaglottis, M. rigidifolia, with the range of Moraea crispa heavily outlined for comparison. ber of x = 10 and karyotype quite different from that of M. crispa and of sect. Polyanthes gener- ally. The basic chromosome number for Moraea is x = 10 and both number and the details of the karyotype of M. rigidifolia conform with species of subg. Moraea as well as with the monotypic Namaqualand Barnardiella. The cytological evi- dence is regarded here as a more reliable indi- cator of relationship, and M. rigidifolia is pre- sumed to be allied to species of sect. Moraea or sect. Deserticola. In its general habit M. rigidi- folia resembles quite closely Barnardiella spi- and sessile lateral inflorescences with relatively short and subequal spathes. Barnardiella differs in having a long slender tube composed of sterile ovary tissue (Goldblatt, 1976c) and a subsessile ovary located in the base of the spathes. Details of the style and anthers of Barnardiella differ only slightly from those of M. rigidifolia and it seems likely that M. rigidifolia is close to the line that gave rise to Barnardiella. The slender ovary of M. rigidifolia, which is sterile in the upper 1- 1.5 mm, is perhaps a link with the unusual ovary found in Barnardiella. Moraea rigidifolia is easily recognized by its 1 1.7 „1 усту unusual у short terete leaf. The swollen leaf, which consists of green tissue surrounding a firm white pith-like parenchyma, is so striking that M. rigidifolia can immediately be identified from this feature alone. Distribution. Moraea rigidifolia is known onl from a single locality in southern Namibia, on 1986] GOLDBLATT— NEW SPECIES OF MORAEA Morphol ogy of Moraea rigidifolia. Whole plant, corm, and leaf transection, x0.5; flower and separated tepals, life size; gynoecium and androecium, х 2; separated style branch, much enlarged. the farm Siid Witpiitz, some 35 km north of the mining settlement of Rosh Pinah near the Orange iver. It was discovered by John Lavranos and Inge Pehlemann in 1982. In the very good spring season of 1983 I revisited the locality and col- lected ample flowering specimens. The habitat of M. rigidifolia is rocky and consists of thin soil and low outcrops of weathered granite. It is lo- nly s lowly buried. Moraea sri Aces occurs elsewhere in the rocky southern Namib Desert, which is poorly known botanically. Owing to its inconspicuous and fugacious flowers that open only in the late afternoon and early evening, it is easily overlooked. Additional material examined. NAMIBIA. Southern rt. 29. 6 (Witpütz) due granite flats ca. ah, farm Witpütz, ca. 1,100 m (DA), Таак. & ГУ ээнин 19104 (МО). SECTION POLYANTHES > Moraea pseudospicata Goldbl., sp. nov. TYPE: South Africa, Calvinia (31.19): Cape Prov- ince, Glenlyon, E of Nieuwoudtville, dry Dwyka tillite flats (AC), Snijman 783 (ho- lotype, NBG; isotypes, K, MO, PRE, S, STE, WAG). FIGURE 7. Plantae 15-40 cm altae, pese o gracili ca- naliculato in latere adaxiali, c terdum infra ra- moso, inflorescentibus iuc abis iiini, floribus caeruleo-malvinis, tepalis 11-16 mm longis, unguibus partem inferiorem column contiguis, antheris 2-4.5 mm longis globoso ca. 2 mm diam., ramis styli 1-1.5 mm longis antheris occultis. Plants slender, 15-40 cm high. Corm 3-5 cm diam., with tough, blackish fibrous tunics usually accumulating to form a thick layer and extending wards in a neck, live corm ca. 1 cm diam. Lea na basal, slender, apparently terete but nar- rowly grooved on adaxial surface and margins tightly inrolled, straight or weakly twisted. Stem simple or 1-2-branched from below, bearing sev- eral sessile inflorescences at the upper nodes, stem bracts dry, brown, to 2.5 cm long, distinctly dark- er on the veins. /nflorescences sessile except the Б inga Pa сар round the base of the filament colu lly; tepals lan- с dI- 17 г шт long, claw ca. 2 mm, limb .5-4 mm wide, widest in the upper third, the inner tepals ca. 1 mm shorter than the outer. Filaments 4—6.5 mm long, united entirely or free in the upper two-thirds but contiguous, forming a slender, smooth cylindric column; anthers 2— 4.5 mm long, erect and contiguous, yellow. Ovary 112 FIGURE 7. Morphology of а pseudospicata. Whole ub and corm 5; flower and separated inflorescence with fruit, life size. globose, ca. 2 mm long; style dividing at about mid anther level, branches narrow, ascending, 1-1.5 mm long, stigmatic apically, initially con- cealed by the anthers, reaching to the upper third or slightly exceeding the apex of the anthers; crests lacking. Capsules globose, 2-4 mm diam., mem- t. not loculicidal but E irregularly; seeds several, dark n, long. Chromosome number 2 = uer (Goldblatt 6543). Flowering time. December to March, flowers opening after 3:30 р.м. and fading in the early evening ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Distribution. Rocky clay flats, in karroid scrub, in the Nieuwoudtville district from the hills north of the town extending south to Loken- berg and possibly to the dry interior valleys of the Cedarberg. Figure 5. Relationships and variation. Moraea pseu- dospicata resembles the widespread Karoo species M. crispa in both flower structure and general aspect. The pale blue-mauve flowers are almost identical in their subequal tepals with short erect claws that form a short, narrow tube, which in- cludes the lower part ofthe filament column. The filaments and anthers are slightly shorter in M. pseudospicata, and the anthers are nearly erect and so conceal the narrow and short style branch- es. In M. crispa the anther and style branches diverge strongly, and although the style branches are shorter than the anthers, they can easily be seen. The two species also have the same chro- mosome number, n = 6, and karyotypes with ac- submetacentric chromosomes that tain that the two are closely allied. Moraea pseu- dospicata differs mainly in its sessile lateral in- florescences and in the dry chaffy stem bracts and spathes. Its capsules also differ from those of M. crispa in being slightly smaller, in having almost membranous walls, and in apparently de- hiscing irregularly rather than along the locule sutures as in nearly all other species of Moraea. Moraea pseudospicata is also distinctive in its flowering time, mid to late summer, in an area of complete summer drought, and at the time of flowering the leaf is usually dry and often broken. In the summer dry western Karoo and adjacent Cedarberg where both M. pseudospicata and M. crispa occur, M. crispa blooms from October to vember and rarely into December. The relatively few collections of Moraea pseu- dospicata eris little variability except for the filaments, may be almost entirely united or free but. pod in the upper half. This feature does not ш to be correlated with geographical distributio unusual collection, ‘Gi 11170 (K, PRE), from Nieuwoudtville, has a strong resemblance to Moraea pseudospicata. It has small but ap- parently deep. blue- purple flowers and the char- f M. pseu- dospicata, but the capsules, of which there are several in nearly mature state, are elongate and fusiform, about 10 mm long, and have a well- developed beak about 1 mm long. This is quite different from the globose capsule of M. pseu- 1986] dospicata. Examination of the poorly preserved and few flowers reveals that the filaments are 3.5 mm long, and at least free above (indistinct be- low), the anthers are contiguous and very short, T 1.2 mm long, and the ovary is about 4.5 m long. Other details are not -o It seems likely ran the collection represe a speci closely allied to M. pseudospicata, pides at least by a different capsule (that contains ap- parently larger seeds), smaller anthers, and dark- er colored flowers. The flowering time noted on the single collection is November, another dif- ference with M. pseudospicata, which has not been recorded in flower earlier than December, and usually flowers later. History. The first known collections of Mo- raea pseudospicata were made by Carl Zeyher, River ‘Reise nach Kamiesberg, Boschmanland, bis zur mundung des Garip.' Several duplicates, distributed as Ecklon & Zeyher Irid. 32 are in herbaria that have good sets of Ecklon and Zey- her duplicates. Only the sheet at Stockholm has complete locality data ‘Onderboksveld, Bluht December, 4 hohe.’ From this it seems almost certain that their gathering was made near pres ent-day Nieuwoudtville, this general area still being sometimes called the Lower B e Other early records of this species include plants from the ‘Cedarberg,’ collected by Thode and from Lokenberg, south of Nieuwoudtville, col- lected by Acocks. All these collections lay un- identified or placed with either Hexaglottis or M. crispa. In 1980 Dierdre Snijman of the Compton Herbarium at Kirstenbosch showed me a collection of a species she had made near Nieuwoudtville that included well-pressed flow- e he specimens struck me as unusual both in morphology and in their flowering time. In 1982 I relocated the species at two sites to the north of Nieuwoudtville and made extensive collec- tions. Once this species had been identified as a new and distinctive Moraea, I was able to assign to it the earlier records cited above. Specimens were found at a third site closer to Nieuwoudt- ville in 1984, and this collection has been se- lected as the type material. Additional specimens examined. SOUTH AFRICA. CAPE: 31.19 (Calvinia) near Nieuwoudtville, Charlies Hoek, between Loeriesfontein Road and Klipkoppies (AC), Snijman 98 (NBG), Goldblatt 6542 (MO); rocky clay flats on the road to Rondekop, E of Nieuwoudt- ville, Goldblatt 6543 (MO, NBG, PRE, S); Glenlyon, E of Nieuwoudtville, dry Dwyka tillite flats (AC), Snij- GOLDBLATT — NEW SPECIES OF MORAEA 113 3 (К, MO, NBG, PRE, S, STE, WAG); Lotenbu re (СА), к = (PRE). HOUT PR v: “Reise nach Kamies- mg odio ‘bis zur Mundung des Garip,” Zeyher as Ecklon & Zeyher Irid. 32 (73.12) (LD, MO, P, S “Bluht December 4 hoh Onderboksfeld’); Cedar- berg, Thode A2076 (PRE). 5. Moraea deserticola Goldbl., sp. nov. TYPE: South Africa. Vanrhynsdorp (31.18): Cape Province, Knersvlakte, farm Quaggas Kop, Zout River (BC), Hall 5089 (holotype, NBG; isotype, MO). Plantae 35-45 cm altae, cormo ca. 1 cm diam., foliis 5 s minusve obovato, 6- venoso, ramis "iis brevibus ae extensis su- pra anther: Plants 35-45 cm high. Corm ca. 1 cm diam., with tunics of dark brown to blackish fibers. Leaves 2-3, linear and channeled, 2-3 mm wide, about as long as the stems, margins becoming tightly inrolled when dry, the lower basal, upper cauline. Stem simple or 1-3-branched from the upper nodes, the branches subtended by sheath- ing bract leaves 2.5-4 cm long. Spathes 3.3-4 cm long, herbaceous, the outer about two-thirds as long as the inner. Flowers pale blue or white with pale blue on the reverse of the tepals and fading pale blue, with yellow nectar guides at the base of the limb of the inner and outer tepals, tepal claws ascending and forming a wide cup including the filament column, limbs extended horizontally; tepals 30-36 mm long, claw ca. 12 mm long, limb 18-20 mm long and 10-11 mm wide, the inner slightly smaller than the outer. Filaments 8-10 mm long, united into a smooth slender cylindrical column; anthers 5-6.5 mm long, erect, contiguous, appressed to and initially concealing the style. Ovary 6-7 mm long, nar- rowly obconic, exserted from the spathes, with conspicuous reddish veins; style dividing into 3 short horizontal lobed branches just above the apex of the anthers; branches ca. 1 mm long, crests lacking. Capsule more or less obovoid, 9— 10 mm long, red-veined, seeds many, angled. Chromosome number unknown Flowering time. June to August. Distribution. Stony slopes and flats in the Knersvlakte between Vanrhynsdorp and Nu- werus. Figure 8. 114 18 19 20 1. 1 1 ANNALS OF THE MISSOURI BOTANICAL GARDEN NA $, Y dz Н РЛ 2 2279252777 III ZO ‘Zr [VoL. 73 Y М. deserticola c M. speciosa A M. worcesterensis а C M. algoensis A A де, 21 22 23 1 1 1 FIGURE 8. Geographical distribution of Moraea deserticola, with the range of the closely allied M. speciosa indicated by the broken line, and M. worcesterensis, with the range of the putatively related M. algoensis indicated by the heavy, unbroken line. Relationships. The large and deeply cupped flower of Moraea deserticola with its nearly equal tepals and long conspicuous anthers is strongly reminiscent of M. speciosa, a species of the drier parts of the western Karoo. It seems certain that the two species are closely allied although they differ greatly in vegetative morphology. The long style surrounded by the contiguous anthers, and short, broadly-lobed style branches held above the anthers when the stigmas are receptive are exactly the same in both species. Except for a slightly smaller size, the flowers and all floral parts of Moraea deserticola are apparently iden- tical to those of M. speciosa. The leaves of M. deserticola are narrow and few, and only the low- ermost is basal while the other two or three are spaced well apart along the stem. Moreover, the leaves are straight and narrowly channeled and the stem bears only a few branches towards the apex. In contrast, M. speciosa has a thick fleshy well-branched stem and several relatively short and broad basal leaves that are often undulate or twisted. Moraea deserticola can also be dis- tinguished by its strongly dark purple-veined ovary, a feature not evident in M. speciosa. History and distribution. The species was first discovered in 1967 by P. A. B. van Breda, in the Knersvlakte, the arid low rolling country in southern Namaqualand that is so extraordinarily rich in low-growing succulents. Due to the lim- ited material available and inadequate preser- vation, the collection could not at first be iden- tified although it did not seem to be any known species. It was recollected in 1981 by Harry Hall and two years later by Jan Vlok but has not been seen since. The collection made by Hall is barely sufficient for a type but it seems preferable to describe the plant now rather than wait indefi- nitely hoping to find more specimens. Perhaps also now that the species is named others will be encouraged to search further for it. The species is probably rare in the wild, but 1986] as it has been found at three sites in the Kners- vlakte, it probably occurs widely in the area. Most likely the plants are scattered and inconspicuous, and added to this, they probably flower only when good rains have fallen in this arid area. Additional specimens examined. SOUTH AFRICA. CAPE: 31.18 (Vanrhynsdorp) 9 km NW of Vanrhyns- D near the gypsum mines, sandy soil, Vlok 65 ; Varsrivier, ca. 11 mi. N of Vanrhynsdorp, van pian 4018 (PRE). SECTION VIEUSSEUXIA 6. Moraea worcesterensis Goldbl., sp. nov. TYPE: South Africa. Worcester (33.19): Cape Prov- ince, stony sandstone flats at Worcester West (CB), Goldblatt & Snijman 69774 (holo- type, NBG; isotypes, K, MO, PRE, S, STE, US). FIGURE 9 Plantae 12-25 cm altae, tunicis fibrosis unguiculatis, folio solitario canaliculato, floribus atropurpureis te- gis, ramis styli angustis apicibus bilobis. Plants 12-25 cm high. Corm globose, 15-20 mm diam., tunics brown, fibrous, often heavily clawed, sometimes accumulating to form a thick covering. Leaf solitary, basal, linear, canalicu- late, somewhat to about twice as long as the stem. tem simple or with a single branch, erect or flexed at the upper nodes. Spathes herbaceous, dry and dark brown above, 28-35 mm long, the outer about half as long as the inner. Flowers dark violet-purple with yellow nectar guides on all tepals; outer tepals 15-17 mm long, claw as- cending, ca. 3 mm long, limb vp extended hor- izontally, 12-14 mm long and ca. 9 mm wide; inner tepals similarly oriented, ыа: 13-15 mm long and 4-5 mm wide. Filaments 5-6 mm long, icum below, free in the upper 2 mm; an- thers 4 mm long, exceeding the style branches, yellow. Ovary ca. 5 mm long, nearly cylindric but narrowing below, style branches narrow, less than 1 mm wide, diverging and appressed to the opposed anther, becoming bilobed apically and stigmatic at the ends of the lobes, crests ети Capsules narrowly obovoid, to 10 mm long. Chromosome number 2n = 12 (Goldblatt ee Flowering time. September. Distribution. Known only from stony flats west of Worcester, in mixed fynbos and renos- terveld. Figure 8 GOLDBLATT —NEW SPECIES OF MORAEA 115 FIGURE 9. Morphology of Moraea worcesterensis. Whole plant, x0.5; flower and separated tepals, life size; gynoecium and androecium, x 2; separated style branch, ak enlarged. 116 Relationships. Могаеа worcesterensis is ap closely related to the widespread south- ern ies M. algoensis, despite its very мы flowers. It has more or less spreading tepals, both whorls of which have nectar guides, a slender filament column, and very reduced, narrow style branches, shorter than the subtend- ing anthers and lacking crests. Moraea algoensis has a typical Moraea flower with large inner te- pals, the claws of which touch the broad petaloid style branches to conceal the anther and stigma lobe. In Moraea algoensis the inner tepals are relatively broad for sect. Vieusseuxia, most species of which have narrow, linear, trifid to ciliate inner tepals. The general flower color of M. worcesterensis and M. algoensis is the same, and the corms conform to the type found in many species of the section. The single and basal leaf and reduced number of branches are also con- sistent with placement in this section. The general similarity of Moraea worcester- nsis to M. polyanthos in sect. Polyanthes is probably due to convergence for the Homeria type of flower. Thus although the flowers of the two species are similar, the vegetative structure is very different. Moraea polyanthos has (two to) three to five leaves, several branches, and black- ish corm tunics of a tough wiry texture, all char- acteristics that seem to be significant in deter- mining the relationships of species of Moraea. History. Moraea worcesterensis was appar- ently first discovered in the spring of 1983, by me, Dierdre Snijman, and Jane Turnbull, both of the National Botanic Gardens, in Worcester West, a rapidly developing suburb of Worcester. It is now extinct at the type locality but probably occurs elsewhere along the foot of the Langeberg Mountains. LITERATURE CITED EDwARDS, D. & О. Н. LEISTNER. 1971. A degree ref- erence system for citing biological records in ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 southern Africa. Mitt. Bot. Staatssamml. Mün- chen 10: 501-509. GOLDBLATT, P. 1971. Cytological and morphological studies in the southern African Iridaceae. J. S. Af- rican Bot. 37: 317-460. . 1973. Contributions to the knowledge of Mo- raea (Iridaceae) in the summer rainfall region of South Africa. Ann. Missouri Bot. Gard. 60: 204— 59 1976a. pi ris cytology and subgeneric classification in Mon a (Iridaceae). Ann. Mis- ouri Bot. . 1976b [1977]. The genus Moraea in the win- Africa. Ann. Missouri ter rainfall area of Southern Bot. Gard. 57-786. 1976c [1977]. Barnardiella: a new genus of the Iridaceae and its relationship to Gynandriris and Moraea. Ann. Missouri Bot. Gard. 63: 309- 977 [1978]. Systematics of Moraea (Irida- ca) in tropical Africa. Ann. Missouri Bot. Gard. 3-295. ae Chromosome cytology and karyotype change in Galaxia (Iridaceae). Pl. Syst. Evol. 133: 161-169. 80. Redefinition of Homeria and Moraea (Iridaceae) in the light of biosystematic data, with Rheome gen. nov. Bot. Not. 133: 85-95. . 198 1a. жошо. phylogeny and evolution of Dietes (Iridaceae). Ann. Missouri Bot. Gard 68: 132-153. 1981b [1982]. о (Iridaceae). 68: 4 Systematics and biology of Ann. Missouri Bot. Gard. РР [1983]. A synopsis of Moraea (Irida- ceae) with new taxa, transfers and notes. Ann. Mis- souri Bot. Gard. 69: 351-369. Gunn, M. & Г.Е. Copp. 1981. Botanical n of Southern Africa. A. A. Balkema, Cape Town JACKSON, R. C. 1975. Chromosomal evolution in Haplopappus gracilis: a centric transposition race. Evolution 27: 243-256. SOLCH, A. 1969. Iridaceae. In H. _ Merxmüller pa tor), Prod 55: 1-12. THE CALYX IN LYCIANTHES AND SOME OTHER GENERA! W. С. D'Arcy? ABSTRACT Vasculature and structure of calyces in Lycianthes and some related genera are analyzed to derive lobes to fuse to higher levels, sometimes right to the top (perfect prefloration). There tendency to fusion of lateral veins to higher levels, which gives rise to ten main m or ribs in the fused area. The flow Whe yx prefloration is complete or nearly so, egress must involve stretching. or tearing. Thus in Diae, Witheringia, a flower or inflorescence, is nearly ubiquitous in vascular plants, calyx teeth are completely fused, they no longer function as teeth. In some grim pr this se is remedied by enation of ‘secondary’ teeth below the sleeve. In Lycianthes they m a by primary traces and fused laterals leading to the ten teeth in two series. These к Жыш of calyx evolution can be seen in some other families such as the Ericaceae. In a group of genera in which the number of parts in a floral whorl is generally five, the fre- quent presence of ten teeth on the calyx of many Lycianthes species seems anomalous. These teeth are often in two alternating series and range in different species from small umbos to elongate, filiform processes many times the size of the ca- lyx body. Closer that these teeth are not apical on the calyx but appear from the side of the calyx below a ring or “‘sleeve”’ of tissue that terminates the calyx wall (Figs. 2-4, 7). Dunal (1852) grouped species like this into subsection Lycianthes of the genus Solanum, which was elevated to generic rank by Hassler b 17). The name Lycianthes was conserved over earlier names (McVaugh, 1973). Bitter (1919) and subsequent workers have expanded the ge- nus to well over a hundred species. Placement of Lycianthes at the generic level was at first justified because of the reduced inflorescences and the presence of large stone cells in the fruit (Hassler, 1917; Bitter, 1914, 1919; Danert, 1969), yet the segregation of the group from Solanum has not met with the acceptance of all botanists. With a desire » ани the nature of the d to reas- sess the validity of this genus, a study of calyx vasculature was undertaken. A considerable range seemingly of material in related genera of Solanaceae, es- pecially in subfamily Solanoidae, was studied, and material from other families was also ex- amined. Some patterns in the evolution of ca- lyces in the Solanaceae are apparent in other un- related families. METHODS Traditional methods using sodium hydroxide and chloral hydrate were used to clear fresh, pick- led, and dried flower buds and calyces of selected Solanaceae, and this treatment was sometimes followed by staining with safranin or basic fuch- sin. These methods gave good three-dimensional views under a binocular dissecting microscope when material was held in vials in liquid but generally yielded poor photographic subjects. For photographs it was sometimes necessary to flat- ten the material on a glass slide and add a cover slip. Series of stained microscope sections of Capsicum annuum L., Solanum seaforthianum Andr., and Oryctes revadensis S. Wats. were used for comparison. In addition, a great many flowers from many families were examined by naked eye and hand lens in the herbarium and in a living state. ! This paper was part of the Second International а of the Biology and Systematics of the Solanaceae presented at the Missouri Botanical Garden on 3-6 Aug 2 Missouri Botanical Garden, P.O. Box 299, St. Louis, gree 63166. ANN. MISSOURI Bor. GARD. 73: 117-127. 1986. 118 ЕСИ 1-4. Lycia Calyx о envelops the inner —2 sleeve. Th h eme ath the eue a an r vasculat ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 nthes maxonii Als pe D’Arcy 5507 (MO)].—1. Inflorescence with emerging buds. rts.— olla Pi die es from the calyx by stretching the e apex into a thin re moves out as they grow.—3. Coroll rea, a ur a open. The teeth have reached full size. The teeth enervated ^ the fused lateral veins 1 th. ead are much smaller (sce en —4. Calyx in fruit showing intact sleeve and teet THE GENERALIZED SOLANACEOUS CALYX In many plant families, including the Solana- ceae, the gamosepalous calyx consists of approx- imately five parts that are fused by their lateral margins to various levels, and the terms sepals and calyx lobes are often used interchangeably. Although the lobes may not in every case be homologous with the free sepals of the Archich- lamidae, the studies of Copeland (1943) and Pal- ser (1951, 1955) in the Ericaceae suggest that they are homologous i in that family. In the Solanaceae these traces maintain their identity throughout their length, although through the pedicel they are intimately associated and in serial sections they appear in this region as a siphonostele. Mur- ray (1945) reported that in the Solanaceae gaps sometimes occur where the calyx traces leave the pedicel cylinder, but that this feature may vary even within a species. Techniques used in the present study did not distinguish parenchyma, and from the evidence seen of the vein architec- ture the presence of gaps could not be ascer- tained. The veins leading into the petals and those leading into the sepals were recognizably distinct from one another throughout their length, even in the relatively constricted base of the pedicel, and they apparently continue independently from the base of the pedicel to the top of the flower Fig. 5). Connections between the two perianth whorls through exchange of sepal and petal traces were not indicated by Murray and in this study were found only as rare exceptions. The nature of connections between this vasculature of calyx and corolla in Solanum boldoense Dun. is dis- cussed below. In one other instance, in only one of several flowers of an ornamental pepper (D'Arcy 7001, MO), a vein was seen running be- tween the calyx and corolla near the base of only one of the sepals. The single trace entering the calyx lobes in the Solanaceae is in contrast to reports (Carlquist, 1961) for many other groups in which the sepals are supplied by three or more traces arising from the stele or stelar plexus but is Similar to some genera of Ericaceae (Copeland, 1943; Palser 1951, 1955), which are slightly more specialized than the most primitive members of the family. Q C 5 FIGURES 5, 6. D'ARCY —LYCIANTHES CALYX 119 anum americanum Mill. [After D’Arcy 15159 (MO)].— 5. Developing buds. Primary traces develop leaf-like minor lateral venation. Note calyx lobes splitting at the sutures and pm PA of vasculature in the suture area. Interconnector vein is at arrow. — 6. Fruiting calyx with lobes split to the bas Almost as soon as the calyx lobes are evident in the bud, a system of minor venation appears in each lobe that is usually soon oriented into two lateral veins flanking the midvein and a vein linking the basal, tangential portions of the lat- erals to form a prominent ‘interconnector’ vein (Figs. 5, 9). The interconnector may later assume structural importance in the fruiting calyx, as in the physaloid genera (Physalis, Nicandra, Wi- thania, etc.). Other minor venation resembles that of a foliage leaf presenting a more or less acropetal, pinnate configuration on each side of moses of the sepal traces, which at a slightly d level broke into usually three separate veins. In the Solanaceae the interconnector often does not appear as such until after the minor venation has taken on its main outline, hence it is seen as secondary in an ontogenetic and perhaps phy- logenetic sense. The pattern ofa single sepal trace expanding into an interconnector and three ma- jor veins can be observed in other families such as the Boraginaceae, Scrophulariaceae, and Ce- rastium in the Caryophyllaceae. In many species of Boraginaceae there are two interconnectors, one at the base of the calyx and the other nearer the top, seemingly formed from venation dis- d fr th ground tissue alone or both ground tissue and vasculature, and the level and nature of fusion may be usually of generic significance. Only one study is known that documents the develop- mental nature of calyx fusion in the Solanaceae. The fusion of ground tissue was investigated in Datura by Satina (1944), who used chimeras of different ploidy levels to show that in that genus the fusion of ground tissue is ontogenetic. The fusion takes place at a very early stage, and stages where fusion has not yet taken place cannot be identified by macroscopic methods. In many other genera of Solanaceae, from the time the flower bud is 1-3 mm long, there is little or no further alteration in length of the free sepal tips. From this it is assumed that by this stage the process of fusion is complete, but there is often a subsequent enlargement of the region of fusion. After fusion of adjacent sepal margins, fusion of adjacent lateral sepal veins may take place, and the fusion is usually for most of the length of the longitudinal portions of the veins produc- ing a second series of longitudinal veins alternate with the primary traces (Figs. 2, 4, 10). Such fusion is always the case in Lycianthes but appears only with irregularity in Capsicum (Fig. 10), often varying from sepal to sepal and from flower to flower. Fused laterals were put to taxonomic use by Waterfall (1958, 1967), Rydberg (1896), and others in the genus Physalis, in which groups of species were distinguished as being five-angled, 10-angled, or round, the condition mainly con- sequent on the state of fusion and development 120 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 7-10. 7. Calyx of Lycianthes maxonii Standl. showing the reduced vasculature in the sleeve and the ры traces Wn have bent out within the teeth. — 8. Calyx of Lycianthes guianensis (Dun.) Bitt. showing w the primary traces bend out 7 the teeth and the minor venation is directed basepetally as а result of this repositioning (a am ow). —9. Calyx of Solanum americanum Mill. showing how the minor venation arising rom the primary trace aas s бс venation of a leaf. The interconnector can be seen at the arrow. — 10. Calyx j mL The rudim apsicum annuu . ес X ee the reduced teeth (arrow). In this calyx the laterals have fused. It is speculated that the main teeth here are actually secondary teeth derived by outbending below the sleeve that have refused back against the primary traces (see text). of the sepal laterals. Such fusion of apparently FLORAL EGRESS IN THE SOLANACEAE homologous laterals is common, and was ob- AND IN GENERAL served to occur in the Phyrmaceae, in nearly all Lythraceae and Labiatae, common in the Car- For the expansion of flowers in many families yophyllaceae (for example, Cerastium), but un- the corolla and often the stamens and stigma common in the Scrophulariaceae. must find their way out of the enclosing outer 1986] 14. Dehiscent Egress: in Solanum, only the tips of the calyx lobes are free in bud, but splitting at the sutures.—15. Distensive Egress: D’ARCY—LYCIANTHES CALYX egress is by stretching of the distal to о the calyx апа а marked sleeve is produced аир — 16. Dehiscent Egress: Witheringia has distensive floral egress, but there is dehiscence for egress of the fruit perianth, which is usually the calyx. The mode of egress can be seen by observing the expansion of flowers in living material and often by ob- serving dried herbarium specimens. In the flower with free sepals, egress of the corolla and other floral parts from the calyx is readily effected by outward movement of the sepals, but where there is fusion of the sepals, other arrangements may be necessary to permit egress of the pistil as it develops at anthesis and into fruit (Figs. 11-16). Where fusion of the calyx involves only the basal portion of the sepals, egress may involve no more than a slight stretching of the calyx without sig- nificant alteration of shape (Fig. 13). This is the case in many species of Verbenaceae (for ex- ample, Clerodendrum splendens), Polemoniace- ae, Plantaginaceae, and other families, but in most Solanaceae there is high enough fusion that oth- er mechanisms are invoked. Where fusion has progressed to unite upper portions of the sepals, egress may require either stretching and major distortion or tearing of the calyx at some point. In Solanum and Lycopersicon the sepals are usu- ally fused except for their tips, and the sutures 122 or regions of fusion between them tear open in a valvate fashion to permit egress (Fig. 14). This is also found in Symphytum (Boraginaceae). The tearing may begin before the tip of the budding corolla reaches the length of the sepal tips or may be delayed until the corolla actually begins to Lycopersicon, and many other solanaceous gen- era, in which floral egress involves tearing at the sutures, there is no vasculature in the region of tearing. Where the calyx is fused to the top of the sepal tips as in Lycianthes, Witheringia, and Anaemopaegma and in many other Bignoni- aceae, instead of by splitting, floral egress is achieved by stretching of the upper portion of the calyx (Fig. 13, 15). The stretched top portion of this calyx is manifestly thinner and smoother than the lower unstretched portion, and its ap- pearance suggests that important stresses have been present as egress was effected. This area is the “sleeve” referred to above in the introductory description of the Lycianthes calyx. Not only is the surface sharply demarcated but the vascu- splitting along the sutures of the calyx would be lessened by the presence of vasculature derived from lateral veins in the region between the pri- mary calyx traces. Where sepals are fused to the top of the calyx leaving no opening, the calyx prefloration is termed complete. Here at least some tearing is necessary to create a circumfer- ence that can bs еше or to provide, by the g for the floral parts to el Possibilities for splitting along e presence of vas- along the sutures, there is no vasculature or fu- sion of lateral veins in the suture area. Where there is fusion of laterals and hence vasculature in the suture areas, the tearing may continue down the calyx wall without regard to the position of vasculature. This sort of splitting without regard to the vasculature is a common situation, oc- curring in many genera with more advanced ca- lyces: Brugmansia, Bourreria pulchra (Boragi- naceae), Silene (Caryophyllaceae), Tabebuia (Bignoniaceae), etc. Where floral egress is by stretching, irregular splitting for egress of the fruit may also occur as in some species of Witheringia (Fig. 16). A special case of splitting without re- ANNALS OF THE MISSOURI BOTANICAL GARDEN | Xm \ [VoL. 73 FiGURE 17. Stylized E of Lycianthes show- ing vasculature running into the teeth and then back to disappear in the sleev gard to the vasculature is the circumscissile or calyptrate dehiscence of the calyx in Datura and in Eschscholzia (Papaveraceae) and Lundia (Bignoniaceae). Species of Margaranthus (Fig. 11) and of Salvia (Labiatae) appear to undergo a prefloral egress in which the floral parts egress at an early stage and undergo much of their de- velopment outside the confines of the calyx. The above listed modes of floral egress are not ex- some species of Cyphomandra and in Cordia po- lycephala (Boraginaceae) prefloration of the ca- lyx is complete except for a minute apical pore that may be stretched or split or both. In some species of Magnolia (Magnoliaceae) floral egress (from enclosing bracts instead of a true outer perianth) is achieved by calyptrate or longitu- dinal dehiscence or sometimes by both on the same flower (Fig. 12). Egress may not be nec- essary in some cleistogamous flowers, and in Aristolochia (Aristolochiaceae) the outer peri- n ring. related to systems of fruit dispersal. In the phys- 1986] aloid genera and in species of Abutilon (Malva- ceae) and Saccellium (Boraginaceae), etc. the bladdery calyx may be a means of adapting a rry or achene-like fruit to wind dispersal. There are various stages from loosely investing calyces, as in Physalis, to tightly fitting calyces, as in 7ra- pa and Globularia (stone fruits), Solanum ros- tratum (dry berry), and Cordia sebestena of the Boraginaceae (stone fruits). In the two last men- tioned cases floral egress is by splitting and stretching, but the portal so formed does not en- large, and the calyx and ovary are in snug contact for much of their development. Such a situation might logically preceed fusion of juxtaposed parts for the evolution of the inferior ovary in groups such as the Rubiaceae, Ericaceae, and Cucurbi- taceae where the tissue surrounding the ovary is considered to be appendicular. “If epigyny in the Rubiaceae has a perigynous ancestry, there is no indication of it in the species alive today" (Cron- quist, 1970), but if evolution proceeded from an enveloping calyx as suggested above, there was probably no need for perigyny as an intermediate stage. Although an inferior ovary usually avoids the need for egress of fruit, a variety of modes of floral egress can be seen in families with epi- gyny. THE WIDESPREAD OCCURRENCE OF TEETH 1 = A The calyx is often an up of several pointed elements. These pointed elements can be viewed as teeth, a term used here for pointed structures accessory to a flower or inflorescence without regard to homology or derivation. It is necessary to emphasize the very widespread presence of such pointed structures in the architecture of inflorescences. In dicoty- ledons and monocotyledons alike the floral parts are regularly provided with a calyx terminating in one or more pointed elements, and where this is lacking an analogous form is often taken by widespread and hence would seem to have im- portant utility or selective value. THE PROVISION OF TEETH IN LYCIANTHES Where fusion has proceeded along the entire length of the sepals and calyx prefloration is com- D’ARCY—LYCIANTHES CALYX 123 plete, the flower bud is no longer provided with teeth. Not surprisingly, some species that have lost their functional teeth in this way have evolved teeth that are not homologous with the teeth of an incompletely fused calyx. In species of Ly- cianthes, Capsicum, and Witheringia these non- homologous or secondary teeth are produced around the side of the calyx and are often vas- cularized by outbending of the calyx venation at the points where they occur. Evidence of the na- ture of these secondary teeth is clear in the pat- tern of vasculature in many species. The primary calyx traces may arch outwards into the second- ary teeth and return again to the calyx wall to end in the terminal or sleeve portion of the calyx. (Figs. 1-4, 7, 8, 17). In Lycianthes (Figs. 8, 17) the pinnate minor venation is consequently now directed downwards with respect to the form of the teeth but is still acropetal with respect to the traces it flanks and to the calyx as a whole. In different species, these secondary structures range from small lumps or umbos to elongate filiform teeth in some Lycianthes groups and to substan- tial acute teeth in Capsicum chacoense. In Ly- cianthes where the lateral sepal veins are fused to form a second series of vertically oriented veins, there is often a second series of secondary teeth vascularized by an outbending of the bases of the fused lateral veins. In some species of Lycianthes, Capsicum, or Witheringia in which outbending of the primary traces is slight, secondary teeth or umbos may not occur with regularity at each primary trace. Calyces of Lycianthes rantonnei may have from zero to ten teeth on different flowers on the same plant. Where the secondary teeth are small, they often shrink on drying and may be missed in n casual viewing of чегин material. Besides the solanaceous cases discusse here, what may be secondary teeth occur in at least a few other flowering plant groups such as Lagerstroemia indica (Lythraceae), Hamelia (Rubiaceae), and Amphilophium (Bignoniaceae). CALYX PATTERNS IN SOME RELATED GENERA Cyphomandra. Adjacent laterals are not fused and sepals are fused to or near to the top of the calyx. Floral egress is by a combination of split- ting and stretching. Brachistus. The sepal tips are not fused and floral egress is achieved by small amounts of tear- ing and stretching. There are no evident swellings or ridges on the calyx wall, and there is no sleeve. It was partly on the basis of this primitive calyx 124 that from Y C арисит,_ al ti ea et al. (1981) separated Brachistus Tn карасин аппиит the minute sep d with minor soit Adjacent sepal laterals are usu- ally close but usually not fused. The base of the calyx may develop a slight swelling and out- bending of the vasculature from its otherwise cylindrical structure. Capsicum chinense has a conspicuous but unvascularized swelling or fer- rule where the pedicel flares into the receptacle or calyx base. Capsicum chacoense and C. bac catum have well-developed secondary teeth, which appear externally like ordinary free sepal tips, but clearing of the tissue reveals that these teeth are secondary, and the primary vasculature runs first to the tip of the teeth and then back down to the calyx and into the sleeve. The ba- sipetal minor venation seen in the teeth of species of Lycianthes was not seen in Capsicum cha- coense. The sleeve in Capsicum is inconspic- uous. Jaltomata. The bud of J. procumbens is a flat- topped cylinder. The calyx lobes are coherent for their full length, there are no discernible free tips, and prefloration is apparently complete. The lat- eral sepal veins are fused as far as the rim of the flat bud, while along both the primary sepal traces and the fused laterals, ridges run from the base of the receptacle to the rim of the bud. These ridges are suggestive of incipient teeth or umbos in the manner of the secondary teeth of Ly- cianthes. Lycianthes. Adjacent laterals are always fused and floral egress is effected by stretching of the calyx rim, which produces a recognizable sleeve. There is a tendency for the development of sec- ondary teeth from the side or base of the calyx wall and for these to be vascularized by out- bending of the primary sepal traces and the lon- gitudinal portions of the fused laterals. Lycopersicon. Calyces are like those in Sola- num, but the free sepal tips are proportionately longer. Margaranthus. Thecalyx is like that in Phys- alis but splitting is very short. Much of floral development seems to take place after egress has been initiated, and this prefloral egress can be used for routine separation of the similar Phys- alis lobata Nutt. Nicandra. Prefloration is complete and floral egress is effected by splitting at the sutures. The interconnector becomes lignified in the accres- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 cent calyx. Two well-spaced laterals appear on each side of the sepal midvein. Physalis. Prefloration is like that in Sola- num, and floral egress is by longitudinal splitting at the sutures. Adjacent laterals are mostly fused, but in some species there are two pairs of laterals as in Nicandra. The interconnector usually be- comes lignified in the accrescent fruiting stage. Solanum. Adjacent calyx teeth are fused partway up, but there is no fusion of lateral veins except for the interconnector. Floral egress is by splitting at the unvascularized sutures. In some cases there is further splitting in fruit. Exceptions are discussed below Witheringia. There is no fusion of adjacent laterals, and floral egress is achieved by stretch- ing, often producing a distinct sleeve. In some species there are umbos at the base of the calyx, and in fruit these species may be difficult to dis- tinguish from species of Lycianthes. Some species split for egress of fruit. This calyx appears to be highly evolved in contrast to the primitive calyx in Brachistus. SOME SPECIALIZED EXCEPTIONS IN SOLANUM In Solanum seaforthianum floral egress com- bines some splitting with stretching, and there are no other apparent deviations from the nor- mal pattern. But in the closely related S. bol- doense of Cuba (Fig. 19) the base of the calyx is expanded into a conspicuous cupule or ferrule, and there is a reticulum of fine venation extend- ing below the connector, which appears to be supplied by traces from both the corolla and the calyx. Solanum surinamense Steud. (Fig. 8) has a fine reticulum like the two species just men- tioned and five prominent apical unvascularized lobes. These three exceptions have floral egress by stretching. Other species with prominent bas- al cupules, S. hazenii and S. antillarum, have no departures from the normal Solanum calyx vasculature. LYCIANTHES AS A GENUS A distinctive feature of plants assigned to Ly- Witheringia. Although a few species of Solanum seem to approach Lycianthes in the manner of portal opening, they are not good candidates for consideration as intermediates between Sola- num and Lycianthes. The calyx in Solanum with 1986] IGURE 18. The exceptional calyx vasculature of olanum surinamense Steud. Note the proliferation of minor venation below the unvasculaturized teeth. its less complicated vasculature should probably be considered primitive among the genera list- ed above and that of Lycianthes to be much more advanced. ther characters separating Lycianthes are generally inconclusive. Chromosomes, wood histology, and pollen have provided little assis- tance. The stone cells invoked by Bitter (1919) occur in other genera and also in Solanum. Re- duced inflorescences, entire leaves, and spine- lessness are useful but are not generic characters. Perhaps improved chemical information will provide better demarcation. There may never- theless be some important biological differences: there are few specimens in herbaria (Nee, 1981; Symon, 1986) and many of the species flower nocturnally (Nee, 1981). Although clearly distinguished only by its ca- lyx, this character indicates a significant advance in the evolution of at least one floral organ over the less specialized genus Solanum. From a prac- tical point of view, it is useful to consider the more than one hundred species of this advanced group as a generic entity with its own name. Thi is helpful in gaining accessibility to the literature and to specimens filed in herbaria. Species of this group are with few exceptions easily identifiable by a glance at the calyx, and confusion in dealing with these exceptions generally relates to their similarity to Witheringia or Capsicum and not Ф D'ARCY —LYCIANTHES CALYX FIGURE 19. The exceptional calyx vasculature of Solanum boldoense. Note connection between calyx and corolla traces. to Solanum. The diagnostic presence of a sleeve and the frequent presence of ten lateral teeth in two series distinguish it readily from Solanum. THE CALYX IN PHYLOGENY AND TAXONOMY Studies in the Lauraceae (Kasapligil, 1951), Ranunculaceae (Tepfer, 1953), and other prim- itive families, usually with distinct sepals, gen- erally show that the sepals are vascularized by a series of traces oriented in much the same way as the traces in leaves of the same plant (see Esau, 1960, 1965). The work of Copeland (1943) and Palser (1951, 1955) in the Ericales lends support to the view that the calyx lobes in the Sympetalae are homologous with those of the more primitive families. Their work outlines a sequence of calyx evolution that parallels the trends in specializa- tion of other organs in those plants. The sequence of evolution of primary calyx vasculature in the 126 Ericales may be summarized as follows (sepals are usually five): — Three independent traces per om (15 gaps) (Befaria, Leucothoe racemosa, etc.). 2. Three traces per sepal (ten die median plus laterals derived from an intermediate sepallary trace common to two sepals (Chamaedaphne). . One trace per sepal that soon divides into a median an Gaultheria, Ledum, etc aS ^ и Mitos P = (five gaps): median с late uy £e or Ww = 1 ре tal bundles Rama pm Clethra, etc.). One trace per sepal (five gaps): no laterals (Cassiope mertensiana). + In the Ericales these branching patterns occur in the receptacles, and at the level of sepal ex- pansion the laterals may branch into five to nine veins. From stage 2, there is the possibility of advancement to either stage 3a or In the genera of Solanaceae here examined, the primary calyx vasculature is all at stage 3a, but at higher levels in the calyx, the vasculature shows two developments not seen in the Ericales: fusion of sepal laterals and enervation of secondary teeth or appendages. Fusion of calyx parts has opened new Рак e evolution of vasculature. The egress are queis in n Table b. ADVANCES AND REVERSALS OF CALYX EVOLUTION One may envision a sequence of evolutionary steps that would yield a condition so like the starting point that the advancement that had oc- curred in between would not be detected. Such a sequence may have taken place in Capsicum. It is probable that the calyx of Capsicum cha- coense evolved from a primitive stage such as one sees in Solanum to its present form by a sequence like that in Lycianthes. If the secondary teeth were not of selective value under new en- vironmental conditions they might become fused back into the calyx wall and sleeve. And if the and probably small teeth, which would be the ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 TABLE l. Degrees of calyx fusion and possibilities for egress Level of Ad- Perianth vance- Mode of Structure nt Egress Outer perianth want- ]. No confinement Outer perianth of free 1. No confinement 2 Expansion 3. Stretching Outer perianth mostly 3. Stretching 4 Splitting at sutures 4 Outer perianth fused Splitting at sutures to the top 5. Tearing irrespec- i ures 6. = tear- Special situations 7. Pal Moon 8. Cleistogamy fusion product of the secondary teeth and the primary traces, these being all that was left of the primary (in the sense of Solanum) teeth. If this were the case, could it be detected? Such a series may well have taken place in Capsicum annuum and other species that have calyces with reduced teeth, inconspicuous sleeves, and floral egress by stretching (Fig. 10). Thus the sequence hypoth- esized here, although supported only by intu- ition, calls for considerable caution in extrapo- lating evolutionary sequences to other groups. Another likelihood is that an evolutionary se- quence like that outlined for Lycianthes has tak- dication that other members of the family also have evolved elaborate calyx structure that can- not be readily assessed or understood on the basis of present calyces. The sleeve and secondary teeth of Lycianthes should probably not be regarded as unique evolutionary etudes or as phylogenetic endpoints but as stages on a larger development route travelled by many advanced plant groups. LITERATURE CITED BITTER, G. 1914. Sektion (8): Polymeris in Weitere Untersuchungen uber das vorkommen von Stein- 1986] | in Fruchtfleisch beerentragen- de r Solanac n. Abh. Naturwiss. Vereine Bremen N w og : 153, t. 1919 20]. Die Gattung Fyelantnes. Abh. 185 а Bremen 24: 2 1961. Comparative Plant Anatomy. olt, Ri nehart & Winston, New York. s 3. Astudy, anatomical and taxo- era of Rhododendroideae. Amer. 533-625. CRONQUIST, A. The Evolution and Classifica- n of Flowering Plants. Houghton Mifflin, Bos- ton DANERT, S. 1 Uber die Entwicklung der Stein- zellkonkretionen in der Gattung Solanum. Kul- turpflanze 17: 299—311. у RCY, W. G. 1972. In Е conservanda рго- posita (323), Taxon x 211. 1972. Gentry & J dy dt rett. 1981. Recog- nition of Brachistus (Solanaceae). Ann. Missouri Bot. Gard. 68: 226- DUNAL, 1852. Solanaceae. In A. P. De Can- dolle (editor), Prodromus Systematis Naturalis Regni Vegctabilis 13(1): Esau 960. tomy of Seed Plants. John Wiley ons, New . 1965. Plant Anatomy, 2nd edition. John Wi- ons, New York. ey HASSLER, dns 1917 ш Austro-Americanae. Annuaire Conserv. Jard. Bot. Genéve 20: 173- KASAPLIGIL, В. 1951. Morphological and ontogenetic studies of Umbellularia californica Nutt. and Lau- rus nobilis L. Univ. Calif. Publ. Bot. 25: 115-239. D'ARCY —LYCIANTHES CALYX 127 MCVAUGH, E . Report of the Committee for perm aophya gee 22: 153- MURRAY, M. A. arpellary and placental s struc- ture in the енты Bot. Gaz. (Crawfordsville) 07: 243-260. . Tips for collecting Lycianthes. Solana- 2: 58-59. 51. Studies in floral morphology in the Eri ¡cales I. Bot. Gaz. (Crawfordsville) 112: 447- 4 —— 1955. Studies in dues floral morphology i in the Ericale es III. proud сао ау : 335-354. RYDBERG, Р. A. North American n of hysalis e uis da pica Mem. Torrey Bot. Club 4: 297-374. SATINA, s 1944. Periclinal chimeras i in atura in re- f the style and stigma (B) of calyx and corolla. ‘Amer. J. Bot. Solanaceae in New Guinea. /n genus Aysalis in North Am Rhodora 60: 107-114; 128- 142: 152-1 . 1967. Hysalis in Mexico, С ntral America and the West Indies. Rhodora 69: 82-120; 202- 239. SEROTAXONOMY OF SOLANUM, CAPSICUM, DUNALIA, AND OTHER SELECTED SOLANACEAE! RICHARD N. LESTER? AND PHILIP A. ROBERTS? ABSTRACT S n ut 50 species of Solanaceae. Immunoabsorption data were obtained Hom reactions of seed ннн Men systems with antibody systems absorbed by stems in elationships of the taxa were computed by newly developed а Twenty species of Solanum, mostly of subg. _Leptostemonum, were compared using six antisera. Solanum nigrum (subg. Solanum) distinct from the others. Several groups of species belonging to different sections of Solanum were recognized. Phenetic and phyletic relationships within sect. Androceras were explored. Re-analysis of a complete cubic matrix of im- munoabsorption data for eight species of Solanum rearranged relationships slightly but emphasized the divergence between different sections of this genus. A broad survey was made using an antibody pg е to Capsicum annuum. This revealed the distinctiveness of Nicandra from other Solana- 1 also some unlikely ones. Further studies revealed Pani lubes of Dunalia species to Capsicum than to Iochroma. Protein characters a are valuable in plant tax- onomy for generic and even inter- family relationships (Jensen & Fairbrothers, 1983). Within the Solanaceae several serological studies have been made by workers such as Ches- ter, Hammond, Tucker, Gray, and Lester (see review ч Lester, 1 Because of the comples physico-chemical properties of proteins, and the even more com- plex phenomena of immunological reactions, many different serotaxonomic techniques have been developed (Jensen & Fairbrothers, 1983). Immunoabsorption resolves subsets of antigenic determinant sites as recognized by corresponding subsets of antibodies after the removal of com- mon antibodies from the antibody system by ab- sorption by other antigen systems. Recent de- of immunoabsorption techniques and the anal- ysis of the resultant data by appropriate numer- ical taxonomic procedures have been discussed in detail (Lester, 1979; Lester et al., 1983). For this paper these procedures were applied to se- lected species of Solanaceae. MATERIALS AND METHODS Seed samples from the Birmingham Univer- sity Solanaceae Collection (Tables 1—3) were used to prepare protein extracts, to induce antibody production in rabbits, and for subsequent im- munological experiments using absorbed anti- body systems. Procedures, especially the scoring and analysis of results, followed those describe by Lester (1979) and Lester et al. (1983). Three sets of experiments were conducted. 1. Twenty accessions of Solanum, mostly of subg. Leptostemonum (Table 1) were com- pared using six antibody systems (data pub- lished in Lester et al., 1983) 2. Eight species from diverse sections of the ge- nus Solanum (Table 2) were compared using eight antibody systems (data published in Lester, 1979). 3. About 25 species of Capsicum, Solanum, and other genera of Solanaceae, mostly of tribe Solaneae (Table 3) were compared using one antibody system (Lester, unpubl. data). Most of these data have now been analyzed by most of the procedures described by Lester et al. (1983), especially by using Jaccard’s and sim ilar coefficients to estimate phenetic eau and subsets of data, some of which are presented herein (Figs. 1-6) ! We are grateful to Sarah Marsh for growing these plants, to the Science and Engineering Research Council (U. K.) for financial support, and to Julie Brean, Barbara Kla auza, and Sarah Sutton for technica l assistance. of the Biology and Systematics of the Solanaceae nical Garden on 3-6 August 1982. 3 Department of Plant Biology, University of Birmingham , Edgbaston, Birmingham B15 2TT, England. 4 Department of Nematology, University of California, Riverside, California 92521. ANN. MISSOURI BoT. GARD. 73: 128-133. 1986. 1986] TABLE 1. sions of Solanum specie Listing and classification of the 20 acces- s that were compared immu- nologically using Si a systems to CAP, ROS, SIS, nd PRN. LESTER & ROBERTS—SOLANACEAE SEROTAXONOMY 129 TABLE 3. List of accessions of Capsicum, Dunalia, and other genera of Solanaceae that were compared immunologically using antibody system to Capsicum annuum. T, TOR, a Taxa Code Acc. No. Code Acc. No. Taxa Subg. Leptostemonum (Dun.) Bitt. ATHEN PIC S.1069 Athenaea picta S. capsicoides All. CAP S.0866 eie imi xn C baccatum annuum S. н Schl. CHL 5.0021 C Sel CHA 24 ba чи 5, viarum Dun IA 5.1418 S 0750 c | СҮРНО ВЕТ S.0045 С звеня betacea Sect. Androceras (Nutt.) Bitt. ex M. DATURSTR 5.0185 Datura stramonium Ser. Androceras DUNALAUS 8.0379 Dunalia australis S. rostratum Dun. ROS S.0097 DUNAL BRE 5.0375 р. breviflora ROS S.0399 DUNALBRF 5.0377 D. breviflora S. fructo-tecto Cav. FTO S.0025 DUNALLOR £5S.0376 D. lorentzii Violacei Whal DUNAL TUB $S.1094 D. tubulosa Ser. Violaceiflorum шеп HYOSC NIG S.0289 Hyoscyamus niger . heterodoxum Dun. HET 5.0593 — ¡OCHR SPE S.1599 Iochroma sp. S. citrullifolium A. Br. CIT 8.0195 IOCHR UMB S.1602 1 umbrosa CIT 5.0127 LYvCIU CES S.0368 208900 aue Sect. Cryptocarpum Dun. LYCOP ESC S.1152 y T icon S. sisymbriifolium Lam. SIS S.1099 y if SIS 5.0136 NICAN PHY S.0082 кош, E Fond NICOT TAB S.0329 Nicotiana tabacu Sect. Lasiocarpa (Dun.) D’Arcy PHYSA ANG 8.0512 Physalis angulata S. hirtum Vahl HIR S.1142 SALPI ORI S.0159 Salpichroa origani- S. quitoense Lam QUT 8.0972 olia S. tequilense A. Gray TEQ S.0973 SARACUMB 5.0117 Saracha umbellata | : SCOPO LUR S.0104 Scopolia lurida Sect. Ol thes (Dun.) Bitt. A One E SOLAN АЕТ $0156 Solanum aethiopicum ‹ . АМС 5.1335 SOLAN CAP $.0263 S.capsicastrum S. prinophyllum Dun. РЕМ 5.0386 TRECHSAT 5.0234 Trechonaetes sativa РЕМ 5.1444 WITHA SOM S.0242 Withania somnifera Sect. Torva Nees WITHE COC 8.1582 Witheringia coc- S. hispidum Pers. HIS — S.0017 coloboides S. torvum Swartz TOR S.0839 Subg. Solanum Sect. Solanum S. nigrum L. NIG S.0498 TABLE 2. List of eight species of Solanum that were compared immunologically using antibody systems to all eight species. Code Acc. No. Species Section TUBERO 8.0952 S. tuberosum L. Petota SCABRU S.0243 S. scabrum Mill. Solanum MAURIT S.0049 S. mauritianum Scop. Brevantherum HENDER S.0167 S. x hendersonii Hort seudocapsicum SEAFOR S.0051 S. seaforthianum Andr Jasminosolanum AETHIO S.0279 S. aethiopicum L Oliganthes SIMILE S.0211 S. simile F. Muell. Archaesolanum CITRUL S.0168 S. citrullifolium A. Br. Androceras ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 ROS 0097 L ROS 0399 FTO 0025 HET 0593 CIT 0195 CIT 0127 | 515 1099 515 0136 САР 0866 | PRN 0386 | PRN 1444 IND 1335 NIG 0498 [ QUT 0972 HIR 1142 TEQ 0973 VIA 1418 CHL 0021 TOR 0839 HIS 0017 Phenogram of immunological similarities of 20 accessions of Solanum species calculated by Jaccard’ $ a and group average clustering (for explanation see text and Table 1). RESULTS AND DISCUSSION SOLANUM SUBG. LEPTOSTEMONUM This set of data has already been described and analyzed in several wayi (Lester et al., 1983): Jaccard’ clusterin is justified on theoretical grounds and produces a taxonomically acceptable phenogram (Fig. 1). 0.6 07 0,8 4 L 1 L 0,4 TUBERO CITRUL SCABRU SIMILE SEAFOR MAURIT HENDER AETHIO FIGURE 2. Phenogram of immunological similari- ties an sieht Sau. species calculated by Jaccard’s coefficient and group average clustering (for explana- tion see text and Table 2). Solanum nigrum showed little similarity to the other species, which supports the major taxo- nomic distinction between subg. Solanum and subg. Leptostemonum. Solanum torvum and S. hispidum of sect. Torvaria were grouped together but were well separated from any other taxa. Solanum hirsutum, S. tequilense, and S. qui- toense, members of the distinctive sect. Lasio- carpa, were grouped together and separated from other taxa. Solanum chloropetalum and S. vi- arum, two morphologically similar and рагі interfertile species of sect. Acanthophora, were grouped together. Solanum capsicoides, of the same section, was placed some distance away, but the antigen system of this species produced nonspecific reactions. The two accessions of Solanum prinophyllum from Australia were placed together and were joined at a low level by S. anguivi from Africa, which is also classified in sect. Oliganthes. The two accessions of Solanum sisymbriifol- ium, sect. Protocryptocarpum, were placed to- gether and were linked with members of sect. Androceras. Two accessions of S. citrullifolium were joined by S. heterodoxum, also of ser. Vio- laceiflorum, and two accessions of S. rostratum were joined by S. fructo-tecto, also of ser. An- droceras. 1986] 5 AE е g z B £ = = in m = P go = @ X FiGURE 3. Cladogram of eight species of Solanum derived from immunological data by the Dollo method (for explanation see text er Table 2). Nearly all of these serologically indicated re- lationships made good taxonomic sense, but in most cases the various sections of Solanum are very distinct and there is little information on the affinities between them. An essay at cladistic analysis (Lester et al., 1983) using only data from the antibody system to S. rostratum suggested an evolutionary se- quence (with distances between evolutionary units as indicated in parentheses): ancestor (17), SIS S.0136, SIS S.1099, (6) HET S.0593 (3) CIT S.0127, (3) CIT S.0195, (5) FTO S.0025 (4) ROS S.0097, ROS $.0399. This is a progression from S. sisymbriifolium, a perennial plant with acti- nomorphic violet flowers and red berries, through to 5. rostratum, a normally short-lived annual plant with strongly zygomorphic yellow flowers and dry capsules. In each morphological attri- bute an evolutionary sequence can be recognized going from S. sisymbriifolium, through S. het- erodoxum, S. citrullifolium and S. fructo-tecto to S. rostratum. These conclusions are the con- verse of Whalen's (1979) evolutionary scheme, but are rueda by spermoderm studies (Les- ter, unpubl. dat t S dicat produced from these im- muno- т орати п data, were published by Lester et al. (1983). One cladogram had successive sim- ple branches to S. quitoense, S. torvum, S. cap- sicoides, S. prinophyllum, and finally S. sisym- briifolium and S. rostratum Cladistic analysis of the total data set was made by computer using the Dollo and Wagner meth- ods (Felsenstein, 1982), but the results are not presented here because the data were analyzed LESTER & ROBERTS—SOLANACEAE SEROTAXONOMY 131 0,0 0,5 1,0 FIGURE 4. Phenogram of immunological similari- ties of 22 accessions of various genera of Solanaceae calculated by Jaccard's coefficient and group average clustering (for explanation see text and Table 3). in only one sequence and the resultant clado- grams were not taxonomically acceptable. EIGHT DIVERSE SECTIONS OF SOLANUM This set of data was described and analyzed in an elementary way in Lester (1979). Re-anal- ysis, using the preferred Jaccard’s coefficient and group но clustering, provided а new den- drogram . 2). Some taxonomic groupings are maintained, such as hoe of Solanum scabrum and S. simile (sects. Solanum and Archaesolanum) and S. hendersoni, S. mauritianum, and S. aethiopicum (sects. Pseudocapsicum, Brevantherum, and Oli- ganthes), but S. citrullifolium (sect. Androceras) is now grouped with S. tuberosum (sect. Petota), which is unacceptable on morphological grounds. Solanum seaforthianum (sect. Jasminosolanum) links with the rest at a very low level. The taxo- nomic relationships between these sections have been discussed previously (Lester, 1979). In gen- eral these results suggest that most of these sec- sion of many more species, even without using any more antibody systems, would improve the taxonomic information and the stability of the classifications. 0,0 0,5 1.0 IOCHR SPE IOCHR UMB LYCIU CES DUNAL AUS ATROP BEL DUNAL BRE DUNAL TUB DUNAL BRF DUNAL LOR CAPSI CHA CAPSI BAC CAPSI ANN FIGURE 5. Phenogram of immunological similari- ties of | 12 accessions ОГ Capsicum, Dunalia, genera calculated by eae ratio (coeff. no. and group average clustering (for explanation see text and Table 3). Cladistic analysis using the Dollo method on Felsenstein’s Phylogeny Inference Package HY sequence with separate branches to S. aethiopi- cum, S. hendersonii, S. mauritianum, S. citrul- lifolium, S. tuberosum, S. simile, and S. seafor- thianum (Fig. 3 It is interesting that the three representatives of subg. Leptostemonum, a group with long thin anthers, stellate hairs, and often with prickles, appear to be more ancestral, whereas four rep- resentatives of subg. Solanum and Archaesola- Cladistic analysis by the Wagner method (Fel- senstein, 1982) produced an unrooted tree with the species in the same sequence as the Dollo method, but this sequence could be read in either direction. CAPSICUM AND OTHER GENERA These were preliminary Lu using antiserum to only Capsicum annuu A study of 19 genera, mostly of k tribe So- laneae, (Fig. 4) showed the distinctiveness of Ni- candra from the other genera, which is widely accepted. Some of the serological relationships, such as Atropa and Withania, Physalis and Sola- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 IOCHR SPE IOCHR UMB LYCIU CES ATROP BEL | DUAL TUB DUNAL AUS | DUNAL BRE DUNAL BRF DUNAL LOR CAPSI CHA CAPSI BAC CAPSI ANN FIGURE 6. Phenogram of immunological similari- ties of 12 accessions of Capsicum, Dunalia, and other genera calculated by dot product and furthest neigh- bour clustering (for explanation see text and Table 3). num, and all these with Lycopersicon, and pos- sibly also the group of Datura, Lycium, and With- eringia, agree with taxonomic dispositions based on morphological data, but others, such as the position of Nicotiana, disagree. The relationship of Dunalia species to Cap- sicum was surprisingly strong and was therefore investigated by a further study including more species of these two genera and also Jochroma, which is sometimes merged with Dunalia. The numeric data were analyzed by Similarity Ratio and Group Average clustering (Fig. 5) and by Dot Product and Furthest Neighbour (Fig. 6). In both cases Capsicum annuum was grouped with C. baccatum and then with C. chacoense, the several species of Dunalia were grouped fairly close to each other and also to Capsicum, and the two species of Jochroma were separated from the other taxa. Lycium cestroides showed a low level of similarity to anything else: the position of Atropa belladonna was different in these two nalys s. The greater similarity of the two cultivated species of Capsicum (C. annuum and C. bac- catum) than to the wild one (C. chacoense) may be significant. The relationship of Dunalia to Capsicum indicated here is interesting, because 1986] although the cultivated peppers are mostly an- nual herbs, the wild relatives are shrubs. Now that these results have suggested it, the similarity of some species of Acnistus/Dunalia/Vassobia to Capsicum becomes apparent. These relation- ships deserve further investigation by sexual or somatic hybridization experiments. The sepa- ration of Tochroma from Dunalia in current rted by these data. аљалл MP CONCLUSIONS The results presented here illustrate the value of immuno-absorption techniques in serotax- onomy, particularly for comparisons of species and genera that are too distinct to allow bio- systematic investigations by hybridization exper- iments. The relationships indicated within and be- breeding, but furthermore they are comparable to relationships of different sections within So/a- num. This emphasizes that the gigantic genus LESTER & ROBERTS—SOLANACEAE SEROTAXONOMY 133 Solanum, which is unified by a few floral char- acters such as poricidal anthers, comprises an assemblage of very diverse taxa, which could be considered as distinct genera. LITERATURE CITED FELSENSTEIN, J. 1982. Numerical methods for infer- ring evolutionary trees. Quart. Rev. Biol. 57: 379- 04. JENSEN, U. & D. E. FAIRBROTHERS (editors). 1983. Nucleic Acids and Proteins in Plant Systematics. Springer-Verlag, Heidelberg and Berlin. LESTER, R. М. The use of protein characters in the taxonomy of Solanum and other Solanaceae. Pp. 285-304 in J. G. Hawkes, R. N. Lester & A. D. Skelding (editors), The Biology and Taxonomy of the Solanaceae. Linnean Society Symposium Series No. 7. Academic Press, London and New York. , P. A. ROBERTS & C. Lester. 1983. Analysis of immunotaxonomic data obtained from spur tec. es — 300 in U. Jensen & D. E Fairbrothers (editors), Nucleic Acids and Proteins in Plant Systematics. Springer-Verlag, Heidelberg and Berlin. WHALEN, M. D. 1979, Taxonomy of Solanum sec- tion Androceras. Gentes Herb. 11: 359-426 NOTES ON THE SYSTEMATICS OF HESPERANTHA (IRIDACEAE) IN TROPICAL AFRICA! PETER GOLDBLATT? ABSTRACT Three species of Hesperantha are here recognized in tropical Africa: H. petitiana is widespread in highland areas from Ethiopia to eastern Zimbabwe; H. ballii is a local endemic of the Chimanimani Mountains in eastern Zimbabwe; and H. longicollis occurs in highlands in Malawi and Zimbabwe, extending into eastern Botswana and the Transvaal, South Africa. Hesperantha petitiana is taxonom ically complex and includes lowgrowing and small-flowered plants corresponding to Ixia pelana and Г. hochstetteriana, and tall and large- flowered forms corresponding to H. volken nsii from Mt. +} th Kilimanjaro, and H. alpina from Mt. Came eroun у, L y allied to the southern African Н. ! hla f£ th African (d to this alliance. Chromo ome numbers are reported for three populations of H. petitiana, all polyploid and either tetraploid or hexaploid, in contrast to all southern African plants so far counted, which are diploids The genus Hesperantha comprises some 55 nial corm-bearing geophytes. 1 widely in Africa (Fig. 1), species are concentrated in southern Africa. There are some 36 species in e winter rainfall region of the Cape Province (Goldblatt, 1984) and about 20 species in the well-watered areas of coastal and montane east- ern southern Africa (Goldblatt, 1982; Hilliard & Burtt, 1979, 1982). Six species of Hesperantha have been recorded in tropical Africa, from Zim- closely allied H. ballii Wild айа Н. longicollis Baker (section Radiata), and the unrelated petitiana (A. Richard) Baker (section Concentri- ca), which is variable and often treated as com- prising two or more species or varieties. Hes- perantha ballii and H. petitiana are found only in tropical Africa, H. ballii being a local endemic of the Chimanimani Mountains of eastern Zim- babwe, while H. petitiana occurs in highland areas above 8,000 ft., almost throughout eastern trop- ical Africa as well as in Cameroun. Hesperantha d " centered in de nade of ie bermeyer, 1980) bu it tenendi into Botswana to ue west and Mise to the no Both Hesperantha longicollis and H. ballii are complex of southern African species centered around H. baurii Baker. The variation pattern and taxonomy of H. petitiana is dealt with in detail in this paper, while H. ballii and H. lon- gicollis are discussed only briefly. romosome number has been determined here for three populations of Hesperantha peti- tiana, the species previously unknown cytolog- ically. A collection from Mt. Kilimanjaro (Puff s.n.) is tetraploid, 2n (4x) = ca. 50. Two more populations are hexaploid, one from Ethiopia (Puff et al. 820911-1/1) 2n (6x) = ca. 72 and the other from Mt. Cameroun (Thomas sub Gold- blatt 7272), 2n (6x) = ca. 76. Basic chromosome number in Hesperantha is x = 13 (Goldblatt, 1984) and all of the many other species so far counted, all from southern Africa, have numbers at the diploid level. The counts for H. petitiana are interesting because they are the first reports nately, the high numbers and small chromo- ! Supported by grant DEB 79-10655 from the United States National Science Foundation. I thank B. L. Burtt for his constructive criticism of the manuscript and Chris Puffand Duncan Thomas for seed of the plants studied cytologically. B. A. Krukoff Curator of African Botany, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. ANN. MISSOURI Bor. GARD. 73: 134-139. 1986. 1986] somes make it difficult to establish an exact count and so a second base number in Hesperantha remains uncertain. HESPERANTHA LONGICOLLIS Hesperantha longicollis, typical in Hesperan- tha in being evening blooming, is closely related to the widespread southern African H. radiata, which extends from Namaqualand on the west coast, through the southern and eastern Cape to Swaziland in the eastern escarpment. The two have, in common, a curved perianth tube and unusual floral bracts, the outer of which have i io partly united around the axis. The two species can readily be identified by a series of diltingulshing features. In H. longicollis the flow- er has a longer perianth tube, 18-25(-30) mm long, well exserted from the bracts; the outer bract is united around the axis only near the base; the leaves are relatively long, usually about half as long as the stem or longer, and plane; and the corm tunics are typically spiny below. In H. ra- diata, the perianth tube is 10-18 mm long, and only slightly exceeds the bracts; the outer bract has margins united around the axis for half to two-thirds its length; the leaves are typically short, about one-third to half as long as the stem, and tend to be thicker in the midrib area; and, at least in populations from eastern southern Africa, the corm tunics are not spiny below (Goldblatt, 1984) although some southwestern Cape forms do have a corm with a spiny base. Hesperantha longicollis grows in moist habi- tats, either in vleis, along streams or in seeps, and it blooms at the end of the dry season, typ- ically in August or September. It is most com- mon in the southern African highveld (Fig. 1) and has been recorded from the Transvaal, Northern Cape, extreme eastern Botswana, cen- tral and western Zimbabwe, and recently from Malawi, where it was collected by R. K. Brum- mitt on the Nyika Plateau (Brummitt 10829) flowering in May. This represents a significant range extension into tropical Africa of what has been regarded as essentially a southern species. The species was recently reviewed for ““Flow- ering Plants of Africa" (Obermeyer, 1980), in which a full description and synonymy were pro- vided. This need not be repeated here; however, it should be noted that of the three synonyms cited, Hesperantha matopensis, H. widmeri, and H. sabiensis, the last does not apply to H. /on- gicollis. It is a later synonym of H. bulbifera GOLDBLATT — HESPERANTHA 135 (Goldblatt, 1984), a rare species, unrelated to H. longicollis, found on damp cliffs and waterfalls in the Eastern Cape and Transvaal (Goldblatt, 1984) HESPERANTHA BALLII Hesperantha ballii is a rare local endemic of the Chimanimani Mountains of eastern Zim- babwe (Fig. 1). It is a small plant, only some 12- 25 cm high, with spikes of one to two flowers to H. longicollis and H shares the distinctive floral bracts characteristic of section Radiata, united to some extent around the spike axis, and curved perianth tube. In H. ballii, the perianth tube is 11-15 mm long, reach- ing a little beyond the apex of the bracts; the leaves are about 1 mm wide; the outer bracts are united for about 3 mm, about one-fourth their length; and the nearly globose corms apparently have tunics without spines below. The species seems most closely related to H. radiata, judging by the similarity of their corms, the length of the perianth tube, and the well-developed union of the outer bracts. It can readily be distinguished from H. longicollis by its small size, 1-2-flow- ered spike, and flowers with a perianth tube only a little longer than the bracts. It is separated from H. radiata also by the few-flowered spike, fili- form leaves, and the outer bracts being united for ca. 3 mm, only about one-fourth their length, while those of H. radiata are united for half to two-thirds their length. HESPERANTHA PETITIANA As outlined in the introduction, Hesperantha only with difficulty, if at all, from some collec- tions from South Africa and Lesotho. The dis- tinguishing features of H. petitiana appear to be its erect, comparatively thick and straight, un- branched stem, straight, 1-3(rarely to 8)-flow- ered spike, and actinomorphic, straight-tubed pink or white flowers. Like the other pink to reddish or purple-flowered species of the H. bau- rii complex, the flowers open during the day and close at night. Whether the white-flowered forms of H. petitiana are also day-blooming is not 136 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 © H. petitiana Ш н. longicollis Ф Н. ballii о 2001400 600 $00 1000 KM FIGURE 1. known. Species or races of Hesperantha with white flowers are often evening-blooming and those with colored flowers are day-blooming (Goldblatt, 1984) but white-flowered plants in eastern south Africa are generally day-blooming (Burtt, pers. comm The variation pattern in Hesperantha peti- ti I ted lauth to admit more than one species or variety in tropical Africa. Originally two species, based on collections of either tall or short plants, were recognized in Ethiopia (Richard, 1850): [xia petitiana and I. hochstetteriana. Baker (1898) reduced the dwarf treatment followed by Cod iud (1972) in his “Enumeratio Plantarum Aethiopiae Spermato- yta Collections from Cameroun and Tanzania were subsequently described as separate species, Hes- Distribution of the tropical African species of Hesperantha. perantha alpina (as Geissorhiza alpina) from Mt. Cameroun by J. D. Hooker in 1864, and H. vol- kensii from Mt. Kilimanjaro by Harms in 1894. Hesperantha kilimanjarica, described by Rendle in 1895, is clearly identical to H. volkensii. Baker (1898) recognized both H. alpina and H. volken- sii in his treatment of the genus in “Flora of Tropical Africa." Later, Foster (1948) reduced H. volkensii to varietal status in H. petitiana, commenting that he had some misgivings about even recognizing the variety. Here, I suggest that H. petitiana be treated as a single variable species including both H. alpina and H. volkensii. An analysis of the variation pattern of H. petitiana is presented below, in which the Ethiopian pop- ulations are discussed first. ETHIOPIAN COLLECTIONS Hesperantha petitiana was first discovered in Ethiopia, where it was collected by Schimper and 1986] by Quartin-Dillon and Petit in the mid-nine- teenth century. In 1850, Achille Вива ае- scribed two species of xia) from their collections, Г. hochstetteriana based on Schimper's Г. uniflora ined., a dwarf form with solitary flowers, and Г. petitiana, based on taller, 2-3- flowered plants collected by Quartin-Dillon d plants, about 30 cm high, with 2 spikes and four plane, soft-textured ies the uppermost almost or entirely sheathing. The flowers have a tube ca. 9 mm long and tepals about 10 mm long and are subtended by her- baceous bracts 10-15 mm lon ost of the collections made by Schimper comprise dwarf plants under 10 cm long, usually with only three leaves, flowers with tepals ca. 10 mm long, and a tube 6-8 mm long. The type of I. hochstetteriana, Schimper 185 from Mt. Bach- it, Semien ‘Bouahit, provinciae Semiene' (also labelled *Hesperantha uniflora Hochst. 1239") consists of such plants. However, a collection from ‘Berg Gunna,’ Schimper 1182 (B), com- prises a range of plants from 7.5 to 18 cm high and with three or four leaves. The taller individ- uals of the collection are interchangeable with plants from the type collection of H. petitiana, while the smaller cannot be distinguished from I. hochstetteriana. Another Schimper collection, 579 from ‘Acallo Meda’ (P), also consists of plants of variable size. It seems likely on the basis of the available information and collections that the tall Hesper- antha petitiana is the dwarf var. uniflora com- n, representing size extremes p that collections of plants of variable size repre- sent a mixture of two species but in the absence of supportive evidence this is unlikely. The ap- parently continuous variation in some popula- tions and the absence of any consistent morpho- logical distinctions leaves little reason for the recognition of var. uniflora. It seems likely that the collections represent the range possible in a а ias фе : А plants аге subject to а variety of soil and climatic conditions that influence their growth into taller plants often with large flowers or shorter plants usually with smaller flowers. mong recent collections of Hesperantha from Ethiopia there are both dwarf and taller speci- mens, but no gathering consists entirely of very GOLDBLATT—HESPERANTHA 137 small plants, as do some of the Schimper collec- tions. Several collections, such as de Wilde 8109 and Westphal & Westphal-Stevels 1652, consist of both small-flowered plants that match Ixia hochstetteriana closely, and taller individuals that have larger flowers with tepals 12-15 mm long. Occasionally collections from Ethiopia are particularly robust (de Wilde 6574 consists of plants with up to eight flowers on a spike) or have flowers that seem beyond the normal range expected for H. petitiana (e.g., de Wilde 6574; Hedberg 4245) with tepals 15-16 mm long, and bracts 12(-15) mm long. Such collections cor- respond well with most specimens of Hesper- antha collected in East Africa that have been described as the separate species, H. volkensii Harms. The significant questions concerning the tax- onomy of Hesperantha in tropical Africa are the following. Are smaller Ethiopian specimens matching H. petitiana different in any taxonom- ically significant way from the taller and very large-flowered plants form Ethiopia, East Africa (H. petitiana var. volkensii of several authors), and the Cameroun highlands (H. alpina)? A sec- ond problem concerns the relationship of the tropical African plants to any southern African species, of which there are several that are ob- viously closely allied. EAST AFRICAN COLLECTIONS Most speci fi East Africa latively uniform in flower size, but variable in height and leaf width and thickness. Plants matching the types of Hesperantha volkensii and H. kiliman- jarica, both from Mt. Kilimanjaro, vary in height but reach a maximum of 45 cm, have four leaves, the lower two basal, the third partly sheathing and inserted near the base, all narrow, 2-3 mm wide and with clearly raised margins and midrib, while the fourth is entirely sheathing and inserted in the upper part of the stem. The bracts are (1012-15 mm long, and like the larger Ethio- pian plants, the flowers are either white or pink- ish purple, with a tube 9-10 mm long, tepals 15 mm long, and anthers 5-6.5 mm lon Plants essentially matching the Kilimanjaro specimens have been collected in highland Ugan- da, southern Sudan, and Kenya, on Mt. Elgon; the Aberdares, especially Mt. Kinangop; Mt. Kenya and elsewhere. Specimens are sometimes very dwarfed (only 4—10 cm high in Gillett 18473, Bickford 34) and with leaves as little as 1 mm 138 wide, but still with raised margins and midribs, although in several collections a whole range of and flower sizes is present (e.g., Hedberg 1953; Gillett 16912) and it seems that the taller plant matching the types of H. volkensii and H. kilimanjarica, as well as the smaller ones, all belong to the same species. Occasionally, as in Battiscombe K7 15 (Kinohop Plateau), Napier 719 (Kinangop), Archer 676 (Namanga Hill), and a few others, the leaves are broader, and the mar- gins and midrib less obviously raised, as in most Ethiopian specimens, but these are also con- nected by a series of intermediates to the typical Kilimanjaro form. n Tanzania, the tall, Kilimanjaro form ap- pears common at higher с throughout the Kilimanjaro Range and 1 outhern Highlands, is sometimes robust ei ibn leaves 3-4 mm wide, but also sometimes small. A col- lection made by Schlieben (4978) on Kiliman- jaro is of especial interest. Plants range in size from 8-20 cm high and the tallest plants appar- ently match H. volkensii in every respect. The shorter individuals have smaller flowers, and in specimens at the Zurich and Stockholm herbaria the tepals may be as short as 10 mm. These plants can barely, if at all, be distinguished form the type material of the dwarf [xia hochstetteriana from Ethiopia. In the Southern Highlands plants tend to have more soft-textured leaves, 3-5 mm wide, and often have spikes with 6 or more flowers, but the fewer-flowered individuals appear to match in all respects plants from Kilimanjaro. In Malawi, the southern Tanzania form has been collected in all the higher areas including the Nyika Pla- teau, the Dezda Mountains, and Mt. Mlange. Further south in Zimbabwe, apparently the same form has been collected along the eastern high- lands from Inyanga to the Vumba Mountains, where it is currently identified either as H. pe- titiana or as the southern African H. baurii, a name in current use for a complex probably in- f H. petitiana, but the entire H. baurii complex awaits further study before the systematics of the genus in this area can be resolved. However, I suggest that, for the present, all collections of Hesperantha occurring from Zimbabwe north to Kenya be assigned to the single taxon, H. petitiana. Hesperantha alpina from Cameroun is poorly ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 known and only a few gatherings have been made. Specimens in the Berlin collection comprise dwarf, 3- or 4-leaved individuals in fruit (Mann 2134) and very э plants with buds or closed flowers (Preuss 968). The dwarf plants match Ixia р в well, while the tall plants are a fair matc H. volkensii but appear to have rather small ribns perhaps not fully de- veloped. The morphology and the range of size in these specimens corresponds closely with the more ample Ethiopian and Kenyan collections and I cannot distinguish the Cameroun plants other than by their origin. Hesperantha alpina is р reduced here їо synonymy in Н. ре- titia The treatment of the variable H. petitiana complex as a single species throughout tropical Africa seems the only consistent way in which to deal with the degree of variability encountered in the complex. None of the variation is strictly geographical, and both tall, large-flowered plants and dwarf, smaller flowered plants may be found almost throughout its ie though the latter appear more frequent in The taxonomy and nie of Hesper- antha petitiana is confusing and the extensive synonymy and a description are presented below: Hesperantha petitiana (A. Richard) Baker, J. Linn. Soc., Bot. 16: 96. 1878 et Fl. Trop. Africa 7: 348-349. 1898; Cufodontis, Enum. Pl. Aethiopiae Sperm. 2: 1588. 1972. Ixia petitiana A. Richard, Tent. Fl. Abyss. 2: 309- 3 a- Sent, ‘Ixia petitiana Nob.” Quartin- Dillon & Petit s.n. (lectotype, P, here des- ignated). Ixia lpr енн A. Richard, Tent. Fl. Abyss. 2: 50. TYPE: Ethiopia, Mt. Bachit: Semien м. Bouahit, Prov. Semiene), Aug., ‘Hesperantha uniflora Hochst.’ Schimper 1239 (lectotype, P, Trop. Africa 7: . 1898; Cufodontis, Enum. Pl. dun dus Sperm . 1972. о uniflora Hochst. ms шон 185 еїс.); ard, Tent. Fl. Abyss. 2: 3 рна alpina Hook. f., J. Р 1864. Hesperantha alpina (Hook. f.) i rika 8 TYPE: Cameroun. Cam de M 1862, я 2134 ea, K; isolec- B). 1986] Hesperantha volkensii Harms, Bot. Jahrb. Syst. 19 (Beibl. 47): 28. 1894. Hesperantha petitiana var. volkensii (Harms) Foster, Contr. Gray Herb. 166: 22. 1948; Р Symb. Bot. Uppsal. 15: 68- 69. 1957. TYPE: Tanzania, ‘Kilimandscharo’: 2,440 m, Volkens 783 (lectotype, B; О ВМ, С). Hesperantha е Rendle, J. Linn. Soc., Bot. 402. 1895. TYPE: Tanzania. Mt. Kilimanjaro: к че i 10,000 ft., 1888, Taylor s.n. (holo , BM). Geissorhiza abyssinica R: Br. ex A. Richard sensu Klatt, a 34: 716. 1866, non G. abyssinica R. Br. mas "Richard (Tent. Fl. Abyss. 2: 308. 1850). [= Lapeirousia abyssinica (R. Br. ex A. Richard) Baker, the types being: Ethiopia. Near 'Maygoua- goua’: Sept., Quartin-Dillon & Petit s.n., Selleuda prope Adowa, Schimper s.n.]. Plants (8—)12-30(445) cm high. Corm 7-12 mm diam. with dark brown, concentric, usually woody tunics. Cataphyll membranous. Leaves (3-)4, the lower two basal, the third usually inserted near the base or in the lower part of the stem and sheathing in the lower half, the fourth leaf, if present, inserted in the mid to upper part of the stem and often entirely sheathing, 2-6 mm wide, usually about half as long as the stems, margins flowered, straight to slightly flexuose; bracts her- baceous, t d the apex, (9-)12-15 mm long, the inner narrower and slightly shorter. Flowers actinomorphic, stellate, white to pink, outer tepals flushed dull red to brown on reverse; perianth tube cylindric, 6—9 mm long; tepals subequal, (9-)12-18 mm long, narrowly ovate to elliptic, 5—7 mm at widest point. Filaments 3-4 mm long; anthers (3-)4-7.5 mm long. Ovary 2-3 mm long, style branches 6 mm long. Capsule obovoid, 7-10 mm long, some- GOLDBLATT—HESPERANTHA 139 what shorter to slightly exceeding y: bracts. Chromosome number 2n = ca. 50, ca. 72- Flowering time. July-December in Ethiopia; May-August and December-February in Kenya, Uganda, and northern Tanzania; March-May in 8. es Malawi, and Zimbabwe; September in Cameroun. а Highlands in Ethiopia, south- ern Sudan, eastern Uganda, Kenya, Tanzania, Malawi, Cameroun, and eastern Zimbabwe. Fig- ure LITERATURE CITED BAKER, G. J. 1898. Irideae. /n W. T. Thiselton-Dyer (editor), Flora of 7 pirak Айтса 7: 337-37 CUFODONTIS, G. 1972. Iridaceae. In Enumeratio Plantarum Aethiopiae Spermatophyta. Bull. Jard. 57. Foster, R. C. ae V. Some new or noteworthy species of Hesperantha. Contr. Gray Herb. 166: 3-27. GOLDBLATT, P. Corm morphology in Hesper- antha (Iridaceae-Ixioideae) and a proposed infra- generic taxonomy. Ann. Missouri Bot. Gard. 69 370-378. 1984. A revision of Hesperantha rer) in ‘the winter rainfall area of southern Afric S. African Bot. 50: 15-14 HiLLIARD, О. M. & B. L. Burtt. 1979. Notes on some plants of southern Africa chiefly from Natal: VIII. Notes Roy. Bot. Gard. Edinburgh 37: 284- 325. 1982. did on some plants of southern Afri from gres IX. Notes Roy. Bot. Gard. Edinburgh 40: 2 OBERMEYER, A. A. 1980. ma longicollis. Fl. Pl. Africa 46: tab. 1810. Krarr, Е. W. . lrideae. /n T. Durand & H а (editors), Conspectus Florae Africae 5: 143- 230. RICHARD, A. 1850. Tentamen Florae Abyssinicae. 2. Be rtrand, Paris. CYTOLOGY AND SYSTEMATICS OF THE MORAEA FUGAX COMPLEX (IRIDACEAE)! PETER GOLDBLATT? ABSTRACT The Moraea fugax complex, widespread in the winter rainfall area of southern Africa, is treated suggests that aneuploid decrease occurred in entiated from the ancestral stock during alternating dry à and = Dn 6, and 5. Subspecies filicaulis has haploid numbers of n = 9, 6, and 5 former perhaps basic for this taxon. Some correlat several morphological lines as geographical races differ- sed tion of morphological and karyotypic variation 5 of the Qua rte ernary. A revis classification ofthe complex is presen the kary rph ted int ological and mo ogical variation. The des of M. juga is not completely known, but the major features are probably reflected in the data presented here The Moraea fugax complex (Moraea sect. Subracemosae sensu Goldblatt, 1976a) is one of the most taxonomically complex groups in this pan African genus of some 120 species. Moraea fugax is the major taxon of the section with one or more distinctive segregates sometimes rec- ognized as varieties or separate species. The sec- tion has been variously treated in the past as comprising one species and four varieties as in “Нога Capensis” (Baker, 1896) or as three species (Baker had added two new species to the complex by 1906). In the most recent revision of Moraea in the winter rainfall area of southern Africa, only two species (Goldblatt, 1976b) were recognized, the uniform and local M. gracilenta Goldbl. and the widespread and variable M. fugax (de la acq. [(7 M. edulis (L. f.) Ker]. The com- plex is distinctive in its vegetative habit. The branches are short and often clustered in a semi- mbellate manner, and the one or two foliage leaves are inserted near the stem apex at the point of branching, usually well above the ground. The S smaller inner tepals, and large flattened style ' Support eres John Rourke and the staff of the Compton Herba elp on several field ааны ш the classification of the complex branches with well-developed crests, but the cap- sules are distinctive in having a well-developed beak. Morphological variation is extensive and confusing. Several races can be distinguished in M. fugax but the existence of many intermedi- ates has made recognition of additional species or subspecies difficult and impractical. extensively in the field. Asa result I have a deeper understanding of the patterns of morphological 1 inn ann h Малая D | 11 J OIC cytological data. This new information has made a review of the complex necessary. I now rec- ognize three species in the complex: M. graci- lenta, the new M. macrocarpa, comprising dwarf, blue-flowered and often unbranched plants with long capsules and nearly sessile spathes; and M. ‚о subspecies, н (Baker) Goldbl., i f plants with small, white to cream or rare ly blue flowers, small capsules, and stalked spathes, and subsp. fugax for larger flowered, robust plants with either white, blue, or yellow flowers and large capsules. Along the west coast several races or forms of subsp. fugax rted by grants DEB 79-10655 and 81-19292 from the United States National Science Foundation. I arium, Kirstenbosch Botanic Gardens, for their hospitality trips in South Africa, especially Dee Snijman, who helped . I also acknowledge with gratitude the collaboration of Margo Branch who provided the plant illustrations used in the aper. ? B. A. Krukoff Curator of African Botany, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. ANN. MIssouRI Bor. GARD. 73: 140-157. 1986. 1986] GOLDBLATT — MORAEA FUGAX COMPLEX 141 TABLE l. Chromosome number and collection information for previously pur and original counts presented in this paper for Moraea sect. Subracemosae. Original counts are in bold p Diploid Number Species (2n) Collection Data M. gracilenta 20 below Piekenierskloof Pass, Goldblatt 3279 ME 20 riverside, Clanwilliam campsite, Goldblatt M. macrocarpa 20 mountains W of Trawal, Goldblatt 5661 (MO); farm Reiers Rus near Worcester, Goldblatt & Snijman 6961 (MO); near Piketberg, Anon s.n. ex hort. Kirsten- bosch M. fugax subsp. 20 Olifants River Bridge, near Klawer, Goldblatt 2828 (MO); mountains W of fugax Trawal, Goldblatt 5667 (MO); NW of Nieuwoudtville, Goldblatt 5843 (MO) 18 near Wallekraal, Namaqualand, Goldblatt s.n. no voucher; Bokbaai road, Gold- blatt 5850 (MO); Donkergat Peninsula, Goldblatt s.n. no voucher; near Rob- ertson, Goldblatt 5860 (MO); near Brandewyn River, Goldblatt 4812 (MO); Cedarberg at Algeria, Viviers s.n. (MO); farm Reiers Rus near Worcester, Golablatt & Snijman 6962 (M 16 8 km N of Malmesbury, Goldblatt 3025 (MO). 16 commonage at Malmesbury, Goldblatt 4813 (MO); 20 km NW of Malmesbury, Golablatt 4081 (MO); sandy soil at epi Goldblatt 5662 (MO); Pakhuis Pass near Leipoldts Grave, Goldblatt s. 14 Cape Point Reserve, Goldblatt 5403 (MO Сава Zwartberg, Swartrivier, Goldblatt 5782 (MO); near Velddrif, sandy plains, Goldblatt s.n. (MO). 12 Hopefield, Goldblatt 152 (J). 12 foot of the Elandskloof Mts., Elandsberg farm, Goldblatt 5851 (MO); near Bellville, Delpierre s.n. no voucher; Silvermine, Cape Peninsula, Goldblatt 164 (MO) 10 (cytotype A, southwestern Cape populations): Malmesbury, near Abbotsdale, Goldblatt 5116 (MO); foot of Houw Hoek Pass, Goldblatt 4292 (MO); slopes of Koeberg, N of Cape Town, Goldblatt 4080 (MO); Kommetje, Cape Pen- insula, Goldblatt s.n. no voucher; Paarl flats, Goldblatt s.n. no voucher; Cape Peninsula, foot of Klaasjagersberg, Goldblatt 5264 (MO). (cytotype B, Olifants River Valley populations): foot of the Winterhoekberg N of Klawer, Goldblatt 5778 (MO); Olifants River Valley, near Clanwilliam, Goldblatt 5936 (MO) M. fugax subsp. 18 Ramskop, Clanwilliam, ipeo da (MO). filicaulis 12 Gifberg slopes, Goldblatt 12 granite hills near is po 4254 (MO); Kamiesberg, Rooiberg slopes, Goldblatt 4308 (MO); Richtersveld, Eksteenfontein road, Goldblatt 5717 (MO). 10 flats below Wildeperdehoek Pass, Goldblatt 5761 (MO); hills W of Trawal, Kleipan road, Goldblatt 5666 (MO). can be recognized, each distinguished by small differences in morphology such as flower color, vegetative size, leaf number, and occasionally in the relative proportions of floral parts. The dif- ferent races displace one another either geo- graphically in suitable habitats or less often tem- rally. complex has been limited (Goldblatt, 1971, 1976a, 1976b). Moraea gracilenta was reported to have a diploid number of 2л = 20, and x = 10 is presumably basic for the complex. But 2n = 16 was recorded in one population of M. fugax and 2n = 12 in two others, one a large white- flowered form and the other corresponding to the dwarf and slender stemmed M. filicaulis Ba- ically uniform (2n = 20), but the more wide- spread M. fugax now app ly variable 142 chromosomally. Chromosome numbers range from 2n = 20 to 2n = 10 in subsp. fugax and 2n = 18, 12, and 10 in subsp. filicaulis. Variation in the karyotype is correspondingly extensive, and karyotypes range from predominantly ac- rocentric complements with 2n = 20 or 18 and medium to small chromosomes, to those with 2n = 16, 14, 12, or 10 and one to three large The tot tof ct | чы constant (Table 2) in all karyotypes, based on a comparison of total chromosome length. Polyploidy has played no role in numerical changes in the complex that thus stands out as a group where aneuploidy alone has apparently resulted in the extreme karyotypic variability. RELATIONSHIPS The complex, as sect. Subracemosae, was as- signed to subg. Moraea (Goldblatt, 1976a) large- ly because of its generalized flower and basic chromosome number, x = , which corre- sponds to the base number for the genus. It is apparently allied to sect. Deserticola (five species), which occurs in the drier parts of the Cape west coast and in southern Namibia. The species of sect. Deserticola all have a basal or nearly basal leaf and an open branching system but resemble sect. Subracemosae in the tendency to develop a short beak on the capsule, especially M. mac- gregorii Goldbl. GEOGRAPHICAL DISTRIBUTION The section is restricted to the winter rainfall region of southern Africa and extends along the west coast from just south of the Orange River to Knysna on the south coast (Fig. D, a distance of some 900 km. Moraea g a limited range at the foot of the mountains and inter- montane valleys from Tulbagh to Clanwilliam in the Olifants River Valley. Moraea macrocarpa is equally restricted in its range and occurs in deep, coarse sand mainly on the west coast from Saldanha to Clanwilliam but also in the Breede River Valley near Worcester. Moraea fugax is found throughout the range of the section. Rec- ords are few and scattered east of the Caledon and Worcester districts but M. fugax has been collected at Knysna on the south coast some 400 km east of Cape Town and has even been re- ported from Humansdorp (Moriarty, 1982), but specimens are needed to confirm this. The com- plex is clearly infrequent east of Caledon and ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 TABLE 2. Mean total length of the chromosome complement in selected populations of Moraea fugax and M. gracilenta. Length Length Species (cm) (um) M. gracilenta 2n = 10 (Clanwilliam) 19.4 87.4 M. fugax subsp. fugax Large white-flowered form 2n — 20 (Klawer) 23.2 104.5 2n = 20 (Trawal) 21.8 90.2 2n = 18 (Donkergat) 19.1 86 2n = 14 (Velddrif) 20.8 93.6 Late-blooming blue-flowered form 2n = 12 (Silvermine) 23.6 106.2 Large yellow-flowered form 2n = 10 (Houw Hoek) 22.4 100.9 M. fugax subsp. filicaulis 2n = 12 (Kamieskroon) 19.8 88.9 evidently less variable there. Subspecies filicaulis is common in Namaqualand, but it extends south along the west coast into the Olifants River Val- ley south to Clanwilliam. Its range overlaps that of subsp. fugax and the two are sympatric at several localities. METHODS CYTOLOGY All chromosome observations were made at mitotic metaphase i in root tip squashes. Root tips from either p were harvested in midmorning ‚апа pretreated i in 8-hydroxyquinoline at refrigerator temperatures for seven to eight hours, before fixation in eth- anol-acetic acid (3: 1) for two to three minutes. The tips were then either stored in 7096 ethanol or immediately hydrolyzed in 10% НСІ for six minutes before being transferred to water. Root apices were squashed in lacto-propionic or- cein (Dyer, 1963) or FLP orcein (Jackson, 1975). With few exceptions, the plants used in the study were collected in the wild by myself (Table 1), and voucher specimens were made for plants in suitable condition. Unvouchered accessions will be grown, wherever possible, to flowerin and then pressed. Vouchers have been placed in the Missouri Botanical Garden Herbarium (MO). The method used here differs from that em- ployed previously (Goldblatt, 1971, 1976a) in 1986] ОЛАРА Д5: “Оһ 1 [| И: A A M. fugax GOLDBLATT — MORAEA FUGAX COMPLEX M. grac. M. macroc. : | | ТЕ ZA | did FIGURE 1. my studies of Moraea cytology, so that mea- surements are not directly comparable. All fig- ured metaphases are drawn at the same scale. Metaphases selected for illustration are repre- sentative of a particular population or karyotype group (Figs. 2, 4, 5). Minor differences between karyotypes of different populations have not been figured for reasons of economy of space, and while these differences may be significant, they are not of the same magnitude as those dealt with here. TAXONOMY Li t ulations was examined before preservation, and illustrations were made from living plants. Spec- imens from the herbaria with important south- ern African collections were also studied. Mea- surements were made from live plants when possible but also include the variation found in ial of all species and Geographical distribution of Moraea fugax, M. gracilenta, and M. macrocarpa. dry material, in which the delicate parts of the flowers shrink some 10-1596. Flower color fades progressively in dry specimens, eventually changing completely, sometimes becoming dark- er and usually disappearing. Color notes on col- lected speci desir: nd are frequently mentioned by collectors. aterial examined is cited according to the grid reference system based on geographical de- gree co-ordinates oflatitude and longitude in cur- rent use in southern Africa (Edwards & Leistner, 1971). KARYOTYPES A chromosome number of 2n = 2x = 20 oc- curs in Moraea macrocarpa, M. fugax subsp. fu- gax, and M. gracilenta, and I believe it is basic for Moraea and for the complex. Moraea fugax subsp. fugax is cytologically heterogeneous and 144 ИН | WL ce A ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 O cuu |! | B) wu usu y инине FIGURE 2. Chromosome cytology of the Moraea fugax comp . B-E. HAE white- (pink adipe: form r Klaw —E. V n= 10. B-I. Moraea fugax subsp. fugax = 10.—C. Donkergat dinates n= 9. — ы form.—F. Cape Point, form aw = 7.—G. "Silvermine, Cape PE n= H, ИГҮ, LU MAA {a нии lex.— А. Mitotic chromosomes of M. macrocarpa, ip Velddrif road, п = 7. F, G. Blue-p 6. I. Small white-flowered —H. Malmesbury, n = 8. т Near Gouda, л = 6. Scale = 10 um. also has 2n = 18, 16, 14, 12, and 10 (Figs. 2, 4). Moraea с subsp. filicaulis has 2n = 18, 12 Ka S vary considerably but are ai similar for each diploid number. They are briefly described in the following para- graphs. Voucher data and chromosome number for all populations examined are presented in Table 2 MORAEA GRACILENTA, M. MACROCARPA & M. FUGAX SUBSP. FUGAX 2n = 20 CYTOTYPE Eight populations, including two of Moraea gracilenta, three of M. macrocarpa (Fig. 2A), and three of M. fugax (Fig. e ds 2n — 20 and ongly acrocentric m S : length is appreciable, the longest being about twice the shortest. Four to five large chromosome pairs (7-8 um long), one to two medium pairs, and four (or five) short pairs (3-4 um long) can be recognized. inser not consistently observed, are on a short p MORAEA FUGAX SUBSP. FUGAX 2n = 18, 16, 14 & 12 CYTOTYPES 1. 2n = 18 karyotypes (tall white- and smaller yellow-flowered forms). The seven populations of Moraea fugax subsp. fugax with 2n = 18 have karyotypes similar to those with 2n = 20. There are five long and four short pairs of acrocentric chromosomes. One individual of a population from Donkergat Peninsula (Goldblatt s.n.) is structurally heterozygous with 11 long and seven ort ch osom wo other individuals of the population had ten long and eight short chro- mosomes. Satellites were seen clearly only in the 1986] Donkergat population on the short arms ofa long chromosome pair (Fig. 2C). 2. 2n = 16 karyotypes (white- and pink-flow- ered forms). Five populations have 2n = 16. Three of these are a distinctive white-flowered form with narrow tepals from the Malmesbury area. The karyotype of m form lo 1976a; Fig. 1G) with conspicuously “large satellites, four only slightly shorter acrocentric pairs, and three short acrocentric pairs (Fig. 2H). Another population matching this form has 2n = 12 (Fig. 21) (see below A second karyotype, with 2n = 16, was found in a single unusual, pale pink-flowered popula- tion from the Olifants River Valley (Goldblatt 5662) and consists of six medium to long acro- centric and two short acrocentric pairs, one of which has small satellites (Fig. 2D). This karyo- type is quite different from the white-flowered Malmesbury form with 2n = 16 and must be of independent origin. yellow-flowered population from Pakhuis Pass 1 has и = 16 (Fig. 4А). The karyotype comprises one long a pair, and the remainder are strongly acrocentric. One of the longer of these pairs has a small satellite located at the end of a long arm, a feature of most pop- ulations of yellow-flowered plants studied. 3. 2n = 14 karyotypes (white, yellow, or blue forms). Three populations, each morphologi- cally distinct, have 2n = 14. One of these, a large yellow-flowered form from Caledon Zwartberg (Goldblatt 5782), has one pair of large metacen- trics (16 um long), five medium-sized pairs (8- 10 um long) two of which are acrocentric, two telocentric, one metacentric, and a tiny acrocen- tric pair (Fig. 4B). There are satellites on one of the telocentrics. Other plants matching this rel- atively short-stemmed but large yellow-flowered form have 2n = 10 and three large metacentric chromosome pairs A second population with 2n = 14 is a blue- flowered and late-blooming form from the Cape pair (Fig. 2F). Satellites were not readily seen but appeared to be on the long arm of one of the medium-sized acrocentric pairs. Other popula- tions of this form have 2n = 12, and their karyo- types differ by a very long metacentric pair. The third population with 2n = 14 was col- lected near Velddrif on the west coast and had GOLDBLATT — MORAEA FUGAX COMPLEX 145 large white flowers. The karyotype comprises a long met pair, two long and one medium pair, and two short pairs (Fig. 2F). Most popu- lations of this tall white-flowered form have 2n = 20 or 18, but 2n = 12 has been recorded from one paga (Goldblatt, 1971). 4. = karyotypes (large white- or blue- flowered peus Four populations have 2n = 12. Three of these, all from the southwest Cape, represent the large blue- (to white-) flowered and moderately robust late-blooming form common in sandy soils. The karyotype (Goldblatt, 1971) is similar in all populations. There are two long metacentric pairs, three medium-sized acrocen- trics, and a tiny acrocentric pair (Fig. 2G). Sat- near Gouda (Goldblatt 5851) s a similar karyo- type (Fig. 2I), but the plants are apparently iden- tical with the slender white-flowered form with narrow tepals from the Malmesbury area having = 16 5. 2n = 10 karyotypes (large yellow-flowered forms). Of the eight populations with 27 = 10, six represent the large yellow-flowered, typical form of Moraea fugax from the southwestern Cape. The karyotype (Fig. 4C) consists of two metacentric pairs 13-14 um long, one submeta- centric 12-13 um long, and two much shorter acrocentrics 7-8 um long. Satellites are on the end ofthe long arm ofa short pair and the longest of the two metacentric pairs. The second karyotype of subsp. fugax with n= 10 occurs in yellow-flowered populations in the Olifants River Valley (Fig. 4D). It com- prises two long metacentrics, a medium-sized meta- to submetacentric, a medium-sized acro- centric, and a pair of tiny subtelocentrics with large satellites. A second pair of satellites was seen, in one population (Goldblatt 5778), on the longer arm of the longest metacentric pair. These populations are notable in having nearly erect inner tepals, and they differ from the southern yellow-flowered form in this feature as well as in details of their karyotype. MORAEA FUGAX SUBSP. FILICAULIS Only one population of subsp. filicaulis has so far been found, with 27 = 18, from Clanwilliam in the far south of its range. This karyotype com- populations have 2n — 12. One ofthese isa small, 146 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 LE 3. Comparison of some critical morphological characteristics of the species of Moraea sect. Subrace- TA mosae. Color key: bl = blue; ye = yellow; wh = white. Spathes Outer Tepal Capsule Anther Species Branches (cm) (cm) (mm) (mm) Color M. gracilenta many, open 2.5-3.5 2-2.6(-3) 8-12 4—6.5 bl M. macrocarpa few, sessile 4—5(-6.5) 2-3 (18-)20-25 4-5 bl M. fugax subsp. several, usually 2-3.5(-4) 2-2.6(3.1) 9-13(-15) 4—5(-6) bl, wh filicaulis (3.5-)4-6 2.7-4 15-22 5-10 bl, wh, ye M. fugax subsp. several, usually fugax congested dark blue-violet-flowered form that corresponds to the type of Baker's M. filicaulis. The other population has white flowers. The karyotype consists of two pairs of long acrocentrics and a graded series of smaller acrocentrics (Fig. 5A). There are very large satellites on the short arm of one of the longest pairs. A population from the southern Richtersveld, in the extreme north of the range of Moraea fu- gax, also nas 2n = 12 but has a different karyo- type (Fig. 5 a a fairly long submetacentric (- acrocentric) pair, gagal small metacentric pair. Small satellites are on the long arms of a medium-sized chromo- some pair. This is somewhat similar to karyo- types in the two populations with 27 = 10, which have one long metacentric and four medium- sized acrocentric pairs and lack the short meta- centric pair (Fig. 5C). The populations with 2л = 10 are from opposite ends of the range of subsp. filicaulis, from the lower Olifants River Valley (Goldblatt 5666) and from northern Namaqua- land (Goldblatt 5761). Satellites, seen only in the latter population, are located on the metacentric air The karyotype previously published for subsp. filicaulis (Goldblatt, 1971: 348), also 2n = 12, corresponds to none of those described above. It has one large metacentric pair, four submeta- The highest number recorded for the subspecies is presumed to be basic, x = 9. „anda an MORPHOLOGY PATTERNS OF VARIATION The characters that unite the taxa of the Mo- raea fugax complex are: (1) the tendency for the branches to be clustered at the insertion of the leafor leaves due to the typically short internodes above the leaves, (2) the elongate basal internode so that the leaf is usually inserted well above ground, and (3) a strongly beaked capsule. In Moraea gracilenta the branches form an open system, and in M. macrocarpa the stem is often very short and reaches only shortly above ground level at flowering time, elongating some- what in fruit. Excluding the peculiarities noted above for Moraea gracilenta and M. macrocarpa, varia- tion within the section is largely restricted to size, relative proportion of characters (Table 3), and flower color, discussed further below. T populations of subsp. filicaulis have two filiform leaves. Some white-flowered populations of subsp. fugax have either one or two equal or unequal, linear, channeled leaves. Plant and flower size, and color are particularly variable in M. fugax, and the variation is some- times correlated with chromosome variation. Races are not sufficiently separable on either morphological or cytological grounds to make an eae eat ASE d only the consistently smaller Namaqualand and northwest Cape forms with 2n = 18, 12, or 10 seem to merit subspecific recognition. Even in this subspecies the degree of chromosomal vari- ation is unusual, and five of the six populations studied can be distinguished by major differences in their karyotypes. Nevertheless, these popu- lations probably have a common ancestry and merit taxonomic recogni ant size. Pus fall roughly into two categories (Table 3): either robust and large-flow- ered with spathes (3.5-)4-6 cm long and outer tepals 2.7-4 cm long (only forms of subsp. fugax) or dwarf and small-flowered spathes 2-3.5(-4) cm long and outer tepals 2-2.6(-3.1) cm long. Flower and spathe size are good indications of overall plant size, because most other characters are similarly proportioned. Moraea gracilenta is 1986] A 4 AY AK ES ANA О > Y 39 1f do Jie ow төр WS e? NA е о 9 x 7 TEN A m ^ ж а FiGURE 3. Geographical distribution of the white- flowered form of Moraea fugax subsp. fugax; circles indicate the large-flowered form and triangles the small- a 1 С, ІІ 1 2 | 1 Г. 21 4 d for pop- ulations where known. Scale = 10 um. | Vf; el Ca у^ 2 у С) > CÓ : ESOS > NEL S ўў 3 ) ИНН Я A 27 2, | © as d SF очам о 0777727 FPD M IA ne METERS "t = e o 58 T <> J: 600 А о 40 BO кт Ш mo а № 2 WY FIGURE 4. Mitotic cl subsp. fugax. — A. Pakhuis Pass, n = 8.—B. GOLDBLATT — MORAEA FUGAX COMPLEX 147 exceptional is having small flowers and spathes, but the plants are usually very tall, robust, and many branched. he two dwarf taxa, Moraea macrocarpa and M. fugax subsp. filicaulis, are similar at first ex- amination but can be distinguished readily by their capsules. Those of Moraea macrocarpa are disproportionately long, (18-)20-25 mm, but they In contrast, M. fugax subsp. filicaulis has short capsules, 10-13(-15) mm long, stalked spathes, and usually two (occasionally one) filiform leaves. The largest flowers are found in the two-leaved white-flowered form (Fig. 6A) of subsp. fugax (n= 10, 9, 7, 6), and individuals may be up to 60 cm high or more and have tepals 3-4 cm long. However, size is variable and forms with yellow or blue flowers have tepals that overlap in size in the lower part of the range for the white form. Thus size is insufficient to define or recognize any of the major forms of the subspecies. Sub- species filicaulis (Fig. 6B) has consistently small tepals (usually 2-2.5 cm long), and it often can be recognized by this characteristic alone. The small tepals are, however, correlated with small capsules and filiform leaves that in combination make confusion with subsp. fugax unlikely. The overlap in the size of individual characters of the two subspecies of Moraea fugax is part of the | инини н и ШЕ ҮТЕ zz —— Xl an > nd geographical distribution of the yellow-flowered form of Moraea fugax Caledon Swartberg, n = 7.—C я Foot of Houw Hoek Pass, п = 5.— D. Near Clanwilliam, n = 5. Haploid numbers are indicated for populations where known. Scale = 10 um. ANNALS OF THE MISSOURI BOTANICAL GARDEN f AL ШИ [VoL. 73 Иланию И клин" >} Nn an FIGURE 5. mieskroon ,n=6.— Mitotic chromosomes and geographical distribution of Moraea fugax subsp. filicaulis. — A. Near B. Eksteenfontein road, Richtersveld, n = 6.—C. Below Wildepaardehoek Pass, n = 5. Haploid numbers are indicated for populations where known. Scale = 10 um. = а +} reason for treating them I than as separate species. In some instances determi- nations may be arbitrary either because the plants in question are poorly grown or poorly pressed, or the plants may be truly intermediate in more than one characteristic. Flower color. Blue-purple flowers are char- acteristic of all populations of M. gracilenta and M. macrocarpa but are less common in Moraea fugax. In the Kamiesberg, populations of subsp. filicaulis (corresponding exactly to the type of M. filicaulis) t deep blue-purpl , but else- where the subspecies has white to cream flowers, sometimes flushed pink or yellow. In subsp. fu- gax, blue to purple flowers occur in the Piketberg area and from Malmesbury south to the Cape Peninsula where they bloom later than yellow- or white-flowered populations. Chromosome numbers in this form are n = 6 and n = 7 (Fig. 2F, G), the latter found in only one population. White-flowered populations of subsp. fugax are most common along the west coast (Fig. 3) and extend from just north of the Cape Peninsula to northern Namaqualand. Chromosome numbers are n = 10, 9, 7, and 6, the latter two numbers infrequent. In the Olifants River Valley, white- flowered populations alternate with yellow-flow- ered populations, the latter consistently differing in having erect inner tepals (and n = 5). Occa- sionally there is a trend for white flowers to be replaced gradually over some distance by blue ones. Populations with smaller white flowers and narrow tepals occur in the Malmesbury district and in the Breede River Valley (Fig. 3). These have n = 9, 8, or 6. They appear to comprise a separate white-flowered race in which aneuploid reduction has occurred independently. Yellow-flowered forms are most common in the southwestern Cape where the type of Moraea fugax, an illustration of a plant of unknown or- igin, was most probably collected. This southern form has n — 5 over much of its area (Fig. 4), but л = 7 is known from one population. Yellow- flowered plants in the Olifants River Valley also have n = 5 but have a somewhat different karyo- type with the smallest chromosome pair minute and metacentric (Fig. 4D). In some ofthe interior mountain valleys of the western Cape, from Pak- huis Pass to Robertson, similar but more slender yellow-flowered plants have either n = 9 or 8. Despite the several different karyotypes found in plants with yellow flowers, there seems to be no consistent and significant morphological feature to distinguish them. It is not clear whether the karyotypically diverse yellow-flowered forms comprise a monophyletic assemblage or evolved 1986] more than once from ancestors that most prob- ably had white flowers. EVOLUTION The cytological diversity in Moraea sect. Sub- racemosae is certainly without parallel in Iri- daceae and is most unusual in plants generally. The section appears to be in a stage of rapid evolution with the differentiation of numerous regional and local morphological and cytological races. Moraea fugax is the most chromosomally variable and probably also the most morpholog- ically variable species in Moraea and its close allies (Goldblatt, 1971, 1979, 1980). Other spe- cies, notably M. tripetala and M. papilionacea, are also morphologically variable and have dif- ferentiated into several distinct forms or races. In these species cytological variability is limited, as would normally be expected, although differ- ent or can be distinguished (Goldblatt, in The chromosomal diversity in the several races of Moraea fugax probably has promoted popu- lation differentiation in the species by restricting hybridization and consequent recombination in forms that have differentiated cytologically. It probably played little role in the origin of pop- ulation differences that most likely developed in isolation at times when more arid conditions pre- vailed along the Cape west coast, and the distri- bution of the various forms or races of the species was more restricted than at present It is especially remarkable that thé extensive chromosomal variation has developed in several lines. The ancestor of the complex probably had small blue-purple flowers, two long leaves, gn many branches іп an open likely resembled M. gracilenta. This species or its immediate ancestor may have given rise to the dwarf and reduced M. macrocarpa and to the larger flowered form of M. fugax, both of which have similar karyotypes with x = 10. Evolution in Moraea fugax presumably fol- lowed a pattern of reduction in plant size, length and number of branches, and diversification of flower color. Distinctive forms or races tend to have either fewer branches, shorter stems, or nar- rower tepals and often have only a single leaf, as in the type form of M. fugax The evolution of the different color forms in subsp. fugax was probably a significant devel- opment, although it is difficult to assess its evo- lutionary importance. The available data do not GOLDBLATT — MORAEA FUGAX COMPLEX 149 favor a single or multiple origin for either blue or yellow color. However, it is clear from the karyology that chromosomal diversification oc- curred at least once in each of the major color forms after their differentiation. Furthermore, flower color in other species of Moraea rarely ies within a species, so it is reasonable to postulate a single (or very few) co events. Thus it is likely that aneuploid reduction from n = 9 to 5 occurred in a monophyletic yellow-flowered form of subsp. fugax and in the late-blooming blue-purple-flowered form at least from n = 7 (the highest number recorded here) to 6. The difference in karyotpes in the yellow- flowered forms with n = 5 from the southwestern Cape and in the Olifants River Valley suggests either an independent origin of this number from a common ancestor with a higher haploid num- ber or major structural change in the karyotype with n = 5. It also seems likely that plants with smaller white flowers and narrow tepals from the Malmesbury district and Breede River Valley comprise a regional race distinct from the large white-flowered form, and the recorded numbers = 9, 8, and 6 indicate separate aneuploid reduction series in the two white-flowered races. о m5, SYSTEMATIC TREATMENT KEY TO THE TAXA OF SECTION SUBRACEMOSAE — a. Spathes short, 2-3. i QA cm long; capsules 8— iform an ndu usually two; branch- es few to several ian ered; flowers open- ing shortly after midday „u EA TO . fugax subsp. ce ma 2b. Leaves linear, usually 2-3 mm wide, itary; branches many ve panic flowers opening after 3:30 P.M. „u “М. gracilenta lb. о longer, (3.5-)4-6.5 cm e capsules 5—28( о 8. a. Flowers small, the outer tepals 2-3 cm long; capsules at least 18 mm long; sile to subsessil а. 2. М. тасгосагра 3b. Flowers large, the outer tepals 2.7-4 cm long; capsules 15-28(-40) mm long: branches usually Ете and stalked .... 3A. M. fugax subsp. fugax 1. Moraea gracilenta Goldbl., Ann. Missouri Bot. Gard. 63: 724. 1976 [1977]; Fl. Pl. Af- rica 44: tab. 1748. 1977. TYPE: South Africa. Cape: sandy flats S of Piekenierskloof Pass, Piketberg district, Goldblatt 3279 (holotype, MO; isotypes, K, NBG, PRE, S). FiGURE 6A. 150 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 FIGURE 6. Morphology of Moraea gracilenta (A) and M. macrocarpa (B). Habits x0.5; flowers, and floral details life size. Moraea edulis L. f. var. gracilis Baker, Handb. Irid. 56. 1892; Fl. Cap. 6: 21. 1896. TYPE: South Africa. Cape: Tulbagh Kloof to Piekeniers Kloof, Zeyher 1647 [lectotype, K, designated by Goldblatt (1976b); isolectotypes, PRE, S]. Plants 30-80 cm high, many branched. Corm 1.5-2 cm diam., deeply buried to 25 cm; tunics of fine to medium pale fibers. Leaf solitary, can- aliculate, linear, exceeding the inflorescence but usually trailing, inserted well above the ground at the base of the first branch. Stem erect and multibranched. Spathes herbaceous, becoming dry, the apices brown, acute; inner spathe 2.5- 3.5(-4) cm long, the outer less than halfthe inner. Flower pale blue-mauve, strongly scented; outer tepals 2-3 cm long, 6-8 mm at the widest point, lanceolate; inner tepals 1.8-2.8 cm long, 5-6 mm wide, spreading. Filaments ca. 5 mm long, united in the lower half; anthers 4-6 mm long, white. Ovary 8-10 mm long, included in the spathes, style branches 7-9 mm long, crests linear-lan- 1986] ceolate, 7-12 mm long. Capsule 8-12 mm long. Chromosome number 2n = Flowering time. Late September to Novem- ber, extending to December at higher altitudes; flowers opening after 3:30 P.M. sometimes as late as 5 P.M. and fading by 7 P.M Distribution. Western coastal belt from Tul- bagh Kloof to Clanwilliam, in sandy flat areas, near streams or at the foot of mountains. Fig- Moraea gracilenta differs consistently from all the forms of the variable M. fugax in its open and many branched habit, small blue flowers, and short inflorescence spathes and capsules. Its floral phenology is also markedly different from that of M. fugax. Blooming occurs in the same months, and the flowers of both species last only a day, but while those of M. gracilenta open be- tween 3:30 and 5 р.м. and last until at least 7 P.M., the flowers of M. fugax open at about mid- day and are usually completely wilted by 6 P.M. The difference in phenology has been observed on several occasions in wild populations and ap- pears to vary little. Moraea gracilenta is appar- ently pollinated by small moths, which I have seen visiting the pale, sweetly scented flowers as they open in the late afternoon. Specimens examined. SOUTH AFRICA. CAPE: 32.18 (Clanwilliam) riverbank near Clanwilliam (BB), Gold- blatt 3076 (MO), Galpin 11482 (B, K, PRE); Eendekuil (DB), Loubser 851 (NBG); sandy areas between Een- dekuil and Piekenierskloof Pass, Goldblatt 3279 (K, MO, NBG, PRE, S), 3029 (MO, NBG, PRE 32.19 (Wuppertal) near Warm Baths, Olifants River Valley (CA), L. Bolus s.n. (BOL-20323, K, PRE); Ci- trusdal, riverbank at campsite, Я do 7 (MO, PRE), 6708 (MO); sandy alluv near river at the entrance to the old Elandskloof Tuus Goldblatt 7122 (MO, PRE); Elandskloof, Compton 16748 (NBG) 33.19 (Worcester) Tulbagh Kloof to Piekenierskloof (AC), Zeyher 1647 Us ipis S); Tulbagh Kloof, Comp- ton 12399 (NBG), E Zeyher Irid. 23 (77.9) (MO, P, S); Gydo (AD). poit 3004 (BOL, K, PRE). 2. Moraea macrocarpa Goldbl., sp. nov. TYPE: South Africa. Cape: S of Redelinghuys, Pi- ketberg district, Acocks 24359 (holotype, PRE; isotype, MO). FIGURE 6B. Plantae 30-80 cm altae, ramis 1-3, cormo 8-12 т diam., folio solitario tereti ad lineari-filiformi, spathis interioribus 4—5(—6.5) cm longis, exterioribus parviori- bus, floribus caeruleis, tepalis exterioribus 2-3 cm lon- gis, filamentis 5-6 mm longis, capsulis (18-)20-25 mm longis. Plants 30-80 cm high, simple or few-branched. GOLDBLATT — MORAEA FUGAX COMPLEX 151 Corm 8-12 mm diam .; tunics of fine pale fibers. Leaf solitary, canaliculate, terete to linear-fili- the spathes, inserted n ground. Stem erect, filiform, amples (to three) sessile or subsessile Su iud di Spathes herbaceous, becoming dry above, the apices brown, acute, the inner (3.4-)4-5(-6.5) cm long, the outer about one-third as long as the inner. Flower blue-mauve, with yellow nectar guides, the claws of the outer tepals yellow with dark spots, scented; outer tepals 2-3 cm long, 7— 10 mmat the widest, lanceolate; inner tepals 1.8— 2.8 cm long, 5-7 mm wide, spreading. Filaments 5-6 mm long, united in the lower half; anthers 4-5 mm long, white. Ovary 17-21 mm long, included in the spathes, style branches 6-7 mm long, the crests linear-lanceolate, 12-18 mm long. Capsule cylindric, (18-)20-25 mm long. Chro- mosome number 2n — 20. Flowering time. (May-)August to early Oc- tober; flowers opening soon after midday and fading at about 5 P.M. Distribution. West coast from Hopefield to the Olifants River mouth and locally in the Breede River Valley, in deep white sand. Figure 1. Moraea macrocarpa has been associated with M. fugax until now, or it has been confused with M. filicaulis, now M. fugax subsp. filicaulis. It bears a superficial resemblance to the latter in its slender stem, small corm, and small flower. Cytological study has, however, provided evi- dence that they are unrelated. Moraea macro- carpa has 2n = 20, the basal diploid number for suggests that they may be derived from different ancestral stock, and the several differences be- n M. macrocarpa and M. fugax subsp. fili- s are significant he differences between Moraea macrocarpa 4 M. fugax subsp. filicaulis include spathe, ovary and capsule length, and leaf number. The majority of specimens that I have examined of subsp. filicaulis have two leaves while M. macro- carpa has only one. The capsules of M. macro- carpa, its most striking attribute, are dispropor- tionately long for the small plants and are in the 20-25 mm range. The ovary and the spathes are correspondingly long because the capsule re- mains enclosed in the spathes throughout de- velopment. The latter range from 4—6.5 mm long. In subsp. filicaulis the capsules are only 8-13 152 (-15) mm long and the spathes 2-3.5(-4) mm long. The flowering stem of subsp. filicaulis is typically branched, and the branches are usually distinctly stalked. In contrast the flowering stems of M. macrocarpa usually have only a single ter- minal inflorescence, rarely one to two branches, and the branches are more or less sessile. foraea long the Cape west coast from Hopefield to Trawal in the Olifants River Valley and locally in the interior in the Breede River Valley near Worcester. It grows in deep, coarse-grained sand. Occasionally it occurs with the taller and more robust M. fugax subsp. fugax. Its range overlaps slightly that of M. fugax subsp. filicaulis in the Olifants River Valley but they have not been observed growing together. Specimens examined. SOUTH AFRICA. CAPE: 31.18 (Vanrhynsdorp) mountains W of Trawal, in sand (DC), Goldblatt 5661 (MO 32.18 (Clanwilliam) 2 km N of the Lamberts Bay road towards cies ay cat E Lund- gren 1552 (MO, S); be mberts Bay (AB-BA), L. к. s.n. . (BOL 23193): pala Elandskloof and Clanwilliam (?BC), Leipoldt s.n. (BOL 20950); ee (BD), Schlechter 5149 (BOL, GRA); 11.7 mi. S of Redelinghuys, white sand (DA), Acocks 2435 9(MO, PRE); 9.5 mi. SW of Redelinghuys, Acocks 19689 (K); near Sauer, Piketberg (DD), Barker 2687 (NBG). 32.19 (Wuppertal) Pakhuis Pass (AA), Barker 1999 (NBG), Compton 9798 (NBG). 33.18 (Cape Town) Oosterwal, Hopefield (?AB), Pamphlett 89 ed in planitie prope Darling (AD), H. Bolus 12833 (BOL, K). 33.19 (Worcester) x de Doorns, farm Reiers Rus, along the Breede River (CB), Goldblatt 6961 (MO, G). we е ode fugax (de la Roche) Jacq., Hort. Bot. indob. 3: 14. tab. 20. 1776; Goldblatt, J. М you Bot. 36: 316. 1970; Ann. Missouri Bot. Gard. 63: 725. 1976 [1977]. Vieus- seuxia fugax de la Roche, Descr. Pl. Aliq. Nov. 33. 1766. TYPE: Illustration in van Ha- zen, Cat. Arb. & Pl. 67. 1759 (lectotype des- ignated by Goldblatt, 1970). FIGURE 7. (For additional synonyms see under the subspecies.) Plants medium to large, 12-40(-50) cm high, branched. Corm (1—)1.5—3 cm diam.; tunics usu- ally pale, rarely dark, of fine to medium fibers. Cataphylls usually two, membranous, pale, be- coming dry and brownish and often broken above. Leaf solitary or two, equal or unequal in length, subopposite, inserted well above ground immediately below the first branch, canaliculate, usually much exceeding the stem and trailing, ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 occasionally loosely twisted distally. Stem erect or more often somewhat inclined, with a con- spicuous, long lower internode, the branches crowded, occasionally subracemose. Spathes herbaceous, becoming dry, the upper margin light brown, apices attenuate; inner spathe (2-)2.5-6 (-8) cm long, the outer one-half to two-thirds as long as the inner. Flowers white, blue, or yellow, strongly scented; tepals lancolate, the limb more or less equal to the claw, the limb spreading hor- izontally or the inner occasionally erect, outer 2- 4 cm long, inner 2-3.5 cm long, 5-8 mm wide. Filaments 5-10 mm long, united in the lower half; anthers 4-10 mm long. Ovary 8-20 mm long, included in the spathes, style branches 1.5— 2 cm long, the crests lanceolate 6-18 mm long. Capsule oblong to cylindric, distinctly beaked, 9-28(-40) mm long. Chromosome number 2n = 20, 18, 16, 14, 12, 10 Flowering time. August to October, occa- sionally extending into December at high ele- vations. Distribution. Northern Namaqualand to the southern Cape, typically in deep sandy or rocky sandstone or granitic soil. Figure 1. As circumscribed here, Moraea fugax is an unusually variable species with several distinct forms or races. Of these only a series of dwarf, small-flowered and usually two-leafed popula- tions, described in the past as separate species, M. filicaulis and M. diphylla, are given taxonom- ic recognition as subsp. filicaulis. This subspecies occurs throughout Namaqualand and extends south to the Clanwilliam district. Although subsp. filicaulis is itself somewhat variable, it appears to constitute a natural assemblage, united by its slender stem, filiform leaves, relatively small flower with tepals 2-2.6(-3.5) mm long and es- pecially small capsules (10-15 mm long). Sub- species filicaulis apparently has x = 9. Some cy- tological variation is evident in karyotypes from several localities, and derived numbers of n = 6 and 5 have been recorded. The pattern of variation is more complex in subsp. fugax, the several forms of which are united by their large size, long broad leaves, and rela- tively large flowers with tepals 23-40 mm long and capsules 15-40 mm long. Chromosomal variation is particularly extensive in this sub- species with numbers ranging from 2n = 20 to 10. The variation within subsp. fugax is dis- cussed in detail after the subspecies description. 1986] KEY TO THE SUBSPECIES la. Spathes short, 2-3.5(-4) cm long; capsules 9- 13 rarely to 15 mm long; flowers small to edium, the outer tepals 2-2.6(-3.5) cm ж pi filiform and usually two; branch clustered 3B. subsp. filicaulis lb. an long, (3.5-)4-6.5 cm long; capsu ) mm long; flowers medium si large, the outer tepals 2.7-4 cm long; leaves channeled, linear, either one or two; branches usually clustered but sometimes subrace- mose 3A. subsp. fugax 3A. M. fugax subsp. fugax. FIGURE 7A. Iris edulis L. f., Suppl. Pl. 93. 1781; Moraea edulis (L. f.) Ker, Bot. Mag. 17: tab. 613. 1803; Baker, Fl. Cap. 6: 20. 1896; Vieusseuxia edulis (L. f.) Link, Enum. Hort. Berol. Alt. 1: 56. 1821. TYPE: se — ae exact locality uncertain, Thunber, n. [lectotype, Herb. Thunb. 1123 (UPS), peta i by Goldblatt, 1976b]. qe о Lam., Tabl. Еосун 1: 114. 1791; ycl. 4: 227. 1797. TYPE: South Africa. Cape, Sonet s.n. (lectotype, P, designated by Gold- blatt, 1976b). Iris longifolia Schneev., Icon. Pl. Rar. 7: tab. a Ie Moraea longifolia (Schneev.) Sweet, ed. 2, 496. 0, nom. illeg., non M. tongifolia (Jacq) Pers. DUE TYPE: South Afric a. Cape: without precise locality (lectotype, illustration in Schneev., Icon. Pl. Rar. 7: tab. 20. 1792, desig- nated by ‘Goldblatt, 1976b). Moraea odora Salisb., Parad. pa 1: tab. 10. = TYPE: Illustration in Salisb ad. Lond. 1: tab. 10. 1805 (lectotype d ded by eden 6b). Plants medium to large, 12-40(-50) cm high, branched. Corm (1—)1.5—3 cm diam.; tunics usu- ally pale, rarely dark, of fine to medium fibers. Leaf usually solitary, occasionally two, equal or unequal in length, canaliculate, usually much ex- ceeding the stem and trailing. Stem sturdy, 2-3 mm thick. Inner spathe 4-8 cm long, the outer ca. one-third the length of the inner. Flowers white, blue, or yellow, strongly scented; outer tepals 2.3-4 cm long, lanceolate, the limb more or less equal the claw; inner tepals 2-3.5 cm long, erect or slight reflexed, 5-8 mm wide. Filaments 6-10 mm long; anthers 4—8 mm long. Ovary 14- 20 mm long, style branches 1.5-2 cm long, the crests lanceolate, 10-18 mm long. Capsule cla- vate to cylindrical, distinctly beaked, 1.5-2.8(-4) cm long; seeds many, angled. Chromosome num- ber 2n — 20, 18, 16, 14, 12, 10. Flowering time. August to November, ex- tending into December at higher elevations. Distribution. Namaqualand to the southern GOLDBLATT — MORAEA FUGAX COMPLEX 153 Cape, frequently in sandy situations. Figures 1, » Variation. Plants corresponding to the type of Moraea fugax have large, yellow flowers, a single leaf, and are of generally moderate size with an umbellate branching pattern (althoug tall racemosely branching plants may occur). Populations of this form occur in the south of the range of subsp. fugax and extend from south east of its range, at Swartrivier, north of Caledon Zwartberg, plants have п = 7. Similar but more slender yellow-flowered populations in the in- terior near Robertson and to the north in the Cedarberg at Brandewyn River and Algeria have n= 9, and a population in Pakhuis Pass has n = 8. Despite this range of otypes, there seem to be no morphological features that distinguish the cytotypes. A yellow-flowered form from the Oli- fants River Valley and adjacent valleys is similar to yellow-flowered plants that occur to the south except that they consistently have erect inner te- pals. These plants also have n = 5 but have a somewhat different karyotype from the southern form that has the same number. A second form that onus in the şouthern part of the range of subsp. fi 15 о tall, and is late blooming. On the e Peninsula and the Cape Flats the yellow- Ad form, described above, blooms in Au- gust and September, while the blue-flowered form blooms only in October or later, and flowering plants have been collected as late as January. Populations of this form usually have n = 6, but n = 7 was found in a population near Cape Point. This form is not well sampled. Along the west coast from near Malmesbury as far north as Hondeklipbaai in Namaqualand (Fig. 3), there is a large, white-flowered form of the species corresponding to Baker's Moraea edulis var. longifolia. It is generally tall and ra- cemosely branched. This robust form also occurs in the e River Valley. Most populations have n = 10 or 9, but n = 7 and 6 are found in the Velddrif ud Hopefield areas (Goldblatt, 1971: 348; Fig. 2E). In the Malmesbury district there is a distinc- tive white-flowered form that has relatively large flowers with unusually narrow tepals. Popula- tions around Malmesbury and to the north to- wards Hopefield have n = 8, but plants corre- 154 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURE 7. Morphology of Moraea fugax.— A. Subsp. fugax, large white-flowered form.—B. Subsp. filicaulis. Habits x0.5; flowers life size. sponding to this form from south near Gouda ation for flower color. Occasionally there is a have n = 6. In the Breede River Valley similar trend for white flowers to be replaced gradually plants have n = 9 (Fig. 3). over some distance by blue ones as in the Pi- There is no recorded intrapopulational vari- — ketberg district. In the Olifants River Valley pop- 1986] ulations of white- or yellow-flowered plants al- ternate over distances of several kilometers without any clear pattern. The yellow-flowered populations consistently differ in having erect inner tepals and as far as is known, n = 5. History. ea fugax is typified by an il- lustration ofa ке flowered plant of unknown origin, most probably collected on or near the Cape Peninsula. It was cultivated in Holland in the mid-seventeenth century and figured in the 1759 “Catalogue des Arbores et Plantes" of the van Hazen, Vakinburg & Company Nursery. The plant was described by Daniel de la Roche in 1766 and the illustration in van Hazen’s Cata- logue is assumed to be the type. Moraea fugax was first assigned to the genus Vieusseuxia but was transferred to Moraea by Jacquin in 1776, who figured the blue-purple-flowered form of subsp. fugax but considered it conspecific with plants with yellow flowers. Subsequently, other forms of subsp. fugax were d ibed from plants in cultivation in Europe, notably M. oe (as Iris) by Schneevoogt in 1792 and M. odor by Salisbury in 1805. Both these species repre- sent the robust white-flowered form of subsp. fugax. Specimens examined. SOUTH AFRICA. aie a (Springbok) Komaggas (CD), Maguire 403 (N 30.17 (Hondeklipbaai) near Soebatsfontein Me van e 268 (N. BG). t in collibus’ (CC), Schlechter 11063 de P). 8 (Van Rhynsdorp) Vredendal commonage (CB- sl pi (NBG, PRE, STE), Bayliss 6126 (MO); Nardouw Pass, Lewis s.n. a 22205); top of Gifberg (DC), cai 6161 (MO); W he Win- terhoek , Goldblatt $778 MO y Olifants River at the bridge m S of Traval, Olifants River Valley, W of Tra on the road MO). (Calvinia) ke places near Nieuwoudtville d ped 11128 n. (Nat. Bot. ie : MO); Doornbosch к, Hall T top of Botterkloof (CD), van Niekerk 3197 (BO L), Hall 3876 (NBG); a uta ca. 42 km SE of Nieuw- oudtville, Goldblatt 7080 (MO). 32.18 (Clanwilliam) sandy slope N of Clanwilliam facing the Olifants River (BB), Goldblatt 5936 (MO); between Clanwilliam and Graafwater, Goldblatt 5161 (MO); sandveld near Velddrif (CC), van Jaarsveld s.n. (MO); hills NW of Moutons Vlei (DC), е fe (BOL); gei (DD), Compton 22989 (NBG), H Bolus s ; N of Piketberg towards о n Goldblatt pi 6 (MO 19 (Wuppertal) Cedarberg Pass near Algeria (AC), Pee 3250 (MO); Cedarberg at Algeria, Viviers s.n. (MO). GOLDBLATT — MORAEA FUGAX COMPLEX 155 33.18 (Cape Town) Langebaan (AA), Lewis s.n. BOL); between Hopefield and Langebaan, L. Bolus s.n. (BOL-20339, K); near Hopefield, Marloth 8207 (PRE); Groenekloof (AD), Zeyher 1646 (K, PRE, S); road to Bokbaai, Goldblatt 5850 (MO); 20 km N Malmesbury on the road to Hopefield (BB), Goldblatt MO); Malmesbury (BC), Barker 2554 (NBG), Goldblatt 4813 (MO); 8 km N of Malmesbury, Gold- blatt 3025 (MO, NBG, PRE); sandy slopes S of Mal- mesbury opposite Abbotsdale, m 5116 (MO); slopes of Koeberg N of Cape Town (DA), Goldblatt 4080 (MO); near Sea Point, Cape T wn (CD), Wolley Dod 1611 (BOL); Wynberg Hill, Pillans 1950 (MO); Kirstenbosch, Barker 2066 (BOL), Verdoorn s.n. (PRE); Table Mountain, Ecklon 823 (PRE-11 33.19 (Worcester) a near Saron (AA), ‘Schlechter 10605 (BR, K, MO, PRE, S); foot of the Elandsrivier Mts. on the farm xb жы (AC), Goldblatt 5851 (MO); Michells Pass (AD), Schlechter 8966 (GRA); Aan de Doorns, farm Reiers Rus, near Moordkuil (CB), Gold- att & Snijman 6962 (MO); near French Hoek (CC), Gillett 1830, 1831 (STE); 15 km from MacGregor on Bonnievale road (DD), Marsh 806 (PRE, STE); hills above Goree, S of the Robertson-Ashton road, Gold- blatt 5860 (MO). 34.18 (Simonstown) beyond Simonstown (AB), Wolley Dod 505 (BOL, K); Fish Hoek, Wolley Dod 1633 (BOL, K); burned lower slopes of Klaasjagers- berg, opposite Cape Point Reserve, Goldblatt 5264 (MO); Silvermine road just N of the Noordhoek tur- noff, Goldblatt 5154 (MO); Olifants Bos (AD), Barker ~ blatt 5403 (MO): Strandfontein (BA), Rycroft 2371 (MO); dunes near Strand (BB), Parker 4141 (BOL, K). 34.19 (Caledon) burned flats at the foot of Houw oek Pass (AA), Goldblatt 4292 (MO); Caledon (AB), Gillett 1111 (STE); N side of Caledon Swartberg, Gold- blatt 5782 (MO); Onrust River (AC), Van Niekerk 300 (BOL); Mainstay, Onrus River, Robertson 111 (MO); roadside ca. 10 km S of Bot River, Goldblatt 2997 (MO); Die Duine, Hermanus, stabilized sand dunes, 50 m (AD), Williams 1010 (MO, PRE, US, WAG) 34.20 (Bredasdorp) near Storms Vlei (AA), Gold- blatt 2923 (MO, PRE); 14 km E of Swellendam (BA), Story 3090 (PRE); Eu Infanta (BD), Blum 64 (E), 214 (Ey; Zoetendals Vlei (CA), Leipoldt 3559 (PRE). 34.21 (Riversdale) pou (AB), H. Bolus s.n. BOL). 34.22 (Mossel Bay) Mossel Bay (AA), Burchell 6295 (K) Ruigte Vlei, near Knysna (BB), Fourcade 1557 ern 2 ~ ECISE LOCALITY: ‘Clanwilliam, am Flus alane und bei Villa Brakfontein’ Ecklon & Zey- her Irid. 16 (76.10) (MO); ‘Worcester, Tulbaghskloof, Tulbaghsthal, am Fuss des Winterhoeksberg’ etc., Eck- lon & Zeyher Irid. 23 (77.11) (MO); ‘bei Puspasvlei, Voormansbosch, Duiwelsbosch’ etc., Zeyher 4079 (70.10) (S). 3B. M. fugax subsp. filicaulis (Baker) Goldbl., nd Kamiesberg between Roodeberg and Ezel- 156 skop, Drége 2605 (lectotype, K, designated by Goldblatt, 1976b; isolectotypes, MO, P). FIGURE 7B Moraea ш Baker, Bull. Misc. Inform. 1906: 42. E: South Africa. Cape: Olifants River, Panther 734 i K, designated by Gold- blatt, 1976b Plants small to medium, 6-12(-20) cm high, few to several-branched. Corm 7-16 mm diam .; tunics usually pale, rarely dark, of fine to medium fibers. Leaves usually two and subopposite, oc- casionally solitary, linear-filiform, channeled, 1— 2 mm wide, equal or и unequal in length, lly loose d distally. Stem slender, to 1 mm thick. dones spathe 2-3.5(-4) cm long, the outer half to one-third as long as the inner. Flowers white, sometimes shaded bluish or pinkish, rarely dark blue-violet, scented, with yellow nectar guides; tepals lanceolate, limbs spreading horizontally, outer 2-2.6(-3.5) cm long, 0.8-12 mm wide; inner 1.8-3 cm long, 5-7 mm wide. Filaments 5-6 mm long; anthers 4-5(-6) mm long. Ovary 8-10 mm long, the style branch- es 8-10 mm long, the crests lanceolate 8-14 mm long. Capsule 10-13(-15) mm long. Chromo- some number 2n = 18, 12, 10 Flowering time. August to October. Distribution. Namaqualand south to the Oli- fants River Valley near Clanwilliam, in sandy or rocky soil. Figure Moraea fugax subsp. filicaulis is a small-flow- ered subspecies with a slender stem, small corm, and usually two filiform leaves. The subspecies ranges from northern Namaqualand to the cen- — 9 in one о п = 6 in three populations, and п = 5 in populations. The type form of M. filicaulis, н the Kamiesberg, has deep blue-purple flowers and n = 6, the karyotype having a conspicuous long acrocentric chromosome pair with a large ter- minal satellite. White-flowered plants from near Kamieskroon, matching the type of the syn- onym, M. diphylla, have an identical karyotype. Plants from the type area of M. diphylla in the Clanwilliam district, also white-flowered, have either n = 9 or 5, the latter karyotype with one long metacentric and five acrocentric pairs. A similar karyotype occurs in plants with some- what larger flowers from northern Namaqua- land. A population from the Richtersveld at the extreme of the range of the subspecies has n = ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 6 and a comparable karyotype in which one very small metacentric chromosome pair stands out. There is probably more variation in the subspe- cies yet to be recorded, and it is premature to attempt to construct a hypothetical pattern of chromosome evolution for subsp. filicaulis. Available data suggest that n = 9 is the ancestral number ecimens examined. SOUTH AFRICA. CAPE: 29.17 T (NBG), 7423 (N DB), H. Bolus bok, Esterhuysen 5879 (B Springbok, Leighton 1173A (BO Wildeperdehoek Pass (DC), viru г. An (MO); 21 km SW of Springbok (DD), Acocks 19576 (K, PRE). 30.17 (Hondeklipbaai) granite hills 2 km N of Ka- mieskroon (BB), ака 4254 (МО); киси, Lewis 5478 (NBG); 1 NNW of Garies ( 14964 (PRE); а River to Soebatsfo ntein, =н ton 1211 (BOL); Brakdam hills, Schlechter 11111 eo 30.18 (Kamiesberg) Kamiesberg, E slopes Rooiberg (AO), Cine 4308 (MO); sandy field below ad Pass, Goldblatt 635 (BOL); о between Roo- deberg and Ezelskop, Drége 2605 (K, oms Ravine (CA), Pearson 6665 (K); алс Bitterfont tein and Garies (CA-CC), Pillans 6346 (BOL), Leipoldt 3846 OL BOL). 31.18 (Vanrhynsdorp) 1 km N of farm Komkans, Kliphuis se Kop (AA), Nordenstam & Lundgren 1710 (MO, NBG, S); 4 mi. N of farm Komkans, Geelkop, Nordenstam & Lundgren 1720 (MO, S), Van Rhyns- dorp (DA), Barker 3642 (NBG); Vredendal Common- age, Hall 3843 (NBG, PRE, STE), 10 km N of Yan Rhynsdorp, Acocks 19499 (M, NBG, PRE); N slopes of Gifberg (DC), Goldblatt 207 (BOL); hills W of Tra- wal on the Kleipan road, Goldblatt 5666 (MO). 32.18 (Clanwilliam) Lamberts Bay (AB), Henrici o (PRE); near Olifants River at Clanwilliam (BB), P ther 734 (K), Ramskop, Clanwilliam, Goldblatt 7376 : 5 А THOUT PRECISE LOCALITY: Richtersveld, Marloth 12207 (BOL); Querung des Olifantsrivier, Penther 554 (BOL); Namaqualand Minor, Scully 111 (BOL). LITERATURE CITED BAKER, J.G. 1892. Handbook of the Irideae. George Bell & Sons, London. 896. Irideae. Jn W. T. Thiselton- d ved itor), Flora Capensis 6: 7-71. Reeve & Co. on. . 1906. 1906: 15-30 Dyer, A. F. 1963. The use of lacto-propionic orcein in rapid squash methods. Stain Technol. 38: 85- In Diagnoses Africanae. Kew Bull. EDWARDS, D. & О. H. LEISTNER. 1971. A degree ref- southern Africa. Mitt. Bot. Staatssamml. Mün chen 10: 501-509 1986] GOLDBLATT, P. 1970. Roche. J. S. African Bot. 36: . 1971. GOLDBLATT — MORAEA FUGAX COMPLEX 157 The Iridaceae of Daniel de la 91- Cytological and morphological d fric in the southern African Iridaceae. J. S. A 0. Bot. 37: 317-4 a. а cytology aad peng classification in vhs a (Iridaceae). Ann. Mis- souri Bot. Gar he . 1976b т ee genus Moraea in the win- ter rainfall area of ме Africa. Ann. Missouri Bot. Ga rd. 63: 1979. 7-7 Chromosome cytology and karyotype change in Galaxia (Iridaceae). Pl. Syst. Evol. 133: —. 1980. Redefinition of Homeria and Moraea (Iridaceae) in the light of ba ¿oe data, with me gen. nov. Bot. Not. 133: Yao R. 1975. Chromosomal evalution in Hap- lopappus a a centric transposition race. Evolution 27: 243-256. Moriarty, A. 1982. Outeniqua, es & Eastern Little Karoo: South African Wild Flower uide 2. Botanical Guy of South Africa, Kir- stenbosch. NOTES ON PERUVIAN PALMS! ALWYN H. GENTRY? ABSTRACT Three new species of Peruvian palms are described: lia? longipetala A. Gentry, Chamaedorea megaphylla A. Gentry, and Chamaedorea smithii A. Gen reported from Peru for the first time, and new a are proposed for Geonoma ddition, the genus Dictyocaryum is trigona (Ruiz & Pavon) A. Gentry, Chamaedorea poeppigiana (Mart.) A. Gentry, and Chamaedorea latisecta (H. Moore) A. Gentry Palms are notorious for their taxonomic dif- ficulty, due in large part to their size and con- sequent unwieldiness as herbarium specimens. Nevertheless, enough recent Peruvian palm col- lections have been generated by the re-activated Flora of Peru project (see Gentry, 1980) to begin to bring order out of their taxonomic near chaos. Each of the new species (plus two newly redis- covered species) described here is an important component of one of the specialized local vege- tation types so characteristic of the interrupted system of parallel ridges that comprise the east- ern slope of the central Peruvian Andes, yet none of them had been collected when Macbride (1960) treatment. Since some of these vegetation types are highly endangered, the palms described here are of more than pass- ing significance to conservation. That most of these species are among the commonest and most obvious plants in their particular type of forest emphasizes the inadequacy of floristic knowl- edge of the complex Amazonian/Andean inter- face in Peru, a lack that will soon be permanent if destruction of these habitats continues at cur- rent rates Dictyocaryum H. Wendl., Bonplandia 8: 106. 1860; ampl. Bot. Zeitung (Berlin) 21: 131. 1863 Dictyocaryum lamarckianum (Mart.) H. Wendl., Bot. Zeitung (Berlin) 21: 131. 1863. Iriartea lamarck- iana Mart., Hist. Nat. Palm. 3(7): 190. 1838. TYPE: Bolivia. Río Beni-Mamore watershed, d'Orbigny s.n. (not "vu леш Palm. Orbign. 18. t. 12: fig. 2. t. 20A. Tree 10-25 m tall, 12-29 cm dbh, the trunk smooth, with faint leaf scars and no central swell- ing, the dense cone of stilt roots to 1 m tall, the ! I thank USAID (DAN-5542-G-SS- 1086-00 from the Latin American and Ca crown shaft 1 m tall, green, abruptly swollen ba- sally. Leaves 4-6, pinnately compound, 3-3.5 m long, with a petiole 50 cm long, the leaflets 26- 35 per side, opposite to subopposite, each split- ting fan-like into ca. 10 stiff laciniae held in dif- ferent planes and conspicuously whitish below from a waxy coating. Inflorescence ca. 2 m long, erect in bud and at anthesis, in bud ca. 2 m long with 5 primary bracts and 3 bract scars, the 4 closed bracts rupturing with growth, the inner- most subwoody and puberulous externally with rather flattened basally swollen trichomes, these largely caducous except the scale-like base, the y paniculate, with ca. 45- 50 erectly subhorizontal oe axes, these mostly 3-4-branched ca. 7-15 cm from base, at maturity downcurving from the central axis from the weight of the fruit, near apex only staminate flowers, with pistillate flowers intermixed in triads with pairs of staminate flowers on basal 60%, the flowers white to yellowish, the male with 6 sta- mens. Fruit ovoid, 2-2.5 cm long when fresh, deep green. Additional specimens о PERU. PASCO: Provincia Oxapampa, 20 k of Oxapampa on road to Paucartambo, 1,970 m (75°28' W, 10%35'S), Smith & Pretel 1655; Oxapampa, 1,860 m (75°21' W, 10?34'S), Smith & Brack 2941; Oxapampa-Cerro de Pasco road, 20 km W of Oxapampa, 1,980-2,000 m, lower mon- tane forest on steep slopes dominated by Dictyocar- yum, ca. 10%45'S, 75°50'W, Gentry, Smith, Vasquez & León 39921. SAN MARTIN: Rioja Provincia, Campa- mento Garcia, km 384.5 Olmos—Moyobamba road, 2,250 m (77°21'W, 5°45'S), Smith 4841 (all MO, USM, and to be distributed). Although this is the first record of Dictyocar- yum for Peru, the species described above is the absolute dominant in an interrupted band of montane forest that extends along much of the ribbean Bureau Office of De- velopment Resources) for support of s research, M. Cunningham for the си ең М. Balick, J. Dransfield, A. Henderson, R. W. Rea D. N. Smith, and N. Uhl for reviewing the ma ? Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. ANN. Missouni Bor. GARD. 73: 158-165. 1986. FIGURE 1. Cerro de Pasco road, 20 k D. lamarckianum trees MODEL upright rescence. — Vasquez & León 40003 (M W of Oxapampa; all p eastern face of the Cordillera Oriental, mostly between 1,800 and 2,200 m altitude (Fig. 1A, B). It is known locally as “basanco” (Oxapampa) and “ропа” (San Martin). In Burret's (1930) syn- opsis of Dictyocaryum, two Colombian species were recognized as well as one each from Bolivia 7 described Ecuadorian species occurs at lower al- titudes (1,200-1,400 m) and is thus unlikely, on GENTRY —PERUVIAN PALMS preanthe esis infloresce . G. trigona, в inflorescence and leaves "nm piliers shears are 22 cm long), Gentry, Smith, 159 A 4 m iniu gd nda and Geonoma. — А. D. lamarckianum dominated forest at 1,900 m on Oxapampa- alms in D are Dictyocaryum. — B. Close view of pair of e and suberect openly paniculate fruiting inflo- oe кырс to be conspecific with the n plan Ithough identification of the aaa a as conspecific with Bolivian D. lamarckianum, which also occurs at some- what lower altitudes, is rather tentative, the Bo- livian and Peruvian populations surely look the same in the field. In Bolivia, Dictyocaryum oc- curs mostly at somewhat lower altitudes [1,500— 1,600 m; Gentry & Solomon 44470 (MO) from La Paz Department, 4 km above Incahuara, 13.5 km above San Pedro, bosque pluvial subtropical sensu Holdridge, 67?35'W, 15?55'S]. Although 160 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 IGURE 2. Wettinia ше А. Сепїгу.—А. Еги infructescence with fruit fallen. A-D. Gentry V Smith Р (МО). TE Young infructescence and subtending bract, Gentry, Smith & рў 42009 (МО). less dominant than іп Peru, the Bolivian plant also is an extremely conspicuous component of its habitat with eight individuals occurring in a 1,000 m? sample area near Incahuara, where it was the seventh commonest species. Wettinia longipetala A. Gentry, sp. nov. TYPE: . Leaf apex.—C. Middle leaf segment.—D. Old Peru. Pasco: Province Oxapampa, Serrania de San Matias W of Puerto Bermudez, near top of fila, 900-1,050 m, ridgetop thicket with lowland tree species, 10%25'S, 74*58'W, 15 June 1983, Gentry, Smith & Jaramillo 42009 (holotype, MO; isotypes, MO, USM). Figure 2. 1986] Caulis singulis. Pinnae foliorum indivisae. Inflores- centia non ramosa, floribus dense aggregatis, petalis florum femineorum (in fructu) 2-3 cm longis. Fructus s dense obtectus. Planta flo- ribus ш кн longissimis a omnibus speciebus bene dis сї irtis rigid Single-stemmed tree ca. 6 m tall with stilt roots at base. Leaves ca. 4 m long with ca. 30 leaflets on each side, the leaflets evenly arranged, un- divided, subopposite at base, alternate at apex, to 50 cm long, broader (to 9 cm across) toward the asymmetrically praemorse apex, gradually and uniformly contracted to the 2-3 cm wide base, with numerous (mostly 10-12) uniformly thickened main veins, these tannish below, the rulous below with numerous short, ilar trichomes, but glabrescent, the rachis sharply angled above, rounde low, densely covered with flat appressed scale-like brownish tri- chomes, scabridulous. Inflorescences erect in bud, arising from an extended section of top of trunk, with ca. 4 bracts, the longest to 20-25 cm long, the outer bracts longitudinally striate, puberu- lous with appressed reddish trichomes, more or less glabrescent, the peduncle ca. 10-15 cm long, more or less rufescent with short erect trichomes, на single spike ca. - 13-1 3 ош long, the petals of str ia te, ета 2-3 ст long and 2-3 mm m wide, exceeding the fruit, the sepals triangular, 5-7 mm long. Fruit densely appressed on the rachis, angular- obconical, densely scabridulous-rufescent with short reddish trichomes, ca. 2.5 cm long at ma- turity, subtended by the persistent tepals, the per- sistent linear style ca. 1 cm lon This species is a distinctive and characteristic element of the ridgetop bamboo thickets of the Serrania de San Matias between: the Rios Pichis and Palcazu, a rain forest under the Holdridge system. Further southwest, toward Villa Rica, it occurs at slightly lower altitudes (down to 700 m) on the steep slopes along the upper Palcazu tributaries. Additional specimen examined. PERU. PASCO: drainage of Río Palcazu between km 51 and 60 of the i de road in ои NW of Villa Rica toward Puerto Berm , wet tropical forest in steep foothills, 10°30’ S. "es Ww. 4 Mar. 1982, Gentry & Smith 36055 (MO). This species is remarkable i in n the genus for the extremely | , muc longer than in aay other species of the genus and GENTRY —PERUVIAN PALMS 161 at least twice as long as in any of the related single-spiked species. In Moore and Dransfield's (1978) treatment it keys out with Wettinia au- gusta Pavon & E. That species (as represented by four Peruvian collections, including topotypic material, at MO) differs, in addition to the much shorter petals, in having densely villous fruits. The short, rather scabrous, fruit indumentum of W. longipetala is highly unusual in subgenus Wettinia and more like that of some species of occurring at lower elevations (below 500 m) in tropical moist forest; the much commoner W. maynensis Spruce, which is very different in its branched inflorescences, occurs from the Ama- zonian lowlands to over 1,500 m, thus spanning the altitudinal range of both W. longipetala and W. augusta Geonoma trigona (Ruiz & Pavon) A. Gentry, comb. nov. Carludovica trigona Ruiz & Pa- von, Syst. Veg. Fl. Peruv. Chil. 293. 1798. sets trigona (Ruiz & Pavon) Pers., Syn. 1. 2: 576. 1807. Salmia trigona (Ruiz & ace Willd., Ges. Naturf. Freunde Berlin Mag. Neuesten Entdeck. Gesammten Na- turk. 5: 401. 1811. TYPE: Peru. Huanuco, Pavon s.n. [FI, not seen; MA (F negative 29563), destroyed fide Harling, 1958]. Treelet 2-3 m tall. Leaves bifid, thick-coria- ceous, very closely plicate with ca. 20 ribs on each side, the blade ca. 30 cm long, 19 cm from apex to apex of midvein, the midvein ca. 12 cm a bract and the enlarged prophyll ca. 30 cm long, the tubular sheath thus formed subwoody, 2-3 cm wide, somewhat compressed, persistent, the peduncle 30-40 cm long, mostly enclosed by the prophyll and bract, the central axis well devel- oped, the rachillae strongly ascending, 8-11 cm long, ca. 6-7 mm diam., densely brownish pu- berulous with mostly branched trichomes be- tween the glabrate grayish lips, the flower pits with single conspicuous, bifid lower lip, spirally arranged in ca. 6 series, adjacent pits separated by 1 mm, the flowers (in bud) about 2 mm long, the stamen filaments linear, fused at base, ca. 1 mm long Additional specimens examined. PERU. PASCO: San Gutardo, Oxapampa—Cerro de Pasco o road 30-35 km W of Oxapampa, elfin en and “pajonal,” 2,650- 162 2,800 m, 3 Feb. 1983, Gentry, Smith, Vasquez & León 40003 (AMAZ, MO, USM). The plant described above apparently repre- sents the rediscovery two centuries later of a species previously known only from the type col- lected near Muna, Peru by Ruiz and Pavon and described by them in the wrong family. Redis- covery of this species thus resolves the long- standing mystery of the identity of Carludovica trigona. Carludovica trigona, incompletely described, apparently from a sterile plant (though said to flower in April, May, and June), was for many years associated with the common Cyclantha- ceae species now known as Evodianthus funifer subsp. peruviana Harling, and that species was treated as Carludovica trigona in the “Flora of Peru." e type material of C. trigona appar- ently was destroyed while on loan from Madrid to Berlin during World War II. In the process of preparing his monograph of Cyclanthaceae, Har- ling (1958: 273) examined a leaf from an isotype of C. trigona preserved at FI and realized that it belonged to Palmae rather than Cyclanthaceae. However, he did not propose the necessary new combination in Geonoma. No Geonoma similar to this species is included in the ““Flora of Peru" (Macbride, 1960) or in Wessels-Boer's (1968) monograph of Geonoma, where it would key out with the very different pinnately compound- leaved Colombian species G. dicranospadix Bur- ret. I originally intended to describe the recently collected plant described here as a new species. However, it seems clear from the type photo- graph that this distinctive palm, with its incred- ibly thick-coriaceous corrugate leaves, is, in fact, the same as Carludovica trigona. Geonoma trigona turns out to be very char- acteristic in a very specialized shrubby xero- morphic high altitude vegetation locally called **pajonal." The distinctive ““pajonal” vegetation type is characteristic of exposed ridges, mostly between 2,800 and 3,000 m, in the Cordillera Oriental ofthe Central Peruvian Andes. The “‘pa- jonal" interdigitates with “‘ceja de la montana" elfin forest and is apparently edaphically limited, a conclusion supported by the extremely coria- ceous leaf textures of many of its species. By far the most dramatically coriaceous “pajonal” species is G. trigona (Fig. 1C), which may have the thickest and most scleromorphic leaf of any palm in the world. The striking texture of the closely and rigidly plicate leaves of G. trigona ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 suggests that it would be of great horticultural interest to palm fanciers, especially since its high altitude habitat indicates that it is probably cold resistant Chamaedorea poeppigiana (Mart.) A. Gentry, comb. nov. Morenia poeppigiana Mart., Hist. Nat. Palm. 3(7): 161. 1838; Hist. Nat. Palm. 3(9): 309. pl. 140, 141. 1849. Nun- nezharoa poeppigiana (Mart.) O. Ktze., Re- vis. Gen. Pl. 2: 730. 1891. TYPE: Peru. Huá- nuco: Río Chinchao, Poeppig 1546 [W, not seen (F negatives 29901, 29902)]. Morenia fragrans Ruiz & Pavon, Fl. Peruv. Prodr. 140. pl 1794; Syst. Veg. Fl. и Chil. 299. 1798. ТҮРЕ: Peru. Huánuco: Muña, Ruiz & Pavon s.n. [G, not seen (F negative TP non Chamae- dorea fragrans (Ruiz & Pavon) Mart., Hist. Nat. 3: figs. 1, 2. Nunnezharia fragrans Ruiz '&P haroa morenia O. Ktze., Revis. Gen. Pl. 2: 1891 (nom. nov. for Morenia fragrans Ruiz & Pavon). This relatively large Chamaedorea species with a 2-7 m tall trunk, is locally fairly common in wet premontane forest between 700 and 1,100 m on the eastern slopes of the Peruvian Andes. It is distinguished from other Peruvian species of the Morenia alliance of Chamaedorea by its narrowly lanceolate leaf segments. Recent col- lections have been made in San Martin, Huan- uco, and Pasco departments, and Macbride (1960) cited earlier collections from Bolivia and from Amazonas Department, as well as the type from mudez, C. poeppigiana occurs in premontane rain forest on the same steep foothills of the upper Palcazu drainage as does Wettinia longipetala. [km 51-60, Villa Rica-Puerto Bermudez, 700 m, Gentry & Smith 36000 (MO, USM)]. In San Martín Department it is known as “‘shucso ne- gro." Ip PIUpPuU5 thisr it ҮҮ of approval of Proposition 458 (Moore, 1979), which has already been unanimously approved & Pavon. Apparently Morenia poeppigiana 1s 1986] conspecific with M. fragrans as already suspected by Macbride (1960), and that specific epithet be- comes the appropriate one for the species in Cha- maedorea Chamaedorea latisecta (H. Moore) A. Gentry, comb. nov. Morenia latisecta H. Moore, ~ Herb. 8: 203. 1949. TYPE: Colombia. : Sibundoy, 2,225-2,300 m, а & Villareal 7676 (not seen, fide Moore, 1949, fig. 87) The Peruvian Chamaedorea poeppigiana is not the only Morenia species for which a new com- bination is needed in Chamaedorea. However, of all the middle elevation species of Morenia, M. latisecta may be the only one adequately dif- ferentiated from Chamaedorea montana (Humb. & Bonpl.) Voss for specific recognition. It is dis- tinguished by its broadly lanceolate leaf seg- ments. If Morenia and Chamaedorea are merged under the latter name, the above combination is needed. I have refrained from proposing other new combinations in Chamaedorea in antici- pation that several of approximately ten Morenia names apparently accepted by Moore (1979) will prove synonymous with Chamaedorea montana (Humb. & Bonpl.) Voss, which has the oldest available basionym and for which the combi- nation in Chamaedorea has already been made. Chamaedorea megaphylla A. Gentry, sp. nov. TYPE: Peru. H : Provincia Leoncio border with Ucayali, 1,620-1,760 turbed cloud forest, 75°48'W, 9°5’S, 10 Aug. 1980, Gentry, Salazar & Horna 29572 (ho- lotype, MO; isotypes, AMAZ, MO, USM). Arbor dioeca, 5 m alta. Folia 2 m longa, pinnis utrinque plus quam 20, lineari-lanceolatis, pro parte maxima 50-58 cm longis. Inflorescentia feminea pan- iculata, ramis lateralibus 19-24. Flores feminei caly- cibus cupula aris 2-2.5 mm longis, petalis ovatis, 3-4 mm longis Dioecious tree 5 m tall, the trunk ca. 4 cm diam., green, ringed. Leaves 2 m long, pinnate, the rachis triangular in section, acutely so toward apex, with over 20 pairs of opposite or subop- posite segments, these averaging a pair every 10 cm except at apex (5 pairs in apical 22 cm), the T linear-lanceolate, not noticeably sig- oid, 50-58 cm long for most of length, the x Se segments smaller and as little as 20 cm long, mostly 9-10 cm wide (the terminal seg- GENTRY—PERUVIAN PALMS 163 ments as little as 3 cm wide), somewhat corru- gated when fresh, drying gray-green with con- trastingly tannish main veins below, mostly with ca. 10 main veins per segment, fewer in narrow ee leaflets, glabrous except for a few minute very inconspicuous scales on underside. In- ee (only female seen) whitish in flower, turning green in fruit, the peduncle 34-36 cm long, semicircular in cross section and 2 cm across at base, bearing 5 chartaceous brownish bracts, these glabrous except for few minute scales, the lowermost attached at base, the second 2-3 cm above base, the third 9-14 cm above base, the fourth ca. 27-28 cm above base (and 8-9 cm below lowermost inflorescence branch), these bracts 3-4 cm wide when flattened, the upper- most bract much smaller and caducous (ca. 3 cm long), 3-4 cm below basal inflorescence branch- es, the rachis 15-18 cm long, with a well-devel- oped straight central axis and ca. 19-24 lateral branches 9-18 cm long, the flowers not sunken into rachillae, adjacent flowers mostly separated by ca. 5 mm. Female flowers with the calyx lobes fused into a 2-2.5 mm long basal cupule, the ovate acute petals 3-4 mm long, sometimes wit a more or less distinct dorsal keel, the thick tri- angular-patelliform ovary ca. 2 mm long and 3 mm across, the stigma lobes sessile at top of ovary, ca. 1 mm long, erect to horizontal. This species is apparently endemic to the iso- lated cloud forest of the upper part of the Cor- dillera Azul between Tingo Maria and Aguaytia. This area, designated by the Holdridge system as premontane rain forest, is well known for its endemism [e.g., Syngonium gentryi Croat, and the frog Dendrobates silverstonei (Myers & Daly, 1979)]. The remnant patch of forest in which megaphylla was discovered has since been c down and we searched in vain for a material in 1982. This species is quite unlike any taxon of Cha- maedorea (or Morenia, which was erroneously differentiated as monoecious and with which it is more closely allied) treated in the “Flora of eru” (Macbride, 1960) in its generally much larger dimensions. The only Peruvian species whose description approaches that of C. mega- phylla is Morenia macrocarpa Bur., known only from the incompletely described type and a sec- ond topotypic collection from a lower altitude (600-700 m) in the Huallaga Valley. From its description, Morenia macrocarpa, of which I have seen no authentic material, differs strongly from 164 M. megaphylla in the linear leaf segments; Mac- bride (1960) suspected that M. macrocarpa might be synonymous with M. linearis (Ruiz & Pavon) Bur., which has similarly narrow leaf segments. While it is clear that this species is new to Peru, m proposed from elsewhere in Latin America are not conspecific, since most descriptions are very incomplete. The only other species of Chamae- dorea sensu lato known to me to have leaves as long as the 2 m ones of C. megaphylla is a col- lection from 2,000 m near Baeza, Ecuador [Bal- slev & Madsen 10346 (MO)] that I have referred to “Morenia” caudata Burret, largely because that species was described from near its collec- tion locality, but which is very different from C. megaphylla in its much narrower leaflets. Un- fortunately leaf length of DK usually cannot d very few of Burret's (1933, 1936) descriptions give leaf dimensions. More important, all of the de- scribed South American species of Chamaedorea (or Morenia) differ from C. megaphylla in having leaves with either many fewer segments or the segments much shorter and/or narrower and/or sigmoid, or the terminal segments broader and confluent. From the description, the only re- motely similar species of Chamaedorea with me- dian leaf segments as broad as those of C. mega- phylla is Morenia latisecta H. Moore of Putumayo, Colombia (see above), but that species differs in shorter (0.3-1.3 m) leaves and the broadly lanceolate shape of its leaf segments. have not attempted to account for the pleth- ora of Central American Chamaedorea names, but assume on phytogeographical grounds that none is relevant to the Peruvian species since it is quite unlike anything known from geograph- ically intermediate Panama or Costa Rica. from herba Chamaedorea smithii A. Gentry, sp. nov. TYPE: nín: Tarma-Chanchamayo border, Río Tulumayo drainage, Rondayacu, N of Monobamba, 45 km S of San Ramón, Podo- carpus forest, 1,800 m (75?25'W, 11?20'S), 15 Oct. 1982, Foster & Smith 9171 (holo- type, MO; isotypes MO(3), USM, to be dis- tributed) bor parva, 2 m alta. Folia ca. 65-75 cm longa, pinnis utrinque 8-9, реА vei lanceolato- ellipticis, non sigmoidei, 15-29 em 1 —5 cm latis. Inflo- р боев са. 30 ст Іоп- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 go, ee pron ramis lateralibus 15-21, fere nn diculari 1.5m longis, an i-a 1-1.5 mm longis. Fructus Em bosus, 0.8-1 cm Slender treelet 2 m tall, the stem green, with prominent rings. Leaves ca. 65-75 cm long, pin- nate with 8-9 pairs of alternate « or i оруш segments, adj ated by 3-7 cm, the segments ‘variable i in Ma Dao. е. elliptic to linear, not sigmoid, 15—29 m long, long acuminate, the lowermost less than cm wide, some of the terminal and middle segments to 5 cm wide, drying grayish green with contrastingly tannish main veins below, mostly with 5-7 main veins, fewer in the narrow basal segments, the “main” veins intergrading into and not very strongly differentiated from the others, more or less minutely lepidote on underside of leaf segments, otherwise glabrous; petiole at least 9-15 cm long below lowermost segments. Inflo- rescence (only female seen) with peduncle ca. 30 cm long, bearing 4 narrow (ca. | mm wide) mem- branaceous bracts, the lowermost attached near base and ca. 8-9 cm long, the second attached ca. 4cm above base and ca. 17 cm long, the third attached ca. 13-14 cm above base and ca. 31 cm long, the uppermost ca. 24-25 cm above base (7-9 cm below basal inflorescence branches) and ca. 5-6 cm long, the rachis 10-13 cm long, rather straight, the 15-21 lateral branches 3.5-5 cm long, nearly at right angles with rachis in flower, the flowers not at all sunken into rachillae (the scars actually slightly raised), adjacent flowers mostly separated by several mm. Female flowers with the calyx lobes fused into a cupule ca. 1.5 mm (to 2 mm when in fruit) long, the broadly obovate petals ca. 1-1.5 mm long, folded over bud, the thick triangular-ovoid ovary ca. 1.5 mm long, the 3 stigma lobes sessile at top of ovary, ca. 1 mm long, recurved. Infructescence with the lat- eral branches to 14 cm long, the fruits globose, turning red at maturity, 0.8-1 cm diam., sub- tended by the persistent and conspicuous calyx Thi ies i tly endemic to the lower montane Podocarpus forests of intermediate el- evations in the central Peruvian Andes. Al- tho C. smithii is locally very common in the remnant Podocarpus forest at Rondayacu, this is one of the last vestiges of this vegetation type in Peru. The new species must thus be regarded as a highly threatened one, helping emphasize the 1986] need for an increased focus on conservation of the last remnants of this endangered vegetation pe. Chamaedorea smithii keys out in the “Flora of Peru" to the vicinity of C. pauciflora Mart. and C. boliviensis Damm. It differs from lowland C. pauciflora in non-sigmoid mostly broader leaf segments with more (5-7) main nerves, the branched inflorescence, and smaller (7-8 mm long and wide versus 10 mm long by 7-8 mm wide) fruits. From C. boliviensis, another Amazonian lowland species, it differs in much narrower (to 5 cm wide versus a described 7-7.5 cm wide) leaf segments, and the globose fruit. This species is also related to C. /anceolata Ruiz & Pavon, a widespread, common, and highly variable species, but differs conspicuously from that species (and its several potential segregates) in having a py- ramidal inflorescence with a straight well-devel- oped central rachis and short (less than 5 cm in flower), numerous (15-21) lateral branches. While a few Central American species (e.g., C. schippii Burret of Belize and C. sartorii rd of Mexico) have somewhat similar inflores- cences, none of these has non-sigmoid leaf seg- ments irregular in width and broad in part. Al- though I have not been able to check all the Central American species of the genus, no Cha- maedorea species of southern Central America is at all similar to C. smithii, suggesting that it GENTRY —PERUVIAN PALMS 165 is phytogeographically highly improbable that it could prove conspecific with any of the poorly known Central American taxa not represented at MO. LITERATURE CITED Воввет, M. 1930. Iriarteae. Notizbl. Bot. Gart. Ber- lin-Dahlem 10: 918—942. 33. Chamaedorea Willd. und verwandte Palmengattungen. Notizbl. Bot. Gart. Berlin-Dah- lem 11: 768. TOS Die Palmengattung Morenia R. & P. Notizbl. Bot. Gart. Berlin-Dahlem 13: 332-339. Neue Arten aus Ecuador III. Palmae. Notizbl. Bot. Gart. Berlin-Dahlem 15: 23-38. GENTRY, A. H. 1980. The flora of Peru: a conspectus. Fieldiana, Bot. n.s. 5: 1-11. Bis oos G. 1958. E quos ofthe Cyclanthaceae. ta Horti Berg. 18: MU J.F. 1960. E In Flora of Peru. Field Mus. Nat. Hist., Bot. Ser. 13: 321-418. MOORE, H. E., JR. maedorea Willd. over Morenia (Palmae). Taxon 27: 555-556. RANSFIELD. 1978. A new species of Wettinia and notes on the NN Notes Roy. Bot. Gard. Edinburgh 36: 259 Myers, C. W. & J. W. DALY. s A name for the poison frog of Cordillera Azul, eastern Peru, with notes on its biology and skin toxins. Amer. Mus. Novit. 2674: 1-24. WESSELS-BoER, J. С. 1968. The Geonomid palms. Verh. Kon. Ned. Akad. Wetensch., Afd. Natuurk., Ser. 3, 58: 1-202. чө 1979. Proposal to conserve Cha- Ruiz et Pavon A GUIDE TO COLLECTING PALMS! JOHN DRANSFIELD? ABSTRACT Because of their frequently bulky nature and unfamiliar morphology, palms need special attention if they are to be adequately collected. A guide is presented to indicate how to go about preparing good herbarium specimens of palms and what notes to make in the field. The making of good herbarium specimens of palms is a laborious, time-consuming, and often rather unpleasant activity, and because of this many palm taxa are very poorly represented in herbaria. The general collector, faced with lim- ited time and money and with the need to collect large numbers of plants to satisfy funding bodies or exchange agreements, tends to shy away from such awkward plants. To the specialist, however, th h sad raraful call fi nfa 1 ti ] can be immensely satisfying. It is probably too much to expect general collectors to use the amount of time spent by a specialist palm col- lector in the preparation of herbarium speci- dequ $ ed. Guidelines for the collection of palms have already been provided by Tomlinson (1965) and, for climbing palms, Dransfield (1979). It is hoped that the present article will о а wider audi g e careful palm collecting. ACCESS TO THE PALM Low forest undergrowth and savannah shrub- like palms do not present many problems of ac- cess. Tall tree palms on the other hand can be very difficult to collect. There are three main methods of attack on such palms: 1. Tree climbers. In some parts of the trop- ics, local people are adept at climbing palms and when such people are employed as field assis- tants, the collecting of tall palms is usually rel- atively straightforward. However, some tree palms cannot be climbed because their trunk di- ameter may be too great, the trunks may be armed with spines, or the trunk and leaf bases may be infested with ants. 2. Climbing irons, loops, and ropes. Such mechanical aids to climbing are usually bulky ! I am grateful to Carol Buckman for preparing the plate and require courage and practice to develop skill in their use. In some parts of the world they may be the only means of getting at fresh material of leaves and flowers without felling the palm. 3. Felling. There is no doubt that the felling of a palm provides the best access for collecting. Where a palm is multiple-stemmed, the felling of one stem for collection will probably do little permanent damage. However, in single-stemmed taxa, the collector is faced with the dilemma of completely destroying the palm. Where such a palm is abundant, the sacrifice of one individual seems justified. Even when a palm is rare, it may be possible to find individuals that can be sac- rificed with a relatively clear conscience to the collector, e.g., if the forest area is being felled in any case. Ifa palm is felled, maximum use should be made of the opportunity provided by making several good collections. In populated areas of the tropics, palms may be the property of local people even if the plants are obviously wild. Con- siderable tact may be required to gain permission for collecting. It is also important to be aware of the dangers involved in the felling of a large palm tree. Palm crowns often provide shelter for a whole range of animals such as ants, hornets, reptiles, spiders, scorpions, birds, and bats. Al- though the dangers are probably not too great, it is certainly advisable to work with care when cutting off palm sheaths and d g the leaves It is also worthwhile to scan the crown with bin- oculars for hornets, wasps, and bees before felling is begun. EQUIPMENT The services of one or two knowledgeable local people make palm collecting much easier, more entertaining, and usually p de ethnobotanical data. A “parang” or machete and a good pair of ? Herbarium, Royal Botanic Gardens, Kew, Richmond, s TW9 3AB, England. ANN. MissouRi Bor. GARD. 73: 166-176. 1986. 1986] secateurs are essential equipment. Secateurs should be of the type with two cutting edges rath- er than the type with a single blade working against a flat surface, favored by some collectors. Strong rough leather gloves are invaluable; it is best not to use gloves consisting of leather palms and canvas backs as these give insufficient pro- tection to the backs of the hands. An axe may be invaluable but of course is yet one more heavy item to carry. A large supply of hanging tags with long strong thread is essential. The quality of collections is greatly enhanced by copious field notes, therefore a collecting book with plenty of room for notes is also essential. Photographs are of great value; the collector may wish to include a supplementary wide-angled lens with the pho- tographic equipment, as it is frequently very dif- ficult to photograph entire tree palms within for- est with only a standard or macro lens. A supply of formalin acetic alcohol (F.A.A.) in wide-mouth plastic bottles and some muslin are important items for the preservation of material for mor- phological or anatomical study. The method of preservation of the herbarium material will vary with local circumstances. It must be said, however, that in humid tropical conditions the modified Schweinfurth method using newspapers, methylated spirits or indus- trial alcohol, and large, thick gauge polythene bags is the easiest way of dealing quickly with the day’s palm collections. Even during the dry season in savannah areas, it may take an inor- dinately long time to dry bulky palm material in the sun or over plant driers, and the longer drying takes, the more the flowers and fruits are likely to become detached from inflorescences and in- fructescences. WHAT TO COLLECT It is not proposed to discuss palm morphology in great detail in this guide, although a certain amount of discussion will be necessary for an understanding of what organs need to be col- lected. Those interested in a more detailed mor- phological account should refer to Moore (1973) and Moore and Uhl (1982). For some genera sterile collections are useless, whereas for others, especially the Asiatic climb- ing palms, they can be of considerable value. Variation in leaf dissection is a feature of several undergrowth genera such as Pinanga, Iguanura, and Geonoma, and in these genera supplemen- tary sterile collections illustrating leaf variation DRANSFIELD — COLLECTING PALMS 167 are useful although not as reliable as when inflo- rescences can be collected too. Dead inflores- cences and infructescences are well worth col- lecting if no fresh material is available or if fresh material is only sufficient for a unicate or one duplicate. When the frustrating circumstances arise of not being able to climb or fell a tree, fallen leaves and inflorescences, accompanied by photographs and good notes, will make a rela- tively good collection. Deciding what parts of the palms should be collected is really a matter of common sense— how can the morphology of the whole palm be represented in the herbarium. A good collection should comprise adequate samples of all the or- gans of the palm available, coupled with notes to allow a reconstruction of the habit, dimen- ha by ecological and ethnobotanical data (see Fig. 7). Ripe seed may be collected at the same time and be grown for research or ornamental pur- poses (see below), and pollen samples also can be conveniently made directly in the field. n making collections, it is advisable to cut all palm pieces to the approximate size of a folded sheet of newspaper. It may not be possible or desirable in some instances (e.g., prophylls or peduncular bracts) to cut the object at all, and such larger objects may always be treated sepa- rately, however, it does help considerably if frag- ments are more or less of the same size. Each fragment must carry a hanging tag bearing the collection number written in pencil or ink in- soluble in water or alcohol. This cannot be stressed too strongly; however, it does mean that that the most satisfactory method is to begin by laying down four double thicknesses of news- paper opened out (Fig. 1а). This becomes the wrapping of the finished bundle (Fig. 1b). It is best to begin with flat leaf fragments, placing them in double or single folds of newspaper, and progress inwards to spiny or bulky fragments in the middle (to prevent puncturing of the poly- thene bags), followed by less bulky, flatter ma- terial at the top. The bundle can then be tied up within the four folds of double thicknesses of newspaper (Fig. 1b). The bundle is then put into a polythene bag, doused with spirit (it may be necessary to make holes in the top end of the newspaper folds to allow spirit to get to the mid- 168 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 O Mes ШШШ 1 li Vau | а Ma: ТТ [ШШ ШЕ yu TT ТВ n LT NH PT li | || fl Mii zu УИ [| ——— e p ier аена > —a. Starting a bundle of palm specimens.—b. with spiri dle of the bundle), and then tied up; the whole can then be put into a second polythene bag for safety. Because of their bulkiness it has been found that many palms, if well pressed and packaged, can put up with a certain degree of rough han- dling if the polythene covered bundles are sewn up in hessian sacks. They can tolerate two to five — : — T — — Z" — = == eee - PL |. - E AA - << ve car x - : = —- = — — 7 ==. = - ES —-— —. — 2 -—- = -_ — — ———— =— pe A l — = | A finished bundle for enclosing in plastic and dousing months in transit if well soaked with spirit (be- fore shipment it is advisable to check the state of the bundles in case drying out has occurred). In the following notes, the general morphology of palms is briefly discussed, the parts necessary for a good collection indicated, and the need for notes explained. 1986] wt er ae E о тиби? P q... han qa? 85е, zm DUET. FIGURE 2.—a. Whole pal leaf sheaths. WHAT TO NOTE AND WHAT FRAGMENTS TAKE General data. Locality, habitat, elevation, vernacular name and uses, and date. DRANSFIELD—COLLECTING PALMS 169 i ий stem.—b. Surface slice of stem.—c. Cross-section of stem.—d. Rattan stem with Habit. Note whether palm is single-stemmed or clustered; if clustered whether the clump is close or diffuse, whether with stolons or not. Note whether the palm is a tall tree, or undergrowth palm, or *stemless" or climbing, whether stilt- 170 rooted or not. Estimate the total height. Photo- graphs showing habit are of great value. Stem. Note whether stem is bare or obscured with leaf sheaths or bases, whether aerially branched or not; if the stem is bellied this should be noted; note any armature, the type of nodal scars and the color. Measure the diameter with and without leaf sheaths, and the length of the internodes; estimate the total length if different from the palm height. In small palms, collect a length of stem (Fig. 2a), which can be split vertically if necessary (Fig. 2b), and in large palms take a sample of the outer stem as if for a regular bark sample (Fig. 2c). Roots. Note any peculiar rooting behavior (e.g., stilt-roots, spine-like adventitious roots, apogeotrophic breathing roots, etc.). Collect samples or roots where appropriate. Leaf in general. Note number of leaves in crown, their arrangement (distichous, tristi- chous, or spiral), whether neatly abscissing, mar- cescent, or persistent, and try to describe general “hang” of leaves. Photographs showing leaf hang are invaluable. Leaf sheath. Note whether tubular (Fig. 5c) or open (Fig. 5d), position of any splits, whether forming a crownshaft (Fig. 5c) or not, whether becoming fibrous; note any armature, indument and color. Measure length and width or diameter. Collect whole sheath when small (Fig. 3a), and where large, collect sections of the sheath to in- clude the base and apex, especially the junction of the sheath with the petiole. In climbing palms it is convenient to take the sheath sample as a section of stem and its surrounding sheaths that need not be removed (Fig. 2d). Petiole. Note presence or absence; if present note any general features of indument and ar- mature. Measure length and, where too large to represent, width. If short, collect whole petiole including the in- sertion of the first leaflet or base of the lamina (in fan palms). If long, collect portions to illus- trate any changes in form or armature along the length and a piece to include the insertion of the first leaflet or base of the lamina (Figs. 2a, 3c). Blade. Note general characteristics and di- mensions; whether palmate, costapalmate, pin- nate, bipinnate, entire, or bifid. Measure the length of the rachis or costa. In pinnate palms note the number of leaflets on each side and how they are arranged (regu- larly, grouped, in one plane or several, pendu- lous, etc.) and any special color features. In palmate palms note the number of segments ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 where leaf too large to collect whole, whether stiff or pendulous, and any special color feature. Collect whole leaf where very small; the blade may be folded up if necessary. When of moderate size collect the base of the blade (Fig. 3a), a mid- portion (Fig. 3b), and the leaf tip (Fig. 3c), the leaflets or segments may be removed from one side of the rachis. When very large, the same samples are required but they may comprise one or two leaflets only; the rachis may have to be split and the leaflets folded up. In fan palms col- lect the tip ofthe petiole with the blade base (Fig. 4c) and hastulae and basal segments on one side, and two pieces of blade to include the central segments and lateral segments (Fig. 4d). In climbing palms. In addition to the above features, note the position ofthe climbing whips— whether an extension of the leaf rachis (cirrus) (Fig. 5a) or borne on a leaf sheath (sterile inflo- rescence = flagellum) (Fig. 5b). Collect a sample climbing organ, which may be folded up (use gloves!). Inflorescence. Note position— suprafoliar, interfoliar, or infrafoliar, whether solitary or multiple, and whether shorter or longer than the leaves. A photograph of the whole inflorescence would be very helpful. Note the sex (if petia whether hermaphrodite, polygamous, mono cious, staminate, or pistillate (in dioecious species never ever mix sexes in the same coll num- ber; give them each a separate number). Peduncle. Note orientation and measure length; note any special features of indument. Primary bracts. Note number; the pro- phyll (Bract 1) is often obscured by leaf sheaths but is of considerable taxonomic importance. Rachis. Note length. Branching. Note whether inflorescence branched or unbranched and count the num- bers of orders of branching. Where first order branches are distant and highly differentiated, so-called partial inflorescences can be distin- guished; when so, their number should be re- corded. Rachillae. Note length, orientation, and color. Flowers. Note color and scent. Close-up photographs of flowers are very useful. Even fallen flowers may be of value. Collect. If small, collect the whole inflo- rescence (thus saving much of the above note- taking), taking great care not to miss the pro- phyll. The prophyll may be caducous; if so it should be searched for on the ground or among the leaf sheaths. If large, then collect the pro- 171 p y ,a por- velopment from the same palm; if not possi n ofthe peduncle, rachis (Fig. 6), and rachil- ble, do not mix numbers. Inflorescences yll and representative primary bracts, a por later so me mou the same palm may be cut up inflorescence represen duplicate has a major NR j os бон 172 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURE 4. Palmate leaf prepared for pressing.—a. Sheath.— b. Petiole section.—c. Base of blade with all segments removed except at one side.—d. Middle portion of blade. tat pled with frag ts (rachillae) at dif- main collection only when there is no doubt as ferent stages. to their provenance. Fruit. Note any color difference between in- florescence and infructescence. Note color of fruit. Close-up photographs of fruit are helpful. Collect infructescence, rachillae, and fruit. Ripe palm seed is easily transported without Seedlings. Seedlings at the foot ofa palm may loss of viability if germination requirements are not belong. They should be included with the ^ understood, and the opportunity to collect seed PALM SEED 1986] DRANSFIELD—COLLECTING PALMS 173 Uu , FIGURE 5.—a. Rattan with а cirrus.—b. Rattan with a flagellum.—c. Stem bearing a crownshaft.—d. Stem bearing open sheaths. should not be wasted. There is a great demand for palm seed for ornamental purposes, both from botanic gardens and amateur enthusiasts. Fur- thermore, it is sometimes possible, though tricky, to obtain chromosome counts from the root-tips of seedling palms, and the seedling stages of many species are not known and could be usefully re- corded by growing seed of known source. The Portion of an infructescence. FIGURE 6. International Palm Society’s Seed Bank acts as a distribution center for palm seed, sending seed to botanic gardens and its members. Collectors contemplating gathering seed should be aware of phytosanitary and conservation regulations and may also wish to consult the Seed Bank. Ripe seed should be cleaned of any soft pulp. This is often most easily effected by trampling on the ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 whole fruit in a bag and then allowing the pulp to rot for a couple of days; the seed may then be easily cleaned in water. Seed should then be packed damp (not wet) in sphagnum or soft toilet tissue in polythene bags and then posted airmail or carried. Seed may happily germinate in the bags and may be stored for several months. Drying out is lethal to most palm seed. 1986] DRANSFIELD—COLLECTING PALMS ARECACEAE Socratea sp. cm below the crownshaft First gece is whi hotel es in Puyo, however, the are eaten whe P HERBARIUM KEWENSE PALMAE Spanish: "Chonta cade" ECUADOR, Prov. PASTAZA: 4 km south of Shell towards Madre Tierra, just west of Puyo (01°30'S; 7 ). Remnants of tropical forest, 1050 m elev, 16 Mar 1983. A soliatry tree in remnants of virgin t forest, but often left in otherwise deforested eas. e base of the trunk is li from the d about 1£ to ng and 5-8 cm in diam, brown, Spiny adventitious roots tha a loose open inverted cone, 3 m high and 140 cm the base runk straight, smooth, ey, 15.4 m long bouth 13 cm in d wter throughout. Crown of 6 expanded leaves and 1 lance-like, erect oung leaf in the center; crownshaft 165 "lorescence insert the peduncle then ye ing into the \ bears long hanging branches of n young and soft Duplicates: QCA, Latinrecu, NY, K, AAU. H. BALSLEV & L. BRAKO No. 4279 (vel vatda aff.) DET. J.DRANSFIELD FIGURE 7. An exemplary herbarium label. MATERIAL FOR LIQUID PRESERVATION Material preserved in formalin acetic alcohol (F.A.A.) is ideal for most morphological and an- atomical studies and for illustrations of flowers. 4.6.1983 F.A.A. is made by mixing 90 parts of 5096 ethyl alcohol, five parts glacial acetic acid, and five parts concentrated formalin. After fixation in F.A.A. for a few days material can remain in F.A.A. or be transferred to 7096 ethyl alcohol, 175 176 allowing the F.A.A. to be used again. F.A.A. can only stand a little re-use, but it is useful to know this ited. Wide- mouth bottles are essential. Fragments for pres- ervation should be clearly labelled with a tag written in pencil and then should be wrapped in a piece of muslin or gauze and placed in F.A.A. in a bottle. The use of muslin allows for the inclusion of several different collections in the same bottle and also helps decrease damage dur- ing inevitable shaking. The most useful frag- ments for preservation are samples of rachillae with flowers at various stages of development and fruits at various stages of maturity. If inflo- rescence primordia are available (search among the young leaf sheaths) these can be of great value ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 and should be preserved entire, where possible. In more specialist collecting, samples of stem, root, petiole, rachis, and lamina should also be included with the reproductive organs in the liq- uid preserved collections. LITERATURE CITED DRANSFIELD, J. 1979. A manual of the rattans of the Malay Peninsula. Malayan For. Rec. 29: 1-270. Moore, JR., H. E 73. The major group of palms and their distribution. Gentes Herb. 11: 27-140. . UnL. 1982. а ofevolution n palm ms. Bot. Rev. (Lancaster) 48: ain con P. B. 1965. Special V dd for col- lecting palms for taxonomic study. 7n F. R. Fos- berg & M.-H. Sachet. Manual for — Her- baria. Regnum Veg. 39: 112-116 CATALOG OF THE MOSSES OF CHINA! PAUL L. REDFEARN, JR.? AND P.-C. Wu? This catalog lists in alphabetical order 2,264 taxa (2,004 species, 21 subspecies, 196 varieties, taxa included in this list are “Genera Muscorum Sinicorum” (Chen et al., 1963, 1978), “Index Bryoflorae Formosensis” (Lai & Wang-Yang, 1976), and “A Glossary of Terms and Names of Bryophytes” (Wu, 1984). Additional taxa, par- ticularly for mainland China, are reported in the publications by Miiller (1896, 1897, 1898), Brotherus (1924-1925, 1929), Thériot (1932), Dixon (1933), Bartram (1935), Varde (1937), Tixier (1966), Gangulee (1969, 1971, 1972, 1974, 1976, 1977, 1978, 1980), Liao-ning Provincial Agriculture and Soil Research Institute (1977), X.-j. Li et al. (1985), Gao & Chang (1983), Ko- ponen et al. (1983), Iwatsuki & Mizutani (1983), S.-h. Lin (1984), Wu (1982), and Wu et al. (1986). A number of other publications were useful in preparing this catalog. Listed by family, these papers are: Archidaceae (Snider, 1975), Bartra- , 1962, 1963), Brachytheciaceae Ochyra, 1982), Bryaceae (Mohamed, 1979; Ko- ponen et al., 1982b), Calymperaceae (P.-j. Lin, 1984), Diphysciaceae (Iwatsuki, 1976b), Dicra- naceae (Noguchi, 1968a; S.-h. Lin, 1981; Tan & Koponen, 1983), Ditrichaceae (Frisvoll, 1985), Encalyptaceae (Horton, 1982), Entodontaceae (Buck, 1980; Hu, 1983; Inoue, 1983), Fabroni- aceae (Buck & Crum, 1978; Taoda, 1977a, 1977b), Fissidentaceae (Iwatsuki, 1980; Iwatsuki & Suzuki, 1982; Z.-h. Li, 1984), Grimmiaceae (Bremer, 1980a, 1980b, 1981), Hylocomiaceae 1973a, 1973b, 1976, 1978; Iwatsuki, 1965), Les- Кеасеае (Iwatsuki, 19762), Leucobryaceae (Iwat- suki, 1977b), Leucodontaceae (Zhang, 1982), Mniaceae (Koponen, 1971b, 1972, 1981; Ko- ponen & Lai, 1978; Koponen & Lou, 1982), Me- teoriaceae (Noguchi, 1976), Myuriaceae (Iwat- suki, 1979), Neckeraceae (Ninh, 1984; Noguchi, 1984), Orthotrichaceae (Vitt, 1972, 1980a, 1980b), Plagiotheciaceae (Iwatsuki, 1970; Ochyra, 1976), Polytrichaceae (Xu & Xiong, 1982; Nyholm, 1971), Pottiaceae (Chen, 1940, 1941; Saito, 1972, 1975; Sollman, 1983; Zander 1978, 1979), Pterobryaceae (Miller & Manuel, 1982), Ptychomitriaceae (Noguchi, 1968b), Rhy- tidiaceae (Ando et al., 1957; Koponen, 1971a), Sematophyllaceae (Iwatsuki, 1977a; Seki, 1968), Sphagnaceae (S.-h. Lin & Lai, 1981), Splachna- ceae (Koponen & Koponen, 1974; Ochi et al., 1974), Thuidiaceae (Iwatsuki, 1963; Noguchi, 1964; Norris & Sharp, 1961; Watanabe, 1972, 1980a, 1980b), and Timmiaceae (Brassard, 1984). Finally, *Index Muscorum" (Wijk et al., 1959— 1969), “Index Muscorum Supplementum" (Crosby, 1977, 1979; Crosby & Bauer, 1981, 1983), and “Index Muscorum Japonicarum" (Iwatsuki. & Noguchi, 1973, A22) were base in d syn- onyms. Crosby and Magill | (1977) were followed in assigning most genera to familie The equal sign (=) "n меч to desta both synonyms. Where some doubt exists about the status of a name, it is preceded by a question mark (?). Synonyms are restricted to those that have been specifically applied to the mosses of China. We have refrained from making judgements as to the validity of the taxa reported for China. Some reports are undoubtedly based upon mis- identifications or a lack of understanding of a taxon throughout its range. Since China borders + ! The senior author wishes to express his thanks to Dr. Z. Iwatsuki for taking se TAF to discuss with н some of the problems associated with the preparation of this catalog жен for sugges deletions, and synonyms and to D is due my colleague Dr. Wu who spen nt m to Cheryl Bauer for a critical review and editing of this paper. n the Hyp naceae. Particular patos, FY n this grateful Financial and logistical support for the senior author is acknowledged to the Е of Botany, Academia Sinica, Beijing; the Missouri Botanical Garden, and Southwest Missouri State Uni rsity. 2 Department of Biology, етене t Missouri State University, Springfield, Missouri 65804-0095. 3 Institute of Botany, Academic Sinica, Beijing, People’s Republic of China. ANN. MISSOURI Bor. GARD. 73: 177-208. 1986. 178 on As 1 (Northern Asia), and As 3 (India, Burma, Indochina) and has historical biogeographical af- finities with North America (Iwatsuki & Sharp, 1967, 1968), one may expect that when careful comparative studies are made of many taxa throughout their range in Asia and North Amer- ica, a large number will be reduced to synonyms. Furthermore, because China encompasses 60 de- grees of longitude and 38 degrees of latitude (17- 55 degrees N) and varies in elevation from sea level along. the coasts to 7, 556 m in western Si- chuan, i ranging ош alpine bong to tropical rain forests and ea Consequently, future field work should produce additional taxa for China. List OF TAXA тн = Miill. (THUIDIACEAE) oth. tanytric richum (Mont.) Broth. in Par. — Wijkia tany- a е! apillat m (Ha rv.) Fl Acaulon üll. Pomaci .) LACEAE) m (Hedw dE Mitt. (SeMATOPHYLLACEAE) diminutum (Brid.) flagelliferum Sak. = come flagelliferum hyalinum Mitt. — Е а пеит oxyporum (Lac.) Fl veh eii (Reim. pe IM ) Fleisch. zukii Sak. A eden Mitt. Actinodontium Schwaegr. in pee (DALTONIACEAE) rhaphidostegum (C. Müll.) B Actinothuidium (Besch.) Broth. ee hookeri (Mitt.) Broth Aerobryidium Fleisch. in ewe и aureo-nitens (Schwaegr.) В filamentosum (Hook.) Cin in Broth. taiwanense Nog. Aerobryopsis Fleisch. egre e (?) auriculata Copp. e ér. brevicuspis Broth. — Pseudobarbell attenuata мес тре “aging ore = А. subdivergens ssp. scariosa imegriolia (Besch. ) B lon ngiss ima ну ` ME ) Fleisch. = A. wallichii parisii (Ca Bro o o ) Broth. ssp. subdivergens = scariosa (Bartr.) Nog. robusta Card. = A. subdivergens ssp. subdiver- gens wallichii (Brid.) Fleisch. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Aerobryum Dozy & Molk. (METEORIACEAE) speciosum (Dozy & Molk.) Dozy & Molk. Aloina Kindb. (POTTIACEAE) aloides (K. Schultz) Kindb. ericaefolia (Lindb.) Kindb. = A. rigida var. ambigua obliquifolia (C. Müll.) Broth as (Hedw.) Li Amblystegiella Loeske (AMBLYSTEGIA jungermannioides (Brid.) Giac. germannioides sinensi-subtilis (C. Müll.) B sprucei (Bruch) Loeske — nioid ~ Platydict ya jun- P ydictya jungerman- les yuennanensis Broth. Amblystegium B.S.G. (AMBLYSTEGIACEAE) campyliopsis Dix. juratzkanum Schimp. = А. кр var. juratzkanum kochii B.S.G. = A. trichop noterophilum (Sull. & pda ex x Sull. ) Holz. riparium (Hedw.) B.S.G. serpens (Hedw.) B.S.G. var. serpens ssp. rigescens (Limpr.) Meyl. — A. serpens var. juratzkanum (Schimp.) Rau & Herv. var. dedo cir m & Arn. tenax (Hedw.) C trichopodium (K. Schult Hartm. varium (Hedw. b Amphidium pm O lapponicum (Hedw.) Schi mougeotii (B.S.G.) Schimp. var. mougeotii var. formosicum Card. papillosum Bartr. sublapponicum (C. Müll.) Broth. Anacamptodon Brid. оо, ате Сага fortun inde т ) Broth. subulatus Bro Anacolia Schimp. (BARTRAMIACEAE) laevisphaera (Tayl.) Flow. in Grout t. Besch. = A. н var. fauriei и Chua likiangensis Chen it in Chen & Wan mamillosula Chen in Chen & Wan rupestris Hedw. var. rupestris var. fauriei (Besch. ) Tak wangiana Chen in Chen & Wan Anisothecium Mitt. (DICRANACEAE) palustre (Dicks.) I. Hag. = Dicranella palustris m Broth. = Dicranella rotundata Lindb. = Dicranella varia varium (Hedw.) Mitt. = Dicranella varia Anoectangium Schwaegr. (POTTIACEAE) 1986] p o Mitt. unt Mi crassinervium Mitt euchloron (Schwaegr.) Mitt. kweichowense Bartr. = A. thomsonii latifolium Broth. = A. clarum perm ipsos Broth. = A. stracheyanum var. stra- eyanum о Mitt. var. stracheyanum var. gymnostomoides (Broth. & Yas.) Wijk & Marg. = A. stracheyanum var. stracheyanum subpulvinatum Broth. = A. thomsonii thomsonii Mitt. Po Jaeg. — A. stracheyanum var. strachey- PE innt Schimp. (BRYACEAE alpinum Zang & X.-j. Li in X.-j. Li (editor) шер «ар Lindb. = А. rie ssp. con- filiforme R ) Solms in A ssp. filiforme ssp. concinnatum (Spruce) Amann ssp. juliforme (Solms) D gemmigerum BÉ eis ) Jaeg. m Broth. & H attenuatus (Hedw. ) Hüb. dentatus Gao giraldii C. Müll. grandiretus Broth. integerrimus Mitt. = A. minor Aii integerrimus longifolius (Brid.) C. J. Hart minor (Hedw.) Fürnr. ssp. mino ssp. integerrimus (Mitt.) Pais rotundatus Par. & Broth. rugelii (C. Müll.) Keissl. sinensis C. Müll. = A. minor ssp. minor thraustus C. Müll. viticulosus (Hedw.) Hook. & Tayl. Antitrichia Brid. (LEUCODONTACEAE) curtipendula (Hedw.) Brid. formosana Nog. Aongstroemia B.S.G. (DICRANACEAE) orientalis Mitt. uncinifolia (Broth. ) Broth. = A. orientalis Fleisch. Flei delicata (Fleisch.) LIED ap. Mp. (Toyama) Seki handelii Broth. heteroclada dod JE Fleisch. planula (M Flei nanensis Broth. Adem Brid. a alternifolium (Hedw.) Mitt. [Not included in As 2 by Snider, 1975] REDFEARN & WU-— MOSSES OF CHINA 179 e Schimp. ex. C. Müll. sinense Dur. in Debeaux = (?) A. ohioense [Snider, 71975 Arctoa B.S. G. oo Astomum Hampe (POTTIACEAE) crispum (Hedw.) Hampe = Weissia crispa Atractylocarpus Mitt. (DICRANACEAE) richum P. Beauv. (POLYTRICHACEAE E) angustatum (Brid.) B . var. angustatum [Not in- cluded in As 2 Th кузен 1971] var. [d (C. Müll. Richards & Wal- lace — A. rhystophyllum dU Schimp. ex crispum (James) Sull. & Lesq. [Not included in As yholm, 1971] gracile (C. Müll.) Par. — A. pallidum var. gracile haussknechtii Jur. & Milde — A. undulatum var. gra- cilisetum henryi (Salm.) Bartr. obtusulum (C. Müll.) Jae pallidum Ren. & Card. Rd т. gracile (C. Müll.) Nyh. rhystophyllum (C. Müll.) P. speciosum (Horik.) Wijk Ру Маг spinulosum (Card.) Miz. = А. crispulum undulatum (Hedw.) P. Beauv. var. iine ш var. gracilisetum var. haussknechtii (Jur. & Milde) Frye in Grout = undulatum var. gracilisetum var. minus (Hedw.) Par. = A. е Раг = (?) A. undulatum var. waegr. (AULACOMNIACEAE) androgynum (Hedw.) Schw. heterostichum (Hedw.) B.S. С palustre (Hedw.) Schwaegr. ssp. palustre ssp. imbricatum (B.S.G.) Kindb turgidum (Wahlenb.) Schwaegr. Aulacopilum Wils. арин oe Mitt icum m Broth. ex Card. Baldwiniella spur ex Fleisch. (NECKERACEAE) tibet o in Gao & K.-c. Chang Barbella Fish in Broth. (METEORIACEAE) asperifo a Card. = B. flagellifera bifor g. chrysonema (C. M üll.) N compressiramea (Ren. & Card) Fleisch. in Broth. enervis (Thwaites & vnd ) Fleisch. ex Broth S (Card.) N ез n .) ена ана : = oo pilifera Ы (Broth. ) la gd Broth. Метонлсло sinensis Broth. = Barbella enervis йл aw ormacen anserino-capitata X.-} 180 asperfolia Mitt. = кн asperfolius constricta Mitt. var. constri var. flexicuspis convoluta Hedw. = oe convolutum C. defos 9А = В. лаг ЖО ditrichoides Bro edw. = mia fallax gigantea Fun gracillima (Herz.) Broth indica (Hook.) Spreng. in Steud javanica Dozy & Molk. = Hydrogonium javanicum longicostata X.-j. Li pallido-basis Dix. perobtusa (Broth.) Chen reflexa (Brid.) Brid. = Didymodon rigidicaulis rigidula (Hedw.) Milde var. rigidula = Didymodon rigidulus var. perobtusa Broth. = B. perobtusa rivicola Broth. = Didymodon rivicola rufa (Lor.) Jur. = Didymodon asperifolius schensiana C. Miill. = Didymodon vinealis var. dag ci Chen üll. che (Brid.) Mitt. = Didymodon tophaceus unguiculata Hedw. var. unguiculata var. trichostomifolia (C. Miill.) Chen ealis Brid. = Didymodon vinealis ланку Hedw. (BARTRAMIACEAE) halleriana Hedw. ee ae orvegi indb. = B. halleriana оо Hedw. ex Okam. var. pomiformis viridissima (Brid.) Kindb. = B. subulata Bartramidula B.S.G. (BARTRAMIACEAE) bartramioides (Griffith) Wijk & Marg. cernua Lindb. roylei (Hook. f.) B.S.G. — B. cernua Bellibarbula Chen (POTTIACEAE) urzi e mittenii (Broth.) Card. Brachymeniopsis Broth. (FUNARIACEAE) gy toma Broth. Brachymenium Schwaegr. о = В. ochianum oth. exile (Dozy & Molk.) Bosch & Lac. immarginatum Gao & K.-c. Chang Jilinense T. Ms Shaw, Lou & Gao muricola B nepalense ien in Schwaegr. var. nepalense ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 var. cum (Mitt.) Ochi Gan ptychothecium (Besch. ) Ochi sinense Card. & Thér. Brachythecium B.S.G. и albicans (Hedw.) В шо: С. Müll. brother buchananii (Hook. ) eia var. buchananii nae ides С. Mill. glaciale B glareosum (Spruce) B.S.G. glauco-viride C. Müll. m C. Müll. sch. h. = i novae-angliae oedistegum (C. Müll.) Jae ] C Müll. piligerum С ll. B.S.G. var. pun var. е Broth & Par.) T rd. populeum (Hedw.) B. s. G. var. populeum var. japonicum Dix. & Thér. var. quelpaertense Г ) Так. var. yamamotoi (Sak.) Tak. соо еш ) Јаер. рувтаеи cuepactense a = B. populeum var. quelpaer- x xum aes rhynchostegielloides Card. — rhynchostegielloides B. rhynchostegielloides thraustum C. Müll. uncinifolium Broth. & Par. = Cratoneurella uncini- olia velutinum (Hedw.) B.S.G. viridefactum C. ud wichurae (Broth.) P. yamamotoi Sak. = B. populeum var. yamamotoi yuennanense Herz. 1986] Braunia B.S.G. d Lim k Hypnum (HYPNACEAE) ta (Mol.) Loeske = Hypnum lindbergii a (B.S.G.) Schimp. ИУ cna epee Flei dicran üll.) seschwanica с = B. po subdeflex. ei yunnanensis aoe Brothera C. Müll. TENE leana (Sull.) C Brotherella Loeske ex Fleisch. (SEMATOPHYLLACEAE) erythrocaulis (Mitt.) Fleisch. & Molk.) Fleisch. elii Broth. henonii (Duby) Fl himalayana Chen pen nud.) = Pylaisiadelpha hi- integrifolia "iom lorentziana (Lor) Loeske in Fleisch. nictans (Mitt.) Broth. Broth. qur E. sinensis e e om. Ani ur. о higoensis Tak. = В. novae-angliae nitida Sak. = Brachythecium че аш уаг. nn” novae-angliae (Sull. & Lesq.) G stokesii (Turn.) Robins. = Kindbergia praelongia sublaevifolia Broth. & Par. in Card. Bryochenea Gao & K.-c. Chang (THUIDIACEAE) ciliata Gao & K.-c. Chang sachalinensis (Lindb.) Gao & K.-c. Chang Bryoerythrophyllum Chen (POTTIACEAE) alpigenum (Vent.) Chen atrorubens (Besch.) Chen = Didymodon atro-rubens brachystegium (Besch.) K. Saito dentatum (Wils.) Chen gymnostomum ( Broth.) Chen hostile (Herz.) Chen obtusissimum (Broth.) Chen = B. brachystegium ergemmascens (Broth.) Chen recurvirostre (Hedw.) Chen rubru h erz.) Chen Watan. & Iwats. зерур Watan. & Iwats. . & Dix. (HYPNACEAE) Fleisch Bryoxiphium Mitt. (BRYOXIPHIACEAE) REDFEARN & WU-—MOSSES OF CHINA 181 bd d Par E S ssp. japonicum (Berggr.) A. Löve Bryum B 9 alpicola Brot alpinum Huds. ex With. var. alpinum var. teretiusculum (Hook.) Podp ambiguum Duby in Moritzi = B. ГИТЕ andrei Card. & Р. de la Varde angustirete Kindb. archangelicum B.S.G. arcticum (R. Br.) B.S.G argenteum Hedw. var. argenteu var. lanatum (P. Beauv.) B atrovirens Vill. ex. Brid. badhwarii Ochi in | Nog. barbuloides Broth. bicolor Dic billardieri rm egr. blandum Hook. f. & Wils. ssp. handelii (Broth.) Ochi calophyllum R. Br. ssp. calophyllum capillare Hedw. cellulare Hook. in Schwae cernuum (Hedw.) B.S.G. = B. uliginosum chrysobasilare Broth. chrysobasillioides Broth. chungii Ba cirrhatum Hoppe & Hornsch. — B. лдын compressidens С. E = B. cellul coronatum Schwae cyclophyllum Gchwaegr) B.S.G. = ваш tortifolium fortunatii Ei B. truncoru funkii Sch gigan pum a Amott = Rhodobryum gi- eae C. Müll. globosum Lindb. = В. wrightii gossypinum dicen & Zang in Gao & K.-c. Chang handelii Broth. = B. blandum ssp. а japonense Gesch. ) Broth. = В. cellul knowltonii leptocaulon Ca leucophylloides Broth lonchocaulon longisetum Bland. ex Schwae C. Müll. var. neelgheriense ed e Itzig. in C. Müll. var. neodamense r. ovatum Lindb. & Arn. Mir Hook. = B. plumosum pallescens Schleich. ex Schwaegr. var. pallescens var. subrotundum (Brid.) B.S.G. pseudo-alpinum Ren. & Card. = B. paradoxum pulchroalare Broth. — B. ко чн handelii purpurascens (R. Br.) B.S.G. 182 ramosum drm ) Mitt. recurvulum рц с & Р. de la Varde rubigineum C. Müll. faits Brid. salakense Card. sauteri B.S.G schleicheri Schwaegr. — B. turbinatum setschwanicum Broth. 5 elh. var. vestitum Broth. = В. pseu- otri riqu uetrum ssp. pseudotriquetrum Brot h. ch var. wichurae wrightii Sull. & Les yuennanense Broth. = В. capillare Buxbaumia Hedw. (BUXBAUMIACEAE) aphylla Hedw indusiata B d. javanica ull puncta hen & Lee ymmetrica Chen & Lee о т т ( (S I ACEAE) m (Grev.) C Calicomell (C. Mall) Mit iene apillata (Mont.) Calliergon paye ) Kindb. (AMBLYSTEGIACEAE) ordifolium (Hedw.) Kindb. Poa chimp.) Kindb = (T. Jens.) Kindb. = Scorpidium turges- Callergonll Loeske nee uspidata (H edw.) Loe a Sw. in Web. ТОРЕНЕ cristatum Н fasciculatum Dozy & Mo fordii Besch. = C. thwaitesii ssp. fordii hampei Богу & Molk. = С. erosum jap ponicum Besch. = Syrrhopodon japonic . Br. ex C. Müll. strictifolium (Mitt. E "i Roth анна е (Sull.) М neru . Müll. яе: Besch. ssp. fordii (Besch.) Fleisc tube jid sum (Thér. & Dix.) Broth. = C. БУРР ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Calymperopsis (C. = ) Fleisch. (CALYMPERACEAE) semiliber (Mitt. ) Fleisch. tjibodensis (Fleisch.) Fleisch. yunfuensis Wu Calyptothecium Mitt. (PTEROBRYACEAE) acostatum Lou caudatum Bartr. cuspidatum (Okam.) я var. cuspidatum hookeri (Mitt.) Broth. japonicum Thér. = C. urvilleanum nitidum (Ren. & Card.) Fleisch. innatum No stricticaulon Lo ри (Broth. & Par.) Broth tumidum (Dicks.) dr = "a urvilleanum urvilleanum (C. Müll.) B wightii (Mitt.) Fleisch. Calyptrochaeta Desv. (HOOKERIACEAE) Japonica (Card. & Thér.) Iwats. & Nog. parviretis (Fleisch. a Iwats., Tan & Touw in Touw spinosa (Nog.) Nin e Reich. LewmoniitActAn arbuscula (J. Sm.) R sinensis Broth. var. ee var. Vine dide (Broth.) Redf. & Wu, comb. n о sinensis var. flagellifera Broth., ш. Sin. . 1929. Camptothecium B.S. e (BRACHYTHECIAC EAE) Campyliadelphus (Kindb.) R. Chopra = Campylium (AMBLYSTEGIACEAE chrysophyllus (Brid.) Kanda = Campylium chryso- рпунит stellatus (Hedw.) Kanda = Campylium stellatum Campylium (Sull.) Mitt. (AMBLYSTEGIACEAE) amblystegioides Broth. chrysophyllum (Brid.) J. ues var. chrysophyllum var. zemliae (C. de н courtoisii Par. & Bro enerve Herz. & Nog. hispidulum (Brid.) Mitt. var. hispidulum var. ee Jada Lindb. protensum а (Card.) Herz. & Nog. som Дан. D J. Lange = C. hispidulum var. om к Hedw. ) C.J Müll. var. uninervium var. Minus И Mü Campylodontium Dozy & Molk. = Mesonodon NTODONTA flavescens (Hook) Bos & Lac. = Mesonodon fla- vescen MG ian (Schimp.) Fleisch. STEGIACEAE halleri (Hedw.) Fl Campylopodium (C. eu ) voa ern euphorocladum (C. Müll.) B Campylopus Brid. о (AMBLY- 1986] alpigena Broth. var. alpigena var. lamellatus Broth. atro-virens De N aureus Bosch & Lac. caud . Müll.) ces in Dozy & Molk. flexuosus (Hedw. )B aeu (Brid.) B.S. chm var. fragilis pyriformis (K. Schultz) Angstr. fusconinidis с ) Dix. & Thér. ex Hong & Ando ilen gracilis (ми ae ha ау Broth. ad ` handelii г setschwanicus Broth. brisas Thé onicus B T ens (Mitt. ) Jaeg. laxitextus Lac. po Gao in Gao, G.-c. Zhang & Cao pinfaensis T po [3 Schultz) Brid. = iform richardii Brid. scabridorsus Dix schimperi Milde = C. subulatus var. schimperi C. fragilis var. pyr- us Sch imp. in Rabenh. var. subulatus var. schimperi (Milde) Husn. is Sak Catharinea Web. & Mohr = Atrichum (POLYTRICH- henryi Salm. = Atrichum henry yuennanensis Broth. = eee undulatum var. Ж etum minor Broth. = Atrichum e AE) Ceratodon Brid. (DITRICHACEAE E) purpureus (Hedw.) Brid. var. purpureus var. formosicus Card var. rotundifolius Berggr. stenocarpus Bruch 8: Schimp. ex C. Müll. Chaetomitriopsis Fleisch. т glaucocarpa (Schwaegr.) Fleisch. Chaetomitrium Dozy & Мок (HOOKERIACEAE) acanthocarpum Bosch & Lac. orthorrhynchum (Dozy & Molk.) Bosch & Lac. Chionostomum C. Müll. (SEMATOPHYLLACEAE) rostratum (Griffith) C. Müll. var. rostratum var. microcarpum Broth. Chrysocladium Fleisch. (METEORIACEAE) ammeum (Mitt.) Fleisch. ssp. flammeum ssp. ochraceum (Nog.) N kiusiuense (Broth. & Par.) Pius — C. retrorsum phaeum (Mitt.) Fleisch. retrorsum (Mitt.) Fleisch. var. retrorsum ar. pinnatum Card. ex Nog. = C. retrorsum var. taiwanense Nog. = C. retrorsum robustu ei nue Sw. in n Schr ere arcticum Schim stygium Sw. in а REDFEARN & WU—MOSSES OF CHINA 183 Cinclidotus P. Beauv. E ы fontinaloides (Hedw.) Р Cirriphyllum Grout pl fia NN cirrhosum (Schwaegr. ex Schultes) Grout crassinervium (Tayl.) Loeske & Fleisch. piliferum (Hedw.) Grout nerve Dix. eS (Lesq. & James) Ren. & Card. (THUIDI- mune (Broth.) Broth. assurgens (Sull. & Lesq.) Card. crispifolium (Hook.) Ren. & Card. fulvellum Herz. gracillimum (Card. & Thér.) Nog. integrum Chen ex Wu, Lou & M.-z. Wang (nom. nud. rv Harv.) Fleisch. = C. prionophyllum papillicaule (Broth.) Broth. = C. pellucinerve pellucinerve (Mitt.) Bes piliferum Broth. = C. pellucinerve sinicum Broth. & Par. = C. aciculum Clastobryella Fleisch. aca cuculligera (Lac.) Fleisc nins ган бүгө (Тоуата) Seki = Aptychel- merato-propaguli kusatsuensis (Besch.) Iwats. = Brotherella yokoha- mae tenerrima Broth. s es Lm & Yas.) Broth. = Brotherella yo- d e Dozy & Molk. (SEMATOPHYLLACEAE) excavatum Broth indicum (Dozy & Molk.) Dozy & Molk. ке oi Broth. = Oedicladium rufescens e Dix Cleistostoma Brid. = p a ee aga be a rid. = hh rrh on ambiguus dede ps. hr (Cr america peros sp Mun n (Lindb. ) J. Perss. — onic dendroides (Hed ) Web. & Mohr aponicum siii rella Robin ns. O cinifolia (Broth. & Par.) R a (Sull.) Spruce муть Е) соттша 5 г. commutatum var. falcatum (Brid.) Mone = c commutatum i Mónk. curvicaule (Jur.) G. Roth = C. filicinum var. curvi- в. nee G. Roth var. falcatum = C. com- mutatu filicinum (Hedw.) Ee var. кош var. curvicaule (Jur.) М var. fallax (Brid.) G. кай = C. filicinum formosanum Bro Crossidium Jur. (POTTIACEA heteromalla (Hedw.) Mohr in Web. е т sinensi Genin oe ) Mitt. (HYPNACEAE) 184 andoi Mishim capillifolium (Mitt.) Broth. plumulosum Bartr. (ho eg.) = C. pinnatum robusticaule Broth. & Par. = c capillum im d.) Broth. var. scaberrimum = Ec- opothecium zollingeri serratifolium (Card.) o ; Cupressina C. Müll. = Нурпит (HvPNACEAE) alaris C. Müll. EH pric ue var. sinensi- leptothalla C. Müll. = Eurohypnum leptothallum leucodontea üll. = Eurohypnum leptothallum minuta . Müll. = Hypnum shensianum sinensi- mollusca C. Mill. = Hypnum plumaeforme . sinensi-molluscum о Müll. = Eurohypnum leptothallum en Müll. = Gollania turgens ulophyila С Müll. = Нурпит subimponens ssp. ulo- байоо (Broth.) Fleisch. (DALTONIACEAE) ensifolia Horik. = C. hookeriana intermedia (Mitt. kyusyuensis Horik. " Nog rigidula Chen (nom. nudis ookeria serrulata Chen e nud.) = v у 1.) Fleisch. taiw waniana bl um ( laete-virens (Hook. & Tayl.) Mitt. nodontium Schimp. (DICRANACEAE) fallax Limpr. gracilescens duis & Mohr) Schimp. Cyrtomnium Holmen (MNIACEAE) hymenophylloides (Hüb.) T. Kop Daltonia Hook. & Tayl. (DARE angustifolia Dozy & Molk. var. iss PE var. strictifolia ad Fleisch. aristifolia Ren. & Ca splachnoides (J. S BUD & Tay Dendrocyathophorum Dix. os intermedium (Mitt.) Herz. Desmatodon Brid. (POTTIACEAE) capillaris Chen cernuus (Hüb.) B.S.G. gemm en latifolius (Hedw.) Brid. Му latifolius эү ри var. laureri ar. а. (Broth.) Chen leucostoma (R. Br.) Be Joa (Schwaegr.) Schimp. = Tortula obtusi- “з ibn j. Li suberectus s (Hook) Timo = D. leucostoma systylius thomsonii с "Müll. ) Jaeg. nanensis Brot j Ren. & Card. R bla rv.) Ren. & C thuidioides | Ren. & Card. = : D. blandus ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Dichodontium Schimp. oo pellucidum (Hedw.) Sc ires (C. Müll.) Schimp. (DICRANACEAE) ustro-sinensis Herz. > рані eile (Hedw.) Sc Мар, coarctata (C. Miill.) Bosch & Lac. grevilleana (Brid.) Schim Tl (Hedw.) Schimp. leptoneura Dix. аана. (C. Müll.) Par micro-divaricata (C. Müll. ) Par. obscura Sull. & Le palustris (Dicks.) Crundw. ex E. Warb. rotundata (Broth.) Tak. squarrosa (Starke) SEE = Dicranella palustris subulata (Hedw.) Schim varia (Hedw.) Schimp. Dicranodontium B.S.G. (DICRANACEAE) m (Mitt.) Broth. nuatum (Mitt.) Wils. ex Jae blindioides (Besch.) Broth. var. Elindioides robustum Brot a (Mitt.) Par denudatum (Brid.) Britt. ex Williams didictyon era ) Jaeg filifolium nitidum ы & Molk.) Fleisch. papillifolium Gao in re G.-c. Zhang & Cao porodictyon Card. & T sinense (C. Müll.) B subintegrifolium Broth. subporodictyon Broth. Ve lois h. & Herz. m (Harv.) Jaeg. Dicranoloma (Ren.) E SEA cylindrothecium (Mitt.) Sak. = D. ки акы m dicarpum (Nees) dis ormosanum Bro mile Dicranoweisia Lindb. ex Milde (Dicrawacens Dicranum Hedw. (DICRANACEAE) assamicum Dix bonjeanii De Not. in Lisa ssp. bonjeanii ssp. angustum (Lindb.) Podp. caesium Mitt. аб нЕ сопапепит Сао і іп Gao & К.-с. Chang. erispifolium C. M üll. cri imp. ex Besch. = D. hamulosum cylindricum 1 Broth. (hom. illeg.) elav о Сао & Аиг rummondii С. МИП. elongatum Schleich. ех Schwaegr. var. elongatum 2 Sp pha x . Jens. = D. groelandicum flagell. w. lila Lindb. 1986] fulvum Hook fuscescens Turn. ssp. fuscescens ssp. congestum ( (Brid. ) Kindb. groenlandicum gymnostomoides Broth. var. gymnostomoides var. microcarpum Broth. е Mitt. linzianum Gao in Gao & K.-c. Chang pan Mitt majus Turn. porc Gao & Aur mayrii muehlenbeckii B.S.G. var. muehlenbeckii ar. neglectum (De Not.) Pfeff. Bebes Besch. orthophyllum Broth. papillidens Broth. perfalcatum Broth. var. perfalcatum (hom. illeg.) var. кзы e Broth. m = D. hamulosum scoparium Hedw. var. scoparium var. integrifolium Lindb. — D. bonjeanii ssp. au- gustum scopellifolium C. Müll. scottianum bo kwangtungens subeylindrorhecium Broth. n Gao & K.-c. Chang sericifoliu etschwanicum Broth. spadiceum Zett. = D. muehlenbeckii var. neglectum spurium Hedw. tauricum Sap. = И strictum thelinotum C. МШ truncicola Broth. undulatum Schrad. e atro-rubens (Besch.) Broth. up apes И wallichii constrictus (Mitt.) K. Saito var. constrictus — D. vi- n var. flexicuspis (Chen) K. Saito — D. vinealis cordatus Jur. = Barbula gigantea E аву (C. Müll.) K. Saito fallax (Hedw.) Zander Ee (Funck) Ju icmadophyllus (Schimp. g Rc Müll.) K. Saito japonicus (Broth.) K. Sai rivicola (Brid.) Zander in T. ar. Gao, Lou & Jar- vinen rubellus B.S.G. — оз ode recurvirostre rufidulus (C. Müll.) B submicrostomus Dix. = TN recur- virostre REDFEARN & WU-—MOSSES OF CHINA 185 tectorum (C. Müll. р E Saito tophaceus (Brid.) L vinealis (Brid. )Za yunnanensis она boot = Bryoerythrophyllum yunnanens Diphyscium Mohr (DIPHYSCIACEAE) ы (Hedw. Mohr form m Hori fubifolium уа var. н var. lev HS in -k. Wang & S.-h. Li Wang & evisetum Gao .S.G. sporu o in X-j. = (editor) Distichophyllum Posi & Mol ааа brevirostratum Thér. m Card. m (Dozy & Molk.) Dozy & Molk. poen (Mitt. fs ie АЫ (C. Müll.) Bosch & Lac. Besch. si = а collenchymatosum stillicidiorum Ditrichopsis Brain. yc c E clausa Brot gymnostoma Bro Ditrichum "d (DITRICHACEAE) aureum Bart brevidens Nog pe ас. (C. Müll.) Par. 3 Msc) Fleisch. diva itt. lb ad ) Hampe = D. crispatissimum flexifolium Hampe = D. difficile heteromallum (Hedw.) Britt. Bi poen... (Hedw.) Hampe = D. heteromallum crocarpum Broth. pallidum (Hedw.) Hampe var. pallidum var. sinense Thér. pusillum (Hedw.) HE var. pem var. tortile (Schrad.) I. setschwanicum я tortile (Schr. rockm. = usillum var. tortile Po MN (Lindb.) Broth. m d. E ymbifolia ас b.) Broth. sE: cymbifolia ntegerrima О Dolichomitriopsis Okam. ъл ipe. (Mitt ) Nog iquel (LEUCODONTACEAE) joy da Lac. in Miquel Drepanocladus (C. Müll.) G. Roth vins dilema, Mx aduncus (Hedw.) Warnst. var. aduncus f. adun Mö . pseudofluitans (Sanio) Mönk. 186 aomoriensis (Par.) Broth. = Sasaokea aomoriensis cuspidarioides n Müll.) Broth. ex Par ulatus (B.S.G.) Warnst. f. exannulatus . angustissimus jode fluitans (He Warn lycopodioides ( (Brid. ) Wars! var. lycopodioides ab о arnst. f. He .) M f. ech ile Sm mulosus бат ) Monk. о (Lindb.) Warnst. m ia Hook. in Drumm. (ORTHOTRICHACEAE) cavaleriei Ee obtusifolia C. Müll. prorepens (Hedw. ) Britt. var. prorepens [Not includ- 2] thomsonii Mitt Duthiella С. ми. in Broth. (TRACHYPODACEAE) а tt.) Zant flaccid а (Card) Broth. on flaccida var. media (Nog var. rigida (roth ) Zant. formosana Nog. media Nog. = D. flaccida var. media pellucens Card. & Thér. = D. flaccida var. flaccida perpapillata Broth. = D. flaccida var. flaccida rigida Broth. = D. flaccida var. rigida robusta Nog. speciosissima Broth. ex Card. wallichii (Mitt.) C. Müll. in Broth. Ectropotheciella Fleisch. л distichophylla (Hampe) Е1 Ectropothecium Mitt. (нут) buitenzorgii (Bél.) Mit circinatum Bartr. — unus plumaeforme dealbatum (Reinw. & Hornsch. : "d intorquatum (Dozy & Molk.) Ja isocladum Dix. = и gom var. minus kweichowense Bart о (Card.) Sa Sak mentorum (Duby) Jaeg. pias Jaeg. planulum Card. = E. zollingeri оа Сага. = E. zollingeri var. zollingeri yasudae i sss (C. Müll. ) Jaeg. var. zollingeri formosanum Broth. var. formosanum А roth. phillippinense Broth. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 esi Schreb. ex Hedw. (ENCALYPTACEAE) alpina J. Sm нета С. Mill. ciliata Hedw erthrodonta e Müll. giraldii C. Müll. rhabdocarpa Schwaegr. var. ы уаг. d (C. Müll.) Hus C. Müll. — E. o edt var. spathulata т P "Müll. (PTEROBRYACEAE) ietrichiae C. Müll. eberhardtii Broth. & Par. in Par. elegans (Dozy & Molk.) Fleisch. in Broth. fauriei (Broth. & Par.) Broth. = E. elegans Entodon C. Müll. слон ) R.-1. aeruginosus C. aL amblyophyllus C. Miill. = E. caligin attenuatus Mitt. = E. sullivantii var. p indb. tt. ee say Card. — E. compressus hloropus ard. Wachs ишы (edv ) C. Müll. cochleatus Broth. = E. luridus compianatulus C. ope in Fleisch. = E. pulchellus essus x Mizush. var. compressus .-]. Hu posse ше (De Not.) Par. rvatiram delavayi 2s = E. macropodus divergens Bro dolichocuculats drummondii Т ) e = к macropodus eurhyn ноа : Herz. & — E. obtusatus excavatus = me macro podus flavescens (Hook giraldii C griffithii n js = m flavescens kungshanensis В.-1. latifolius jee E. ostatus Buck = E запі var. versicolor b (C. Müll.) Jae microthecius Broth. — E. compressus var. compres- SUS morrisonensis Nog. myurus (Hook.) Hampe var. myurus [Not included in China by Hu, 1983 var. hokinensis Besch. [Not included in China by 3 . Müll. var. nanocarpus = E. com- pressus var. comp var. zikaiwiensis (Par. ) Gao = E. compressus var. is nepalensis ВЫ in Nog. obtusatus ч атигае Broth ex Card. = E. lu Asien rthoc Lindb. = E. con И E ) Nog. ad TE ¿A 1986] plicatus C. Müll. prorepens (Mitt.) Jaeg. u ocarpus C. Müll. — E. caliginosus pu ата (Griffith) Jaeg. pylaisio SEV a u & Y. и mulosu = E. flav rubicundus ( (Miu Ja aeg. = E ec — E. perichaetialis scariosus o 8 Card. schensianus C. Müll. seductrix (Hedw.) C. Müll. [Not included in As 2 by Buck, 1980] serpentinus C. Müll. — E и smaragdinus Par. & Bro crush Thér. = E. sien epens mulosus Broth. — E. caliginosus sullivantii (C. Müll.) Lindb. var. sullivantii var. versicolor (Besch.) gy in rn & Marg. nsis C.-k. Wang & S.-h. L d. yunnanensis Thér. [Not included in China by Hu, 1983 zikaiwiensis Par. = E. compressus var. zikaiwiensis buseanus Dozy & Molk. = Funaria busea digi dem (Mont.) Mitt. — сга о mitrioi. Ephemerum Fue (EPHEMERACEAE) apiculatum Chen cohaerens (Hedw.) Hampe serratum (Hedw.) Ham Epipterygium Lindb. (BRYACEAE) tozeri (Grev.) Lindb. wrightii (Sull.) Lindb Eriopus (Brid.) Brid. (HOOKERIACEAE) cristatus (Hedw.) B japonicus Card. & ip — Calyptrochaeta japonica pe Fleisch. = Calyptrochaeta parviretus spinosus N Erpodium (Brid. ) Brid. er biseriatum (Aust.) A Erythrodontium Hampe ca julaceum (Schwaegr.) Par. fe tine, (C. Miill.) Nog. hallum glial dll Hampe p. [as Gollania tereticaulis Broth. (Ando et al., 1957)] Eucladium verticillatum (Brid. )B Nog. о sinicum н (Ми Мо ar. flagelliferum (Sak. )N BR ACHYTHECIACEAE) | angustirete (Broth.) T. Kop. = E. stri arbuscula Broth. var. acuminatum Tak. = = indies gia arbuscula asperisetum (C. Müll.) Bartr. — Eurohypnum lep- laxirete Broth. in Card. = Oxyrrhynchium laxirete muelleri (Ja Ө! al polystichum pu yo llu m (Hedw) Je riparioides (Hedw.) л - E a REDFEARN & WU—MOSSES OF CHINA savatieri Schimp. ex Besch. serricuspis stokesii (Turn.) B. SG. ex Schimp. = Kindbergia striatum sig ) Sch Eurohypnum Ando as E) leptothallum (C. Müll.) Ando f. и blumii (C. Müll.) Fleisch. sullivantii (Dozy & Molk.) Fleisch. Fabronia Raddi be pilae ) anacamptodens 26 п Сао, С-с. Zhang & Сао ciliaris (Brid.) Bri enervis Broth mariei Card. in Broth. (nom. nud.) matsumurae Besch. var. matsumurae var. yunnanensis Thér. microspora Gao in Gao, G.-c. Zhang & Cao papillidens Gao in Gao, G.-c. Zhang & Cao patentissima C. Müll. perpilosa Broth. — Rhizofabronia perpilosa cunda Mont. Fauriella Besch. a aa robustiuscula Broth. tenerrima Bro tenuis (Mitt.) Lm in Broth. arevensis adelphinus Besch. d Hedw. an us Mont. areolatus C ви beckettii posoriensis Fleisch st . bryoides var. bryoides p. bryoides var. UDIN var. esqui uirolii (Thér) Iw bu & T var. hedwigii Limpr. — ¡gres var. bryoides var. lateralis (Broth.) Iwats. & T. Suz. var. ramosissimus (Thér.) Iwats. & T. Suz. (hom. ille var. schmidii (C. Müll. А E ems & S. Kumar ssp. Sipe (Sw.) Kind capitulatus ceylonensis Dozy & Molk. crassipes Wils. ex B.S.G. diversiretus Broth. = F. grandifrons var. grandifrons esquirolii Thér. = F. bryoides var. esquirolii formosanus Nog. garberi Lesq. & James = F. microcladus oe Dozy & Molk. var. ani ME (Besch.) wats. Nd Herz. & P. de la Varde — F. protone- maecola ii Fleisch. giraldii Broth. 188 grandifrons Brid. var. grandifrons var. planicaulis (Besch.) а = F. grandifrons guangdongensis Iwats. & Z.-h. Li mnogynus hetero-limbatus vv = Ee tosaensis hollianus Dozy & M hyalianus Hook. & wi incognitus incrassatus Sull. & Lesq. — F. zippelianus intromarginatulus Bartr. involutus Wils. ex Mitt. irrigatus Broth. = F. я е PE & Mo Sull. & Le laxus sq mangar Mont microcladus Thwaites & M micro- -serratus Sak. = F. obscurirete mitten. vede = F. lax nobilis RUE Broth. & Par. in Broth. osmundoides Hedw. papillosus Lac. perdecurrens Besch. . Müll. plagiochiloides Besch. ryukyuensis Broth. = F. bryoides var. ramosissimus saxatilis Tuz. & Nog. = F. strictulus schmidii C. Müll. = F. bryoides var. schmidii schwabei splachnobryoides Broth. in Schumann. & Laut. f. subbr. ut (Thér. & P. de la Varde) Herz strictulus subinteger Broth. = F. plagiochiloides ie ssilis subxiphioides Broth. = E. EN si s Griffith = F. taxifo taelingensis Gao = (?) F. лн S taxifolius Hedw tosaensis Brot wichurae Bro xiphioides Fleisch. - y К шшш Мор. = yamamotoi Sak. = F. pola var. geen yunnanensis Besch. = F. grandifrons var. grandi- TONS zippelianus Dozy & Molk. in Zoll. zollingeri Mont Fleischerobryum Loeske (BARTRAMIACEAE) longicolle (Hampe) Loeske [May be F. macrophyl- m, Lin, 1984] macrophyllum Broth. роон Fleisch. (METEORIACEAE) oth. = F. setschwanica aurea vie ) Broth. ssp. aurea nipponica (Nog.) Nog dii ыы (C. Müll.) Broth. = F. sparsa var. sparsa chrysonema (C. Müll.) Broth. = Barbella chrysone- bic E lb idi obsc din ma floribunda (Dozy & Molk.) Fleisch. hookeri Broth. — F. sparsa horridula Broth. var. horridula = Chrysocladium flammeum ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 var. rufescens Broth. = (?) Chrysocladium flam- meum ssp. flammeum intermedia Thér. = F. thuidioides Nog. = F. aurea ssp. nipponica isch. sparsa (Mitt.) Broth. var. E var. pilifera ге" thuidioides Flei torquata C.-k. Wang & S.-h. Lin walkeri (Ren. & Card.) Broth. Fontinalis Hedw. (FONTINALACEAE) antipyretica Hedw. var. antipyretica ssp. gothica (Card. & Arn.) Podp. = F. antipyretica var. gigantea (Sull.) Sull. var. m d Schim ar. livonica (G. th & Bock) № in Н. Moll. = F. anii var. gigan d. & Ат. = F.a gothica la var. gracilis hypnoides Hartm. var. hypnoides var. es (Card.) Gao squamosa w. Forsstroemia Lindb. (CRYPHAEACEAE) cor cryphaeoides Card. cryphaeopsis Dix filiformis M.-x. Zhang in Wu, Lou & M.-z. Wang ud. a neckeroi r recurvimarginata Nog. var. di ain Su . filiformis Nog. in Chen (nom. nud.) dolia ат іп h.) Раг. var. sinensis FEER EZ ri Funaria Hed CEA attenuata (Dicks.) db. eana (Dozy & Molk.) Broth. calcarea Wahlenb. connivens C. Müll. tata Crome - F. ca gracilis (Hook. f. & Wils.) Br ybi i Hedw. var. À ygrometrica microstoma Bruch ex Schimp. muehlenbergii Hedw. f. ex Lam. & DC. pallescens (Jur.) Lindb. in Broth. pilifera (Mitt. ) Broth sinensis Dix. in C. Yang (nom. nud.) submicrostoma C. Müll. = F. microstoma Gammiella Broth. и oe (Griffith) B ш arc и С. МШ. Dan AE mosa (Dozy & Molk.) Wijk а яр = G. flexuosa е Endl. (PTEROBRYACEAE) scula Card. = Pterobryopsis crassiuscula 1986] plicata (Brid.) Bosch & Lac. taiwanensis Nog. ji baies Schimp. (POTTIACEAE) gigantea (Funck) Boul. = Barbula gigantea Giraldie'la x Mill. (HYPNACEAE) Oirgensohnia (Lindb.) Kindb. = Pleuroziopsis (PLEU- AE ruthenica (Weinm.) Kindb. = Pleuroziopsis ruthe- Glossadephus Fleisch. A h. & Yas alaris ano oot т attenuatus Broth P (Card.) Broth. — Ectropothecium zol- E жи ы» a & Par.) Fleisch. = Ectropothecium zolli j киеп (C. Müll.) Fleisch = Ectropothecium zol- ingeri Glyphomitrium Brid. (PTYCHOMITRIACEAE) inatum calycinum (Mitt rd daviesii (With. formosanum Iwats. var. formosanum var. serratum I Sau in Iwats. & Sharp = G. oo issimum minutissimum (Okam rot gave sh = ) rae var. warburgii ense Thér. & Hen Glyphot uM пое (PTYCHOMNIACEAE) sciuroides (Hook.) Ham Gollania ae (RHYTIDIACEAE) arisanensis Sa clarescens (Mitt |.) Bro coclearifolia ne ex vo Taxiphyllum alternans densepinnata densifolia Dix. — Tax pu Broth. horrida oe (Broth. & Par.) Broth. neckerella (C. Müll.) Broth. var. neckerella var. coreensis (Card.) Broth. Philippinensis (Broth.) Nog robusta Bro ruginosa (Mitt. ) Broth. sinensis Broth. & Par. subtereticaulis Broth. & Yas tereticaulis Broth. = Erythrodontium sp. [Ando et al., 1957 turgens (C. Müll.) Ando varians (Mitt.) Broth Grimmia ше (GRIMMIACEAE) iphyllum aomoriense affinis Hornsch. alpestris (Web. & Mohr) Schleich. ex Nees var. al- pestris = G. donniana var. donniana var. iiio (Card. & Thér. E - Jones in Grout = G. donniana var. holzi alpicola Sw. ex Hedw. = Schistidium agassizii REDFEARN & WU -— MOSSES OF CHINA 189 anodon G apiculata Hornsch. carpa Hedw. = Schistidium apocarpum var. gracile Röhl. = Schistidium apocarpum aspera C. Mü atrata Miel. ex Hornsch chenii S.- commutata Hiib. = G. affinis ecalvata . decipiens каш Lindb. in Hartm. dimo = a C. pae donnia . Sm ar. do onnia var. holzingeri (Car, & Ther) Wijk & Marg. t De N . Ka urm filicaulis C. Müll. - = Mun apocarpum handelii ш hartmani himp. himalayana Mitt. = G. chenii longicapusula Gao & Cao in Gao, G.-c. Zhang & Cao mammosa Gao & Cao in Gao, G.-c. Zhang & Cao А Broth. montana B.S.G. = G. donniana var. donniana obtusifolia Gao & ae. in Gao, G. -c. Zhang & Cao ovalis Шык A А = С. affinis pilifera Р. гра С. Müll. = Schistidium apocarpum subconferta Broth. = Schistidium apocarpum е C. Müll. te tergestina Tomm . ex B.S.G. torqu rn unicolor Hook. in Gre SON Fleisch. DONE lon nginerv is vernicosa Hook; Fleisch. = G. longinervis Gymnostomum Nees & Hornsch. (POTTIACEAE) aeruginosum J. Sm. angustifolium K. Saito = Tuerckheimia angustifolia recurvirostre Hedw. = NEM recurvirostre subrigidulum (Broth.) C Gyroweisia Schimp. bs brevicaulis (C. Müll.) Broth. tenuis (Hedw.) Schimp. yuennanensis Broth Habrodon Schimp. (FABRONIACEAE) leucotrichus (Mitt.) H. Perss. = Iwatsukiella leuco- tricha perpusillus (De Not.) Lin piliferus Card. = Bini leucotricha 190 Handeliobryum Broth. (NECKERACEAE) setschwanicum Broth. = Sciaromiopsis brevifolia Haplocladium (C. Müll.) C. Müll. (THUIDIACEAE) gap ens (Hampe & C. Müll.) Broth. = Bryo- ium angustifolium discolor (Par & Broth.) Broth. i . & Par.) Watan. = H. strictulum rvum th. рена (Broth. & Раг.) Bro кы (Hedw.) Broth. pl eee = ocladium microphy SSp. capillum (Mitt.) Reim. Mu bxc crophyllum mido: Thér. paraphylliferum Broth. = Bryohaplocladium micro- strictulum (Card.) R Haplodontium Ha ampe = — Mielichhoferia (BRYACEAE) Shaw & Crum ] p. [Koponen et ^ 1983] Hi aplohymenium on & Molk. (THUIDIACEAE) flagelliform ormosanum sid og. longinerve (Broth.) B oth. microphyllum вош gs — Broth. = H. triste pseudo-triste (C. Müll.) B sieboldii (Dozy & Molk.) ed & Molk. v. (HEDWIGIACEAE) . ciliat Sio > ead Ehrh. ex P. Beauv. var. ciliata f. cil- f. be is S.G.) G. Jones in Grout var. viridis B З Hedwigidium В. s. A (HEDWIGIACEAE) integrifolium (P. Beauv.) Dix. in C. Jens. elodium Warnst RE Mohr) Warnst. S ae Herzogiella Broth. ma tifoli (Di papillosum (Lindb. e Mii d deban (бтр, ) Kindb. in Fleisch. LLAC foliolatum (Card. ) Wijk & & Marg haldan m Fleisch. = Callicladium hal- Да! тапи nemorosum (Brid .) Kindb. Himantocladium (Mitt.) о кыны cyclophyllum (C. Müll.) Fl as. loriforme (Bosch & ed ) Fleisch. plumula (Nees) Fleisc Holomitrium Brid. is PR cylindraceum (P. Beauv.) i a Marg. densifolium (Wils.) Wijk & griffithianum Mitt. = H. in ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 japonicum Card. = H. и perichaetiale (Hook.) Brid vaginatum (Hook.) Brid. = H. cylindraceum Homalia Schimp. (NECKERACEAE) japonica Besch microdendron (Mont. ) Jaeg. = Homaliodendron mi- crodendron Нее jos a лл T Homaliadelphus Dix. & P. de а Varde (NECKERACEAE) oe (Okam.) Iw arpii (Williams) Sharp var. scone a (Nog.) Iwats. resti lo (Mitt.) Dix. & P. de la Varde var. tar- nus var. laevidentatus (Okam.) Nog. = H. laeviden- Homaliodendron Fleisch. (NECKERACEAE) exiguum (Bos isch. 1 Lac.) Flei flabellatum (J. Sm.) Fleisch. handelii Brot interm javanicum (C. M üll.) Fleisch. ) microphyllum Gao in Wu pps & M.-z.Wang (nom. nud.) = . exiguum montagneanum [s Müll.) Fleisch. neckeroides opa pd Broth. pygmaeum Herz. & eed reci clio! lium (Mitt.) Fl хара оит (Mitt. ) Fish var. scalpellifolium olium Flei audacia (C. Müll. " Fleisch. un Homalotheciella (Card. a Broth. (BRACHYTHECIACEAE) iS li ан Hedw. ) Brot ass sli S.G. as аа ш leucodonticaule (© Müll.) Broth. longicuspis я (Mont ) Robins. = Palamocladium nilgheri nitens (Hed wR perimbricatum rae var. perimbricatum tokiadense (Mitt. ) Besch. = H. laevisetum Homomallium (Schimp.) Loeske (HYPNACEAE) connexum (Card.) Bro denticulatum Dix. = H. connexum incurvatum (Brid.) Loeske japonico-adnatum (Broth. ) Broth. icc de (C. Müll.) Nog. var. leptothallum — rohypnum leptothallum END E (C. Müll.) Broth. yuennanense Broth. Hondaella Dix. & Sak. (HvPNACEAE) 1986] aulacophylla Dix. & Sak. = H. S brachytheciella (Broth. & Par.) A Hookeria J. Sm. ee acutifolia Hook. & Gre lucens (Hedw. ) J. Sm. nipponensis (Besch). Broth. — H. acutifolia Hookeriopsis (Besch.) Jaeg. (HOOKERIACEAE) geminidens Broth leiophylla (Besch. ) Jaeg. pappeana (Hampe) Jaeg. utacamundiana (Mont.) Bro Horikawaea Nog. (PHYLLOGONIACEAE) nitida Nog йл (C. Müll.) EAE (POTTIACEAE) amplexifolium (Mitt.) Che anceps (Card.) Herz. & Nog. arcuatum (Griffith) Wijk & M consanguineum (Thwaites & Mitt. ) Hilp. on Chen ehrenb (Lor.) Jae gracilentum (Mitt.) Chen nicum (Doz y & Molk.) A a var. javanicum va convolutifolium (Dix.) Che levium (Broth. & Yas.) Chen culum (C. Müll. en -ehrenbergii me E Chen 0да es Legh pros subcomosum (Broth.) subpeltucidum (Mitt.) Hip var. subpelludicum r. me oloma Herz william ii C yaroamblystegium Loeske (AMBLYSTEGIACEAE) fluviatile E w.) Loeske ssp. noterophilum (Sull. & out = Amblystegium noterophilum Ma, (Hook. £ Wils.) Loeske = Amblystegium tenax (Hedw.) Jenn. = Amblystegium tenax Hygrohypnum Lindb. (AMBLYSTEGIACEAE) alpestre (Hedw.) Loeske dilatatum (Wils $ ium Е Broth fontinaloides Che luridum Шик ) Jenn. var. luridum var. ehlei (Arn.) Wijk & Ma var. ed pac Mdh (Brid.) C. Jens. in Podp. molle (Hedw.) Loeske montanum (Lindb.) Broth. palustre Loeske — tat luridum var. luridum poecilophyllum D — pa. Bath. var. smithii oulardii (Schimp.) Wijk & Marg. Hyiocomiopsis Card. (THUIDIACEAE) ovicarpa (Besch.) Car mon B. $. G. (HYroCOMIACEAR) m (Lesq. & James) Aust. и (Впа.) B.S.G. var. brevirostre var. cavifolium (Lac.) N cavifolium Lac. = H. е var. cavifolium himalayanum (Mitt.) Jaeg proliferum (Brid.) iia — H. splendens pyrenaicum (Spruce) Lindb. splendens (Hedw.) B.S.G. REDFEARN & WU—MOSSES OF CHINA 191 umbratum (Hedw.) B.S.G. yunnanense Besch. — оон yunnanensis R. Br. с АЕ) Be = Weissia exserta m ma вне Fleisch. = Barbula crueger um (Hedw.) R. Br. in yis ү" Hornsch. Hymenostyliella Bartr. (POTTIACE K. Saito = Didymodon japonicus ca (Nees & Blume) Brid. pesi Sa Broth. setschwanica (Broth.) 8 ех Сһеп spathulata (Harv.) Ja stenophylla Card. = EN bonn: platyphyllum аа (C. Müll.) Lindb. ex Mitt.(Hyp- desc im formosicum Card. = H. v и (Reinw. & Hornsch.) Lindb. in Dozy & Molk. га Mitt. in Seem m Hedw. [m жө (С. же» 1. zm = H.r ofc var. Sinensi- mollus аы Lindb. — H. lindbergii calcicolum Ando callichroum dios А Hedw den ndo жы cum Sull. & Lesq. fertile Sendtn filare (С. Müll.) P аг. = ө» pc flaccens seer H. macrogynu hamulos homaliaceum Besch. ) Doign. = H. erectiusculum kushakue кол ra Mb Par. = Eurohypnum lepto- thallum lindbergii Mitt. macrogynum Besc minutum (C. Müll.) Par. sinensi-molluscum oldhamii (Mitt) Jaeg. pallescens (Hedw.) P. Beauv. var. pallescens var. reptile pd Husn. — H. pallescens var. pallesce plumaeforme Wit var. plumaeforme var. alare (Par.) Iis. var. gracile Broth. var. minus Broth. ex lis var. sinensi-molluscum (C. Müll.) Ando var. strictifolium Broth. pratense (Rabenh.) W. Koch ex Hartm. procerrimum Mol. — Pseudostereodon procerrimum — H. plumaeforme var. 192 pseudorevolutum Reim. = Stereodontopsis pseudo- reptile Michx. = H. pallescens var. pallescens revolutum (Mitt.) Lindb. setschwanicum (Broth.) Ando shensianum Ando siuzewii (Broth.) Broth. = Brotherella yokohamae subimponens Lesq. ssp. subimponens ssp. ulophyllum (C. Müll.) Ando submolluscum Besch. р Sull. & Lesq. = Rhaphidostichum bos- ssp. thelidictyon ый С (Broth.) Par. ill.) Par. vaucheri Le zickendrahtii Ren. & Card. = H. macrogynum Hypopterygium Brid. лынын aristatum Bosch & ceylanicum Mitt. = H. tenellum fauriei Besch. flavo- desde C. Müll. sinicum Mitt. — ОН. tenellum tenellum C. Müll tibetanum Mitt. Indusiella Broth. & C. Müll. in Broth. (GRIMMIACEAE) thian-schanica Broth. & C. Müll. in Broth. Ishibaea Broth. & Okam. = Lescuraea (BRA- С АСЕАЕ ms (Besch. & Card.) Toyama = Lescuraea sax- Iopterygiopsis Iwats. ee muelleriana (Schimp.) Iw doc um Mitt. m albescens (Hook.) Jae, applanatum Fleisch arquifolium yay & Lac.) Jaeg. = I. pohliaecarpum bancanum (Lac.) Jaeg. courtoisii ои & Раг cuspidifolium mae = ОАО cuspidifolium expallescens Lev. in Okam laxissimum Card. lioui Thér. & P. de la V micans (Sw.) Kindb. var. ye (Jaeg. й и ex Iwats. minutifolium Card. & Thér. = J. a ns muellerianum (Schimp.) Jaeg. = Isopterygiopsis leriana ee yes ex lis. = I. pohliaecarpum piliferum (C. m.) Loe pohlecarpum cul & lu ) Jaeg. deis um e sin th & P subalbidum vc rus ) Mitt. = 7. minutirameum tenerum Sw.) Mitt. textori (Lac.) Mitt. — . pohliaecarpum Isotheciopsis т. (LE Isothecium Brid. (LEMBOPHYLLACEAE ) yosuroides subdiversiforme Broth. var. subdiversiforme ar. filiforme Nog. ar. formosanum Nog. a Lindb. = Г. myurum ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ишш Buck & е о leucotricha dcs ru Jafuelio bryum Thér. as marginatu H Juratzkaea Lor. (FABRONIACEAS) seminervis (Schwae sinensis Fleisch. Kiaeria 1. Hag. (DICRANACEAE) blytii (B.S.G.) Broth falcata (Hedw.) I. Ha starkei (Web. & Mohr) I. Hag. Kindbergia Ochyra (BRACHYTHECIACEAE) oen (Broth.) о ra var. arbuscula ar. acuminata (Tak.) Ochyra iis (Hedw.) Ochyra Leiodontium Broth. (HvPNACEAE) gracile A: m robustum Leptobryum m * G. ) vied ecu lutescens щш )M pyriforme (Hedw.) W m pode Fleisch. = Leptohymenium (HvLo- COMIACEAE) psilura (Mitt.) Fleisch. = Leptohymenium tenue Leptocladium Broth. (THUIDIACEAE) sinense Broth т (Schimp.) Warnst. (AMBLYSTEGIACEAE) kochii (B.S.G.) Warnst. = Amblystegium trichopo- dium m (Hedw.) Warnst. = Amblystegium er Leptodontium (C. Müll.) Hampe ex Lindb. (Po AE) ur NE Nog. handelii Thér. nakaii Okam pergemmascens Broth. gemmascens scaberrimum Broth. subfilescens od — L. handelii = Bryoerythrophyllum per- viticulosoides (P. Beauv.) Wijk & Marg. var. viticu- losoides var. deo rd (C. Müll.) Wijk & Marg. warnstorfii sch. mean ro (HYLOCOMIACEAE) brachystegium Besch. hokinense eje macroalare = Macrothamnium psilurum tenue ree Sc hwaegr paced, ade C. Müll. (THUIDIACEAE) ustro-alpinum C. Müll. in resin um Broth. stricticaule Broth. кири КАА in ) Broth. tenellum Lescuraea B. ES ў (LESKEACEAE) incurvata (Hedw.) Lawt. morrisonensis (Tak.) Nog. (comb. inval. basion. non cit. mutabilis (Brid.) Lindb. ex I. Hag. saxicola (B.S.G.) Milde Leskea Hedw. (LESKEACEAE) arenicola Best consanguinea (Mont.) Mitt. polycarpa Ehrh. ex Hedw. scabrinervis Broth. & Par. 1986] REDFEARN & WU—MOSSES OF CHINA 193 subacuminata Nog Loeskeobryum Fleisch. in Broth. = Hylocomium subfiliramea Broth, & P (HYLOCOMIACEA ela (Limpr.) E aa brevirostre (Brid.) Fleisch. in Broth. = Hylocomium nervosa (Brid.) Loe revirostre tectorum (Brid.) t. Lopidium Hook. f. & Wils. in Hook. f. D dd Leucobryum Hampe (LEUCOBRYACEAE) concinnum (Hook.) Wils. in Hook. f. aduncum Dozy & Molk. javanicum Hampe = L. struthiopteris angustissimum Broth. nazeense (Thér.) Broth. armatum Broth. struthiopteris (Brid.) Fleisch bowringii Mitt. trichocladon (Bosch & Lac.) Fleisch. brevicaule Besch. = L. neilgherrense Lyellia R. Br. (POLYTRICHACEAE) canadidum (P. Beauv.) bius in Hook. f. var. pen- crispa R. Br. tastichum (Dozy & Molk.) Dix minor Xu & Xiong glaucum (Hedw.) | Hirn in Fr. platycarpa Card. & Thér. japonicum (Besch.) Card. ex Broth. = Г. neilgher- Macrocoma (C. Müll.) Grout (ORTHOTRICHACEAE) rense чаш (Mont.) Grout = М. tenue ssp. sul- javense (Britt.) M juniperoideum (Brid) C. Müll. tenue e (Hook. & Grev.) Vitt ssp. sullivantii (C. Müll.) neilgherrense C. Müll. scaberulum Card. in Ren. & Card. Macromitium Brid. (ORTHOTRICHACEAE) scabrum Lac. gustifolium Dozy & Mo scalare C. Miill. ex Fleisch. жы ар аайы Broth. & Par. ex Broth. = M. pro- textorii Besch. = L. neilgherre Leucodon ный die dad i UN brevituberculatum Dix. angustiretis Dix chungkingense Chen coreensis Card. courtoisii Broth. & Par. denticulatus Broth. in C. Müll. = L. exaltatus cylindrothecium Nog. esquirolii fasciculare Mitt exaltatus C. Mül ferriei Card. & Thér flagelliformis C. Müll. = L. pendulus formosae Card. giraldii C. Müll. = L. exaltatus fortunatii Thér. var. fortunatii lasioides C. Müll. = Forsstroemia laisoides var. nigrescens latifolius Broth. foxworthyi Broth. u i gebaueri Broth. luteus Besch giraldii C. Müll. = M. japonicum mollis Dix goniostomum Bro m Nog. mnostomum Sull. & Lesq. pendulus Lindb. handelii Broth а. ОКат. = L. pendulus heterodictyon Dix. radicalis M.-x. Zhang in X.-j. Li & M.-x. Zhang holomitrioides No: sciuroides (Hedw. + eae hymenostomum Mont. = (?) Macrocoma tenue ssp. secundus (Harv.) sap sinensis Thér. incurvum (Lindb.) Mitt. = M. japonicum squarricuspis Broth. & Par. = L. esquirolii involutifolium (Hook. & Grev.) Schwaegr. subulatulus Broth. japonicum Dozy & Mo subulatus Br A makinoi d th. A т = M. japonicum tibeticus M.-x. Zhang melanostomum Leucoloma Brid. een moorcrofti (Hook p Grev.) Schwae aegr. um (Bri eel e C. Müll. = M. sulcatum var. neel- molle ( Müll.) Mitt. comium Mitt. (HYPNACEAE) nepalense (Hook. & Grev.) Schwae TRU (C. Müll.) Jaeg. okamurae Broth. — Macrocoma fiai ssp. sullivantii Leucophanes Brid. (LEUCOBRYACEAE) perottetii C. Müll. = Macrocoma tenue ssp. sulli- EE, (Schwaegr.) Lindb. vantii octoblepharioides Brid. prolongatum Mitt Levierella C. Müll. (ENTODONTACEAE) quercicola Broth. var. quercicola fabroniacea C. Müll. var. angustifolium Brot Lindbergia Kindb. (LESKEACEAE) reinwardtii Schwae brachyptera (Mitt.) Kindb. var. brachyptera ee Nog. var. austinii (Sull.) Grout sinense Bart brevifolia (Gao) Gao in X.-j. Li (editor) sulcatum (Hook. ) Brid. var. neelgheriense (C. Müll.) japonica Card. magniretis = — Broth. var. roue sullivantii C. Müll. = Macrocoma tenue ssp. sulli- var. yunnanensis Thér. & C П ovata Gao in са G.-c. Zhang fi Cao (nom. illeg.) VÀ eium Acn Thér. & P. de la Varde sinensis (C. Müll.) Broth. aiheizanense N 194 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 taiwanense Nog. tuberculatum Dix. uraiense Nog. Macrothamnium Fleisch. ee Ve nie Mi ii Gao & A delicatulum ed javense Flei ета (Reinw. & Hornsch.) Fleisch. psilurum (Mitt.) Nog setschwanicum Broth, = M. macrocarpum Meesia Hedw. (MEESIACEAE) longiseta Hedw. triquetra (L. ex Richt.) Angstr. f. triquetra f. crassifolia Kab. ү Hedw. var. piii alpina (Bruch) Ham Xu Mitt. су з шша angustirete Broth. Merceya Schimp. = Scopelophila (POTTIACEAE) gedeana (Lac.) Nog. = Scopelophila cataractae ligulata (Spruce) Schimp. = Scopelophila ligulata thermalis Fleisch. ex Broth. var. compacta Fleisch. = Scopelophila ligulata Merceyopsis Broth. & Dix. (POTTIACEAE) sikkimensis (C. Müll.) Broth. & Dix. = Scopelophila cataractae Mesonodon Hampe (ENTODONTACEAE) flavescens (Hook.) Buck Meteoriella Okam. (METEORIACEAE) soluta (Mitt.) Okam. f. soluta f. flagellata Nog. Meteoriopsis Fleisch. ex Broth. (METEORIACEAE ancistrodes (Ren. & Card.) Broth. = M. reclinata var. ancistrodes formosana Nog. = M. reclinata var. formosana reclinata (C. Müll. ) Fleisch. ex Broth. var. reclinata var. ancistrodes (Ren. & Card.) Nog. sinense se (С. Müll.) Broth. — M. reclinata var. ceylon- ensis squarrosa (Hook.) Fleisch. in Broth. var. squarrosa Meteorium (Brid.) Dozy & Molk. (METEORIACEAE) acutirameum Thér. & Copp. = M. buchananii ssp. chananii buchananii (Brid.) Broth. ssp. buchananii ssp. helminthocladulum (Card.) No helminthocladulum (Card.) Broth. = M. buchananii ssp. helminthocladulum helminthocladum (C. Müll.) Fleisch. = M. subpolyt- козе = М. subpolytrichum ssp. horikawae poris о (С. МШ.) Fleisch. ex Broth. ssp. mi- ssp. atrovariegatum (Card. & Thér.) Nog. papillarioides Nog. piliferum Nog. = M. subpolytrichum ssp. horikawae rigidum Broth. = M. buchananii ssp. buchananii ee (Besch.) x ssp. subpolytrichum p. horikawae (Nog.) N и Metzlerella I. Hag. = Atractylocarpus AREE RE Microcampylopus (C. Müll.) Fleisch. (DICRANACEAE) longifolius Nog. f. longifolius f. densifolius Nog. Microctenidium Fleisch. (HvPNACEAE) assimile th. heterophyllum T leveilleanum (Dozy & Molk.) Fleisch. Microdendron Broth. (POLYTRICHACEAE) sinense Broth. Microdus Schimp. in Besch. LADEN austro-exiguus (C. Müll.) P. brasiliensis (Duby) Thér. laxiretis Broth. (hom. ille pomiformis (Griffith) Besch. in Par. = M. brasiliensis sinensis Herz M Aust. (EPHEMERACEAE) tenerum ( Mielichhoferia М ees & Hornsch. (BRYACEAE) mielichhoferi (Hook.) Wijk & Marg. var. mielichho- eri var. Japonica (Besch.) Wijk & Marg. sinensis Mitthyridium Robins. ( ) fasciculatum (Hook. ^ pase ) Robins. flavum (C. Müll.) Robins. undulatum (Dozy & Molk.) Robins. [Not included n China by Lin, 1984] Мес Broth. (THUIDIACEAE) fruticella ip 5 rotundifolia Mniobryum тшм (BRYACE AE) albicans (Wahlenb.) г = Pohlia wahlenbergii ludwigii (Schwaegr.) Loesk lutescens (Limpr.) i — Leptobryum lutescens pulchellum (Hedw.) Loeske — Pohlia lescuriana tapintzense (Besch.) Broth. аа (Web. & Mohr) Jenn. = Pohlia wah- lenbergii Mnium Hedw. (MNIACEAE) ambiguum H. Müll. arbuscula C. Müll. = Plagiomnium arbuscula areolosum X.-j. Li & Zang = Plagiomnium arbus- cula arisanense Sak. = M. laevinerve cinclidioides Hiib. = Pseudobryum cinclidioides cuspidatum diea var. cuspidatum = Plagiomnium cuspidatu var. subintegrum Chen ex X.-j. Li & Zang = Pla- mnium acutum var. trichomanes (Mitt.) Chen ex X.-j. Li & Zang — Plagiomnium acutum ee Schimp. ex Salm. = Plagiomnium japon- ый на d ex X.-j. Li & Zang = Plagiom- ium succulen drummondii Bruch & Schimp. = Plagiomnium excurrens Par. & Broth.= Plagiomnium acutum 1986] Ee Sull. & Lesq. = Аи flagellaris m Card. = Plagiom handelii- oth. — Orthomnion п handel hetero dag ) Schw: zomnium oe ва inatum Broth. = Trachycystis ussuriensis c mnium dida um p ети japoni laevinerv atilimbatum X. -j. Li & Zang = Plagiomnium ellip- leucolepioides Chen ex X.-j. Li & Zang = Trachy- ussuriensis loni cond tum X.-j. Li & Zang = M. lycopo- dioi longispinum X.-j. Li & Zang = M. lycopodioides luteolimbatum Broth. = Plagiomnium succulentum lycopodioides Schwaegr. dici (Lindb. & Arn.) Kindb. var. а т X.-j. Td gd M. marginat marginatum (With.) P. B maximoviczii и уаг. maximoviczii = Plagiom- nium maximovi var. angustilimbatum Dix. = Plagiomnium max- imoviczii var. о йлн Сһеп ех Х. с). Li & Zang = agiomnium maximovi medium B.S.G. = Plagiomnium microphyllum Dozy & Molk. — Tob micro- a nakanishikii Broth. : — Plagiomnium succulentum = M. thomsonii — Rhizomnium parvulum üll & Kindb. in Ma- — Plagiomnium rhyncho- u = rhynchophorum Hook. rostratum Schrad. f. rostratum = Plagiomnium ros- f. coriaceum (Griffith) Kab. = Plagiomnium rhyn- f. laxirete Kab. = Plagiomnium succulentum f. microovale jos Müll) Kab. = Plagiomnium = Pandan ellipticu рен x. -J. Li & Zang = ое arbus- culum spinosum (Voit) Schwaegr. spinulosum B.S.G. stellare Reichd. ex Hedw striatulum Mitt. = Rhizomnium striatulum subundulatum Dix. ex Kab. = Plagiomnium max- imoviczii succulentum Mitt. var. integrum Nog. = Plagiom- nium succulentum tezukae Sak. — н tezukae thomsonii Schim trichomanes Mitt. = Plagiomnium acu undulatum Weis ex Hedw. var. iar icm A Plagiomnium arbuscula vesicatum Besch. — Plagiomnium vesicatum yunnanense Thér. — Plagiomnium maximoviczii Molendoa Lindb. (POTTIACEAE) REDFEARN & WU—MOSSES OF CHINA 195 hornschuchiana (Hook.) Lindb. ex Limpr. f. horn- uchiana f. barbuloides Broth. japonica Broth. = Didymodon japonicus roylei (Mitt.) Broth. —' (B.S.G.) Limpr. var. sendtneriana var. nica Gyórt. in Thér. жолы Broth. = М. sendtneriana var. yun- nanica Myurella B.S.G. oo brevicosta Lou julacea (Schwaegr.) B.S.G. sibirica (C. Müll.) Reim. sinensi-julacea C. Müll. = M. julacea tenerrima (Brid.) Lindb. Myuriopsis Nog. = Eumyurium (MYURIACEAE) myurium sinicum ) foxworthyi (Broth.) Broth. = Vig ag le a fragile (Card.) Broth. = Oedicladium frag. rufescens (Reinw. & Hornsch.) Fleisch. = а dium rufescens tortifolium Chen = Oedicladium e Myu — Besch. (BRACHYTHECIACEA concinna Besch. = M. maximowiczii maximowiczii (Borsz. Д Steere & Schof. Nanomitrium Lindb. = Micromitrium (EPHEMER- ACEAE tenerum (B.S. Lindb. = Micromitrium tenerum Neacroporium Iwats. & Nog. flagelliferum (Sek. ) Iwats. & Nog. . (NECKERACEAE) brachyclada Besch. brevicaulis Broth. ex Card. = Neckeropsis obtusata Ss (Hedw.) Hiib corean diss Harv. ex Nog. crenulata Harv. in Hook. crispa He decurrens Broth. lir decurrens var. rupicola Bro fauriei Card. жиш Card. var. ниш attenuata nate Nog. konoi Broth. ex Card. morrisonensis Nog. pennata Hedw perpinnata Card. & Thér. gracilenta (Bosch & p A Fleisch. lepineana (Mont.) Fl nitidula (Mitt.) Нева obtusata (Mont.) erra in B sinensis Chen jo A rE targionianus undulata (Hedw.) Reich. Neobarbella Nog. (METEORIACEAE) 196 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 attenuata Nog. comes (Griffith) N pilifera (Broth. & Y m Nog. Ne iig ed ra Nog. RHYTIDIACEAE) giga uei (Broth. ) Nog. yunnanensis (Besch.) T. K dec i ipei Xu » Xiong nd ti u Octoblepharum + Hedw. (LEUCOBRYACEAE) um Oedicladium Mitt. (MYURIACEAE) doii ( OXW orthyii Broth. = O. fragile fragile C. rufesce ar inw. & n Mitt. кү (Chen) Iwat Oka roth. (кнутолсел coleando (Card.) N hako oniensis E fro. var. hakoniensis f. ha- kon f. multiflagillifera (Okam.) Nog. var. ussuriensis Mey )N тн, m. DC. (POLYTRICHACEAE) aligerum aristatulum Bod: armatum Broth. peer eio m & Wan ex Xu & Xiong alcatum ч їеег form = O. suzukii fecum (Hedw.) Lam. & obtusatum Broth. semi- um (Hook. f.) Mitt. var. semi-lamel- lat ar. i ES Broth. ex Chen (nom. nud.) se o & Wu Oncophorus (Brid.) Brid. (DICRANACEAE) crispifolius (Mitt.) Lindb. var. crispifolius var. brevipes аса hér. virens (Hedw. wahlenbergii Brid. var. wahlenbergii serrulata (Funck) De Not. = O. torquescens setschwanica Broth. torquescens (Brid.) Wijk & Marg. weisioides Broth Orthoamblystegium Dix. & Sak. (LESKEACEAE) longinerve (Card.) Toyama Orthodicranum (B.S.G.) Loeske seis Оне flagellare (Hedw.) Loeske = Dicranum flagell fragilifolium (Lindb.) Podp. = Dicranum fragilifo- montanum (Hedw.) Loeske scottianum (Turn.) G. Roth in Cas.-Gil. = Dicranum scotti strictum Bro Orthodontium э Ж (BRYACEAE) infractum Pe isda Wils. (MNIACEAE) bryoides (Griffith) UE & Marg. dilatatum (Mitt.) C о (Broth.) T. m ais artr pili етт В Кор. e T. Kop . Li s Zang Orthomnionss iod im onica Broth. = Or nnno dilatatum иона Broth. (MYURIACEAE) h HE (Brid.) B.S.G. ic Lor. Orthotrichum Hedw. (ORTHOTRICHACEAE) nage ex Brid. Hedw. callistomoides Broth. m B.S.G. = O. striatum posten C. Müll. acounii ust. ssp. macounii ssp. japonicum Iwats ruspestre Schleich. ex Schwaegr. scaberrimum Bro soridum Sull. & Lesa. in Aust. speciosum Nees in Sturm stria ipit vds szuchua hen Очетчаййейа Fleisch. ex Broth. (PTEROBRYACEAE) on stricta Fleisch. ex Broth. охутіутнит (B.S.G.) Warnst. (BRACHYTHECIACEAE) biform hiang i a xe um mU ieri = Eurhynchium hians laxirete (Broth.) Broth. polystictum piedi ‘Broth, = Eurhynchium savatiert ть w.) Warnst. var. praelongum ergia jb il var. locke (Turn.) Podp. = Kindbergia praelon- protractrum (C. Miill.) Broth. savatieri (Besch.) a = Sai ii Savatieri Oxystegus (Limpr.) Hilp. (POTTIACEAE) cuspidatus (Dozy & Molk.) Chen cylindricus (Brid.) Hilp. = O. tenuirostris tenuirostris (Hook. & Tayl.) A. J. Sm. var. tenuiros- is sensu lato var. stenocarpus (Thér.) Zand Palamocladium C. Müll. Go ee arg. euchloron (C. Miill.) Wijk & Ma 1986] m (Sull. & Lesq.) Iwats. & Tak. in Tak., og. var. macrostegium vatum (Dix. & Sak.) Tak. & Iwats. iie ce (Mont.) C. Müll. f. nilgheriense f. luzonense (Broth.) Tak., me & Nog. sciureum (Mitt.) нано іп Р var. macrostegi Palisadula Toyama (SEMA TOPHYLLACEA®) chrysophylla (Card.) Toy katoi auch. non. Girth) T Eo [Iwatsuki, 1979] — Oedicladium ru ры Brid. EE squarrosa (Hedw.) Brid. Papillaria (C. Müll.) C. Müll. in Ångstr. (METEORI- — P. macrostegium Pret Nog. = P. crocea chrysoclada (C. Müll.) Jaeg. ) Jae, cusipidifera (Hook. A & Wils.) Jaeg. = P. crocea feae C. Miill. ex Fleisch. flexicaulis (Wils.) Jaeg. fuscescens (Hook.) Jaeg helminthocladula Card. - — Meteorium buchananii ssp. helminthocladulum nigrescens auct. non (Hedw.) Jaeg. [Noguchi, 1976] — Meteorium TN semitorta (C. M Paraleucobryum me) Loeske (DICRANACEAE) enerve (Thed.) Lo fulvum (Hook.) pon in Podp. = Dicranum fulvum longifolium (Hedw.) Loeske Pelekium Mitt. (THUIDIACEAE) bifarium (Bosch & Lac.) Fleisch. velatum Mitt. Penzigiella Fleisch. (PTEROBRYACEAE) Broth. — Neodolichomitra yunnanensis pidat e Philonotis Brid. i (BARTRAMIACEAE) appresifolia falcata с Mitt. var. a var. rae (Mitt.) Och кекеч уаг. pone f. fontana 6 Я y Ki indb. hastata Tu Wijk & Marg. lancifolia longiseta Michx ) Britt. lut =P. п var. fontana f. fontana Brid. palustris Mitt. = P. turneriana papillatomarginata Lou & Wu plumulosa Card. & Thér. = P. nitida d Bosch & Lac. vatieriana (Besch.) Broth. = : thwaitesii seschuanica (C. Müll.) Par. — P. nitida Mitt. — P. thwaitesii ía (Griffith) Mitt. thwaitesii Mitt. turneriana (Schwaegr.) Mitt. var. turneriana var. robusta Bartr. vitrea Herz. & Nog REDFEARN & WU —MOSSES OF CHINA 197 Physcomitrium (Brid.) Brid. (FUNARIACEAE) i G. = P. eurystomum sinensis-sphaericum C. Müll sphaericum (C.F. Ludw.) Fürnr. in Hampe spurio-acuminatum Dix. — P. eurystomum subeurystomum Card. = P. japonicum systylioides C. Müll. = P. sphaericum Pilotrichopsis Besch. (CRYPHAEACEAE) dentata (Mitt.) Besch. var. dentata var. hamulata N hen Pinnatella Fleisch. (NECKERACEAE) makinoi (Broth.) Broth. pusilla No robusta Nog. taiwanensis Nog. Pireella Card. (PTEROBRYACEAE) ormosana Broth. Plagiobryum Lindb. (BRYACEAE) demissum (Hook.) Lindb. giraldii (C. Müll.) Par zierii (Hedw.) Lindb. : var. а C. Müll. . (MNIACEAE) arbuscula (C. Müll.) T. Kop confertidens (Lindb. & г. Amel T. Kop. cuspidatum (Hedw.) T drummondii (Bruch & Schi ) T. Kop. decidi (Brid.) T integro-radiatum е Сао & К.-с. Chang = succulentum integrum (Bosch & 2 T. Kop. japonicum (Lindb.) T. Kop. maximoviczii E y a Kop. medium (B.S.G.) T rhynchophorum на т. Кор. ит ioe rad. ssp. у san (Besch. ) Nog. in Nog. & Iwats. — РУ „иШ ы Ын ) Т. Кор. tezukae (Sak.) Т. Kop. trichomanes (Mitt. ) T. E — P. acutum ederi pinus (Schwaegr.) Dalla Torre & Sarnth. Шы о B.S.G. (PLAGIOTHECIACEAE) aomoriense Besch. = Taxiphyllum aomoriense brevicuspis Broth. = Taxiphyllum alternans sale Sig (Brid.) Iwats. ar. fallax (Card. & Thér.) Iwats. cundum Schleiph. ex Limpr. denticulatum (Hedw.) B.S.G. euryphyllum (Card. & Thér.) Iwats. var. euryphyllum var. brevirameum (Card.) Iwats. formosicum Broth. & Yas. var. formosicum var. rectiapex D.-k. Li 198 сото Broth. handelii aetum B.S.G. laevigatum Schimp. ex Besch. = (?) Entodon com- longisetum ms — P. sylvati .G. var ie N ) Iw th. .S.G. = Isopterygium piliferum — Taxiphyllum pilosum roseanum B.S.G. = P. cavifoliu rotundifoli ium D.-k. Li = (?) Р. cavifolium splendens Schimp. ex Card. var. splendens - P. eu- r. brevirameum Card. = P. euryphyllum ae Bro succulentum (Wils) B d.) B.S. a var. sylvaticum .) Koppe = Taxiphyllum alternans = (?) P. neckeroideum eum Plandictva В erk. (AMBLYSTEGIACEAE) ger MER rid.) Crum Fleisch. arioi =E urhynchium riparioides aan cullen ok "Fleisch. = Eurhynchium ri- rioides Pleuridium Rabenh. У ситіпаіит Lind julaceum Besch rego (Hedw. ) Rabenh. un Ф е. = e Mitt Раги Lindb. (POTTIACEAE) indb. Palamocladium (BRACH- ACEAE) euchloron (С. Müll.) Broth. = Palamocladium eu- chlor oron fenestratus Griffith = Palamocladium nilgheriense sciureus Mb ) Toyama = Palamocladium macro- PE Lim. « ex awe (POTTIACEAE) schliephackei L Pleuroziopsis Kindb. € ex x Britt. (PLEUROZIOPSACEAE) ruthenica (Weinm.) Kindb. Pleurozium (Sull.) pen (ENTODONTACEAE) schreberi (Brid.) M Pogonatum P. ао (Фогутюснаскл) akitense Besch. = P. n aloides (Hedw.) P. B ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 capillare (Michx.) Brid. cirratum (Sw.) Bri contortum (Brid.) Lesq. m Chen ex Wu, Lou & M.-z. Wang (nom. June Mitt. formosanum ik. grandifolium (Lindb.) Jaeg. = P. japonicum hetero-contortum Horik. = P. spurio-cirratum hetero-proliferm Horik. = P. nudiusculum iliangense Chen & Wan (nom. nud.) inflexum (Lindb.) a japonicum Sull. & Le А (Dozy & Molk.) Dozy & Molk. neesii iud e. Broth. longicollum Chen & Wan (nom. nud.) macrocarpon Broth. ee Dozy & Molk. = P. microstomum manchuricum Horik. = Polytrichum manchuricum din (C. Müll.) Par. = P. urnigerum microstomum (Schwaegr.) Brid. var. microstomum var. ciliatum Xu & Xiong inus Xu & Xiong muticum Broth. neesii (C. Müll.) Dozy nudicaule (Wright) Par. nudiusculum Mitt. oligotrichoides Horik. = P. nudiusculum otaruense Besch. paucidens Besch. = P. microstomum pergranulatum Chen perichaetiale (Mont.) Jaeg. Par. = P. urnigerum = P. perichaetiale ] SU submicrostomum Broth. takao-montanum ие (Mitt.) Jaeg. var. thomsonii r. tibetanum шы (Mitt.) Jaeg. urnigerum (Hedw.) P. Beauv. var. urnigerum var. subintegrifolium (Arn. & C. Jens.) H. Móll. yunnanense Besch. Pohlia Hedw. (BRYACEAE) acuminata Hoppe & Hornsch atrothecium (C. Miill.) Broth camptotrachela (Ren. & Card.) Broth. аге оо Chen ех Wu, Lou & M.-z. Wang inval. basion. non cit.) cilifera B Ahn a (Hedw.) Lindb. Masseria (Sull. & Lesq.) t var. crudoides exuosa Hook. ати (Card. & Thér.) Chen (comb. inval. ba- sion. non cit. gracillima (Card.) Horik. & Ochi 1986] е (Broth.) Chen (comb. inval. basion. поп nts (Sull.) Ii longicollis (Hedw.) Lindb ludwigii (Spreng. ex d Vids minor Schleich. ex Schwaegr. v | ssp. acuminata (Hoppe & Horae )Wijk & Marg. nemicaulon (C. Müll.) Broth. orthocarpula (C. Müll. ) Broth. pol ee Hoppe & Hornsch. = P. minor var. mi- lus (Kindb.) Bro ‘a m. ~ 2 = ur des (Broth.) Chen ‘amb. ovat ae non aes (C. Müll.) bid sphagnicola (B.S.G.) Bro subcompactula ria ) Wijk & Marg. subflexuosa Bro timmioides Brot rà ex Chen (nom. nud.) ss. var. fragariformis (Xu & Xiong) Redf. & Wu, nov. Polytrichum alpinum Hedw. var. fragariformis Xu & Xiong, Acta Bot. Yunn. var. ие (Broth.) Redf. & Wu, comb. nov. um alpinum Hedw. Yi leptocar- „ Symb. Sin. 4: 136. 1929. var. secundifolium (Broth.) Redf. " Wu, comb. formosum (Hedw. ar. formosum var. densifolium (Mitt. | мез, & Nog. . Sm Mb enim Hedw. (POLYTRICHACEAE) Bon — Polytrichastrum alpinum var. al- = РА (В. Br.) C. Müll. = Polytrichas- rum alpinum var. brevifolium chingdingense Chen ex Wu, Lou & M.-z. Wang (nom. nud commune Hedw. var. commune Xio ns Limpr. = Polyrichasrum ohioense pisani Wils. ex formosum = = Polyrichasirum formosum var. densifolium (Mitt.) O Yano = Poly- richastrum a var. densifoliu gracile d in Menz. — Polytrichastrum longise- jensenii I. Hag. = P. commune var. jensenii ju hnianum Dozy & Molk. = SOLIS neesii pd Willd. ex Hedw. var. juniperinum ar. affine (Funck) Brid. REDFEARN & WU—MOSSES OF CHINA 199 var. — Wahlenb. = P. juniperinum var. af- var. plferides Xu & Xiong longisetum Sw manchuricum Horik) Gao & K.-c. Chan ohioense Ren. & Card. = | ohioense d h sphaerothecium (Besch. ) C. Mill. trictum Menz. ex Brid. — P. juniperinum var. affine subformosum Besch. var. pnm & Copp. tibetanum Gao in Gao & K hang xanthopilum Wils. ex Mitt. Porotrichodendron Fleisch. (LEMBOPHYLLACEAE) c trip Гон о ) Ehrh. ex Fürnr. (POTTIACEAE) intermedia (Turn.) т, lanceolata (Hedw.) С. М latifolia со) С. 3 — Stegonia latifolia üll. in a о Card. Фтиснаскл) арй =. ides С roth. Prionidium Hilp. (POTTIACEAE) и e Müll.) Chen = Didymodon lenticu ie da (Broth.) Hilp. = Didymodon eroso- ticulatus Pseudatrichum a (POLYTRICHACEAE) spinosissimum Pseudisothecium pom = BOPHYLLACEA a (Brid) Grout = Isothecium (LEM- Isothecium myosu- Pseudobarbell Nog. (METEORIACEAE) gustifolia Nog. assimilis (Card.) Nog. = P. attenuata uata (Thwaites & hangs ) Nog. = P. си = = P. atten formosica (Broth.) Nog. = Barbella compressrames kiushiuensis (Broth.) Nog. = P. atte laosiensis (Broth. & Par.) Nog laxifolia Nog. levieri (Ren. & Card.) N mollissima (Broth.) Bl. = - P. laosiensis niitakayamensis Nog. — Floribundaria setschwanica (Nog.) Nog. = Chrysocladium flammeum ssp. ochraceum Pseudobryum (Kindb.) T. sped (MNIACEAE) cinclidioides (Hüb.) T. K speciosum (Mitt.) T. K Pseudoleskea B.S.G. она NR 200 denudata (Kindb.) Best = P. radicosa incurvata (Hedw.) Loeske = Lescuraea incurvata papillarioides C. Miill. radicosa (Mitt.) Mac. & Kindb. yu roth. Pseudoleskeella Kindb. (LESKEACEAE) catenulata eros ) Kindb. жоо id d.) Kindb. ex Broth. = Leskeella tec- Pseudoleskeopsis Broth. (LESKEACEAE) decurva ан em oth. — P. zippelii —— japo ca (Sull. 2 Lesq.) Iw ats. — P. zippelii DR (Mitt.) Broth. var. orbiculata — P. zippelii var. Pepe pea r. — P. zippelii Thér. appel ory ju Molk.) Broth. Pseudop Tak. (BRACHYTHECIACEAE) morrison is Ta Pseudopohlia Wiliams (BRYACEAE) apa yunnena тна и Broth. (PTEROBRYACEAE) laticuspis Bro tenuicuspis Broth Pseudoracelopus Broth. (POLYTRICHACEAE) philippinensis Broth. Pseudoscleropodium (Limpr.) Fleisch. ex Broth. NTODONTACEAE levieri (C. Miill.) Broth. purum (Hedw.) Fleisch. in Bro о (Broth CHYPODAC horrida (Card) Fleisch. f. horrida f. laxifolia Nog. Pseudostereodon (Broth.) Fleisch. (HvPNACEAE) procerrimum (Mol.) Fleisch. in Broth. Pseudosymblepharis Broth. (POTTIACEAE) Fleisch. (TRA- sinense P. de la Varde Pterobryon Hornsch. in Mart. (PTEROBRYACEAE) arbuscula Mitt. о Broth. var. subarbuscula ar. lon Раа» Fleisch. (PTEROBRYACEAE) acuminata (Hook.) Fleisch. angustifolia Nog. arcuata Nog. lesion A Müll.) Fleisch. rd. 2 Broth. le var. tsinlingense Chen ex M.-x. Zhang ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Ptilium De Not. (HyPNACEAE) crista-castrensis (Hedw.) De Not Ptychomitrium Fürnr. (PTYCHOMITRIACEAE) dentatum (Mitt.) Jae evanidinerve (Broth.) Broth. = P. wilsonii faueri Besch. formosicum Broth. & Yas. linearifolium Reim. & Sak. longisetum gr T Sak. = P. polyphylloides mairei (Thér.) B polyphylloides (С. "Mb ) Par sinense ens ) Jaeg. var. sinense var. humile Nog var. microcarpum (C. Müll.) Card. in Broth. = tortula an ine: wilsonii Sull. & Les а. = Schimp. (Hy AE) ri Bes ch. = Pylaisiella brotheri cu е C. Müll. = — Bryosedwickia entodontea plangiangia C. Müll. = Homomallium plangiangium drm Broth & Par. in P. de la Varde = Pylaisiella robusta Pylaisiadelpha Card. (SEMATOPHYLLACEAE) imalayana in Pylaisiella Kindb. ex Grout (HYPNACEAE) brotheri (Besch.) Iwats. & Nog. extenta (Mitt.) Ando falcata (B.S.G.) Ando var. falcata var. recurvatula (Broth.) Ando - P. extenta polyantha (Hedw.) Grout robusta (Broth. & Par.) Gao e K.-c. Chang schim ri Grout = P. selwyni selwynii (Kindb.) Crum, Steere & Ander: Pylaisiopsis (Broth. ) Broth. ola) eciosa (Mitt.) Broth. Racelopodopsis Thér. (POLYTRICHACEAE) usii Thér. Racomitrium Brid. (GRIMMIACEAE) aciculare (Hedw.) Brid. albipiliferum Gao & Cao in Gao, G.-c. Zhang & Cao d. barbuloides Card. = R. canescens var. epilosum brevisetum Lindb. canescens (Hedw.) Brid. var. canescens var. epilosum H. Miill. ex Milde ericoi ides ee Hampe = R. ericoides var. и carinatum Car crispulum (Hook. : & Wils.) Dix. cucullatulum Bro delavayi Broth. ч Tm dicarpum Broth ericoides (Hedw.) Brid. fasciculare (Hedw.) Brid. var. fasciculare У " var. occidental Ren. & Card. 1986] var. ramulosum (Lindb.) Corb. var. sudeticum (Funck) E. Bauer himalayanum (Mitt.) Jaeg. = Grimmia chenii javanicum Dozy & Molk. in Zoll. var. javanicum = lanuginosum (Hedw.) Brid. microcarpum (Hedw.) Brid. molle Card. puo Broth. ex Wu, Lou & M.-z. Wang (nom. Sess (A. Br.) Hüb. = R. с subsecundum (Hook. & Grev.) М sudeticum (Funck) B.S.G. = R. oM var. sudeticum varium (Mitt.) Jaeg. yakushimense Sak. — R. fasciculare var. atroviride Racopilum P. Beauv. (RACOPILACEAE k. spectabile Reinw. & Horn Rauiella Reim. и. Regmatodon Brid. (LESKEACEAE) declinatus (Hook.) Brid handlelii Broth. longinervis Gao in Gao, G.-c. Zhang & Cao orthostegius Mont. т Herz. = К. declinatus rulatus (Dozy & Molk.) Bosch & Lac. gax (H edw.) азе, E subdenticulata (Boul.) De- - ri. pa Broth. = A laa sinensis Chen = R. crispata striata Sy Lindb. var. subdenticulata (Boul.) g. = R. crisp Rhachithecium Broth. ex Le Jol. (ORTHOTRICHACEAE) roth к (C. RS ) Par. vagi roth. Rhaphidostichum Fleisch. SS > ай оту & : piliferum (Broth.) B Rhizofabronia (Broth.) Fleisch (FABRONIACEAE) perpilosa (Broth.) Bro Rhizogonium Brid. ниочем се Fleisch. = А. longiflorum dozyan Кын (Mitt. ) Jaeg. spiniforme (Hedw.) Bruch in Rhizomnium (Broth.) T. m ee horikawae (Nog.) T magnifolium (Horik. ) T cer minutulum (Besch.) T. Kop. - — R. parvulum REDFEARN & WU— MOSSES OF CHINA arvulum (Mitt.) T. Kop. striatulum (Mitt.) T. Kop. tuomikoskii T. Ko Rhodobryum (Schimp. ) Hampe (BRYACEAE) giganteum (Schwaegr.) Par laxelimbatum (Ochi) Iwats. & T. Kop longicaudatum Zhang & X.-j. Li in x. Jj. Li (editor) = Bryum billardieri ка (Hedw.) Limpr. ulatum (Hornsch.) Pécs in Biz & Póc Riynchostegiella (B. js G. ы ай о Brid.) L lepton tenella (Dicks.) Limpr. Rhynchostegium B.S.G. (BRACHYTHECIACEAE) w.) B.S.G pallidifolium (Mitt.) Jaeg. plumosum Thér riparioides (Hedw.) Card. in Tourr. = Eurhynchium riparioides rubro-carinatum Я serpenticaule (C. Müll.) Broth. in Lev. sinense (Broth. & Par.) Broth. subspeciosum (C. Müll.) C. Müll. vagans (Harv.) Jaeg n (трг) Warnst. (RHYTIDIACEAE) loreus (Hedw.) Warn squarrosus que eile Warnst subpinnatus Leite ) i ae triquetrus )W yunnanensis ES, pes Neodolichomitra yun- Rhytidium (Sull.) Kindb. (RHYTIDIACEAE) dietis Broth. f. киы f. elongata Buck & Cru myura pteroganicides (Harv.) Jaeg. B Saelania eed (DITRICHACEAE) glaucescens (Hedw.) Broth. in Bom. & Broth. Sakuraia Broth. (ENTODONTACEAE conchophylla (Card.) Nog. = Entodon conchophyl- Sanionia Loeske — Drepanocladus (AMBLYSTEGIACEAE) в (Hedw.) ГоезКе = Drepanocladus uncina- See Broth. (AMBLYSTEGIACEAE) aomoriensis (Par.) Kanda Scabridens Bartr. (LEUCODONTACEAE) sinensis Bartr. Schistidium Brid. (GRIMMIACEAE) 202 ANNALS OF THE MISSOURI BOTANICAL GARDEN agassizii Senk : Agen in Sull. [Not included in As 2by B 1980 р (Hedw. " Limpr. var. alpicola = 5. agassizii r. rivulare (Brid.) Limpr. m Sull. = S. apocarpum var. apocarpum pum var. gracile (Róhl.) B.S.G. = S. apocarpum var. apocarpum gracile (Róhl.) Limpr. = S. apocarpum var. apocar- um Perla (C. Müll.) D ritimum (Turn.) B.S. се [Not included in As 2 by Bremer, 1980] strictum im ) Loeske W tibetanum u гоне Mohr (SCHISTOSTEGACEAE) со Вгіа. (ORTHOTRICHACEAE) charrieri Thér. & P. de la Varde и Mitt. japonica Besch. & Card. latifolia Card. & Thér. pungens purpurascens Par. Schwetschkea C. Müll. (FABRONIACEAE) brevipes (Broth. & Par.) Broth courtoisii Broth. & Par matsumurae Besch. schweinfurthii C. Müll. sinica Broth. & Par. sublaxa Broth. & Par Schwetschkeopsis Broth. (FABRONIACEAE) denticulata (Sull.) Broth. = S. fabronia fabronia (Schwaegr.) Broth. formosana Nog japonica (Besch. ) Broth. = S. fabronia subulata Chen ex Wu, Lou & M.-z. Wang (nom. nud. Sciaromiopsis Broth. (AMBLYSTEGIACEAE) brevifolia Broth. sinensis (Broth.) Broth. Scleropodium B.S.G. (BRACHYTHECIACEAE) coreense Card. illecebrum B.S. am — S. touretii touretii (Brid.) L. Koch Scopelophila (Mitt. ) Lindo МТ Spruce . Müll. in Ren. & Card. = S. cataractae Scorpidium Poea ) pen (AMBLYSTEGIACEAE) gin Drumm. S И Seligeria B.S.G (SELIGERIACEAE) pusilla (Hedw.) B.S.G. Sematophyllum Mitt. (SEMATOPHYLLACEAE) batanense Broth. — Acroporium stramineum caespitosum (Hedw.) Mitt [VoL. 73 со (Wils.) Mitt. henryi (Par. & Broth.) Bro japonicum (Broth.) Bn = S subhumile ssp. ja- к їсеит (С. Miill.) Fleisch. ln (Card.) 2 robustulum (Card.) Brot subhumile (C. MR ssp. subhumile ssp. о A ee ) Seki ha oa (M Sem arbula Herz. ex Hilp. (POTTIACEAE ) & Marg. = Barbula indica seligeri (Brid. selig spinulosa (Sull. & Lesq.) Iwats. = ИА per- Solmsiella C. Müll. = Erpodium age IACEAE) m biseriatum EAE) Sphagnum L. (S PHAGNACEAE) acutifolioides W rnst. acutifolium Ehrh ex Schrad. = S. nemoreum ше: а n Warnst. = S. recurvum var. mblyp angustifolium de Jen. ex Russ.) C. Jens. aongstroe mii Hart auri fein Schimp. — S. subsecundum var. auricu- atu beccarii H mpe oense Warnst. = _ ыы illos capillifolium (Ehrh) A = $n nemoreum compactum b var. compactum var. imbricatum Warnst. contortum K. Schultz var. subsecundum Bayrh. = subsecundum cuspidatulum C. Müll. cuspidatum Ehrh. ex G. F. ffm cymbifolium (Ehrh.) Hedw. = S. palustre cladum bist falcatulum Besch. fimbriatum Wils, in Hook. imbricatum Hornsch. ex Russ. incertum Warnst. & Car inundatum Russ. var. inundatum = S. subsecundum var. inundatum ar. perfibrosum P. de la Varde jensenii Lindb. f. junghuhnianum Dozy & Molk. x* м ssp. pseudomolle (Warnst.) Н. $ khasianum Mitt. multifibrosum X.-j. Li & Zang in X.-j. Li mori cop. Vie al Lindb. ex Warnst. obtus IR о ае е arnst. oligoporum Warnst. & a = S. microporum ovatum Hampe in C. Mül 1986] devi Warnst. & Card. papillosum En qu ucip = 5. ~ RA (Lind) Warnst. = S. subsecundum ngst robus m (Warns) )R rnst squarro. m Cro n Hop sibacutolum Ship. in Waist subbicolor ae VORN KA Nees in Sturm spp. subsecundum var. var. auriculatum (Schimp.) Schlieph. var. inundatum ss.) s. arnst.) C. Jens. in Broth. = 5. var. luzon (W zon r. yuennanense C. Jens. in Broth. bed planphvllum wg 2 n tenellum Ehrh. teres Gamp) ae in = Е. Hartm. W ens "es Nees SPIRIDENTACIAD nwardtii Splachno bryum c. Müll. (POTTIACEAE) giganteum Broth. obtusum "Ва, С.М Splachnum Нейм. es E — А Hedw. var. ampullaceum r. brevisetum Gao шы. Montin ех Hedw ovatum Dicks. ex Hedw. = S. sphaericum rubrum Montin ex Hedw. uta (Reim. киш (Broth.) Ando = Hypnum setschwan- Stereophvilum Mitt. (PLAGIOTHECIACEAE) ps (Bosch & Lac si Broth. iran (Bél.) Mitt ligulatum Jaeg. schwanicum Broth. Stokesiella (Kindb.) Robins. = Kindbergia (BRA- ) arbuscula (Broth.) Robins. = Kindbergia arbuscula Streblotrichum P. Beauv. ача СЕАЕ) i & M.-x. Zhang Struckia C. Miill. (SEMATOPHYLLACEAE) argentata (Mitt.) C. M Symblepharis Mont. pes ВИР REDFEARN & WU-—MOSSES OF CHINA cali RE Broth. 203 поса Mont. = S. vaginata t üll. vaginata (Hook.) Wijk & Marg. Symphyodon ise (SYMPHYODONTACEAE) perrottetii M bs mouthioides Card. & Thér. nesis Broth. Syrrhopodon Schwaegr. E. Pn mbiguus (H Spr шо; eri i (Ho ok.) Schwaegr. japonicus (Besch.) Broth. konoi larminatii loncho mperi tosaensi Car tsushimae Card = S. larminatii Taiwanobryum Nog. (PRIONODONTACEAE) pe Broth. & Par phyllus Dix. = S. japonicus üll Nog. Fleisch. (HYPNACEAE) W aomoriense Besch. ) Iwats. cuspidifolium (Card.) Iwats. H for а (С. 1 э Fleisch. = Т. taxirameum inundatum bo ig "e Yas.) Iwa ts. m (Br splendescens (C. Müll.) Fleisch. squamatulum (C. Müll.) Fleisch. — 7. cuspidifolium subarcuatum ii ) Iwats taiwanense taxirameum (Mitt. ) Fleis ры Im ex Mitt (SEMATOPHYLLACEAE) bat e Bart и т (Brid. ) Broth. in Ren. & Card. nepalense (Schwaegr.) Broth. rvulum (Broth. & Par.) Seki Tayloria Hook. (SPLACHNACEAE) delavayi (Besch. ) Besch. indica id Reim. = 7. indica lingulata (Dicks.) Lindb. recurvi- marginata Nog. sinensis C. Müll. splachnoides (Schwaegr.) Hoo ens ДЫМ Mitt. var. на = Bryonoguchia (THUuI- molkenboeri gus ) Broth. — E sachalinense о pen sadae — Helodium "is n edges tv DAC x Milde аи B.S. S. erp diga lcd (He B.S.G. var. angustatus ar. d (Herz.) Gao 204 bryoides =; Т. mnioides .) B.S. С. var. mnioides im var. sin aloe Gao in 1 Wu, Lou & M.-z. Wang nud. ) adu (Hedw. ) B.S Thamnium B.S.G. = Thamnobryum (NECKERACEAE) кс (Hedw.) B.S.G. = Thamnobryum alo- pecuru laevinerve Broth, = Thamnobryum laevinerve n alopecurum (Hedw.) Nieuwl. ex Gan incurvum (Nog.) Nog. & Iwats. in Iw latifolium (Bosh & Lac.) sent lavinerve Broth. ex Chen (nom. nud.) pandum (Hook. f. us zt ) ee & G. Scott plicatulum (Lac.) Iw: T pandum .) Nog. Theriotia Card. (BUXBAUMIACEAE lorifolia Card. Thuidium B.S.G. a, assimile (Mitt.) Ja bipinnatulum Mi = T. sparsifolium gs — T. submicropteris ba (Don & Molk.) Dozy & Molk. var. cymbifo var. japonicum (Lac.) Sak. = T. cymbifolium var. ша A (Hedw.) Mitt. var. delicatulum var. radicans (Kindb.) Crum, Steere & Anderson fuscatum B glaucinoides glaucinum eder ‚Ж & Гас. lejeuneoides lepidoziaceum minutulum (Hedw. ) B.S.G. [May occur in As 2] orientale Mitt. ex Dix. = T. glaucinoides perpapillosum Watan. — T. philibe pr. ^ T. oo var. radicans non ber (C. Müll.) P pygm Nog. a Hae (Hedw. ) Lindb. nh song (Mitt. ) Jaeg. subglaucinum Card. submicropteris m Var paciente ie talongens ipie Husum pa Müll.) Bosch & Lac. tibetanum venustulum Besch. vestitissimum Besch Thyridium Mitt. = Mitthyridium (CALYMPERACEAE) fasciculatum (Hook. & Grev.) Mitt. = Mitthyridium fasciculatum flavum (C. Müll.) Fleisch. = Mitthyridium flavum Thysanomitrion Schwaegr. = Campylopus (DICRANA- CEAE blumii (Dozy & Molk.) Card. = Campylopus um- bellatus ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 nigrescens (Mitt.) P. de la Varde = Campylopus ri- rdii Timmia е ни о Hedw. ssp. теши t bavarica (Hessl.) В schensiana с Müll. = T. megapolitina ssp. mega- Timmictia (De Not.) oe (POTTIACEAE) Li a la To menthypnum Loeske (BRACHYTHECIACEAE) nitens (Hedw.) Loeske = Homalothecium nitens Tortella (Lindb.) Limpr. ee E) yuennanensis Broth. = mba subduri- uscula Tortula Hedw. (POTTIACEAE) caninervis (Mitt.) Broth. d eserto t inermis ont longimucr a X.-j. L mucronifolia Schwaegr. ralis H var. muralis var ti d. ex var. obcordata (Schimp.) Limpr nko ana Nog. obtusifolia (Schwaegr.) Math planifolia Х.-). Li rinceps De Not мое ^ Müll. ) Broth. reflex ruralis e ) bees ee & Schreb. schmidii (C. Müll.) B sinensis (C. Müll.) Broth ex Lev. ne (Mitt.) Brot subuluta Hedw. dice (Ehrh. ex. Hedw.) Limpr. = Tortella tortuosa yuennanensis Chen Trachycystis Lindb. (MNIACEAE) аа (Sull. & Lesq.) Lindb. marginata (Bro = T. ussuriensis microphylla (Dozy & Molk. ) Lindb planifolium (Hedw.) Brid tula (Ho ok.) Fleisch. = T. serrulata var. cris- densifolia n c T. serrulata var. crispatula ^ . Beauv.) Fleisch. var. crispatula (Hook.) Zant var. guilbertii (Thér. & P. de la Varde) Zant. Trachypus Reinw. & Hornsch. (TRACHYPODACEAE) 1986] p Reinw. & Hornsch. var. bicolor ar. brevifolius Broth. — T. bicolor var. hispidus var. sinensis (C. Müll.) Broth. — 7. bicolor var. hispidus var. viridulus (Mitt.) Zant. humilis Lin No longifolius subbicolor e Müll. ex Card. — T. bicolor var. viri- dulus Trachythecium Fleisch. (HvPNACEAE) micropyxis (Broth.) Bartr. verrucosum (Jaeg.) Fleisch. var. verrucosum var. binervulum Herz. Trematodon Michx. (DICRANACEAE) acutus C. Müll. = 7. as ambiguus (Hedw.) H drepanellus Besch. = T longicollis longicollis Mich Trichodon Schimp. (DITRICHACEAE) cylindricus (Hedw.) Schimp. muricatus Herz Trichosteleum Mitt. (SEMATOPHYLLACEAE aculeatum Broth. & Par. = Rhaphidostichum boscii ssp. thelidictyon boschii (Dozy & Molk.) Jaeg. = Rhaphidostichum өрү n thelidictyon fissum E Богу & Molk.) Jaeg. mammosum (C. Müll.) Jaeg. parvulum Broth. & Par. — Taxithelium parvulum Trichostomum Bruch in F. Müll. (POTTIACEAE pou ны Fleisch. = Pseudosymblepharis tat aia (Broth. ) Hilp. ex кү Bo illeg.) barbuloides (Broth.) Chen ipe ill brachydontium Bruch in F cylindricum LM var. cylindricum tenuirostri var. Esc HEN Broth. — Oxystegus cuspidatus involutum Broth. (hom. illeg obtusifolium Broth. — Streblotrchum „шаш parvulum Broth. = Oxyste, id platyphyllum (Broth. ex Iis. УС Trismegistia (C. Müll.) Broth. e NND Oxystegus undulata . & Yas. Tristichella Dix. (SEMATOPHYLLACEAE) glabrescens Iwats Thistichium C. Mi (DITRICHACEAE) lorentzii sin Tuerckheimia Broth. (POTTIACEAE) crispa (Hedw.) Brid. var. crispa var. longifolia (Dix. & Sak.) Iwats. drummondii (Hook. & Grev.) Brid. eurystoma Nog. macrocarpa Broth. morrisonensis Horik. & Nog. REDFEARN & WU— MOSSES OF CHINA Venturiella C. Miill. (ERPODIACEAE) и (Vent.) C. Müll. var. sinensis angusti-annulata Griffin & Sharp йо (C. Müll.) С. site (HYPNACEAE) apiculata Broth. = V. ferri chlorotica (Besch.) Brath ferriei (Card. & Thér.) Broth. flaccida (Sull. & Lesq.) Iwats. marginata Thér. montagnei (Bél.) B reticulata (Dozy & ME ) Broth. sasaokae Oka shimadae Okam. stillicidiorum Broth. qe Hornsch. var. nivalis enocar, Warburgiella C. Müll. in 1 Broth, (SEMATOPHYLLACEAE) cupressinoides C. Miill. ex B Webera Hedw. = Pohlia о ) acuminata (Hoppe & Hornsch.) Schimp. = Pohlia ] sp. acuminata ciliifera Broth. = Pohlia ciliifera commutata Schimp. = Pohlia drummondii cruda (Hedw.) Fiirnr. = Pohlia cruda elongata (Hedw.) Schwaegr. = Pohlia elongata graciliformis Card. & Thér. = Pohlia graciliformis laticuspis Broth. = Pohlia laticuspis polymorpha (Hoppe & Hornsch.) Schimp. = Pohlia minor pygmaea Broth. = Pohlia pygmaea timmioides Broth. = Pohlia timmioides Weisiopsis Broth. (POTTIACEAE) anomala Broth. & Par a Hedw. var. controversa var. тш (Раг.) Wijk & Marg. crispa (Hedw.) Mitt edentula Mitt. = Hymenostomum edentulum exserta a ) Chen longiden microstoma Hedw.) C. Müll. = Hymenostomum . Müll. m (SEMATOPHYLLACEAE) ral e roth.) Crum deflexifolia (Ren. & Card.) Crum extenuata (Brid.) Crum hornschuchii (Dozy & Molk.) Crum juliformis (Herz. & Dix. ex Dix.) Crum semitortipilia (Fl. S.-h. Lin (comb. inval. basion. inval. tanytricha (Mont.) Crum Wilsoniella C. Müll. (DICRANACEAE) decipiens (Mitt.) Alst. in Dix. var. = уаг. о eo ) Wijk & M karsteniana C. M Zygodon Hook. & et (ORTHOTRICHACEAE) brevisetus Wils. ex Mitt. conoideus (Dicks.) Hook. & Tayl. 206 obtusifolius Hook. reinwardtii (Hornsch.) A. Br. in B.S.G. viridissimus (Dicks.) Bri yuennanensis Malta LITERATURE CITED ANDO, H. 1956. The Hypnum aud restricted to Japan and еа areas (1). J. Sci. Hiroshima , Div. 2, укн 7: 143-152. т species restricted to Sci. Hiroshima Univ., Ser. B, Div. 2, Bot. 8: 1-18. 1958. The Hypnum species restricted to Japan and adjacent areas (3). í zi Hiroshima Univ., Ser. B, Div. 2, Bot. 8: 167-208. 1963. A second species ze к a unique genus of the Hypnaceae. Hikobia 295-299. 1964. A revision of the east-Asian species of Homomallium, a genus of Musci I. Hikobia : 1965. A revision of the east-Asian species of Homomallium, a genus of Musci II. Hikobia 4: 161-176 . 1966. A revision of the Chinese Cupres- sinae described by C. Mueller. Bot. Mag. (To- yko) 79: 759-769. 1969. es aa Asiae Ori- entalis I. Hikobia 5: 8. . Studies on йс, pa Hypnum Hedw. 0, » Sci. Hiroshima Univ., Ser. B., Div. 2, Bot. 53-73. 1973a. Miscellanea bryologica Asiae Ori- entalis = Hikobia 6: 203-216. 73b. Studies on the genus Hypnum Hedw. (II). T Sci. Hiroshima Univ., Ser. B, Div. 2, Bot. 14: 165-207. . o [1977]. Studieson the genus Hypnum Hedw. (III). Y ih Hiroshima Univ., Ser. B, Div. 2, Bot 1978. Pylaisiella p ee (B. S.G.) Ando, a moss of east Asia-Latin American distribution. Phyta 1: n , H. PER E. M. SHERRARD. 1957. The first record of Gollania’ in North America. Bryol- t 60: 326-3 . Additions to the moss flora ia. ed 10: BREMER, B. 1980a. A taxonomic revision of Schis- n AA Bryophyta) 1. Lindbergia . EU . A taxonomic revision of Schisti- dium (Grimmiaceae, Bryophyta) 2. Lindbergia 6: 89-117. A taxonomic revision of Schistidium (Grimmiaceae, Bryophyta) 3. Lindbergia 7: 7 Beets У.Е. 1924-1925. Musci. In A. Engler & K. Prantl, Die Natiirlichen Pflanzenfamilien, 2nd edition. Volumes 10, 11. Leipzig, Berlin. 1929. Musci. Jn M. Handel-Mazetti, Sym- bolae Sinicae Botanische Ergebnisse der Expe- dition der Akademie der Wissenschaften in Wein ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 nach dpi E 1914/1918, Part 4. Julius A generic revision of the En- todontaceae. J. Hattori Bot. Lab. 48: 71-159. 1978. A re-interpretation of the Fabroniaceae with notes on selected genera. J. Hattori Bot. Lab. 4 CHEN, P.-C. 1940. Studie n über e: ostasiatischen Arte ЕТ ег Pottiaceae. І. Hedwigia 80: 1-76. 41. Studien tiber die osasiatischen Ar- ten ped Pottiaceae. II. Hedw 1-322. -L. er . GAO, x.. -J. LI P.-c. — Genera Muscorum a apical lea Publ. Soc., Pekin M. -5. Kuo & w. -C. ‘SHIH (editors). 1978. pon era Mu m Sinicorum, Pars Secunda. Sci- entific Publ. =: Pekin Crossy, M. R. 1977. nde; ex Muscorum Supple- mentum: о Тахоп 26: 285-307. 1979. Index Muscorum Supplementum: 1976-1977. Magus 28: 237 Rs 981. x Muscorum а 1978-1979. ns 30: 667-693. 1983. ex Muscorum Supple- mentum: 1980-1981. Теа 32: 670-691. & .M Dictionary of is. M Dixon, H. N. Kong wit other Chinese mosses. Hong Kong Naturalist, : 1-31. .А. del Lectotypifications е nomenclatural and taxonomical itri- chum phas sensu lato. Bryologist 1 88: 31- GANGULEE, H. C. 1969. Mosses of Eastern India and Adjacent Regions. Fascicle 1. Sphagnidae, Andreaidae : Nematodonteae. Publ. by the au- thor, eie 1971. Moses of Eastern India and Adja- cent Re En s. Fascicle 2. Archidiales, Dicrana- les & шасы. Publ. by the author, Cal- tta . 1972. Mosses of Eastern India and Adja- cent Regions. Fascicle 3. Syrropodontales, Pot- tiales & Grimmiales. Publ. by the author, Cal- cutta 1974, Mosses of Eastern India and Adja- cent Regions. Fascicle 4. Funariales & Eu- bryales. Publ. by the author, Calcutta. 1976. Mosses of Eastern India and Adja- cent Regions. сое 5. Isobryales. Publ. by the author, Calcut 77. Mosses 7 Eastern India and Adja- cent Regions. Fascicle 6. Hookeriales. Publ. by the author, Calcutta 78. Mosses of Eastern India and Adja- cent Regions. Fascicle 7. Arce oia (Les- keineae). Publ. by the author, Calcu 1980 Mosses of Eastern India mu y Regions. Fascicle 8. Hypnobryales (Hypnineae): Floristic ае Index: Errata. Publ. by the au- thor, Calcut Gao, C. & K.-C. p 1983. Index Muscorum Chinae Boreali- DLE J. Hattori Bot. Lab. 54: 187-205. 1986] REDFEARN & WU—MOSSES OF CHINA 207 Horton, D. G. 1982. The status of taxonomic research in the Encalyptaceae. Nova Hedwigia -237. Hou, H. -Y. 1983[1984]. Vegatation of China with reference to its geographical distribution. Ann. Missouri Bot. Gard. 70: 509-549 Hu, R.-L. 1983. A revision of the Chinese species of Entodon (Musci, Entodontaceae). Bryologist 86: 193-233. INOUE, S. 1983. Karyological studies on some species of Entodontaceae (Musci). J. Hattori Bot. Lab. 54: 299-305 IwATSUKI, Z. 1963. A revision of the east Asiatic species of the genus Anomodon. J. Hattori Bot. 27-62. А 1965. Notes on the genus Dolichotheca, with special reference to the Japanese species. J. eer Bot. Lab. 28: 202-208. . Arevision of Plagiothecium and its related а from Japan and her adjacent areas. I. J. Hattori Bot. Lab. 33: 331-380. . 1976a. Bryological miscellanies. XXI. What is Pseudoleskeopsis zippelii (Doz. et Molk.) = ? J. Hattori Bot. Lab. 40: 139-143. . Notes on Asiatic species of Di- physcium Mohr. J. Hattori Bot. Lab. 40: 143- 49. 77a. Tristichella glabrescens, a new species of Sematophyllaceae (Musci) e a = Bull. Natl. Sci. Mus. 3: 13-22. 1977b. Bryological е XXIV. Notes on J apanese Leucobry m with special ref- 1ce to tk p ] Hattori Bot. Lab. 43: ‚359—364. . 1979. Re-examination of Myurium and its J. Hattori Bot. Lab. 46: 257-283. 1980. A preliminary ore of POM in China. J. Hattori Bot. Lab. 48: -l & M. MIZUTANI. 1983. Anal a of bryophytes from Mt. Yushan, Taiwan. Misc. Bryol. Lichenol. 9: 194-195. & A. NocucHr. 1973. Index Muscorum Japonicarum. J. Hattori Bot. Lab. 37: 299-418. & Index Muscorum Japon- icarum Supplement ae J. Hattori Bot. Lab. 46: 2 A.J. D 1967. The коки еп сисара between eastern Asia a pe Nort rica, I. J. Hattori Bot. Lab. 30: 152- a 1968. The озова ка ге- lationships between eastern Asia and North merica. II. J. Hattori Bot. m 3r 55-58. & T. Suzuki. 1982 taxonomic revision id Japanese species of Fissidens! (Musci). J. Hat- ri Bot. Lab. Zr 329-508 ко Т. 1971а. The taxonomic status of Rhy- tidiadelphus iae ч Hikobia 6: 53-56 1971b. A preliminary report of Plagiom- nium section uda in Asia. Hikobia 6: 57- 1972. The east Asiatic species of Plagiom- nium section Rostrata (Bryophyta). Acta Bot. Fenn —— 1979. PATA to the East Asiatic bryoflora. III. Hylocomium himalayanum and H. poene Ann. Bot. Fenn. 16: 102-107 A synopsis of Mniaceae (Bryophyta). E diendo Asian taxa. Acta Bot. Fenn. 117: 1-34. —— —& A. KOPONEN. 1974. Tayloria subgenus ) 1 st Asia. Ann. Bot Fenn. 11: 216-222. & M.-J. Lar. 1978. Rhizomnium tuomi- koskii T. Kop. (Mniaceae) newly recorded from Taiwan. Quart. J. Taiwan Mus. 29: 497-498 & J.-S. Lou. 1982. Miscellaneous notes on Mniaceae. XII. Revision of specimens in the Institute of Botany, Academia Sinica, Beijing, China. Ann. Bot. Fenn. 19: 67-72. , X.-J. Li & M. ZANG. 1982a. Micellaneous notes on Mniaceae (Bryophyta). XIII. Orthom- nion yunnanense, species nova. Ann. Bot. Fenn. 9: 73-74. — 19 A synopsis of Rhodob ryum (Musci, Bryaceae) in China. Ann Bot. Fenn. 19: 75-80. , C. Gao, J.-s. Lou & I. JARVINEN. 1983. Bryophytes from Mt. Aue Bai, Jilin Province, Northeast China. Ann. Bot. Fenn. 20: 215-232. Lar, M.-J. & J.-R. WANG-YANG. 1976. Index bry- oflora formosensis. Taiwana 21: 159-203. Li, X.-7., M. ZANG, S.-H. ZENG, С. GAO, G.-C. ZHANG, T. Cao, P.-c. Wu, J.-s. Lou, R.-1. Hu, Y.-F. Wana, W.-x. Xu, R.-1. XIONG, M.-x. ZHANG, P.-J. LIN, Z.-w. Ao, Z.-H. Li & D.-k. Li (editors). 1985. Bryoflora of Xizang. Science Press, Beij- ing Li Z.-H. 1984. Fissidens obscurus Mitt. (Musci, Fissidentaceae), new to China. Lindbergia 10: 57-59. LIAO-NING PROVINCIAL AGRICULTURE AND SOIL RE- SEARCH INSTITUTE. Flora Muscorum Chinae Boreali- Orientalis. Science Press, Bei- jing. Lin, P.-j. 1984. А study of Chinese oe J. Hattori Bot. Lab. 55: 267-272 Lin, S.-H. 1981. Dicranoloma blumii, a new record for the moss flora of Taiwan. Misc. Bryol. Li- chenol. 9: 11-12. 1984. Notes on some Formosan moss col- lections at the Taiwan Museum. Bryologist 87: 45-348. M.-J. Lar. 1981. The peat moss genus Sphagnum L. in Taiwan. Quart. J. Taiwan Mus. 34: 95-104. MILLER, N. G. € M. G. MANUEL. 1982. Trachy- loma (Bryophytina, Pterobryaceae): a taxonom- ic monograph. J. Hattori Bot. Lab. 51: 273-321. MOoHAMED, M. A. H. 1979. A taxonomic study of Brvum billardieri Sct and related 1 J. Bryol. 10: 401—465. MÜLLER, C. 1896. Bryologia Provinciae Schen-si Sinensis. Nouvo Giorn. Bot. Ital. n.s. 3: 89-129. Bryologia Provinciae Schen-si Si- nensis II. Nuovo Giorn. Bot. Ital. n.s. 4: 245- 1898. Bryologia Provinciae Schen-si Si- nensis ex Collectione Giraldiana III. Nuovo Giorn. Bot. Ital. n.s. 5: 158-209 208 NiNH, T. 1984. A revision of Indochinese Homali- endron. J. Hattori Bot. Lab. 57: 1-39. NOGUGHI, А. 64. A revision М y pu Cla- podium. J. Hattori Bot. Lab. 2 ———, оба. А а of Japanese UMEN J. Hattori Bot. Lab. 1 otulae со. УШ. Sta of some e Asiatic mosses and the pel of a new species of Glyphomitrium. J. Hattori Bot. Lab. 31: Weed xonomic revision of the d Meteoriaceae of Asia. J. Hattori Bot. Lab. 4 231-357. 1984. рая bryologicae, XI. J. Hattori t. Norris, D. J. H 1961. The known distribution of Herpetineuron toccoae (Sull. & Lesq.) Card. J. Hattori Bot. Lab. 24: 110-114. NYHOLM, d 1971. E udies in the genus Arch P. Bea ey ofthe genus and specie Lindbergia 1: 1-33. Осні, H. 1962. Contributions to the mosses of (1). Nova Hedwigia. 4: 87-108 . 1963. Contributions to the mosses of Bar- tramiaceae in Japan and the adjacent regions (2 and last). Nova Hedwigia 5: 91-115. TO : 184- А . Taxonomic position and geo- hical distribution "of | оос» muel- leriana (Schimp.) Iwats. Fragm. Florist. Geobot. E: E ; 1982. Kindbergia ENT hie hom n. illeg. Lindbergia 8: 53-54. ROBINSON, H. 1 eneric revisions of North American Brachytheciaceae. Bryologist 65: 73. SAITO, K. 1972. Taxonomic and phytogeographic us sendtneriana. J. Hattori Bot. Lab. Е 170. ———, 1975 Am monograph of Japanese Pottiaceae (Musci). J. Hattori Bot. Lab. 39: 373-5 SEKI, T. 1968. e family phyllaceae of Japan with special pia to a statistical Medien a the family. J. Sci. Hi- shima Univ., Se 2, E de 1-80. SHAW, J. & H. CR RUM. teh Peristome homology in Mielichhoferia and a taxonomic account of North American species. J. Hattori Bot. Lab. 57: wee ics J. A 5. A revision of the genus Ar- chidium Pre J. Hattori Bot. Lab. 39: 105- 201. SOLLMAN, P. 1983. Notes on Pottiaceous mosses. I. Bryologist 86: 271-272. TAKAKI, N. 1955a. Researches on the Brachythe- ciaceae of Japan "a its adjacent areas. I. J Hattori Bot. Lab. -28. 955b. i on the Br аске. асеае of Japan and its adjacent areas. II. J. Hat tori Bot. Lab. 15: 1-69. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 . 1956. Researches on the Brachytheciaceae of Japan and its adjacent areas. III. J. Hattori Bot. Lab. 16: 1-71. TAN, В. C. & T. KoPoNEN. 1983. Dicranoloma іп southeast Asia, with special reference to the Philippine taxa. Ann. Bot. Fenn. 20: 317-334. ; 7a. Studies on the Fabroniaceae. I. Hikobia 8: 46-58. . 1977b. Studies on the Fabroniaceae of Ja- an. II. Hikobia 8: 298-321. ir c I. 1932. жег de la Chine Orientale. n. Cryptog. Exot. 5: 9. биш. Р. 1966. аа du Vietnam. Récoltes de a Petelot et V. Demange au Nord Vietnam (Relictae Henryanae). Rev. Bryol. Lichénol. 34: 127-181. VARDE, R. Р. DE LA. 1937. Contribution a la flore bryologique de la Chine. Rev. Bryol. Lichénol. 10: 136-145. Упт, D. E. 1972. A monograph ofthe genus Drum- ет Canad. J. Bot. 50: 1191-1208. 1980 a. The genus Macrocoma 1. Typifi- y of species. Bryol- Ogist ie 405-4 —— The к Macrocoma П. Geo- grap ea variation the Macrocoma tenue— M. sullivantii Eis complex. Bryologist 83: 1972. Arevision ofthe family а adjacent areas. J. Hatto ЖОЛАМАН, R. diaceae in Japan an Bot. Lab. = 171-320. 1 A note on Thuidium tibetorum Sa Imon from Tibet, China. Misc. Bryol. Li- chenol. 8: 145. ————. 1980b. eee г эрен collected by Dr. suki in Taiwan in WIJK, VA ARG FLORSCHUTZ. -1 I А Volumes 1-5. Int tional Bureau vs Plant Taxonomy and Nomenclature, Utrec Wu, P.-C. 19 of b ия from Mt. Wuyi. II. Wuyi Sci. 1 Glossary of Terms and Names of Bryophytes. чар Press, Beijing. у & В. E. AGILL. 1986. А и. and phytogeographical survey of the bryop Mt. Shennongjia, Western Hu- a. denk Missouri Bot. Gard. 73: (in bei, China SS). Xu, W.-X. & R.-L. XIONG. 1982. Taxa nova ge- neris Pogonatum. Acta Bot. Yunn. 4: 51-53. ZANDER, R. H. 1978. New combinations in Di- dymodon (Musci) and a key to the taxa in North America north of Mexico. Phytologia 41: 11- 2. 79. Notes on Barbula and Pseudocros- sidium (Bryopsida) in North America and an annotated key to the taxa. Phytologia 44: 177- 214 ZHANG, M. -X. 1982. The taxonomy of Chinese Leucodontaceae. Acta Bot. Boreal. Occid. Sin. 8-27. NOTES A REMARKABLE NEW SELAGINELLA FROM VENEZUELA Despite a recent treatment of South American spike mosses (Alston et al., 1981) in which 133 species are recognized, Selaginella in the Neo- tropics remains taxonomically difficult, with many undescribed species. In Venezuela alone, 74 species are now known, and many more will eventually be recorded. We describe one new one here, perhaps the largest known Selaginella in terms of girth and certainly the most robust American species. Selaginella gigantea grows in colonies of sev- eral hundred plants in wet evergreen primary forest in the Coastal Cordillera of Venezuela. The many endemic taxa of vascular plants (Steyer- mark, 1979, 1982). Selaginella gigantea Steyermark & A. R. Smith, sp. nov. TYPE: Venezuela. Carabobo: entre Los Тай у La Toma, a lo largo del Rio San Gian, al sur de Borburata, arriba de la Planta Electrica, 750 m, 1 Apr. 1966, Stey- ermark & Steyermark 95418 (holotype, hii 2549293; isotype, VEN). PARATYPES: sam locality, 27 Mar. 1966, Steyermark & pni ermark 95148 (G, VEN); 7 km S of Bor- burata, 10?23'30"N, 67%58'30"W, 16 Dec. 1983, Steyermark, Berry & Manara 129691 (UC, VEN). Figures 1, 2. Planta 1-2 metralis, caulibus erectis robustis arti- culatis 10-15 mm diam. glabris, parte inferiore 0.5 m rhizophora I" ferenti; internodiis usque 10 mm diam. efoliatis v integerrimis 4+ 75 mm; parte frondosa ambitu late ovato- lanceolata 4-5- -pinnata ca. 70 x 30 cm, ra- gua ovatis vel oblongo-ovatis 7-14 x 3.5-7 cm 15-30 divisiones secundarias lineares; foliis late- тай divisionum secundariarum horizontaliter pa- pire 13-29 paribus plerumque 4-5 x 1.25-1.5 mm lineari-subfalcatis oblique insertis apice acutis per to- tam latitudinem decurrentibus, marginibus integer- rimis vel apicem versus remote serrulatis utrinque gla- bris; foliis axillaribus ellipticis acutis 3.5 x m integerrimis exauriculatis; foliis intermediis 1-1.7 х .2-0.5 mm acutis basi paullo asymmetricis exauri- culatis per latitudinem adnatis; strobilis usque 20 x 1.5-2 mm in apicibus ramulorum singulatim dispo- sitis; megasporophyllis basi strobili solitariis; micro- sporophyllis numerosis late deltoideis vel subrotun- datis apice acuminatis vel breviter aristatis ca. 1.25 x ANN. MIssouRI Bor. GARD. 73: 209-215. 1986. 1.25 mm; is cremeis ca. 1.2 mm diam. gros rugosis; micros po ris ca. 30 um longis dense sar Tats. prominentiis leviter capitatis. Plant 1-2 m tall, erect, with many strongly ascending branches arising from the lower 0.5 m of main stem; stem and main branches stout, firm, with swollen, bronze-colored nodes 10-15 mm diam., glabrous; nodes 10-15 cm apart to- ward base of plant, drying brown and contracted, sending out aerial roots (rhizophores) to 6.5 mm diam. at the lower nodes; internodes to ca. 10 mm diam., stramineous, lacking leaves or leaves very remote, spreading, 3-4 mm by 1.75 mm, ovate, decurrent, acute, entire; stele T-shaped or X-shaped in cross section, with arms 2-3 mm long; frondose branch system broadly ovate-lan- ceolate in outline, 4-5 times divided with pen- ultimate divisions pinnately arranged, each of the penultimate divisions 7-14 cm by 3.5-7 cm with 15-30 secondary linear divisions; pes leaves of the secondary divisions 13-29 p horizontally spreading, linear- ad © obliquely inserted, acute at apex, 4-5 mm by 1.25-1.5 mm, decurrent and attached along their width, contiguous or slightly imbricate at the base, but spatially separated for most of their length (at least in the dried state), entire or remotely serrulate along acroscopic margin, linear-subfal- mm, elliptic, acute, entire, exauriculate; median leaves 1-1.7 mm mm, acute, glabrous, slightly asymmetric at base, exauriculate, adnate their width; strobili to 20 mm by 1.5-2 mm, solitary on the apices of penultimate branches, maturing simultaneously; megasporophylls sin- gle at base of strobilus, ca. 2 mm by 2-2.5 mm, broadly ovate to subrotund, rounded to subacute at tip; microsporophylls numerous, ca. 1.25 mm .25 mm, broadly deltoid to subrotund, acu- or short-aristate at apex; megaspores cream-colored, ca. 1.2 mm diam., coarsely ru- gose, 1-4 per sporophyll; microspores beige, ca. 30 um long, densely papillate with blunt, slightly capitate projections, shed singly. Figures 3, 4. The Steyermark, Berry and Manara collection, cited with coordinate details, was collected from 210 TABLE 1. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Differences between Se/aginella exaltata and S. gigantea. Selaginella exaltata Selaginella gigantea Stem mainly clambering, sprawling, scandent, or re- clining Stem, in some portion, pubescent Lateral leaves 3-4 mm long Lateral leaves (in dried state at least) contiguous Lateral leaves with short and narrowly decurrent bases Axillary leaves auriculate Median leaves of ultimate and penultimate branch- systems 0.5-1 mm wide, narrow | white-margined, and more rounded on the outer Lateral ч non-articulate or ill weakly Stele basically 3-lobed or T-shaped with xylem in sev- eral patches, each patch surrounded by phloem Penultimate branch-systems lanceolate, 7-17 cm by 1.5-5 cm, 3-4 times longer than broad Penultimate branch-systems with 30—50 divisions Stem erect Stem completely glabrous Lateral leaves 4-5 mm lon Lateral leaves (dried state at least) well separated Lateral leaves with long and broadly decurrent bases Axillary leaves exauriculate Median leaves 0.2-0.5 mm wide, not white-margined, and less curved on the outer edge Lateral branches decidedly articulate Stele (at least in the main axes) with xylem in a solid strand, either X-shaped or T-shaped (see Mickel & Hellwig, 1969)... tem t oblong-ovate, 7- m by 3.5-7 cm, 1.5-2 times longer than broad Penultimate branch-systems with 15-30 divisions the type locality. Specimens have also been col- lected from the type locality by Aristeguieta and by Tamayo, but are not presently available. This very distinctive and apparently taxonom- ically isolated species is perhaps most closely re- lated to S. exaltata (Kunze) Spring, which is known from Costa Rica to Peru and western Bra- zil (Alston et al., 1981). That species differs in many respects from S. gigantea. These differ- ences are presented in Table 1 (also see Figs. 1- 4 Somers (1978), in an unpublished thesis on the articulate Selaginellas (subg. Stachygynandrum ser. Articulatae), considered Steyermark 95418 (the type) to represent a new species or possibly be of hybrid origin between S. exaltata and a partly gial dehiscence. According to Somers (1978, 1982), S. exaltata and other articulate species have a unique and more complex sporangial de- hiscence. However, S. gigantea has the simpler bivalvate condition characteristic of the nonar- ticulate species. We confirm the bivalvate nature in S. gigantea. This casts some doubt on the megasporangia, very large megaspores, beige mi- crospores, and rhizophores orignating dorsally (Somers, 1982). Even if the hybridization hypothesis could be shown to be true, the event most likely was an- cient and the characteristics of S. gigantea now so с that we believe it warrants descrip- Hybrid origin is difficult to invoke for yet р reason: S. exaltata, a most distinctive species itself and not likely to be overlooked by collectors, has never been found in Venezuela or the Guianas. c. 1986] NOTES 211 T NACIONAL DE VENEZUELA ЖА " VENEZUELA FIGURE 1. Paratype of Selaginella = (Steyermark, Berry & Manara 129691), showing a portion of the net stem with an elongated rhizophor [Vor. 73 ANNALS OF THE MISSOURI BOTANICAL GARDEN 2 n a ` vmm : E THE UNIVERSITY OF TENNESSEE , 1 involwir Solar ineplan mal: tata (Kunsa) Spring culate taxon, ger] RÀ pos Paul Somers {л ^ NL | Au лалы ——— MINISTERIO DE AGRICULTUR RA Y CRIA E Jal sht. — HERBARIO NACIONAL DE VENEZUELA no. 951 Seleginella Stem erect, fleshy, 1-2 m. tall \ ' кат ми »« Ope "we siempre verde а № largo del Rio San Ms de la Pianta lis Avion, entre Lo y tin Tos ^ altura: T30 metros UNITED STATES 549293 NATIONAL HERBARIUM FIGURE 2. Paratype of Selaginella gigantea (Steyermark, Berry & Manara 129691), showing an upper fron- dose bran 1986] NOTES 213 y y Y Чу ше. © ДШ S (que e e ge S N A Y AG к а ў by SY « P ZA j NS X к= S) [ We LLE «С | = MKS P b FIGURE 3. Selaginella gigantea.—a. Portion of lower stem with rhizophore.—b. Penultimate division of branch-system. x1. 214 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 b. S. exultat Lateral leaves from a port ion of the ultimate branch-system, ventral view.— c. $ igantea. ater leaves from a portion of the ultimate ysg -system, dorsal view. —d. S. exaltata. tale leaves from a portio of the ultimate branch-system, dorsal vie 1986] LITERATURE CITED ALSTON, A. H. G., А. C. JERMY & J. M. RANKIN. 1981. The genus Selaginella i in tropical South America. Bull. Brit. s. (Nat. Hist.), Bot. 9: 233-330. MICKEL, J. T. ni R. L. HeLLWIG. 1969. Actino-plec- "a a complex new "pus pattern in Selagi- a. Amer. Fern J. 59: 4. 78. A o survey of the Arti- culatae series of the genus Selaginella and mono- graphic treatment of the S. sulcata group (sensu str.). Unpubl. Ph.D. thesis. Univ. of Tennessee, Knoxville 982. A unique type of microsporangium in Selaginella series Articulatae. Amer. Fern J. 72: 8-92. NOTES 215 STEYERMARK, J. A. 1979. Plant refuge and dispersal centres in Venezuela: ig relict and endemic ele- ment. Pp. 185-221 in K. Larsen & L. B. Holm- Nielsen (editors), E Botany. ты Press, ondon. . 1982. Relationships of some Venezuelan for- p. 182- fication in the Tropics. Columbia Univ. Press, New York. —Julian A. Steyermark, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166; and Alan R. Smith, Department of Botany, Uni- versity of California, Berkeley, California 94720. BOURRERIA RUBRA (BORAGINACEAE), A NEW SPECIES FROM COASTAL JALISCO, MEXICO Bourreria rubra, a new species from coastal Jalisco, Mexico, is described and illustrated. The new species is distinguished by its shrubby habit, long slender pedicels, few-flowered inflores- cences, short-petiolate oblanceolate leaves, and showy red flowers. It does not appear to be closely related to any other Mexican species. Intensive collecting in the Estacion de Biologia Chamela (U.N.A.M.), Jalisco, and vicinity, and critical revision of these collections for a florula of the Station (Lott, in prep.) have resulted in the discovery of several novelties. Among these is a striking new red-flowered species of Bour- reria. Bourreria rubra E. J. Lott & J. S. Miller, sp. nov. TYPE: Mexico. Jalisco: Mpio. La Huerta, Es- tacion de Biologia Chamela, ca. m 1,200 de la Vereda Tejon, en Selva Baja Caducifolia achaparrada, con dominante de Plumeria rubra, Caesalpinia coriaria, Jatropha stand- leyi. 22 Sept. 1981 (fl), Lott 507 (holotype, МЕХО; isotypes, CAS, GH, MO, US). Fig- ure 1 rutex usque ad 3 m altus. Folia brevipetiolata, ob- lanceolata, strigillosa. Inflorescentia cymosa laxa, pe- dicellis ad 1.5-3.8 cm longis. Flores 1-7(-20); calyx campanulatus, sparse strigillosus, 5-lobatus, lobis acu- tis vel acuminatis, adaxialiter dense strigillosis; corolla rubra, late infundibuliformis, 1.4—2 cm longa; stamina exserta, filamentis in duabus tertiis partibus inferio- ribus hirsutis, alibi glabratis. Fructus globosus. Shrub to 3 m, base 3 cm diam. or more, the bark gray, the stems pale, with scattered ovate lenticels ca. 0.5 mm long, strigose to glabrate. Leaves petiolate, alternate, the internodes mostly 1-2 cm long, or the leaves crowded on short shoots; blade oblanceolate, 2-6.5 cm long, 0.8- 2.5 cm wide, the apex obtuse to acute, minutely apiculate with acumen 0.4—0.5 mm long, re curved abaxially, the base obtuse to slightly oblique, the sides more or less unequal, the mar- in entire, revolute, the upper surface scabrous and pustulate, more or less strigillose, shiny olive een, the lower surface sparsely strigillose and paler dull green, the midvein sulcate above, prominent beneath, the 4—7 pairs of secondary veins much less prominent; petiole 1-2 mm long, broadly sulcate on the adaxial surface, strigillose. Inflorescence terminal or subterminal, laxly cy- ANN. MISSOURI Bor. GARD. 73: 216-218. 1986. mose, sparsely strigose throughout; flowers 1-7 (-20), erect, bracts leaf-like, narrowly oblanceo- late, 4-10 mm long, peduncle slender, mostly 0.3-1.5 cm long; pedicels slender, mostly 1.5- 3.8 cm long, articulate at the base of the calyx, bracteoles linear to linear-lanceolate, 0.5-2 mm long, restricted to the lower one-fifth of the ped- icel; calyx valvate, subglobose, acute in bud, 8- 12 mm long at anthesis, short-stipitate, campan- ulate, with a few scattered strigillose hairs out- side, the lobes 5, triangular, acute to acuminate, 3-4 mm long, 2-3 mm wide, subequal, thickened at the tips and densely white-strigillose within with hairs ca. 0.5 mm long; corolla red, broadly funnelform, 1.4-2 cm long, the tube 9 mm long, the lobes 5, rounded, 8-9 mm long, 4—7 mm wide, glandular-pubescent on the inner surface; stamens the same number as the corolla lobes, exserted, inserted on the corolla tube ca. 4 mm above the base, the filaments hirsute at and just above the point of insertion, with hairs ca. 0.6 mm long, the upper one-third glabrous, the free portion 6-9 mm long, the anthers oblong, 2-2.5 mm long, bithecous; ovary long, glabrous; disc annular; style ca. long, bifid, the branches 3-5 mm long, stigmas capitate. Fruit bluish black, globose, ca. 8 mm diam., nutlets 4, ca. 6 mm long, ca. 4 mm wide, the outer surface rounded, deeply fluted with fine vertical ridges, the inner surface biconvex, dark reddish brown. Mice dn specimens exami ined. MEXICO. JALISCO: y; 21 Feb. 1977 (f 1) Solis Magallanes 523 кы, vs LA 1981 (ЕІ), Solis Magallanes 2759 (Е, MEXU); 2 Sept . 1981 (fl), Solis Magallanes 3103 (GH, MEXU, MO, 4 Nov. 1982 (#1), Solis — 3928 (MEXU); Las Alamandas, centro turistico a 2.5 km al W de Que- maro, a 5 km al W de la реи Puerto Vallarta- Barra de Navidad, 29 Oct. 1981 (#1, fr), Lott 683 (CHA- PA, MEXU, MO); 1 Aug. 1983 (f1, fr), Lott et al. 1712 (CAS, ENCB, MEXU, MICH, MO, US) Distribution. Thus far known only from the Estacion de Biologia Chamela and surrounding coastal area, the new species is apparently con- fined to tropical deciduous forest (Selva Baja Caducifolia) at 10-150 m elevation. Where ex- posed to constant strong coastal winds it forms part ofa thick dwarf shrubby forest with Bursera 1986] NOTES 217 Ficu fruits. B REI A, B. Bourreria rubra E. J. Lott & J. S. Miller. — A. Branch with inflorescence. — B. branchlet with oth from Lott 683. 218 instabilis, B. excelsa, E ср schlechtendalii, aesalpinia platy. loba, a ow- ers are visited by la bees. Flowering and fruiting occur from August to February. A genus in need of revision, Bourreria com- prises nearly 50 species, all from the Neotropics Airy Shaw, 1973). Perhaps 15 species occur in Mexico (Standley, 1924), but the genus is best represented in Cuba where there are probably close to 20 species. Bourreria rubra is unique in the genus in having red corollas. The majority of Bourreria species have either white- or cream- are reported to have bluish corollas (Leon Alain, 1957). Bourreria rubra bears a superficial resemblance to the white-flowered B. spathulata (Miers) Hemsl. of Guerrero but differs in having longer and more slender pedicels, larger flowers, and essentially eglandular filaments. It is unusual among Mexican species in its densely shrubb habit, which resembles more closely that of many of the Cuban species, to which it is perhaps more ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 closely allied than it is to any of the species known rom Mexico or Central America. We thank Fernando Chiang for the Latin di- agnosis and F. Chiang and David Lorence for their comments on the manuscript. Elvia Espar- za is gratefully acknowledged 1 for ner illustration, as is Arturo Solis N that aided in preparation of the drawing. LITERATURE CITED AIRY SHAW, H. K. 73. A Dictionary of Flowering Plants and Ferns, 8th edition. Cambridge Univ. Press, Cambri LEON, BR. & BR. ALAIN. 1957. Boraginaceae. /п Flora e Cuba 4: 252-278. STANDLEY, P. 24. Trees and shrubs of Mexico. "iva naceae. Contr. U.S. Natl. Herb. 23: 1216- — Emily J. Lott, Herbario Nacional, Instituto de Biologia U.N.A.M., Apdo. Postal 70-367, Dele- gacion Coyoacan, 04510 Mexico, D.F., Mexico; and James S. Miller, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166. NOTES ON DEINBOLLIA SPECIES FROM CAMEROON This note describes two new species of Dein- bollia (Sapindaceae) from the Southwest Prov- ince of Cameroon and describes the fruit of D. saligna Keay for the first time. The collections were made by the author in the vicinity of the P.A.M.O.L. plantation, Ndian, which is the type locality for several river bank shrubs with very restricted ranges, including D. saligna. The two new species are closely related and differ from u branched shrub found in deep shade in the fond understorey, whereas the other is a bushy, steno- phyllous shrub that grows on exposed, rocky riv- er banks. A collection by De Wilde (# not known) from near Kribi, Centre-Sud Province, appears to be a third undescribed unijugate Deinbollia with distinctive, cordate leaflets. Deinbollia unijuga D. W. Thomas, sp. nov. TYPE: Cameroon. Southwest Province: near Mun- demba, 4?56'N, 8?55'E. Wet forest on steep river bank, ca. 1 km from Center Last Bush on path to Meka, ca. 50 m, 31 May 1984 (fl.), Thomas 3496 (holotype, MO; isotypes, K, P, YA). Figure 1A-H. tex ad 3 m altus, plerumque simplex, maturitate taber. Folia alternata, Ре 1-4 ст longi, crassiculi, n sect ngu ares foliola 2, Muay aaa coriacea, ати 17- cm longa, 5-14 cm lata, apice acuminata, basi cu- neata ad a petiole crassi, ad Hp ongi; cos inl c bbs principales rumque late- vae, triangulares, pedicellis 1 oblonga, 2-2.5 mm longa, extra puberula; petala 5, albida, obovata, 2. 5 mm longa, 2 mm lata, marginibus A forest shrub to 3 m tall, though usually less than 1 m, usually unbranched. Leaves few, clus- tered at the branch tips, alternate, compound, glabrous when mature; petioles 1-4 cm long, stout and triangular in section, green when young, with wide, acuminate, base unequal, cuneate to ob- ANN. MISSOURI Bor. GARD. 73: 219-221. 1986. tuse, margin entire, slightly к blades yel- low-green above, paler below, mi at above, prominent and keeled below, main Mal nerves 12-18 pairs, looped close to the margin, reticu- late venation prominent on both surfaces, blade yellow-green above, paler below. Inflorescences lateral, borne among the leaves. Panicles stout, to 6 cm long, with few short lateral branches, densely pubescent with short, stout, appressed, -fl pals 5, inp broadly oblong, 2-2.5 mm long, densely pubescent on the surfaces exposed in bud. Petals 5, white, obovate, 2.5 mm long, 2 mm wide, ciliate with a fringed lip within. Stamens 8-12, surrounded by an annulus, filaments pi- lose, 0.8 mm long, anthers 0.8 mm long. Ovary sparsely pubescent, of 2-3 carpels united at the docarp not woody; embryos large and green, usu- ally solitary. Additional material examined. CAMEROON. SOUTHWEST PROVINCE: type locality, 10 July 1983 (fl. & imm. fr.), Thomas 2208; 5 km S of Ilor on Ekondo- Titi-Mundemba Road, forest on E side of road, 4°48'N, 8°54'Е, 50 m, 20 Nov. 1983 (ripe fr.), Thomas 2552; 1 km S of Ekumbako, along forest footpath, 4°53’N, 8*53'E, 100 m, (sterile), Thomas 2717 Deinbollia angustifolia D. W. Thomas, sp. nov. TYPE: Cameroon. Southwest Province: near Mundemba, 4°56'N, 8°52’E. Rocky bank of Idu River at Bulu on path to Ekumbako, 10 m, 7 Mar. 1984 (fl. & fr), Thomas 3253 (holotype, MO; isotypes, K, P, YA). Figure Affinis D. unijuga D. W. Thomas. Frutex ad 1 m altus, ramus. Petioli 0.6-1.5 cm longi. Foliola 2, op- posita, lineari-lanceolata, 10—30 cm longa, 0.8-2. 2 cm acuminata, basi cuneata. Paniculae sparse e, ngae, plerumque en Sepala vix puberula. Fructus circiter 1.5 cm dia This species is closely related to the preceding species and differs in the following characters: a much branched shrub of frequently inundated rocky river banks, to 1 m tall; petioles 0.6-1.5 cm long; leaflets linear-lanceolate, 10—30 cm long, 0.8-2.2 cm broad, base cuneate, panicles usually terminal, sparsely pubescent, to 12 cm long; se- 220 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ў @ Щи, p NW SN ИШ 53 v d “8. A NEST JF y — 6,L | —1 B.C,D.E Н.К | | 4 cm 2 cm 1 cm 0.25 cm FIGURE 1. A-H. Deinbollia unijuga (Thomas 2208). — A. Branch, leaves, and inflorescence. — B. Flower. — C. Flower with two sepals and petals removed. — D, E. Sepals. — Е. Petal interior. – С. Leaf margin. —H. floral diagram. I-L. D. angustifolia. (Thomas 3497 except J)—I. Branch, leaves, and inflorescence. — J. Fruit (Thomas 3253).—K. Flower buds. — L. Leaf underside. pals sparsely pubescent on the surfaces exposed Deinbollia saligna Keay in bud; ripe fruits ca. 1.5 cm diam. Ripe fruits were collected from the type lo- cality (banks of the Ndian River, less than 30 m Additional material exami CAMEROON. .), elev.) by the author (Thomas 2205, 10 July 1983). ned. SOUTHWEST PROVINCE: type locality, 1 June 1984 (fl Thomas 3497. 1986] The mature fruit usually consists of a single mer- icarp with one or two aborted carpels at the base. Occasionally a second mericarp is present. The sepals are persistent but not enlarged. The mer- glabrous, smooth, orange-col- ored and 1 cm in diameter. The mesocarp is yellow and firm-textured. Each mericarp con- tains one (rarely two) green embryos, which ap- parently begin germination on the parent plant. NOTES 221 I would like to thank J. Dwyer, Т. В. Croat, and R. Letouzey for comments on the manu- script, and M. Bucy for the illustration. Field work in Cameroon was supported by the Na- tional Geographic Society, with permission from M.E.S.R.E.S. Yaoundé. — Duncan W. Thomas, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166. DATES OF PUBLICATION [VOLUMES 1 (1914)-72 (1985)] CHERYL R. BAUER (COMPILER) Listed below are the actual publication dates of the individual numbers in each volume of the Annals of the Missouri Botanical Garden. The publication date of each number was published in a subsequent number; those with question marks never have been printed in the journal itself, and at this time are not known. The entries followed with a date in brackets are ones where our records differ from the date we actually pub- lished as having been correct. In these cases, the date in brackets should be considered correct. 1 (1) 1-156. 1914. 2 (2) 157-262. 1 July 1914 (3) 263-356. 30 Sept. 1914 (4) 357—432. 30 Jan. 1915 2 (1-2) 1-402. 1915. 17 May 1915 (3) 403—658. 8 Oct. 1915 (4) 659-841. 20 Dec. 1915 3 (1) 1-200. 1916. 7 July 1916 (2) 201-308. 30 Sept. 1916 (3) 309-402. 4 Nov. 1916 (4) 403-517. 12 Jan. 1917 4 (1) 1-92. 1917. 9 Mar. 1917 (2) 93-232. ??? (3) 233—288. 20 Sept. 1917 (4) 289-368. 8 Dec. 1917 5 (1) 1-108. 1918. 3 Apr. 1918 (2) 109-176 24 May 1918 (3) 177-224. 20 Sept. 1918 (4) 225-377. 23 Dec. 1918 6 (1) 1-92. 1919. ?? (2) 93-170. 77? (3) 171-252. 11 Oct. 1919 (4) 253-320. 2 Mar. 1920 7 (1) 1 Eon 1920. mM (2-3) 81-248. 8 Dec. 1920 (4) 249-333. ? P 8 (1) 1-96. 1921. M (2) 97-236. 9 Jan. 1922 (3) 237-342. ??? (4) 343-400. 6 July 1922 9 (1) 1-78. 1922. 31 July 1922 (2) 79-232. 27 Nov. 1922 (3) 233-332. 16 Feb. 1923 (4) 333-442. ??? 10 (1) 1-110. 1923. 28 Sept. 1923 (2) 111-190. 12 Oct. 1923 (3) 191-298. d (4) 299-427. 7?? 11 (1) 1-98. 1924. 25 July 1924 (2-3) 99—388. 5 Jan. 1925 (4) 389—465. 22 May 1925 ANN. Missouni Bor. GARD. 73: 222-224. 1986. 12 — W 1-132. 133-212. 213-358. 359-424. 1-100. 101-172. 173-354. 355-489. 1-86. 87-210. 211-358. 359-436. 1-112. 113-240. 241-332. 333-437. 1-128. 129-226. 227-388. 389-519. 1-212. 213-476. 489-612. 1-176. 177-374. 375-538. 569-817. 1-230. 231-432. 433-608. 609-719. 795-853. 1925. 1926. 1927. 1928. 1929. 1930. 1931. 1932. 1933. 1934. 1935. 1936. 1937. 1938. 28 Sept. 1925 18 Feb. 1926 999 8 Мау 1926 77? 20 Sept. 1926 8 June 1927 8 Oct. 1927 999 30 Арг. 1928 22 Dec. 1928 13 Маг. 1929 99) 30 Dec. 1929 14 May 1930 27 Dec. 1930 30 June 1931 28 Oct. 1931 24 Dec. 1931 15 Apr. 1932 23 July 1932 15 Nov. 1932 29 Apr. 1933 10 July 1933 6 Oct. 1933 19 Dec. 1933 12 Apr. 1934 5 June 1934 20 Sept. 1934 12 Dec. 1934 25 Mar. 1935 25 May 1935 30 Sept. 1935 20 Mar. 1936 10 June 1936 1 Sept. 1936 8 Jan. 1937 20 Mar. 1937 30 Apr. 1937 27 Sept. 1937 23 Nov. 1937 27 Dec. 1937 6 May 1938 28 Sept. 1938 28 Nov. 1938 1986] 165-256. 257-433. 1-118. 119-258. 259-370. 371-606. 281-404. 207-376. 255-353. 1-62. 63-182. 1939. 1940. 1941. 1942. 1943. 1944. 1945. 1946. 1947. 1948. 1949. 1950. 1951. 1952. 1953. BAUER-— PUBLICATION DATES ?m 29 Apr. 1939 79 30 Nov. 1939 29 Feb. 1940 10 May 1940 25 Sept. 1940 10 Dec. 1940 26 Feb. 1941 28 Apr. 1941 20 Sept. 1941 27 Nov. 1941 18 Feb. 1942 20 Apr. 1942 18 Sept. 1942 18 Dec. 1942 22 Mar. 1943 15 June 1943 30 Sept. 1943 20 Nov. 1943 31 Mar. 1944 30 Sept. 1944 30 Nov. 1944 28 Feb. 1945 999 15 Sept. 1945 30 Nov. 1945 8 Mar. 1946 30 Apr. 1946 7 Dec. 1946 29 Mar. 1947 ??9 31 Oct. 1947 97 22 Mar. 1948 22 June 1948 20 Sept. 1948 21 Mar. 1949 30 Nov. 1949 31 Mar. 1950 15 Dec. 1950 22 Mar. 1951 183-258. 259-412. 1-212. 213-262. 263-350. 351-424. 1-102. 103-194. 195-302. 303-418. 355-399. 1-124. 125-194. 195-270. 271-360. 1-92. 93-202. 203-304. 305-354. 1-194. 195-256. 257-363. 1-80. 81-204. 205-262. 263-372. 275-349. 1-136. 137-259. 487-604. 1-114. 115-264. 265-388. 1-94. 95-200. 201-421. 1-80. 81-170. 171-402. 135-264. 265-388. 1954. 1955. 1956. 1957. 1958. 1959. 1960. 1961. 1962. 1963. 1964. 1965. 1966. 1967. 1968. 1969. 1970. 223 30 Sept. 1953 18 Dec. 1953 22 June 1954 999 24 Mar. 1955 23 June 1955 17 Nov. 1955 ??? 15 Арг. 1956 19 July 1956 31 Oct. 1956 22 Dec. 1956 14 Mar. 1957 29 Jan. 1958 20 Mar. 1958 27 May 1958 28 Oct. 1958 6 Jan. 1959 17 June 1959 15 Oct. 1959 28 Jan. 1960 31 Mar. 1960 5 Aug. 1960 27% 7 Feb. 1961 3 Apr. 1961 222 28 Dec. 1961 19 Apr. 1962 27 May 1963 31 Jan. 1964 30 Nov. 1964 31 Mar. 1965 27 July 1965 15 Oct. 1965 5 Jan. 1966 6 May 1966 7 Nov. 1966 30 Dec. 1966 12 June 1967 27 Oct. 1967 11 Mar. 1968 9 Sept. 1968 22 Oct. 1968 30 Apr. 1969 13 Oct. 1969 18 Nov. 1969 17 July 1970 2 Dec. 1970 16 Feb. 1971 14 June 1971 224 (1) 1-98 (2) 99- (3) 267-369 (1) 1-104 (2) 105-322 (3) 323-478 (1) 1-168 (2) 169-572 (3) 573-977 (1) 1-262 (2) 263-538 (3) 539-907 (1) 1 (1) 1-208 (2) 209-384 (3) 385-656 (3) 381-656 (4) 657-748 (1) 1-366. (2) 367-782 (3) 783-998. (4) 999-1258. ANNALS OF THE MISSOURI BOTANICAL GARDEN 1971. 1972. 1973. 1974. 1975. 1976. 1977. 1978. 14 July 1971 21 Jan. 1972 25 May 1972 27 July 1972 28 Mar. 1973 15 May 1973 22 Sept. 1973 19 Dec. 1973 3 July 1974 8 Aug. 1974 11 Oct. 1974 24 Dec. 1974 [19 J P 1976] 17 Aug. 1976 7 Sept. 1976 18 Jan. 1977 23 Mar. 1977 14 June 1977 26 July 1977 2 Feb. 1978 26 May 1978 14 June 1978 12 Oct. 1978 1 Feb. 1979 23 Mar. 1979 13 Aug. 1979 66 67 68 71 72 ERRATA 591—903. 1-256. 257-522. 523-818. 819-1059. 1-238. 571-149. 1-346. 347-630. 631-986. 987-1187. 591-880. The acronym MAD is used erroneously for TAN in the paper titled “A Monograph of the Monimiaceae (Laurales) in the Malagasy Region (Southwest Indian Ocean),” by David H. Lorence [72(1): 1-165. 1985]. 1979. 1980. 1981. 1982. 1983. 1984. 1985. [Vor. 73 13 Aug. 1979 [29 Aug. 1979] 10 Oct. 1979 15 Jan. 1980 22 Feb. 1980 12 Mar. 1980 2 May 1980 299 [4 Mar. 1981] 29 June 1981 17 Nov. 1981 29 Mar. 1982 15 Dec. 1982 [16 Dec. 1982] 5 Jan. 1983 3 June 1983 8 Aug. 1983 14 Nov. 1983 4 Apr. 1984 17 July 1984 9 Oct. 1984 31 Dec. 1984 19 Mar. 1985 19 Apr. 1985 14 May 1985 10 June 1985 Volume 72, No. 4, pp. 591-880 of the ANNALS OF THE MISSOURI BOTANICAL GARDEN, was published 5. on 5 November 198 INFORMATION FOR AUTHORS | The ANNALS publishes original manuscripts | insy Vittatic botany and related к poes ) п order | to expedite editing and publication. Minus on not prepared properly may be returned for re- . vision prior to review. If an author feels that his manuscript presents special problems, he should write the editor concerning the best way to han- dle these before submitting the manuscript. Page costs are $60.00 per page; charges may be re- duced under special circumstances, and the ed- itor will help authors to seek additional funding if necessary. Corrections on manuscripts in the proof stage will cost $2.50 per line and are not | subject to negotiation. Manuscripts must be typewritten on one side of good quality (not erasable) 8%” x 11" paper. The manuscript should have at least one inch margin all around and be double-spaced through- out, including the abstract, notes, legends, ta- bles, lists of specimens, and literature cited. Fig- ure legends and tables should be typed separately and placed at the end of the text. Legends for cach group of figures should be t Paragraph (consecutive paragraphs on o more pages separate from figures themselves), using the style found in the current issue of the as one ANNALS. Authors should indicate in the margins * approximate places for illustrations and ta- les. Include author's name and page number in ZR ic copies of the manuscript is required, and the Mise should retain a copy of the final, m | Е sion of the original and two carbon or xerograph- Acknnulaj iy right-hand corner of each page. Submis- Aes author's full mailing address should е second footnote. These should be he | CE used in the manuscript. = An abstract | ег than “Notes.” Th e abstra Summarize the sions of the р; у b чу. illustrators, and technical assistants may | Conveniently placed as a footnote on page k i must accompany each paper oth- ; tract : ould s tly style found in other ANNALS issues, e.g., in Lor- ence, 1985 (72: 1-166). Keys should be dichot- omous, indented, with numbered co i gensen 865 (BM, G, к vere se herbarium ac- ronyms as listed in Ind: Veg. 106). Use parser abbreviations as listed in Draft Index of Author Abbreviations compiled | at the Herbarium, Royal Botanic илем, Kew |. (1980). tency and to make sure they are unambiguous. Periods are used after all abbreviations except measures, compass directions, and herbarium designations. Send copies of illustrations; origi- nals will be requested after the manuscript is accepted. All illustrative — should be mounted on a printed illustration is e x 8". Presize il- lustrations to fit either column width (2%” or ca. 68 mm) or full page width (542” or ca. 140 mm). Figures are numbered consecutively with Arabic numbers, in the order they are referred to in the text. Photographs should be sharp, glossy, black- and-white prints. Abut edges of photographs in composite plates. Do not mix line copy and con- tinuous-tone illustrations | on one plate. Maps should include a metric scale and reference to - latitude and longitude; where appropriate, size and scale should be included in photographs or dra , not in figure legends. All original il- jostrations will be returned after publication. The bibliography should be compiled with care. The author should check to make sure each entry in the bibliography is referred to in the text, each reference toa paper in the text is entered e. Publisher. Pics of aspects. of style, consult a recent ALS; ‚ The Chicago Manual of Abbreviations should be checked ie consis- m Notes on Deinbollia Species from Sate Duncan W. Thome Dates of Publication Cheryl R. Bauer EE Contents continued from front cover Catalog of the Mosses of China Paul Г. Redfearn & P.-C. Wu ........ NOTES A Remarkable New Selaginella from Venezuela Julian A. Steyer & Alan В. Smith ............... Bourreria rubra (Boraginaceae), a New Species from Consta Jalisco, Emily J. Lott & James S. Miller „m ERRATA SOUR] BOTANICAL GARDEN LUME 73 1986 NUMBER 2 CONTENTS Phylogeny of the Hamamelidae: An Introduction David L. Dilcher & d Michael S. Zavada E Commentary on the Status of the Hamamelidae Arthur Cronquist 22 Evolution of the Fagaceae: The Implications of Foliar Features Jay ay H. Jones n 228 Fossil Evidence Regarding the Evolution of Nothofagus Blume Ec- 76 gardo J. Romero 276 Evolution and Reproductive Biology of Inflorescences 1n Lithocarpus, > Y ; zs um CE aram obert 20 Castanopsis, Castanea, and Quercus (Fagaceae) v Floral Structure, Systematics, and Phylogeny in Trochodendral Peter K. Endress The Floral Morphology and Vascular Anatomy ofthe Hamamelidaceae: ise Subfamily Liquidambaroideae {. Linn Bogie А SA 'omparative Pollen Morphology and Its Relationship to Phylogeny 5 Pollen in the Hamamelidae Michael S. Zavada & David L. C | nl 348 Dilcher „+ | | "11 roo in Extant and Morphology and Development of Pistillate Inflorescences in Extant ano — : E x - . zd Г "пир Ar Path 1 Sto key 382 Fossil Cercidiphyllaceae Peter В. Crane & Ruth A. Stocke. " te enntinuerl on back cove | Contents continued on раск c Fruit and Seed Dispersal and the Evolution of the Hamametidae t м. > r] „Уг ж i : Bruce H. Tiffney p OF THE j MISSOURI BOTANICAL GARDEN The ANNALS, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be ac- EDITORIAL COMMITTEE Nancy Morin, Editor Missouri Botanical Garden CHERYL R. BAUER, Editorial Assistant Missouri Botanical Garden Kathy Ityes, Editorial Assistant Missouri Botanical Garden MARSHALL R. CROSBY Missouri Botanical Garden Gerrit DAVIDSE Missouri Botanical Garden 5 - v Jonn D. DY ER mu Missouri Botanical Garden & Saint Louis University — PETER GoLpBLATT Missouri Botanical Garden. ANNALS MISSOURI BOTANICAL GARDEN VOLUME 73 1986 NUMBER 2 PHYLOGENY OF THE HAMAMELIDAE: AN INTRODUCTION The collection of papers published in this issue of the Annals of the Missouri Botanical Garden are primarily those papers delivered at a sym- posium, Phylogeny of the Hamamelidae, pre- sented at the annual meeting of the Botanical Society of America, August 5-9, 1984, in Fort Collins, Colorado The Superorder Amentiferae presented by Takhtajan in 1959 and later recognized as the subclass Hamamelididae by him in 1969 was adopted by Cronquist (1981) in his presentation of the subclass Hamamelidae. The roots of this classification lie in the early recognition by nine- teenth century systematic botanists that these plants share many similar features and may be considered a natural group (Engler, 1886; Stern, 1973). Stern (1973) presented detailed docu- mentation of the history of the ““Amentiferous Concept." The papers published in this sympo- sium volume present an a of several as- pects of this group of plan With a few exceptions en 1973, 1983), most researchers recognize the Hamamelidae as a grouping of closely-related plants. The ques- tions that exist concerning the phylogeny of the Hamamelidae are: Can this group of plants still be considered a natural group if examined in detail by new methods and techniques? What can detailed analysis of particular orders or families tell us about the relationships within these taxa and between them and related orders or families? Can we develop a clear understanding of prim- itive versus advanced character states in the Hamamelidae? The papers presented in this symposium volume address many of these ques- ons. ANN. MISSOURI Bor. GARD. 73: 225-226. 1986. It is clear that some taxa of the Hamamelidae are very ancient. The subclass can be recognized back into the Lower Cretaceous (Upchurch, 1984; Dilcher & Eriksen, 1983). Also other taxa, such as the Chloranthaceae, that were once considered to be very closely related to this group (Engler, 1886), have been reported from the Lower Cre- taceous based upon pollen data and leaf data (Walker & Walker, 1984; Upchurch, 1984). The very early occurrence of taxa that bear so-called reduced simple flowers (Dilcher, 1979; Dilcher & Eriksen, 1983), as illustrated by platanoid fos- sil flowers and fruits from the Lower Cretaceous, raises a basic question about the extent of floral elaborations versus floral reductions as impor- tant factors in floral evolution. It n dd that at least some very early d very small flowers. This may have been much more com- mon than was previously thought. Certainly, flo- ral elaborations as well as reductions of floral parts played a significant role in evolution of the various taxa of the Hamamelidae. Many ofthe taxa in this subclass are effectively wind pollinated, some accommodate both ane- mophily and entomophily, and a few taxa are exclusively entomophilous (Faegri & van der Pijl, 1978). Pollination by wind has changed very lit- tle, if at all, since the Cretaceous and probably the insect vectors of the pollen in the Hama- melidae have changed very little as well. There- fore, in the absence of pressures to change their reproductive structures, the Hamamelidae prob- ably represent one group of flowering plants that has changed very little since the end of the Cre- taceous with the exceptions of extinctions of taxa and radiations resulting from animal dispersal 226 of fruits and seeds (Tiffney, 1984). This is why the Hamamelidae are such an excellent group to use in testing hypotheses of angiosperm evolu- tion. In addition, because of the abundant fossil representation of the group and the availability of many of the extant taxa in the populated areas of the Northern and Southern Hemispheres, it should continue to yield useful data on angio- sperm evolution and phylogeny. LITERATURE CITED CRONQUIST, А. 1981. An Integrated System of Clas- ification of Flowering Plants. Columbia Univ. Press, New York. DiLCHER, D. L. 1979. Early angiosperm reproduc- tion: an ТОРДЫ report. Rev. Palaeobot. Pal- ynol. 27: camores are ancient o Quart. 9(2): 8-9. n konigiglich bo- sd cd; VAN . 1978. The Principles of Pollination Ecology, 3rd edition. Oxford Univ. Press, Ox m. STERN, W. L. 1973. Development ofthe amentiferous concept. BE 25: 316-333. TAKHTAJAN, A. 1959. Die Evolution der Angiosper- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 men. iniu Fischer Verlag, Jena. [Translated by W. Hoppne 1969. Flow wering Plants: Origin and Dispers- al. ‘Oliver and Boyd, Edinburgh. [Translated by C. Jeffrey.] THORNE, R F. 1973. The “Amentiferae” or Hama- melidae as an artificial group: a summary state- ment. Brittonia 25: 395-405. 1983. Proposed realignments in the angio- sperm s. Nordic J. Bot. 3: 85-117. TIFFNEY, B. H. 1984. Seed size, dispersal syndromes, and the origin of the angiosperms: evidence an буро. Ann. Missouri Bot. Gard. 71: 551— 576. UPCHURCH, G. R., JR. 1984 [1985]. Cuticle evolution in Early Cretaceous angiosperms from the Poto- mac Group of Virginia and Maryland. Ann. Mis- souri Bot. Gard. 71: 522-550. WALKER, J. W. & A. G. WALKER. 1984 [1985]. UI- trastructure of Lower Cretaceous angiosperm pol- len and the origin and early evolution of flowering plants. Ann. Missouri Bot. Gard. 71: 464-521 — David Г. Dilcher and Michael Zavada*, Department of Biology, Indiana University, Bloomington, Indiana 47405. * Current address: Department of Botany, University of the Wit- watersrand, 1 Jan Smuts Avenue, Johannesburg, 2001 South Africa. COMMENTARY ON THE STATUS OF THE HAMAMELIDAE ARTHUR CRONQUIST! Some years ago there was a symposium at the AIBS meetings in Amherst, called “What ever happened to the Amentiferae?" My joking an- swer at the time was, “They are alive and well under the name Hamamelidae.” A few of En- gler’s families, notably the Salicaceae, have been pruned off, but the bulk of the group remains together as Engler had it. Most of us now believe that the group has reduced rather than primi- tively simple flowers, and therefore a basal po- sition within the dicots is no longer supported. Still, Engler would feel quite at home with the Hamamelidae as now conceived. The frequent reference in papers of the present symposium to the “ makes me a bit uneasy. I stand by the interpre- tation I presented in 1981, only slightly modified from that of 1968, but I cannot take too much credit for it. The phyletic diagram at the back of Takhtajan's German-language book of 1959 ved from the Magnoliales. He formally id and named the Hamamelidae in his Russian- language book of 1966. I well remember his pointing out to me (in 1965) the phyletic signif- icance of the distally unsealed carpel in Platanus. = дә) 5 jo] O 4 “a 5 Lnd x O Е O [21 O 3 = “A < 3 jo] о @ [=] 3 б -n Ф Q e thought regarding the Hamamelidae. I can take some comfort in the degree to which the data and their interpretation are compatible with my 1981 scheme. It is interesting that Zavada and Dilcher (1986) consider that the pollen mor- phology supports inclusion of the Daphniphyl- lales, Juglandales, Leitneriales, and Urticales in the Hamamelidae. The position of all of these has been questioned (and sometimes vigorously debated) by other authors. Leitneria, in partic- ular, has recently been associated on serological grounds with families ofthe Sapindales (Petersen & Fairbrothers, 1983 I take note of Giannasi's (1986) reference to the effort by Kubitzki and Gottlieb (1984) to present a general scheme for the evolution of secondary metabolites in the angiosperms. He is quite right that if that scheme is accepted, then the status of the Hamamelidae as derived from the Magnoliidae must be questioned. I intend to present a different scheme for secondary metab- olites in a forthcoming book. LITERATURE CITED — A. 1968. The Evolution and Classifica- of Flowering Plants. Houghton Mifflin, Bos- a “1981. An Integrated System of Classification of. Flowering Plants. Columbia Univ. Press, New York. GIANNASI, D. E. 1986. Phytochemical aspects of phy- openi in Hamamelidae. Ann. Missouri Bot. Gard. 73: 417-437. KUBITZK1, го p В. GOTTLIEB. 1984. Phytochem- ical aspects of angiosperm origin and evolution. Acta Bot. Neerl. 33: 457-468. PETERSEN, F. P. & D. E. PAIRBROTHERS, 1983. dnd neria —two Le at taxa of Rutiflorae FA idae). Syst. Bot. 8: 134-148. TAKHTAJAN, A. L. 1959. Die Evolution der Angio- spermen. Gustav Fischer Verlag, Jena. ; 1966. Sistema i filogenia tsvetkovykh ras- tenii. Soviet Sciences Press (Nauka), Moscow and ZAVADA, M. S. & D. L. DILCHER. 1986. Comparative pollen morphology and its MARR to phylog- eny of pollen in the кү melidae. Ann. Missouri Bot. Gard. 73: 348- ! Herbarium, New York Botanical Garden, Bronx, New York 10458. ANN. MISSOURI Bor. GARD. 73: 227. 1986. EVOLUTION OF THE FAGACEAE: THE IMPLICATIONS OF FOLIAR FEATURES! JAY H. JONES? ABSTRACT The leaf architecture and cuticular e of extant брае һауе been examined and com- pared to orms. The various lea t Fagaceae appear at different times in the fossil nl The most recent nno to evolve are typical lobed oak forms (Oligocene-Miocene) whereas coarsely toothed oaks and regularly craspedodromous chestnut- like leaves are found as early as the Eocene. There are n > = E O [e] L we e © 3 po] = £m T O + о Fagaceae іп pre-Eocene sediments but es presence of well defined vegetative and reproductive ma- terials of two distinct ern milies in the Eocene suggests that this family arose in or befo the Paleocene s generally comparable to those of the Faga r in the Paleocene a common ori r gus leaves are distinct from е | all othe er Fagaceae suggesting that this taxon should be considered a separate family as propo nova and Nixon on other grounds. Nothofagaceae also seem to be closely related to the lee and probably arose from the same fagalean stock. Suggestions that modern lobed oaks are atavistic expressions of primitive palmately lobed-compound forms, variously assigned to Debeya, Dewalquea, and уре ve although possibly correct, are poorly sup- ported by foliar data. Foliar features also fail to provide convincing support for the origin of the Fagaceae from Cretaceous platanoids. The Fagaceae comprise a large and important family of trees and shrubs (Lawrence, 1951), yet we know little about their evolution. The fossil record of the family reportedly extends back to records been questioned in recent years (Wolfe, 1973). Nineteenth and early twentieth century reports are especially questionable be- cause they usually were based on the superficial resemblance of fossil materials, most commonly leaves, with those of extant plants (Dilcher, 1973, 1974). Because any complete evolutionary study must include information from both fossil and modern organisms (Simpson, 1953; Wolfe, 1973), it is important to study fossil materials in suffi- cient detail to assure accurate taxonomic place- ment. This requires not only a detailed study of the fossil material but also a detailed knowledge ofthe nature and variation ofthe same characters in similar modern plants. Much work has been done on modern Fagaceae (Trelease, 1924; Ca- mus, 1928-1929, 1934-1954; Brett, 1964; For- man, 1966a; Elias, 1971; Soepadmo, 1972; and others), yet even Camus, in her monumental monographs of the chestnuts and oaks (1928- 1929; 1934-1954), did not provide much of the information on foliar characteristics necessary for the proper interpretation of fossil leaves. A primary objective of the research upon which this paper is based (Jones, 1984) was to examine thoroughly the foliar characteristics of extant Fa- gaceae using modern methods of leaf architec- tural and cuticular analyses (Stace, 1965; Hickey, 1973,1977, 1979; Dilcher, 1974; Hickey & Wolfe, 1975). Information obtained from this study (Jones, 1984) has provided a basis for the inter- pretation of putatively fagaceous fossil leaves as well as useful information for testing classifica- tion schemes of modern Fagaceae, which are based primarily on nonfoliar features. The goal of this paper is to integrate newly acquired foliar information with other vegeta- tive, reproductive, and fossil information in or- der to provide a view of the evolution of and ! This work is part of a doctoral к submitted to the Indiana University Graduate School. The help and comments of my advisory committee Lane, and Garland Upchur ch for valuable discussions and constructive criticism of the manuscript. Financial support in the form of associate instructorships, fellowships, a grant-in-aid of research (Indiana University), a faculty travel grant (Ripon College), a grant from the Kentucky Academy of Science Foundation for Botanical Research, and funds from N.S.F. Grants 79-00898 and BSR 85-16657 to David L. Dilcher is also gratefully acknowledged. Special thanks go to Karl Longstreth and Melinda Jones for their assistance in many aspects of this project. ? Indiana University, Bloomington, Indiana 47405. Present address: ARCO Resources Technology, 2300 West Plano Parkway, Plano, Texas 75075. ANN. MISSOURI Bor. GARD. 73: 228-275. 1986. 1986] within the Fagaceae. Past taxonomic treatments of the Fagaceae are reviewed first, followed by a discussion of the classification and evolutionary history of each genus with emphasis on the con- tribution of foliar information. Discussion of re- productive aspects, although taken into consid- eration, has been kept to a minimum because these have been reviewed in great detail else- where (Hjelmqvist, 1948; Brett, 1964; Forman, 1964a, 1966a, 1966b; Soepadmo, 1968a, 1968b, 1970; Abbe, 1974; Fey & Endress, 1983; Kaul & Abbe, 1984). The final section of this paper provides an assessment of the contribution of foliar information to classification, primarily above the generic level, and to understanding of the origin and evolution of the family within the Amentiferae. Keys for confirming fagaceous af- iCal d generic affinities within the Fagaceae are found in Appendix I and Appendix II respectively. П- lustrations and descriptions of trichome types found in this family are presented in Appendix TAXONOMIC REVIEW OF THE FAGACEAE The constituent genera of the Fagaceae have been surveyed in numerous works. From the time of Ray (1682, 1703) members ofthe genera Quer- cus, Fagus, and Castanea have been grouped with other amentiferous plants (Stern, 1973). De Jus- sieu (1789) formalized the concept of the Amen- tiferae and included the above genera in the "order" Amentaceae. Dumortier in 1829 was al- legedly the first to recognize the family (Soepad- mo, 1972), and de Candolle (1868) was the first to circumscribe the family in its present form. He applied the name Cupuliferae, which has been used since in numerous ways by various authors. Oersted (1871) subdivided the Cupuliferae, used here at the familial level, into three subfamilies along lines approximating those drawn by recent students of this family (e.g., Forman, 1966a). Bentham and Hooker (1880) treated the Cupu- liferae as an “order” with three tribes. The cir- cumscription of the tribe Quercineae conforms to that of the Fagaceae of modern authors. How- ever, Bentham and Hooker did not subdivide this tribe further, above the generic level, and thus their scheme does not provide taxa corre- sponding to modern subfamilies. The name Fa- gaceae was first used by Prantl (1894), who ab- sorbed the ““subfamily”” Quercineae, sensu Oersted, into the Castaneae and reduced the number of genera in the family from seven to JONES—FAGACEAE EVOLUTION 229 five. Brett (1964) retained Prantl's subfamilial classification but recognized, at least tacitly, nine genera. Schwarz (1936a) appears to be the most extreme "splitter" among western botanists. He recognized 11 genera distributed among the tra- ditional three subfamilies. With the possible ex- ception of Schwarz, modern treatments of the Fagaceae are remarkably uniform in spite of sig- nificant intrageneric variability and the remark- able similarity between some elements of differ- ent genera. The suprageneric classifications of Forman (19662), Hutchinson (1967), and Elias (1971) are identical. Melchior's (1964) system differs in that the genus Trigonobalanus is placed in the Fagoideae rather than the Quercoideae. Lozano-C. et al. (1979), in a report of a third species of Trigonobalanus, established a fourth subfamily, the Trigonobalanoideae, to accom- modate this problematic genus. Most recently Smiley and Huggins (1981) adopted Forman's (19662) classification with the addition of the subfamily Pseudofagineae to house the fossil ge- nus Pseudofagus. The analysis of another fossil form (Manchester & Crane, 1983), Fagopsis, pre- sented an equally good case for the erection of a new subfamily if not a new family. However, proposed by Kuprianova (1962), also seems to be gaining support (Nixon, 1982), but no com- prehensive scheme of classification containing this innovation has been published. A summary of the more important classification schemes, to- gether with a proposed classification based on the integration of foliar information, is found in Table 1 GENERIC ANALYSES FAGUS L. The genus Fagus exhibits very little morpho- logical variation. This is not only true of leaves (Jones, 1984) but also extends to reproductive features (see, for example, Praglowski, 1982) and nonfoliar vegetative features such as wood (Shi- maji, 1962). The homogeneity of this genus has apparently made it an unattractive candidate for monography. It is the only genus of Fagaceae for which there is no recent comprehensive mono- graph. The most complete recent works dealing with Fagus are the palynological studies of Hanks and Fairbrothers (1976) and Praglowski (1982). No formal subgeneric groups of Fagus species have been established in these or other works. 230 ANNALS OF THE MISSOURI BOTANICAL GARDEN TABLE 1. Taxonomic treatment of fagaceous plants various authors. Inclusions noted are relative to the classification of Forman (1966a) and others. de Candolle 1868 Family Cupulifera Fagus (incl. some CODES Noth dead Castane a (incl. Chrysolepis) Quercus (incl. Lithocarpus) Oersted 1871 Family Cupuliferae is mily Fagineae Fagus Nothofagus Subfamily Castaninae astanea (incl. Castanopsis & Chrysolepis) Pasania Cyclobalanus (incl. Lithocarpus) Subfamily Quercineae Quercus Cyclobalanopsis Prantl 1894 Family Fagaceae “Subfamily” Fageae Fagus Nothofagus “Subfamily” Castane Castanea (incl. Canes & Chrysolepis) Pasania Quercus Schwarz 1936a Family “Cupuliferae” Subfamily Fagoideae Fi agus Nothofagus Subfamily Castaneoideae Tribe Castaneae astanea Castanopsis (incl. Chrysolepis) Tribe Pasanieae Pasania Cyclobalanus Lithocarpus pce) о Тя Sici daa C Nori Erythrobalanus Tribe Querceae Macrobalanus Quercus TABLE l. Continued. Brett 1964 Family Fagacea еш Fagoideae Fagus Nothofagus Subfamily Castaneoideae Trib eae Tribe с dd Lithoca Tribe uoc Pasania Tribe Querceae Quercus Cyclobalanopsis Melchior 1964 Family Fagac Subfamily кыша Fagus Nothofagus Trigonobalanus Subfamily Castaneoideae Tribe Castaneae Castanea Castanopsis (incl. Chrysolepis) Tribe Pasanieae asania Lithocarpus Subfamily Quercoideae Quercus Hutchinson 1967 Elias 1971 Family Fagaceae Subfamily Fagoideae Fa, gus Nothofagus Subfamily Castaneoideae astanea Castanopsis Chrysolepis Lithocarpus (Pasania) Subfamily Quercoideae uercus Trigonobalanus Soepadmo 1972 Forman 1966a Family Fagaceae Subfamily Fagoideae Fa gus Nothofagus 1986] TABLE 1. Continued. Subfamily Castaneoideae astanea Ln. (incl. Chrysolepis) Lithoc Шс re Trigonobalanus Lozano-C. et al. 1979 Family Fagaceae ien Fagoideae Fagus Nothofagus Subfamily Castarieoideae Castanea Subfamily Quercoideae Quercus Subfamily Trigonobalanoideae Trigonobalanus Smiley and — 1981 Family Fagac е Fagoidese NU MS Subfamily Pseudofagineae Pseudofagus Subfamily Castaneoideae Subfamily Quercoideae uercus Trigonobalanus Present Study Family Nothofagaceae Nothofagus Family Fagaceae Subfamily Fagoideae Fagus Subfamily Castaneoideae Lithocarpus Subfamily Trigonobalanoideae bal, Trigonobalanus Subfamily Quercoideae uercus JONES—FAGACEAE EVOLUTION 231 Paleobotanists have made informal groupings based on relatively minor differences in leaf char- acteristics. Tralau (1962), for example, divided the beeches into Fagus 'grandifolia' and 'sylvat- ica’ types in his investigation of fossil Fagus from Europe. Pollen data seem to at least partially support this distinction (see Praglowski, 1982). Tanai (1974) distinguished three groups based primarily on a leaf index (i.e., length/width x 100) and the number of secondary veins but again avoided the establishment of any formal infra- generic taxa. Tralau (1962) suggested that the ‘sylvatica’ and ‘longipetiolata’ groups intergrade p as well. In fact, Hanks and Lippen (1976) found that none ofthe groups proposed by Tralau (1962) and Tanai (1974) clustered tightly on the basis of pollen features. The lack of formal subge- neric taxa, therefore, seems quite reasonable. Foliar information confirms the relative uni- formity within this genus and can be used only in a very limited fashion for classification below the generic level (as in Cooper & Mercer, 1977). The observed intraspecific variation in foliar characteristics often exceeds interspecific varia- tion. There are no clear cut qualitative differ- ences that can be used to distinguish subgroups. In contrast, foliar information is very useful for the recognition of Fagus, whose leaves can be distinguished from those of other amentiferous amilies as well as other genera of Fagaceae (Ap- are the most characteristic features of these leaves (Jones, 1984) The fossil record of Fagus was thought to ex- tend far back into the Cretaceous (Berry, 1923). However, most reports of Cretaceous Fagus have been based on isolated leaves that are often poor- ly preserved and frequently bear only a superfi- cial resemblance to those of Fagus (see Lesque- reux, 1874). Some describe leaves that do not even resemble modern Fagus. Others describe leaves from sediments that have been shown to be much younger than originally thought. Wolfe (1973) suggested that there are no lena reports of any Fagaceae prior to the Paleocene. Muller (1981) suggested that the earliest reliable pollen record is that of Fagus granulata (Piel, 1971) from the Lower Oligocene of Canada. There are 232 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Quercus [^] a = [7] = о| 2] ol g]| o o o S o| èl Заз] о el2|3|8|2 о © Ф | > о = a Е с| o els 2 = сі s/s] 89| sla[|3|s|5 > 5 oa ы a ee ee Е ва! | ч. = a|<|u|=|2|o0|2lolol|elols оо зе] Е а|о|=|= пиши >|o|2|o О) > Unicellular Oolo|=|>|=|aljlu 1. Solitary e|jejej|e ојо|о| ө ojojojojojo 2. Conical ojojojo e о оо 3. Papillae e e 4. Appressed laterally attached e e 5. Fasciculate e о | о | о | о! о | о | о | е | е | о 6. Stellate ele e ellelele ele 7. Fused Stellate е e!| e o 8. Stipitate fasciculate е еї о јо | о e 9. Appressed parellel tuft е 10. Multiradiate e e Intermediate 11. "Thick"-walled peltate e 12. Curly thin-walled 13. "Thin"-walled peltate о | © 14. Rosulate e elle e Glandular 15. Simple uniseriate e e e|ejejejejejejej|e 16. Capi pedibus AN ө e|ejejejej|e 17. Large globular e |o |o 18. Branched uniseriate “lolo elle ele 19. “Glandular” peltate e TEXT-FIGURE 1. tions and ica б of each type. 'Not obse observed in this study but reported in Camus (1928-1 reports of earlier occurrences that may be ac- curate but need to be studied further. For ex- ample, Krutzsch (1970) reported fagoid pollen from the Eocene of Europe, but provided neither detailed descriptions nor figures. Kedves (1967) also reported fagoid pollen from the Eocene of Distribution of trichome types within the Fagaceae. See Appendix III for thorough descri rved in this study but reported in Camus (1934-1954). "Not 929). Europe. Pollen with the general form ofthat from Fagus has been reported from late Cretaceous sediments (Srivastava, 1969; Jarzen & Norris, 1975) but Muller (1981) has rejected these as not representing Fagus. Fagus-like wood has been reported from the Upper Cretaceous of Japan 1986] (Stopes & Fuji, 1910), the Lower Senonian of western Germany (Vater, 1884), and other Eu- ropean and North American localities. There is also one unconfirmed report of Fagus wood from South America (Salard, 1961). The first pre- sumed fruits of Fagus appear to be those reported from the Middle-Upper Eocene “Wilcox” Flora of southeastern North America (Lesquereux, 1859). This report is based solely on gross ex- ternal features and no subsequent reports of Fa- gus fruits have been made in spite of continuing work in the area. Brown (1944) considered a quadripartite structure, previously assigned to Diospyros aspera Berry, to be a Fagus cupule but Berry (1945) vigorously refuted the reassignment of this form. Brown (1944) also associated the fossil leaf form Dryophyllum tennesseense with the purported Fagus cupules. Upon reexamina- tion, however, these leaves have been shown clearly not to be Fagus but rather a castaneoid type most likely associated with the genus Cas- tanea (Jones, 1984). Thus, this report of Eocene Fagus is questionable. Manchester and Crane (1983) suggested that there are no confirmed re- s of Fagus earlier than the Oligocene, yet, tainly valid reports of this genus in the Oligocene (e.g., Conwentz, 1886; Chandler, 1957), and there are numerous reports of beech leaves and fruits in the Neogene (Unger, 1850; Kirchheimer, 1957; Tralau, 1962; Tanai, 1974; Van der Burgh, 1978a; Smiley & Huggins, 1981; Takhtajan, 1982). In summary, the fossil evidence currently available suggests that Fagus had evolved, certainly by the Oligocene, and perhaps as early as the Upper Paleocene NL; geographical distribution of Pagus tanao 1 1 fthe CI 1116 )st Northern Hence It is notably absent in some areas, such as western North America, where it is found only in the fossil record. There are no native Fagus in the Southern Hemisphere and reports of fossil Fagus from there are most likely in error. Many early reports result from the historical assignment of Nothofagus forms to the genus Fagus (see Ettingshausen, 1891). Oth- ers probably result from misidentifications based on superficial resemblances between Fagus and вее similar leaf forms such as №. ales- sandrii. In t orthern Hemisphere, Fagus ap- pears to eni been uniformly distributed over the major land masses with Neogene occurrences even in Alaska (Hollick, 1936) and Greenland JONES—FAGACEAE EVOLUTION 233 (Нег, 1883). The distribution of forms has ction, particularly die the Pleistocene. The origin of Fagus appears to have been in the Northern Hemisphere (Berry, 1923). Hanks and Fairbrothers (1976) placed the site of origin in south central Asia but Tanai (1974) reported older Fagus fossils from North America, imply- ing a North American origin. Takhtajan’s (1982) reports of Oligocene, Eocene, and Upper Paleo- cene Fagus in Asia, together with greater diver- sity, tend to support an Asian origin. In his 1969 treatise on the origin of flowering plants, Takh- tajan in fact, explicitly suggested an east Asian origin for an Steenis (1971a) is more specific, asserting that the Fagoideae arose some- where between Yunnan and Queensland with Fagus radiating from the Northern part of the region. Opponents (e.g., Schuster, 1972), have contended that the region circumscribed by van Steenis is not consistent with current geological interpretations, which suggest that this area was widely separated until the late Tertiary. Although van Steenis' hypothesis has largely been dis- missed, the site of origin for Fagus is still a sub- ject of debate. A recent paper by Hickey et al. (1983) is par- ticularly germane to the discussion of centers of origin. These authors suggested that, on the basis of new paleomagnetic data, the age assessments of many high latitude floras, modified in recent years on the basis of paleofloristic similarities with “younger” floras of lower latitudes, may be as old as originally thought. If their conclusions are accurate, many plants and terrestrial animals occur at high latitudes up to 18 million years before they occur at mid-latitudes. Although there has been much concern over some of the paleo- magnetic and other interpretations made in this particular study (Kent et al., 1984; Norris & Miall, 1984; Hickey et al., 1984), the once popular be- lief that temperate floral elements differentiated at high latitudes may be valid for many species. It seems that Fagus fits this pattern very well. I find this hypothesis to be very appealing but a better knowledge of the fossil floras of southeast Asia as well as those of higher latitudes is needed to confirm or refute it. Both Tralau (1962) and Tanai (1974) found Fagus grandifolia-like fossils antedating the F. sylvatica types. The F. grandifolia-like forms are also the most widely distributed, occurring in the fossil record of Europe, Asia, and North America until the end of the Tertiary. The F. sy/vatica and 234 F. longipetiolata types have not been found out- side Europe and Asia. At the end of the Tertiary the raros types became extinct in all but North America. Available paleobotanical data therefore, А that the ‘grandifolia’ type is primitive within the genus Fagus Foliar data, from extant slants: do not con- tribute much to the identification of a primitive form within the genus. The basic leaf architecture (i.e., characteristics of the margin, secondary veins, and gross form) of F. grandifolia ap- proaches those of some species in the Castaneoi- deae and Quercoideae more closely than the 'syl- vatica' or ‘longipetiolata’ types do, but such similarity in form is probably best explained by convergence rather than close proximity to a common ancestral form. NOTHOFAGUS BLUME Unlike Fagus, Nothofagus is not tightly bound together as a group on the basis of leaf characters. In fact, there are several different leaf types rep- resented in this genus. The trichome comple- ments are consistent, however, across most of the genus. None of the species bears tufts and nearly all are supplied with large, often resinous, globular to peltate trichomes. Short conical tri- chomes are also much more abundant in Noth- ofagus than in any genus of the Fagaceae sensu stricto. n spite of these common diede the w features characterize the genus as a whole. г of this variability I differentiated four general leaf groups, each with a fairly distinct assemblage of char- acteristics that allows it to be distinguished from other groups of Nothofagus as well as other gen- era and families (Jones, 1984). The first (type I) can be briefly characterized as being evergreen and entire-margined. The second (type II) is ev- ergreen with well developed serrations. The third (type III) is a rather heterogeneous group that is com- plement characteristic of the above groups. A list of the species studied that belong in each group is given in Text-Figure The diversity and distribution of Nothofagus has made it an att t fori (see, for example, Langdon, 1947; Camus, 1951; van Steenis, 1953, 1971a, 1971b; Cookson & Pike, 1955; Soepadmo, 1972; Hanks & Fair- brothers, 1976; Philipson & Philipson, 1979; ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Romero, 1980; Ragonese, 1981; Humphries, 1981; Aguirre & Romero, 1982; Romero & Aguirre, 1982; Praglowski, 1982). The group has been variously subdivided on the basis of repro- ductive and vegetative features. The classifica- tion of van Steenis (1953) is generally accepted. This scheme is built, in part, on foliar features, and thus there is some correlation between it and the division by leaf types found in Jones (1984) (see Text-Fig. 2). Palynologists divide the genus into three groups based on pollen characteristics (1.е., ‘menziesii,’ ‘fusca,’ ang ‘brassi’ types), but these too show van Steenis’s scheme. In the final analysis, Notho- fagus appears to be a heterogeneous assemblage. There is little doubt they are all related, at least at the family level (see Nixon, 1982), but the Meis within e group are more obscure. related to the phy- Dit distribution. Although restricted to the Southern Hemisphere, Nothofagus is fairly wide ranging with occurrences in temperate South America, Australia, Tasmania, and New Zealand as well as tropical New Britain, New Caledonia, and New Guinea. The tropical species form the most homogeneous and well defined group. This group collectively and exclusively possesses type I leaves with conduplicate vernation (Philipson & Philipson, 1979) as well as ‘brassi’ type pollen.? Species of this type also possesses bipartite cu- pules, a character that can be found, only rarely, outside this group. In addition, the wood of these tropical forms appears to be consistent in form and distinguishable from those of temperate species (Dadswell & Ingle, 1954) he temperate forms do not possess such uni- formity even within “reproductively isolated" floral regions. For instance, the Nothofagus of South America, New Zealand, and Australia all possess more than one leaf type as well as mul- tiple types of vernation and pollen (Text-Fig. 2). Both deciduous and evergreen forms can also be found in all but New Zealand. The distribution of these characteristics implies the existence of multiple parallel evolutionary events in several 3 Van Steenis (197 1a) ое ни type pollen from that, according to Cranwell (1939), bears ‘fusca’ type pollen. Cranwell’s sociated with the tropical Nothofagus from New Guinea and New Caledonia 235 1986] JONES—FAGACEAE EVOLUTION van Steenis ical Species Msn Subsection | Habit | Vernation | Pollen e Geographica section Cupule parts | distribution Leaf type ! Nf. aequilateralis | Calusparassus Bipartite e Conduplicate brassi 2 NC? (Trop Nf. balansae Calusparassus Bipartite e Conduplicate brassi 2 NC (Trop Nf. brassi ;alusparassus ipartite e Conduplicate brassi 2 NG? (Trop Nf. codonandra Calusparassus Bipartite e Conduplicate brassi 2 NC (Trop Nf. arenat Calusparassus Bipartite e Conduplicate brassi 2 NG (Trop. Nf. discoidea Calusparassus Bipartite e Conduplicate brassi 2 NC (Trop Nf. grand Calusparassus Bipartite e Conduplicate brassi 2 NG (Trop Nf. pullei Calusparassus Bipartite e Conduplicate brassi 2 NG (Trop. Nf. rubra Calusparassus Bipartite e Conduplicate brassi 2 NC (Trop Nf. starkenborghi | Calusparassus Bipartite e Conduplicate brassi 2 NG (Trop Leaf type Il Nf. alessandri Nothofagus Antarcticae d Plicate 2(4)! SAm? (Temp.) Nf. alpina Nothofagus Antarcticae d Plicate menziesii 4 SAm (Temp. Nf. antarctica Nothofagus Antarcticae d Plicate fusca 4 SAm (Temp.) Nf. glauca Nothofagus Antarcticae d Plicate 2(4)! SAm (Temp) f. gunnii Nothofagus Antarcticae d Plicate fusca 4 Tas? (Temp.) Nf. obliqua Nothofagus Antarcticae d Plicate menziesii 4 SAm (Тетр.) Nf. procera Nothofagus Antarcticae d Plicate menziesii 4 SAm (Тетр.) Nf. mil Nothofagus umiliae d Plicate fusca 4(2)! SAm (Temp.) Leaf type Ill Nf. betuloides Calusparassus | Quadripartite e Plane 4 SAm (Temp. Nf. cunninghami | Calusparassus | Quadripartite e Plane menziesii 4 Aust? (Temp. Nf. dombeyi Calusparassus | Quadripartite e Plane fusca 4 SA Temp.) Nf. fusca Calusparassus | Quadripartite e Revolute 4 NZ? (Temp. Nf. menziesii Calusparassus | Quadripartite e Plane menziesii 4 NZ (Temp.) Nf. nitida Calusparassus | Quadripartite e Plane fusca 4 SAm (Temp.) Nf. truncata Calusparassus | Quadripartite e Revolute fusca 4 NZ (Temp.) Leaf type IV Nf. cliffortioides | Calusparassus Tripartite e Revolute fusca 3 NZ (Temp.) solandri Calusparassus Tripartite e Revolute fusca 3 NZ (Temp.) TExT-FIGURE 2. Comparison of proposed classification, by leaf type, with that of van Steenis (as reported in 5осрайто, 1972) and some individual properties of Nothofagus species. 'Disparant values for this variable ist for these ms thos Philipson (1979). 2 South America, My = Tasman character groups and/or the presence of the cur- rent diverse set of characters before reproductive isolation occurred (estimated by tectonic data to have been ар by the mid-Cretaceous; Smith and Briden, , Postulates seta on e tectonic informa- and Briden data would indicate the existence of the ancestral group from which Nothofagus and other Faga- ceae arose during the late Jurassic or early Cre- taceous yet the earliest confirmed report of Noth- ofagus or any other Fagaceae does not occur until the Santonian (Muller, 1981) or perhaps Conia- cian (Stover & Evans, 1973). This and all other Cretaceous reports of Fagaceae are based on pol- se in parentheses are from van Steenis (1953) and the others are from Philipson and ust = ~~ NC = New Caledonia, NG = New Guinea, NZ = New Zealand, SAm = len. The oldest record of Nothofagus (Dettmann & Playford, 1969) is from southern Australia. This pollen, Nothofagidites senectus, is of the *brassi' type, which is the only type found until the middle to late Maestrichtian when the ‘fusca’ and “menziesii” 1 with the ‘brassi’ “type, in South America (Ro- mero, 1973, 1977; Archangelsky & Romero, 1974; Menendez & Filice, 1975). The strati- graphic ranges of the pollen types in each major "egies region (Text-Fig. 3) imply a radia- tion of the ‘brassi’ type from southern Australia to E on America and other points in its current range. The pollen record further suggests that the genus underwent considerable differentiation in or enroute to South America where the *menzie- sii’ and ‘fusca’ types first appear. These forms alor 15 236 N. Guinea N. Britain Australia New Zealand Antarctica S. America N. Caledonia Tasmania bf m btm b m b fm b m Quaternary ——— M Pliocene PA Paleocene L + Maestrichtian Campanian Santonian Coniacian TEXT-FIGURE 3. кш of Nothofagus pollen types (b = ‘brassi’; f = ‘fusc *menziesii') through space and time; gaps are n Modified and revised from Muller (1981) and Humphries (1981). may have subsequently spread back across Ant- arctica to New Zealand and Australia. This hy- pothesis is highly conjectural at this point and must be tested along with a myriad of other hy- potheses as we broaden the body of fossil evi- dence and determine, with greater accuracy, the stratigraphic position of previously reported forms. It should be noted that this and many other models require free routes of migration between South America and Australia at least into the Maestrichtian. Again, this is not in agree- ment with the current paleogeographic recon- structions. The cladistic representation of land mass separations constructed by Rosen (1978), however, indicates that sea boundaries were not f ] til l ] ] l diffe ti 1 atlea had occurred. Van Steenis (197 1а) circumvented discrepancies between the time of barrier for- mation and the emergence of various groups by suggesting the existence of some as yet unre- ported land bridges linking South America to Australia through Antarctica. Archipelagos or migrating microplates would account for the pos- tulated migration. Schuster (1976) included re- ports of cordilleras that might satisfy van Steen- is’s requirements. The strength of biological evidence for the existence of free paths for dis- persal supports either the existence of these dis- continuous pathways or the later estimates for the separation of the “Nothofagus” province. If the earlier estimates of separation are accurate and van Steenis's land bridges did not exist, I would accept the occurrence of long distance transport, after the wide range of characters had evolved, rather than the multiple parallel evo- lutionary events that would be required to ex- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 plain the current distribution of characters. Al- though unlikely, long distance dispersal may not be as ae as it might first appear. Iti = ue h body ofe that ius of Nothofagus, and for that matter all extant Fagaceae, are not suited for long distance dispersal (Holloway, 1954; Cranwell, 1963, 1964; Preest, 1963). Yet, Tiffney (1984) has shown that substantial changes in fruit and seed sizes have occurred from mid-Cretaceous into the early Tertiary. The trends in size are undoubtedly cor- related with changes in modes of dispersal and the evolution of various animal groups. Thus, the dispersal strategies exploited by fossil Faga- ceae may have differed substantially from those of today. The presence of Fagopsis, a wind dis- persed fagaceous plant from the Eocene and Oli- gocene of North America, underscores the need for flexibility when ач the dispersal mechanisms of fossil form An intriguing phyiogeosraphi problem is that of the center of origin and/or radiation. Postu- lated centers of origins ig New Caledonia (Croizat, 1952), southeast Asia (Takhtajan, 1969), southeast Asia-Queensland (van Steenis, 1971a, 1971b), North America (Oliver, 1925; Schuster, 1976), Asia (Darlington, 1965), Northern Hemi- sphere (Raven & Axelrod, 1974) and “southern” (Couper, 1960а, 1960b; Cranwell, 1963). Al- though many postulate a northern origin, it is clear that most of the differentiation occurred in the Southern od A northern origin of Nothofagus (no morphologically dissimilar ancestor) would Tes leave a recognizable fossil record in that region. Although there have been several reports of Nothofagus or Nothofagus pol- len from the Northern Hemisphere (Mtchedlish- vili, 1961; Sein, 1961; Ames & Riegel, 1962; Kedves, 1964; Penny, 1969; Hopkins, 1969; El- sik, 1974), none provide convincing evidence for the existence of Nothofagus in that region. Pollen forms attributed to Nothofagus from the Maes- trichtian of Siberia (Mtchedlishvili, 1961) and the Eocene London Clay (Sein, 1961) are said to be chloranthaceous (Kuprianova, 1967). She and others (e.g., Soepadmo, 1972) doubt the exis- ence of Nothofa in the Northern Hemi- sphere. Elsik (1974) refuted Kuprianova's as- sessment on the basis of exine sculptural differences between the fossil forms she studied and extant Chloranthaceae. It is interesting to note, however, that the pollen Elsik assigned to Nothofagus has an exine sculpture that is quite distinct from all living members of this genus. ee 1986] As he tangentially suggested, this form may rep- resent a fagaceous or pre-fagaceous ancestral form since the sculpturing is more like that found in extant Fagus and Quercus than in Nothofagus. Megafossil evidence of Nothofagus is also lack- ing in the Northern Hemisphere. Bandulska (1923, 1924) reported Nothofagus leaves from the Eocene of Bournemouth, but her description of the leaf architecture was very brief and did not contain any information to indicate that the specimens under consideration were actually Nothofagus. She admirably ш informa- tion and an illustra ese leaves but again the description was very ca and the illustration did not show any characteristics that would confirm the generic affinities of this leaf type. The T pieces or “dagger-like” structures of which she spoke are common to many fagaceous genera and their presence in cuticle preparations is often related to leaf age and the length of maceration during cuticle preparation. Complex gland bases, common to nearly all Nothofagus, were not mentioned in the description and were not evident in the illustration. No subsequent reports confirming or augmenting her report have been made. There are other leaves from the Northern Hemisphere with architecture similar to those of Nothofagus. Phyllites kryshtofovichii (Klimova) Iljinskaja et Ablaev, for instance, has been compared to Nothofagus moorei (Takhta- jan, 1982). No information on the cuticles of leaves such as these is available. Leaves of other members of the Hamamelidopsidae also possess the architectural characteristics of this and other similar leaf types, so until cuticular evidence is produced they cannot be considered as an indi- cation of the presence of this genus in the North- ern Hemisphere. There are no reports of wood, flowers, fruits, or other plant parts attributable to this genus from the Northern Hemisphere. A northern origin, for fully differentiated Notho- fagus, therefore is unlikely. Raven and Axelrod's (1974) postulate that Nothofagus must have mi- grated to the Austral regions through Africa lacks corroborative fossil evidence. Raven and Axel- rod (1974) recognized this deficiency, citing only two reports, one of which was attributed to con- taminated drilling mud (Puri, 1965) and the oth- er to windblown contaminants from South America (Schalke, 1973). Fossil evidence there- fore strongly supports a southern origin for rec- ognizable Nothofagus. This is not to say that a remote hamamelidopsid ancestor did not devel- op in the Northern Hemisphere and migrate to JONES — FAGACEAE EVOLUTION 237 the south via Africa as suggested by Raven and Axelrod (1974) or North America as proposed by Oliver (1925) he fossil record of Nothofagus is well repre- sented in the Southern Hemisphere (e.g., Couper, 1953, 1960a, 1960b; Cookson, 1958; Cranwell, 1959, 1963; Romero, 1973; and many others). Pollen has been extensively studied and is found in the remnants of Gondwanaland except India, disregarding one questionable report from the Miocene (Ramanujam, 1966), and Africa (Axel- rod & Raven, 1978). These two land masses were the first to become isolated during the fragmen- tation of Gondwanaland. Axelrod and Raven (1978) suggested that the evergreen southern beech-podocarp forests must have extended into assumption. Surely pollen evidence would have been found if this were the case. Nothofagus is known to produce massive amounts of pollen (van Steenis, 1971a; Soepadmo, 1972) and it often occurs as a dominant element in fossil assem- blages. To me the absence of fossil forms on these continents argues for an origin in the Austral complex after the separation of these two land asses. Megafossils are also abundant, yet they have been largely neglected. Fossil wood (Nothofa- goxylon Gothan) has been described from the Upper Cretaceous or Tertiary (age uncertain) of Antarctica (Gothan, 1908) and the Oligocene to Miocene of southern South America (Kráusel, 1924; Cozzo, 1950; Boureau & Salard, 1960; Sa- lard, 1961; Ragonese, 1977). The wood de- scribed is all of temperate type. Leaves are very abundant (Berry, 1938; Menendez, 1971) and have been described from Antarctica (Dusén, 1908; Orlando, 1964; Zastawniak, 1981), Aus- tralia (Ettingshausen, 1888), New Zealand (Et- tingshausen, 18872; Oliver, 1936), Tasmania (Hill, 19832, 1983b), and South America (Dusén, 1899— 1902; Frenguelli, 1941). Many of the leaf types described in the early literature were assigned to Fagus in accordance with the plant classification schemes for modern Fagaceae in vogue at that time. Although some leaf types do not appear to be Nothofagus or any other Fagaceae, most are comparable with Nothofagus on the basis of the information available. Except for Hill (1983a, none of the studies provide cuticular in- tio pa the abundance of the leaf record and the early and widespread occurrence of the 238 ‘brassi’ type pollen, one might expect to find en- tire-margined evergreen (type I) leaves in the fos- sil record, but this is not the case. Yet, the ap- parent absence of these leaves may be due to difficulties involved in identifying entire-mar- gined leaves rather than a true absence. Cuticular material, when available, is a great asset when working with nondescript entire-margined leaves Close examination of austral fossil floras may I iin the presence of these Nothofagus "АП sil Nothofagus leaf materials reported thus far have been from the Tertiary. The ac- curacy and precision of the age determinations are poor in most cases. Thus, at present, the leaf record can tell us little more than that deciduous and/or evergreen temperate (types II and III) Nothofagus were present in South America, Ant- arctica, New Zealand, and Australia during the Tertiary. A comprehensive review of fossil Noth- ofagus leaf forms is desperately needed. The leaf architectural and cuticular morphol- ogy of extant Nothofagus has been the subject of several recent papers. Romero and co-workers (Romero, 1980; Aguirre & Romero, 1982; Ro- mero & Aguirre, 1982) have surveyed the leaf architecture of nearly the entire genus, providing keys for the identification of leaves to the species level. Bandulska (1924) surveyed the cuticle of some Nothofagus and Fagus species, but her analysis was limited to only a few species and employed a restricted set of characters. Ragonese (1981), however, provided a much more com- plete analysis in her anatomical assessment of Nothofagus leaves, but she investigated only the South American forms. The results of Jones (1984) are in general agreement with those of the above studies. Little can be said about intrageneric evolution on the basis of extant leaf characteristics alone. Romero (1980) stated that Nothofagus antarctica leaves were the most primitive among those of the South American species but did not provide foliar evidence to support this assertion. It seems he was echoing the assertions of van Steenis and others based on the presence of "primitive" re- productive features such as a two-parted cupule. Type IV leaves, characteristic of N. solandri and N. truncata, are specialized and do appear to be advanced. They possess a rather unique trichome complement that lacks large glands, common in the rest of the genus, and possess a dense cover of curly, thin-walled, solitary trichomes on the abaxial surface. This trichome type is restricted ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 to this group with the exception of Nothofagus glauca, in which they are much less abundant and occur with thick-walled solitary trichomes and large glands. Nothofagus gunnii of Tasmania also lacks glands and possesses a rather distinct a specialized form, but witho fossil evidence we cannot know for sure. often SISEDCHVE leaf architecture and rather unique cuticular cl hould be of great value in m mcn of fossil Nothofagus leaf forms. The analysis of these forms will provide much valuable information about the evolution- ary history of this important group particularly with regard to its origin and relationship to the Fagaceae sensu stricto and the Betulaceae. CASTANEA MILL. Castanea is a small genus of about 12 species, restricted to temperate regions of the Northern Hemisphere. It is a relatively homogeneous group foliar vegetative features (see, for example, Shi- maji, 1962) also exhibit little variation in this genus. Unlike Fagus, however, this genus has been studied fairly thoroughly. Camus (1928- 1929) presented a thorough monograph of the genus in which she adopted the subgeneric clas- sification proposed by Dode (1908). According features (i.e., those ofthe fruit and cupule). Foliar characteristics provide some support for this classification. The leaves of section Eucastanon, for example, generally have more prominent teeth with distinct bristle tips. The uniseriate (type 15) and capitate (type 16) trichomes also have less prominent basal cells in species of this section. Capitate trichomes with multicellular heads are abundant in section Eucastanon whereas simple uniseriates are more abundant in the other two sections. The monotypic section Hypocastanon can be roughly distinguished from the other sec- tions by the lack of abundant trichomes and sub- tle teeth, which often lack bristles. Some speci- mens of C. henryi, in fact, possess upturned setaceous tips. None of these features, however, can be relied upon to clearly differentiate sub- a The intraspecific variation, once again, s or exceeds that between most species. nes (unpubl. data) faced this problem in his 1986] attempt to construct a key to the species of Cas- tanea based on foliar features. He emphasized the variability of the foliar features and stressed the importance of relying on a number of char- acteristics rather than one or only a few. In the final analysis he was unable to construct a key to the species based entirely on foliar features. The variability is probably due, at least in part, to the reproductive properties of the group. Fac- ile formation of hybrids is one of the salient char- acteristics of the Fagaceae, and Castanea is no exception. The formation of natural hybrids has ong been noted and has caused the taxonomy, particularly of the “chinkapin” species, to be quite unsettled. Castanea species are discontinuously distrib- uted in temperate regions of the Northern Hemi- sphere. This group does not appear to compete as favorably in the cooler parts of the North tem- perate zone as Fagus, Quercus, and many other elements of the temperate forests. Fossil records indicate a typical **holarctic" distribution with records from Greenland (Heer, 1883), Alaska (Hollick, 1936), western North America (La- Motte, 1952), and other regions in which Cas- tanea is no longer found. Since there is undis- puted proof of the existence of Castanea (1.е., inflorescences, cupules, pollen, and leaves) by the Eocene (Crepet & Daghlian, 1980; Jones, 1984), the current distribution is in harmony with plate tectonic evidence Reports of castaneoid fossils span from the Santonian to the present. The first records are in the form of pollen. Castaneoid (Tricolpollenites type) pollen has been reported from the Lower Campanian of the Netherlands as well as the Santonian/Campanian of western Canada (Rouse et al., 1971). Cupuliferites pusillus, another taxon associated with the Castaneoideae, has been re- ported from the Maestrichtian of California (Chmura, 1973). Pollen corresponding to this type was later reported from Eocene Castanea inflo- rescences from southeastern North America (Crepet & Daghlian, 1980). Castaneoid pollen has been reported from the Upper Paleocene (Gruas-Cavagnetto, 1978) and Lower Eocene (Kedves, 1978) of Europe as well. Reports from the Eocene to the present abound (Muller, 1981). Palynological data alone cannot be used to in- dicate the presence of Castanea but rather only the Castaneoideae. Wood resembling that of Castanea (i.e., Cas- tanoxylon Navale) has been reported from the Neogene of Europe and Eocene-Miocene of India JONES—FAGACEAE EVOLUTION 239 (Navale, 1964; Selmeier, 1970c; van der Burgh, 1978b). Wood, like pollen, is not generically spe- cific in this group and similar wood is present in Castanopsis as well as some species of Quercus and Lithocarpus (Navale, 1964). Castanea flow- ers, which provide definitive proof, have been reported by Conwentz (1886) in the Eocene-Oli- gocene Baltic Amber and by Crepet and Daghlian (1980) from the Eocene of southeastern North America. There are very few reports of Castanea fruits in the tese record. Kirchheimer (1957) listed only d (B y , 1940). Cupules have been found in the Eocene of south- eastern North America (Crepet & Daghlian, 1980; Jones, 1984) but the only report of actual fruits is one of aborted fruits from the Pliocene of Ger- many (van der Burgh, 1978a). The rarity of Cas- tanea as well as Quercus fruits in the fossil record is probably related to the lack of hard, durable ovary walls, or parts thereof, and the presence of abundant substrate for decomposer organ- isms. That the Pliocene fruits mentioned above were aborted fruits is consistent with this hy- pothesis. Most of the reproductive structures that ered to be Castanopsis (Kirchheimer, which is commonly reported from the Neogene of Europe. The basic Castanea leaf form first appears dur- ing the Eocene in the form of Dryophyllum tennesseense (Jones, 1984). Similar leaves can be found in the Paleocene (e.g., Laurent, 1912; Crane, 1978), but few possess the regularity of secondaries and teeth characteristic of Castanea leaves, and those that do lack cuticle needed for identification. Many “‘castaneoid” leaf forms have been reported from the Eocene to the present (e.g., Nathorst, 1888; Berry, 1916; Knobloch, 1969; Takhtajan, 1982; Hummel, 1983). None of nae sii has been unquestionably associ- ated v how- ever, and similar forms can be found in other Many work- ers (e.g., Laurent & Marty, 1909: Ferguson, 1971) have specifically mentioned the difficulties in- volved when working with isolated Castanea- like leaves. The Castanea-like Dryophyllum fur- cinervis (Rossm.) Schmalh., for example, is thought by some to be Castanopsis (Kráusel & Weyland, 1950), whereas others hold it to be a distinct ancestral type (Raniecka-Bobrowska, 1962). Mai (1970) believed that the morpholog- ical and anatomical features are consistent with those of the genus Trigonobalanus. He further fagaceous and f: n 240 suggested that these leaves are stratigraphically correlated with fossil fruits of this genus. Fer- guson (1971) chose to leave it in the genus Quer- cus as suggested by Heer. Realistically, this leaf type could belong to this or to one of many other extant genera to which it has been assigned. It could just as easily represent an extinct form whether ancestral or not The foliar features of Castanea species do not provide any hint as to possible primitive forms within the genus. It may be possible to infer the primitiveness on the basis of correlated fruit and other features but even this would be tenuous when one considers the relative plasticity of leaf characters and the interspecific mixes character- istic of this taxon. CHRYSOLEPIS HELMQV. Chrysolepis consists of two sympatric species native to the Pacific western United States. This genus is recognized by many, but not all, prom- inent students of the Fagaceae (Hjelmqvist, 1948; Hutchinson, 1967; Forman, 1966b; Elias, 1971; Abbe, 1974). Others, Soepadmo (1972) for ex- mple, do not feel that the Chrysolepis species nopsis. Inacom 5, ро allegedly stated that if Chrysolepis is at least one species from : southeast Asia would have to be included (van Steenis, 1971a). Lawrence (1951) and Mel- chior (1964) did not recognize this genus but failed to support their position. Brett (1964) ev- idently did not recognize Chrysolepis, which is not too surprising since he also favored the fusion of Castanopsis and о thus reducing the Castaneoideae to two gen Separation of Chrysolepis ndn Castanopsis is based identical. Erdtman (1943) for example, found the pollen of Chrysolepis (Castanopsis) chrysophylla to be indistinguishable from that of Castanopsis and Castanea. This has been supported recently using scanning electron microscopy (Crepet & Daghlian, 1980). The leaves of Chrysolepis and Castanopsis species are also very similar in gross form, venation, and cuticular characteristics. There are some differences, however, that can be the generic distinction. generally smaller an somewhat xerophytically modified. Most tend to ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 be elliptical to oblong rather than lanceolate as n most Castanopsis. The most consistent dif- кишене is in the trichome complement. Even these are similar, but both species of Chrysolepis pos- sess thick-walled peltate trichomes with radially aligned cap cells (type 11, see Appendix III), which are totally lacking in Castanopsis species. Similar peltate trichomes are found in other Castaneo- о cell walls, and the cap is composed of nonradially aligned cells. The taxonomic significance of these differences can be debated, but they do allow one to distinguish Chrysolepis leaves from those of other genera. I believe Chrysolepis, although very closely related to enisi should be recog- nized at the generic le The phytogeograpy of Croat can be in- terpreted in two different ways depending on the basic assumptions made about the group. Nei- ther of these can be confirmed or rejected with the data currently available. If the genus is an ancestral form not derived from or closely as- sociated with Castanopsis, as suggested by Elias (1971), it simply has a very restricted relict dis- tribution. Yet, if Chrysolepis species are closely allied derived forms, perhaps even congeneric with those of Castanopsis, which are restricted today to southeast Asia and surrounding islands, the disjunction must be explained by one or more of the following: 1) long distance dispersal, 2) a widespread (““holarctic””) past distribution, and 3) plate tectonic anomalies (i.e., rafting on mi- egies adopted by this and other groups of extant Fagaceae. In contrast, the fossil record indicates that the range of Castanopsis was indeed much greater in the past. There have been reports of Castanopsis in western North America, Asia, and Europe (Kráusel & Weyland, 1950; LaMotte, 1952; Kirchheimer, 1957; Takhtajan, 1982). If these reports are accurate they, collectively, in- dicate a near pan north temperate distribution for Castanopsis. The absence of reports from eastern North America and other portions of the north temperate zone may be more a function of biases in the identification of fossil leaves than true distributional gaps. Leaves, which are the bases for most of the reports, are mostly entire- margined in Castanopsis. They are difficult if not impossible to distinguish from a host of other entire- and nearly entire-margined forms on the basis of leaf architecture alone. Thus, when ana- 1986] lyzing leaves using only leaf form and architec- ture, one could quite likely assign Castanopsis leaves to other taxa with similar leaves partic- ularly when encountered in regions where Cas- tanopsis leaves are not expected. One only has to examine the synonymy lists in LaMotte (1952) to recognize the taxonomic diversity of forms that resemble Castanopsis. Cuticular features are very valuable in distinguishing forms with sim- ilar leaf architecture, yet, they are often neglected even when cuticle is preserved. Regardless of whether or not these gaps in the distribution are filled, the reports that exist indicate a sufficiently wide distribution of Castanopsis to account for the presence of Chrysolepis in its present range. though the current distribution is easily ex- plained on the basis of paleoenvironmental in- formation and the movement of large continen- tal plates, some workers have suggested that western North American-southeast Asian dis- juncts may have arisen by rafting on microplates such as those that form the southern portions of Alaska. Such explanations need not be invoked in this case, and little evidence for rafting has been presente The fossil record of Chrysolepis is scanty. Since the pollen of all Castaneoideae are similar, the earliest records may range back to the Santonian- Campanian (Muller, 1981). Reports from Cali- fornia (Chmura, 1973) and western Canada (Rouse et al., 1971) indicate castaneoid pollen was present in and near the current range of Chrysolepis by the late Cretaceous. Chrysolepis leaves (Castanopsis chrysophylloides Lesq.) have been reported m the Eocene to Pliocene (LaMotte, 1952). Dorf (1938) reported a leaf form, which he believed was most closely related to Chrysolepis (Castanopsis) sempervirens, but this report, like all other reports of species as- signed to this genus, is based on gross form and architecture alone. Some of these reports, par- ticularly those of Upper Miocene and Pliocene forms, are reasonable, yet cupules, upon which definitive identification must rest, have not been found. Because of uncertainty in generic identi- fication it is impossible to tell, with the infor- mation available, precisely when Chrysolepis first evolved. If, as Elias (1971) suggested, this is an ancestral form, we might expect it to have evolved very early in the history of the family. It might account, at least in part, for the early reports of castaneoid pollen. However, if it is an advanced offshoot of Castanopsis (Brett, 1964) that evolved independently only after geological and environ- JONES— FAGACEAE EVOLUTION 241 mental factors created reproductive isolation would we expect a post-Eocene, probably late Oligocene or early Miocene, origin. At present both an early origin, that is, Upper Cretaceous or Paleogene, and a late origin, that is, Neogene, seem equally plausible. CASTANOPSIS (D. DON) SPACH. Castanopsis is a fairly large genus estimated to contain from “approximately 30 species or more” (Hutchinson, 1967; Abbe, 1974) to about 120 species (Camus, 1928-1929; Forman, 1966b; Willis, 1973). The actual number of species is probably around 100, extrapolating from recent revisions (Soepadmo, 1972) of the Malaysian pecies. The lower estimates appear to come from sae century works and/or the works of Soepadmo on Malaysian Castanopsis. The great variation in the estimated size of this genus is, in part, an indication that few have studied Cas- tanopsis in its entirety. The only modern com- prehensive treatment of the genus is that of Ca- mus (1928-1929). Soepadmo (1968a, 1972) provided a more up to date treatment, but it is restricted to the Malaysian species and is not nearly as thorough as that of Camus. There are numerous, less comprehensive assessments in- cluded in papers dealing with the systematics of the Fagaceae in general, but these do not contain much information on foliar features. Oersted (1871) and Camus (1928-1929) di- vided Castanopsis species into groups. Oersted (1871) absorbed the species of Castanopsis into Castanea, segregating them into three groups on the basis of foliar and rep tive features. Two groups were placed in the subgenus Castanopsis, within which he placed Castanopsis (Castanea) indica A. DC. ex Seem., a species with acute serrate leaves, in one section and those species with entire cr paucidentate leaves in a second. N (Miq.) Oersted. De Candolle (1868) separated the species of Castanopsis in a similar fashion but held this genus to be distinct from Castanea. Camus (1928-1929) divided the species of this genus into three sections using only reproductive features. Unfortunately, there are no foliar fea- tures that can be used to clearly distinguish these three sections. ee of variation in the leaf architectural and cuticular features of Castanopsis leaves var- ies considerably. The margins, for example, range 242 from entire to completely serrate. The venation varies similarly from camptodromous to cras- pedodromous. The fine venation is fairly con- sistent within the group except as modified by gross architectural differences. There is some variability in the cuticle as well but nearly all have a rather distinct speckled appearance. The serrate forms tend to possess tufts (type 5 tri- chomes, see Appendix III), which are otherwise rare in this genus. Nearly all species possess thin- walled peltate to *'stellate" (type 13) trichomes abundantly distributed over the lower epidermis. These trichomes often cover the lower surface and obscure other epidermal features. It is the bases of these trichomes that give Castanopsis cuticles their characteristic speckled appearance. gaceous leaves. It does occur in a few Lithocarpus species though, making absolute determinations of these leaves impossible at the generic level. Similar trichome forms occur in Trigonobalanus doichangensis (type 19) and Chrysolepis (type 11), but these can be distinguished readily from those of Castanopsis. In spite of the variability in foliar features within the entire genus, most species of Casta- nopsis bear leaves of a single general type. The basic form of these leaves is similar to those of mesic tropical species in other fagaceous genera. The abundance of this generalized leaf type in Castanopsis is probably due, in part, to the rather limited geographic distribution of this genus. Castanopsis is native only in southeast Asia, Indonesia, Korea, and Japan but the phytogeo- graphic distribution is reputed to have been much wider in the past (Soepadmo, 1972). Kirchhei- mer (1957) suggested that fruits assigned to Cas- tanopsis salinarum (Unger) Kirchheimer from the Oligocene of Europe are definitively Casta- nopsis. Many others have reported Castanopsis fruits, leaves, and/or wood from the middle to late Tertiary of Europe (e.g., Notzold, 1961; Sza- fer, 1961; Raniecka-Babrowska, 1962; Mai, 1964; Selmeier, 1970a, 1970b, 1972; Jung et al., 1971; van der Burgh, 1973). Ogura (1949) has reported wood of this genus from Japan, and Takhtajan (1982) included a number of Castanopsis leaf forms from eastern Europe and Asia. Wolfe (1968) reported a new species of Castanopsis from the Eocene of North America and further sug- gested (1973) that leaf material (i.e., that of MacGinitie, 1941) is truly Castanopsis (non Chrysolepis Hjelm.). Dorf (1938) reported a Cas- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 tanea-like form, which he believed to be Cas- tanopsis sensu stricto from the Tertiary of Idaho. Chrysolepis forms are also present in western orth America (Axelrod, 1944; Chaney, 1944). Yet, these North American reports, those in Takhtajan (1982), and most others dealing with leaf material fail to give any information on the cuticle of the fossils in question. Even with cu- ticle, the leaves of this genus are often indistin- guishable from some species of Lithocarpus and Quercus (see for instance Forman, 1964a, 1966a, 1966b). Therefore, fossil forms cannot be as- signed at the generic level with certainty. Kráusel and Weyland (1950) were the first to use cuticular features to classify material as Castanopsis, but few have utilized these important features since. There are no reports of fossil Castanopsis fruits, or other reproductive macrofossils, in either North America or Asia to corroborate question- able leaf data. There are no credible megafossil records of Castanopsis prior to the Eocene. Yet, pollen comparable to that of Castanopsis has been re- ported as early as the Upper Cretaceous (San- tonian/Campanian of Canada; Rouse et al., 1971). Castanopsis-like pollen has also been reported from the Eocene of Great Britain (Chandler, 1964) and Europe (Kedves, 1978). Reports of Casta- nopsis-like pollen are, in fact, quite common throughout the Northern Hemisphere since the Eocene (Muller, 1981). As previously men- tioned, however, the pollen of Castanopsis is not distinct but rather belongs to a generalized “cas- taneoid” type. Therefore, the а of this pollen type ence of Castanopsis. The fossil record can be interpreted as indi- cating an origin perhaps as early as the Upper Cretaceous but more likely in the early Tertiary. This genus was apparently widely distributed during the Eocene to Miocene and was lost in North America, Europe, and most of Asia during the Pliocene. Modern phylogenies suggest a fairly long evolutionary history for this genus, but forms included by the modern circumscription of Cas- tanopsis are not generally considered to be prim- itive within the family. The foliar features of this genus do not appear to be particularly advanced or primitive. The peltate (type 13) trichomes seem, intuitively, to be somewhat advanced yet the speckled appear- ance, lent by the bases of these trichomes, is present in some early forms of Dryophyllum and similar leaf types (Jones, 1984). Although the t necessar ily P 1986] foliar characteristics do not tell us much about the evolutionary status of this group within the family, they do support the separation of Cas- tanea from Castanopsis, instead of the union sug- gested by Oersted (1871) and Brett (1964). In addition to morphological differences indicated by other authors (e.g., Camus, 1928-1929; For- man, 1964a, 1966a, 1966b), Castanea species do not possess the peltate trichomes characteristic of Castanopsis but rather bear capitate trichomes and tufts that are rare or absent in extant Cas- tanopsis. LITHOCARPUS BLUME Lithocarpus is a large complex genus estimat- ed, by some, to contain about 300 species (Willis, 1973). It has been split into two or more genera by several investigators (Oersted, 1871; Schwarz, 1936a; Nakai, 1939; Masamune & Tomiya, 1948; Brett, 1964). Others have distributed “I ithaca: pus” species among sections of the genus Quercus (de Candolle, 1868). Camus (1934-1954), For- man (1964a, 1964b), and Soepadmo (1970, 1972) suggested that there was no basis for splitting this xon. The boundaries between Lithocarpus, Castanopsis, and Quercus have long been a sub- ject of controversy. Modern workers agree that Quercus and Lithocarpus are phylogenetically distinct, even though they show many parallel traits (Brett, 1964; Forman, 1964a, 1966a; Soe- padmo, 1970, 1972). Luong (1965) and Forman (1966b) further suggested that Castanopsis and Lithocarpus can be clearly separated, at least phylogenetically, using specific vegetative and reproductive features. It is clear, however, that considerable mosaic and parallel evolution has occurred in this family. There is little correlation between foliar fea- tures and most intrageneric classification schemes, which are based primarily on cupule morphology and other reproductive features. For example, very few of Camus’ 14 subgenera are distinct and homogeneous with respect to foliar features. Lithocarpus leaves, in general, show rel- atively little variability. They are usually entire or only partly toothed. Only L. densiflorus has strongly serrate margins and then only in a few forms (e.g., L. densiflorus forma attenuato-den- tatus, Tucker et al., 1969). Leaves of Lithocarpus are persistent to tardily deciduous. The basic leaf form is, in fact, very similar to those of other tropical fagaceous genera (i.e., Castanopsis and Quercus particularly those of the subgenus Cy- JONES— FAGACEAE EVOLUTION 243 clobalanopsis). Fortunately, most leaves of this genus can be distinguished from those of other genera using epidermal features. Appressed lat- erally attached (type 9, see Appendix III) tri- chomes are present in nearly all Lithocarpus species, and it is interesting to note that recent revisions (Soepadmo, 1970, 1972) involve the transfer of many species lacking this trichome type to Castanopsis and Quercus. As the taxon- omy of Lithocarpus is further refined, this char- acter may pire perfect agreement with those that are definitiv relative A of Lithocarpus foliar байа is probably a function of its rather lim- ited distribution. With the exception of L. den- siflorus, this genus is restricted to southeast Asia, Indonesia, and southern Japan. Lithocarpus den- siflorus, which is found only in the Pacific south- western part of North America, has rather dis- tinctive foliar characteristics. Leaves of this species lack acuminate tips as well as other fea- tures common to the tropical oriental species of this genus. These sclerophyllous leaves invari- ably lack the appressed laterally attached (type 9) trichomes, common to most Lithocarpus species, and possess unique multiradiate (type 10) trichomes. The differences between the leaves of this species and those of the eastern species suggest rather distant affinities. Forman (1964a, 980) and ever, that the reproductive and nonfoliar vege- tative features clearly associate this species with Lithocarpus. Although L. densiflorus may fit within Lithocarpus, the disparity in the foliar characteristics suggests a significant period of re- productive isolation in dissimilar environments. The presence of Lithocarpus on both sides of the Pacific Ocean implies that this genus had a more extensive range in the past that must have bridged Asia and North America. Thus fossil Lithocarpus might be expected to occur at high northern latitudes on both continents. One might also assume that the distribution would extend to Europe as well since this is a pattern common to many contemporary floral elements. Yet, re- ports of this genus from the Northern Hemi- sphere are extremely rare. LaMotte (1952) listed only one North American species, L. klama- thensis (MacGinitie) Axelrod, which occurs in the Miocene and Pliocene of the Pacific North- west. Axelrod (1966) subsequently reported Lithocarpus leaves and a cupule from the Eocene (age uncertain) of Nevada. Axelrod and Raven 244 (1978) suggested that Miocene Lithocarpus gave rise to forms of L. densiflorus that occur in Cal- ifornia today. Cuticular information, currently unavailable, should help to clarify the affinities of these fossil leaves. Reports are equally rare from Europe. Saporta and Marion (1878) established the genus Pasa- niopsis to contain Pasania (Lithocarpus)-like fos- sil leaves from the Paleocene Gelinden Flora. Only leaf architectural information is given for this material. It is consistent with placement near Lithocarpus, but cuticular information is needed before this leaf type can be allied to this genus with confidence. Andreánsky (1966) reported "undoubted" Lithocarpus from the Oligocene of Hungary. Yet, the published description and fig- ures do not reveal a basis for this assignment. Only two probable fossil Lithocarpus species have been reported from Asia (Takhtajan, 1982). These two leaf forms are from Kazakhstan, and their ages were not reported. The figures of these leaf types reveal intersecondary veins and biconcave teeth that are not characteristic of Lithocarpus leaves. : is unlikely that these leaf forms belong to this genus. The ib of Lithocarpus, as previously men- tioned, is of a generalized type and cannot be distinguished from other castaneoidean forms. Records of this pollen type, however, are com- mon throughout the Northern Hemisphere. The first records of this type of pollen are from the Upper Cretaceous (Santonian/Campanian) of Canada (Rouse et al., 1971). No additional re- ports of wood, fruits, flowers, or other fossil ma- terials have been made. The paucity of fossil Lithocarpus in present North Temperate zones may be a function of difficulties in the identification of isolated organs of this genus rather than a true absence. MacGinitie (1969) acknowledged these difficul- ties when describing a leaf form from the Eocene Green River Flora. He contended, however, that suitable distinctions could be made in most cases if fourth order venation was available. Unfor- tunately, he did not mention the specific char- acteristics with which these distinctions can be made. I did not observe any reliable differences in the higher order venation of these genera. Foliar features generally confirm the integrity ofthe genus as it has been recently circumscribed (Camus, 1934-1954; modified by Soepadmo, 1970, 1972). However, they do support the dis- tinction of Limlia as suggested by Masamune and Tomiya (1948) on other grounds. The fea- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 tures upon which this genus was originally based seem rather weak and when considered alone would not, in my opinion, justify separation at the generic level. However, Limlia leaves are distinct from those of nearly all Lithocarpus in that they lack type 9 trichomes and possess a dense coat of type 13 trichomes as in Casta- nopsis. Interestingly, additional work on wood anatomy (Lee, 1968) has shown that the closest affinities of the species in question lie with Cas- tanea and Castanopsis rather than Lithocarpus. This is consistent with the aforementioned leaf data. Hence, the recognition of this somewhat intermediate taxon at the generic level may be warranted. The same epidermal characteristics support the retention of the Castanopsis fissa- group in Castanopsis rather than inclusion in Lithocarpus as suggested by Camus (1934-1954). Finally, the significance of the foliar features of Lithocarpus densiflorus deserves consider- ation. The peu of а unique trichome com- plement and disti ts this species apart from all other Lithocarpus. Foliar features would provide very strong support for generic distinction if other, more definitive, fea- tures so indicated. The intraspecific variability in foliar characteristics would also support rec- ognition of some varieties as separate species (e.g., L. densiflorus var. echinoides). TRIGONOBALANUS FORMAN The genus Trigonobalanus was established in 1962 by Forman with a complement of one species, 7. verticillata, and was expanded to two species, with the reassignment of Q. doichangen- sis Camus, in Forman's detailed description of the genus (1964a). Lozano-C. et al. (1979) further expanded the genus to three species as the result of the discovery of T. excelsa in South America. The small size and proposed primitive position of this taxon (see Forman, 1964a) has made it an attractive subject for investigation (see For- man, 1962, 1964a, 1966a; Cutler, 1964; Erdt- man, 1967; Forman & Cutler, 1967; Hou, 1971; Soepadmo, 1972; Lozano-C. et al., 1979; Men- nega, 1980; Hernández-Comacho et al., 1980; Baas, 1982). This genus has been assigned to various subfamilies (Forman, 1964a, 1964b; Melchior, 1964; Lozano-C. et al., 1979) but has not been formally subdivided above the species level. Trigonobalanus, small as it may be, does show considerable intrageneric variation. Erdt- man (1967) for example, felt that the pollen mor- 1986] JONES— phology of T. doichangensis and T. verticillata differed sufficiently that inclusion in separate sec- tions or even separate genera should be consid- ered. Mennega (1980) found substantial differ- ences between the wood anatomy of these two species as well but felt that they were congeneric and closest to the Quercoideae. The male inflo- rescences of Trigonobalanus species vary as well, ranging from rigid, as in the Castaneoideae, to flexuous, as in most Quercoideae. The leaves of Trigonobalanus also exhibit ma- jor differences. Trigonobalanus verticillata, for example, possesses whorled leaves, a character- istic unique to this species within the Fagaceae. The leaves of all Trigonobalanus species are sim- ilar in form to those of many other tropical Fa- gaceae (i.e., some Castanopsis, Lithocarpus, and Quercus). All species possess entire or mostly entire leaves of similar gross form. Trigonobal- anus doichangensis, however, has leaves that are distinct from the other two species in that they are invariably entire and possess cuticles bearing papillae and rather exotic multicellular “peltate” (type 19) trichomes. Baas (1982) correctly con- cluded that Т. doichangensis is an outlier with respect to leaf anatomy. In fact, the leaves of T. excelsa (of South America) and T. verticillata (of Malaysia) share many more common foliar fea- ures than either do with 7. doichangensis (of Thailand). In spite of substantial intrageneric variability, leaf architectural and cuticular fea- tures together still allow leaves of Trigonobala- nus species to be distinguished from other trop- ical Fagaceae (see Appendix II). he intrageneric variability is undoubtedly re- lated to the wide “disjunct” distribution of this taxon. Trigonobalanus is found in Thailand, the Malay Peninsula, Borneo, Celebes, and in Co- lombia, South America. These ranges may well be extended in the future since the genus is so new and many of the areas in which it might likely be found are poorly explored. The known range of this genus is quite wide but patchy, sug- gesting it is a relict of a much broader past dis- tribution. A once broad distribution is almost extensive today, is quite similar to that of Tri- gonobalanus. Both were probably circumboreal in the past. Trigonobalanus probably entered South America during the Miocene along with Quercus. Neither genus has penetrated past Co- lombia in South America. Unlike Quercus, Tri- gonobalanus seems to have become extinct in FAGACEAE EVOLUTION 245 temperate regions, perhaps due to intolerance of the late Tertiary climatic deterioration. We must turn to the fossil record to confirm the presence or absence of this genus in higher latitudes of the Northern Hemisphere. The fossil record of Trigonobalanus is very scanty. There are no confirmed reports of fossil Trigonobalanus pollen although it is very similar to that of Quercus and therefore may have been assigned to typical quercoid form genera. No Tri- gonobalanus wood, flowers, or fruits have been confirmed but paleogene reproductive materials under investigation by W. L. Crepet and S. R. Manchester may prove to be from plants of this genus. In addition, Forman (1964b) has sug- gested that the fossil fruits assigned to Fagus suc- cinea Goeppert & Menge are quite similar to those of this genus. Mai (1970) assigned four fruit types from the Eocene and Miocene of Europe to Trigonobalanus. Forman (pers. comm., 1980) suggested that some of the types probably rep- resent Fagus and others, although possibly Tri- gonobalanus, cannot be assigned to this genus with certainty using available information. Mai (1970) also suggested that these fruits were cor- related with Castanea-like leaves previously as- -— to Dryophyllum furcinervis Schmalhau- en. The leaves have been placed in several genera dus Phyllites, Castanea, Quercus, and Cas- tanopsis (Krausel & Weyland, 1950). The cuticle, as illustrated by Kráusel and Weyland (1950), indicates that the leaves probably do not repre- sent Castanea. The trichome bases are scattered and simple, resembling those of Castanopsis. Mai (1970) correctly suggested that the epidermis is also similar to that of 7. doichangensis. Yet, the leaf architecture is much more similar to extant members of the Castaneoideae than to those of extant Trigonobalanus. The nature of the tri- chomes themselves, when known, will probably shed some light on the affinities of this leaf type. The reports of Trigonobalanus (Mai, 1970; Gre- gor, 1980; Takhtajan, 1982), if correct, would support the hypothesis of a Laurasian distribu- tion during the Tertiary. We still do not have any significant indication that Trigonobalanus was present in North America. A search for this genus in the fossil record of this continent is in order. Trigonobalanus must have evolved prior to the Eocene (most likely in the latest Cretaceous or Paleocene) if it is an ancestral form within the Quercoideae (see Forman, 1964a, 1966a; Elias, 1971). If Trigonobalanus evolved from a casta- 246 neoid ancestor, as Forman (1966a) and Elias (1971) imply on the basis of cupule character- istics, and if the cupule characteristics are linked to foliar characteristics, we might assume that leaves of the most primitive Trigonobalanus would most closely resemble those of the sup- posed ancestors (ie., near Chrysolepis, Elias, primitive species in this genus. This species pos- sesses entire leaves and multicellular branched trichomes that sometimes resemble the scales of Castanopsis. This hardly seems to be the case however, since this species is more advanced with regard to reproductive features that are usually considered to be more conservative (1.e., flexible male catkins and cupules usually bearing a single fruit). However, the scale-like trichomes may be derived, as they appear to be based on trichome data alone. We must remember that the evolu- tionary rates and even directions of various or- gans and characters can differ. Thus, primitive foliar characters cannot be determined with cer- tainty using information gathered from extant species alone. It seems advisable to wait for fur- ther information on the nature of fossil forms before making firm conclusions about evolu- tionary trends in leaf characteristics of this group. QUERCUS L. Quercus is the largest genus in the гадавое, 1971) to 600 (Soepadmo, 1972) spec genus has been studied often, but, because of its immense size and wide distribution, few have studied the genus in its entiréty: Camus (1934— hed the ge- nus in three В | volumes. Although some disagree with finer points of her classifi- cation scheme, most modern taxonomists gen- erally accept Camus’ circumscription of the ge- nus. Historically, however, the circ umscription nopsis, as a pin Schwarz (1 Dio. went so far as to divide Quercus into four separate genera. A significant number of taxonomists accept the recognition of Cyclo- balanopsis (e.g., Hsu & Jen, 1979) but few sup- port the extensive splitting proposed by Schwarz ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 (see, for example, Muller, 1942a). In contrast, many have suggested that Quercus should be united with morphologically similar Lithocarpus (e.g., de Candolle, 1868; Bentham & Hooker, 1880; Corner, 1939). Most modern workers (e.g., Camus, 1934-1954; Forman, 1964a, 1966a; Melchior, 1964; Soepadmo, 1968b, 1972) have suggested that there is a clear basis for distin- guishing these as two genera. This certainly seems to be the case from a palynological point of view, but the intrageneric variability and extensive parallelism in wood, leaves, fruits, and other fea- tures often make discrimination difficult without ко examination of fertile material. oliar variation in Quercus is not tightly cor- related es established major infrageneric clas- sification. Leaves of the subgenus Cyclobala- nopsis (sensu Camus, 1934-1954) can be separated from those of the subgenus Euquercus, but there is considerable overlap among the sec- tions of the latter. The sections often can be dif- ferentiated in a particular geographical area (Tre- lease, M Dyal, p Set 8, Barnett, 1944), b disti species on other geographical areas are con- sidered. Numerous examples of parallel and con- vergent evolution of leaf forms have been noted in in this genus pc 1974). Foliar variation in when ly associated with tal factors than with the definitive reproductive features that are the basis for classification at the section level. Quercus has the widest distribution of all fa- gaceous genera. Its species are found in nearly all north temperate forests and woodlands, often as dominant elements. They are also abundant in subtropical and tropical regions of the North- ern Hemisphere extending into the Southern Hemisphere only in northern South America and Indonesia. The wide geographical distribution probably accounts, at least in part, for the large variation in leaf form observed within Quercus. Brenner (1902) noted correlations between many foliar features and the geographical origin of the specimens. He also experimentally demonstrat- ed substantial intraspecific effects of the envi- ronment on leaf form There is more foliar variability in this genus than in any other genus of Fagaceae. The leaves range from entire, nearly glabrous, tropical forms with drip tips and a coriaceous texture, to lobed, hairy, chartaceous forms characteristic of many temperate species. Camus (1934—1954) and oth- ers have noted the existence of specific leaf types = 1 + d A 1986] among Quercus species. There are entire- mar- iny sclero- phyllous “holly” oaks, он" *chestnut" oaks as well as typical lobed “white” and “red” oaks. Unfortunately, the types are not section specific, making classification at this level diffi- cult when using leaf material alone (see Text-Fig. 4). Even trichome complements fail to allow clean separation of sections. This foliar variability often extends to lower taxonomic levels. The foliar variation within species is frequently dramatic, and foliar characteristics of species sometimes intergrade. Such intergradation is often the result of actual gene flow between species. The oaks are noted for their propensity toward hybridization (Trelease, 1917; Muller, 1942a, 1942b, 1952; Stebbins et al., 1947; Palmer, 1948; Bray, 1960; Benson, 1962; Hardin, 1975; Jensen & Esh- baugh, 1976), calling into question strict appli- cation of the biological species concept (Muller, 1942a; Burger, 1975; Crovello, 1976; van Valen, 1976). The capacity for genetic exchange may be one of the factors that makes the oaks very com- petitive in a wide range of environments. The present broad distribution of Quercus is in harmony with the fossil record of this group. Quercus, like most fagaceous genera, had a “hol- arctic" distribution during the Neogene. There are abundant reports of fagaceous fossils from North America (see Trelease, 1918, 1924; LaMotte, 1952). These include reports of leaves (e.g., Lesquereux, 1878; Edwards, 1931; Chaney & Sanborn, 1933; Axelrod, 1956, 1964; Dorf, 1960, 1964; MacGinitie, 1969; Daghlian & Cre- pet, 1983; Manchester, 1983), wood (e.g., Platen, 1908; Schuster, 1908; Eames, 1910; Boeshore & Jump, 1938; Beyer, 1954; Prakash & Barghoorn, 1961a, 1961b; Wheeler et al., 1978), and repro- ductive materials including inflorescences (e.g., Daghlian & Crepet, 1983), nuts (e.g., Newberry, 1898; Manchester, 1981a, 1983), and cupules (e.g., Bones, 1979; Daghlian & Crepet, 1983; Manchester, 1983). Quercus remains are also abundant in Europe (Andreánsky & Kovacs- Sonkody, 1955; Kirchheimer, 1957). Leaves are reported most often (e.g., Schimper, 1870-1872; Kovacs, 1961; Andreánsky, 1966; Gazeau & Koeniguer, 1968; Sitar, 1974; Petrescu, 1976; Gros, 1983; Hummel, 1983), fruits (e.g., Heer, 1869; Saporta & Marion, 1878; Raniecka-Bob- rowska, 1962; Hummel, 1983), cupules (e.g., Goeppert, 1855; Gregor, 1980; Hummel, 1983), and wood (Miiller-Stoll & Madel, 1957; Brett, 1960; Selmeier, 1971; Prive, 1975), are also very JONES— FAGACEAE EVOLUTION 247 General leaf type Cyclobalanopsis Euquercus n mn 3 2 El 212151513 © о © «азер E D a o - - Ф Ф Ф 5 e > о > 4 = a ul Entire or partially Nu M € oo to la evergreen e Entire, broad tradily deciduous to evergreen e e Xeromorphic, oe small, mm margined evergre ° е е Sclerophyllous holly- like e. е ә е e Craspedodromous “chestnut'’-like (@) е Ф е (e) Rounded coarsely toothed, deciduous е e Roundly lobed deciduous e e e. е Sharply lobed deciduous e e e TexT-FIGURE 4. Distribution of general = types among the subgenera and sections of Quercu common. Takhtajan (1982) listed and illustrated many Quercus leaves, woods, and reproductive materials from eastern Europe and Asia. Hu and Chaney (1940) reported Quercus-like leaves of the castaneoid type from the Oligocene to the Pliocene of China. More recent Chinese reports (Chinese Neogene Plant Editorial Group, 1978; Shuangxing, 1980) listed several fagaceous fossils including Quercus leaves from the Eocene. A large amount of work has been done on the Tertiary flora of Japan, and many species of Quercus and other fagaceous genera have been found (Nat- horst, 1888; Tanai & Onoe, 1961; Chaney, 1963; Tanai & Suzuki, 1965; Matsuo, 1968; Tanai, 1970; Tanai & Yokoyama, 1975). Quercus has been reported also from various arctic and other high latitude localities (Heer, 1870, 1883; Hol- lick, 1936; Koch, 1963). It appears that Quercus is a recent immigrant to Africa for it is found only near the Mediterranean from the late Neo- gene to the present (Kráusel, 1939; Raven & Ax- elrod, 1974). There have also been reports from Australia and other Southern Hemisphere local- ities, beyond the current range (Ettingshausen, 18872, 1887b, 1888; Krasser, 1903; Berry, 1923), but these appear to be in error. The fossil record of Quercus parallels that of many other north temperate arborescent forms. Quercus appears to have been an important part of the ““Arcto-Ter- tiary” geoflora. During the course of the Tertiary, species of this group have migrated latitudinally in response to climatic conditions (see, for ex- ample, Tanai, 1961, 1967; Tanai & Huzioka, 248 1967), but they have maintained their ability to compete favorably with other broad-leaved forms. Unlike Lithocarpus and Castanopsis, Quercus has survived in Europe as well as Asia and across North America. The time and place of the origin of Quercus is uncertain. Pollen of the Quercus type has not been found in sediments older than the Oligo- cene (Muller, 1981). Yet, many reports of Cre- taceous leaves exist (e.g., Quercophyllum chin- kapinense, Ward, 1905, Cenomanian). Most are poorly preserved and none have proven to be Quercus. Trelease (1918, that many fossil leaves assigned to this genus are placed there for lack of a better place to put them. This is particularly true of the fossil leaf types he assigned to form groups that have no counterparts among modern Quercus (e.g., Fraxinifolae, Distinctae ). The tendency to assign fossil leaves to Quercus with- out solid morphological or anatomical evidence is probably a function of two intrinsic properties of the oaks. First, they are so widespread and abundant that oaks might be expected in almost any Northern Hemispheric arborescent angio- sperm flora. Second, Quercus leaves are so vari- able that counterparts to many fossil leaf forms can be found among the highly sn and wide ranging leaf types found amon rn mem- bers of this genus. The first believable, but still not conclusive, record of Quercus (Bones, 1979; Manchester, 1981a, 1983) is from the Middle Eocene Clarno Formation of Oregon. Leaves and wood, as well as nuts and cupules can be found here, strongly suggesting the presence of Quercus in fairly modern form at that time. Floral ele- ments are needed to confirm the generic affinity, however, since the features found in the fossils are not definitive. Similar wood and fruits can be found, for example, in Lithocarpus. Crepet and Zavada (1983) have an inflorescence that is essentially like those of Quercus from the Middle however, is different from that of typical Quercus pollen. The significance of the differences are cur- rently being evaluated and may be develop- mentally rather than taxonomically related. It is likely that this will prove to be the earliest de- finitive evidence for the existence of Quercus or a close relative thereof. Flowers of this genus have been reported from the Eocene-Oligocene ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Catahoula Formation (Daghlian & Crepet, 1983). Based on this evidence we can say that Quercus was definitely present by the Oligocene and was probably present in the Eocene or perhaps ear- ier. When the leaf record is examined, we find that the earliest apparently valid reports of Quercus are of entire- and serrate-margined forms. Bren- ner (1902) felt that the entire-margined forms were indeed the most primitive within the genus. There are lobed forms reported from the Cre- taceous, but the nature of the leaves themselves does not provide convincing evidence of faga- ceous affinity. In fact, they are most often closer to other extant families. The first plausible lobed oak is that reported by Daghlian and Crepet (1983) from the Oligocene. Lobed forms do not become common, however, until the Miocene (Tanai & Yokoyama, 1975). Rüffle and Knappe (1977) suggested that lobation in oaks is an ata- vistic phenomenon that recalls the palmately lobed nature of distant ancestors, presumably of the Debeya- Dewalquea complex (see also Rüffle, 1978, 1980). This is an interesting hypothesis, but there does not seem to be much evidence to support it at this time. Holly-like Quercus leaves have been reported from the Cretaceous and throughout the Tertiary. Ettingshausen (1896) felt that this was the primitive leaf form within the genus and that all other forms were derived. This is a common form in the section Protobalanus, which is held to be primitive among American oaks. Trelease (1918, 1924) believed that all the American oaks evolved from a single holly oak ancestor (i.e., a mid-Tertiary Q. chrysolepis-like form). A few of the early holly oak fossils may well be ancestral within Quercus, yet, some have been shown to belong to Лех or Nyssa when examined more closely (Куабек & Walther, 1981). This leaf type became common during the Miocene, when drier climates, for which this leaf type is best suited, prevailed. Castanea-like leaves are also among the oldest to be assigned to Quercus. However, forms with regularly spaced, uniform teeth and secondaries do not appear until the Eocene. These “chestnut” forms may be Quercus, but leaves of this type are very difficult to distinguish from similar forms in other genera of Fagaceae as well as some non- fagaceous taxa. The problems of determining the generic affinities of these forms have been dis- cussed by numerous authors (e.g., Trelease, 1924; Schwarz, 1936b; Camus, 1934-1954; Mac- Ginitie, 1969) and have been confirmed in Jones 1986] (1984). There are slight differences in the modern forms (particularly the nature of the multicellular “glandular” trichomes) that usually allow mod- ern forms to be distinguished, but there are no consistent generically correlated features that can be used to assign a fossil form to an extant genus with confidence. As with all forms, cuticular in- formation greatly enhances the chances of cor- rectly classifying a fossil leaf type. Kráusel and Weyland (1950), for instance, used cuticular evi- dence to justify transfer of a fossil putative chest- nut oak Q. furcinervis (Rossm.) Unger to the ge- nus Castanopsis. Although I do not share the certainty with which Kräusel and Weyland made the transfer, the their position. If the nature of the trichomes, ‘not just the trichome bases, were known, such a transfer might be made with assurance. Although the true affinities of many putatively fagaceous fossil leaves remain to be established, different leaf forms appear to have arisen at dif- ferent times in this family. The entire and holly- like leaves from the Upper Cretaceous and Pa- leocene may be from the first Quercus species. Coarsely round-toothed and regular chestnut oak types do not appear until the Eocene. Lobed forms first occur in the Oligocene and become common in the Miocene. A thorough fossil fagaceous leaves is needed, confirm the affinities of these fossils and help resolve questions about the origin and subse- quent evolution of Quercus. FINAL DISCUSSION AND CONCLUSIONS Leaves of the Fagaceae vary considerably within broad limits. All extant Fagaceae have simple leaves of high rank with obliquely ori- ented tertiary veins, anomocytic and/or cyclo- cytic stomatal complexes, and specific trichome complements, which depend on the intrafamilial taxon in question. Nearly all fagaceous leaves can be classified into one of the following major mor- photypes: 1) entire or partially toothed, elliptical to lanceolate evergreen forms; 2 үнөн. broad, tardily deciduous to basin Mid xe morphic, р — forms; 5) io a “chestn " like forms; 6) coarsely round-toothed, cen forms; 7) roundly lobed deciduous forms; 8) sharply lobed deciduous forms; and 9) narrow willow-like leaves. Many of these leaf types, how- ever, cross taxonomic boundaries, and there is JONES—FAGACEAE EVOLUTION 249 considerable foliar variability among intrafa- milial taxa. Leaves of extant Fagaceae can be distinguished from those of other families when cuticular an architectural features are available, but the in- trafamilial taxonomic utility is somewhat lim- ited because of the variability of foliar characters. Leaves of most Fagaceae can be classified at the species or species complex level. Unfortunately, leaf character differences often do not coincide with established subfamilial, generic, subgeneric, and even sectional boundaries. Thus, it is some- times impossible to separate these intermediate level taxa without including more than one char- acter assemblage of subordinate taxa. Some fea- tures or groups of characters are unique to spe- cific taxa whereas other foliar syndromes can be found among several. Entire-margined evergreen leaves with type 9 trichomes, for instance, are found only in the genus Lithocarpus whereas reg- ularly toothed, craspedodromous, chestnut-like leaves with types 5, А апа 15 and/or 16 tri- chomes, are found in four genera and two subfamilies. Thus, Pr ston among variou intrafamilial taxa can be easy to impossible, = pending on the uniqueness of the character as- semblage of the leaf at hand. Foliar information, when 1 with other taxonomically useful data, contributes to our un- derstanding of the taxonomy of this family. A classification scheme based on foliar and other ata is presented in Table 2. This proposed clas- sification ела oe in some respects from the widel ose ed by Forman (1966a). Leaf data, for а indicate that Nothofagus is distinct from all other Fagaceae and is, instead, closer to the Betulaceae. The dou- bly serrate margins and large glandular tri- chomes, common to Nothofagus and members of the Betulaceae, are rare or absent in all other modern Fagales. A distinction between Notho- fagus and the Fagaceae sensu stricto is also ap- parent when pollen (Praglowski, 1982), chro- mosome numbers (Armstrong & Wylie, 1965), inflorescences (Reece, 1938), vegetative features such as wood (Shimaji, 1962), and phytochem- ical data (Oesch, 1969) are examined. Foliar in- proposed by Kuprianova (1962) and Nixon (1982) and suggests the НЕ и Р ships illustrated in Text-Figure infor- mation tells us little, however, abet a intra- generic taxonomy of this genus. The entire- 250 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 / TABLE 2. Proposed classification of genera traditionally considered to be fagaceous. Nothofagaceae Fagaceae Fagoideae Castaneoideae Trigonobalanoideae Quercoideae Nothofagus Fagus tan Trigonobalanus Quercus Chrysolepis Castanopsis Lithocarpus а Trigonobalanus is а heterogeneous assemblage that should be divided when a more complete knowledge of ed. he component species is obtain margined tropical forms (i.e., those of the ‘brassi’ pollen group) do seem to be a tightly knit unit, but little correlation can be found between the various lines of morphological and anatomical evidence in other groups. The lack of correlation tween various lines of morphological and an- atomical evidence has been noted by previous authors (Hanks & Fairbrothers, 1976). Thus, if one accepts the Nothofagaceae, recognition of the ‘brassi’ group as a separate genus could be justified but it would be hard to support further division of the remaining species at this level. Fagus, often thought to be closely related to Nothofagus, is clearly distinguishable from this and other genera of Fagaceae and thus should be retained in its own subfamily, the Fagoideae. The uniformity of the leaves in this genus supports the lack of formal division into subgenera or sec- tions. The leaves of the Quercoideae and Cas- taneoideae are most often distinguishable. How- ever, certain forms like the “‘chestnut” leaf type are found in both subfamilies. This foliar overlap may be used in support of the recognition of a single subfamily, the Castaneoideae, as suggested most recently by Brett (1964). Overlap in wood anatomy, fruit, and cupule structure also ob- scures the distinction between these two subfam- ilies. Pollen morphology provides convincing evidence, however, for separating these two taxa. Another problem, at the subfamilial level, is the proposed recognition of the Trigonobalanoideae (Lozano-C. et al., 1979). The foliar characteris- tics of вы (e. g., tooth na trichome tly distinct from those of other Quercoideae to rationalize ша: in a separate subfamily when considered in light of differences in chromosome numbers as well as the morphological and anatomical features re- viewed by Lozano-C. et al. (1979). Removal of Trigonobalanus from the Quercoideae would also elp sharpen the distinction between the Quer- coideae and the Castaneoideae. Within Trigo- nobalanus there is sufficient foliar diversity to support further separation of this genus into dis- crete subgenera or even separate genera, as pro- posed by Erdtman (1967) on the basis of pollen orm and structure. Quercus presents numerous taxonomic prob- lems. The division of Quercus into several gen- era, as proposed by Schwarz (1936a), is not war- ranted by foliar data. The intrasectional variability is great in nearly all sections and there is a great deal of overlap. The subgenus Cyclo- balanopsis as proposed by Camus (1934-1954) is a rather discrete unit and probably merits its subgeneric status. The sections of the subgenus Quercus show general trends in foliar features, but the sections cannot be discretely separated using leaf characters alone. Foliar features generally confirm the current consensus concerning generic delimitation with- in the Castaneoideae. Chrysolepis is clearly dif- ferentiable from Castanopsis on the basis of dif- erences in leaf architecture and trichome complements. This supports the conclusions of Hjelmqvist (1948), Forman (1966b), and many other modern authors, who base their conclu- sions primarily on reproductive features. Cas- tanea is a uniform | and well defined taxon. The rly all organs of тта genus. . There i is little difference e general leaf architectural and cuticular features making identification to the generic level diffi- cult. In Castanopsis there are few differences among the leaves of most sections, and thus foliar fea- tures cannot be used to support these sectional divisions. Foliar data do support retention of the Castanopsis fissa group within Castanopsis rath- er than placement in Lithocarpus. This is in ac- 1986] Quercus Recent Pleistocene Pliocene \ Miocene | \ x 1 Oligocene \ а. Quercoideae — Trigonobalanoideae Castaneoideae Eocene \ \ N ү Paleocene JONES— FAGACEAE EVOLUTION Trigonobalanus Lithocarpus Castanopsis Chrysolepis Castanea Fagus 251 Nothofagaceae Betulaceae Pseudofagus I Pseudofaginae / Fagopsis Fagoideae / d P. P IM ancestral complex A Cretaceous € T um cmm uml m m — TEXT-FIGURE 5. cordance with the findings of Luong (1965) and Forman (1966b) based on infructescence studies. It is apparent that Lithocarpus is a large and relatively unstudied genus that requires further taxonomic revision. The leaves of this group, as currently circumscribed, are fairly uniform in gross morphology but trichome differences, par- ticularly the absence of type 9 trichomes in rel- atively few specimens, provide incentive for the much needed revision of this group. It does ap- pear that Lithocarpus densiflorus stands apart from all other members of this genus and there is some indication that Limlia may deserve rec- ognition as a separate genus. Additional conclu- sions concerning the impact of foliar data on the classification of other infrageneric taxa in this and some other genera of Fagaceae must await analysis of a more complete sample. The potential for identifying and correctly classifying fossil fagaceous leaves diminishes as the age of the material increases. Although this is substantially intuitive, the age at which one must use extreme caution when assigning ma- terials to modern genera or subfamilies is much more recent than one might have expected only a decade ago. The recent discovery of Pseudo- fagus (Smiley & Huggins, 1981) and the reinter- pretation of Fagopsis (Manchester & Crane, 1983) serve to illustrate that leaves similar to those of extant genera can be found on plants of extinct genera and subfamilies, in sediments as young as the Miocene. The strength of any determi- nation based on leaves must therefore rest sub- stantially on correlation with fossils of other more definitive organs. The ability to determine the affinities of more recent materials will depend, as with modern leaves of this family, on how unique the particular leaf type is within the fam- ily The time and place of origin of the Fagaceae is still largely a mystery. No leaves or other or- gans definitely assignable to the Fagaceae have Possible phylogenetic relationships within the Fagales. been found prior to the Eocene. The diversity present in the Eocene, however, suggests an or- igin at least as early as the Paleocene. The ex- istence of primitive Fagaceae, or possible pre- cursors thereof, in the Cretaceous (Santonian) of North America is indicated by the recent dis- covery of a “fagalean” flower by Friis (pers. comm.) and Tiffney (pers. comm.). The ability to identify early remains as Fagaceae may well be limited by the existence of vegetative and re- productive strategies, such as wind and/or water fruit dispersal, alien to modern Fagaceae with which we are familiar. The modern large animal dispersed fruits, that is, acorns, chestnuts, and beechnuts, do not appear until the Eocene (Tiff- ney, 1984). Their development was probably a function of coevolution with small mammals, for example, rodents. Rodents first appear in the up- permos st Paleocene. and squirrels, which com- monly evolved in the late Eocene (Romer, ‚ 1966). It is reasonable to assume then that Paleocene and earlier forms would h of modern members ofthe family. Similar trends in fruit evolution occur in the Juglandaceae (Manchester, 1981b) and most likely the Betu- laceae as well. Most modern authors contend that the Faga- those о this conclusion if one considers the relation of this taxon to the Fagaceae sensu stricto and the Betulaceae. If Nothofagus is closer to the Betu- laceae an its disti ti E їе Ё Пу, it more than likely evolved independently from a common fagalean ancestor during the late Cre- taceous. This ancestor or ancestral complex probably extended its range southward, or north- ward according to Schuster (1972), at or prior to this time. This migration may well have occurred through Africa, as Raven and Axelrod (1974) 252 have suggested. Examination of the Upper Cre- taceous floras of Africa for evidence of fagalean fossils is needed to help test this hypothesis. The known fossil record suggests that the Nothofagaceae evolved from this primitive fa- galean complex in the Southern Hemisphere (probably in southern Australia or Antarctica) while the Fagaceae sensu stricto evolved from representatives of the same complex in rn Hemisphere. The fossil record indi- cates, with minor recent exceptions, that neither family has occurred in any hemisphere other than that in which it is found today. It is extremely difficult to be more specific about the place in which the Fagaceae sensu stricto evolved. It has been noted that American forms of both Fagus a O erica. It also seems that the greatest diversity has existed on the Eurasian land mass. Whether diversity is a good indicator of the site of origin of a taxon remains to be proven. The fossil record, upon which final resolution of this problem must depend, is not known well enough to confirm southeast Asia, as many have suggested, or any other Е locus, as the site of origin for the Fagaceae. We know little more about the nature of the first Fagaceae. When the first firm record of the Fagaceae appears (in the Eocene) we have both the Castaneoideae and Quercoideae represented in fairly modern form. The first records of mod- ern genera are Castanea and Quercus, but neo- botanists have generally considered these to be derived forms within the family. Lithocarpus, Castanopsis, Chrysolepis, and Trigonobalanus are the genera that are usually cited as candidates for evolutionary ‘Urpflanzen’ within this group. Given the different selective pressures, particu- larly those associated with fruit dispersal, it seems likely that the earliest Fagaceae would be differ- mplex. en the mutual characteristics of the Betu- laceae, Nothofagus, and the Fagaceae sensu stric- to are considered, we might expect the ancestral plexus to consist of wind dispersed ‘‘amentifer- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 us” forms with irregularly toothed (perhaps doubly serrate), craspedodromous, simple leaves. Such leaves have been found in the late Creta- ceous and have been attributed to the Betulaceae (Wolfe, 1973). Crane (1981), however, indicated that the first solid evidence of the existence of the Betulaceae occurs in the Paleocene. Whether or not these leaves are betulaceous, fagaceous, Nothofagus, or some fagalean intermediate can be determined only when reproductive materials are found and analyzed. There does seem to be a pattern to the ap- pearance of various leaf forms in the Fagaceae. If we disregard the Cretaceous leaf forms as sug- gested by Wolfe (1973), the first fagaceous leaves appear to be irregularly serrate, simple, pinnate forms with percurrent tertiaries obliquely dis- posed in relation to the midrib. Many of these occur in the Paleocene and have been placed in the genus Dryophyllum by previous workers. During the Eocene we have the appearance of the “chestnut” leaf type and coarsely round- toothed oak leaves. The first sharply lobed oaks outed in the Oligocene and become abundant n the Miocene. The earliest record of round- lobed leaves occurs in the Miocene where they are particularly abundant in Eurasia. Because of numerous homeomorphies, it is difficult to know which of the fossil holly-like forms and laurel forms are truly fagaceous. In spite of their alleged antiquity, fagaceous holly-like forms may not have arisen until the climatic deterioration at the end of the Eocene and did not become abundant until the Miocene. It is also impossible to know which of the laurel-like and other entire-mar- ature of their gined fi Iorms are cuticles has been determined. Rüffle and co-workers (see Rüffle, 1980) have proposed the only phylogenetic model for this oaks, for example, Q. falcata, are atavistic expressions of the early platanoid types. This is a very intriguing hypothesis, which may be cor- rect, yet the cuticular evidence that these workers cite is not very strong. No trichome data are 1986] offered and the characters that are consistent with those of the Fagaceae (e.g., anomocytic stomatal complexes) are common to many other taxa as well. The validity of this proposal must be tested by rigorous examination of relevant fossil leaf forms in conjunction with the pioneering work on reproductive materials being conducted by Crepet, Daghlian, Manchester, and others. LITERATURE CITED ABBE, E. C. 1974. Flowers and inflorescences of the “A mentiferae.” Bot. R caster) 40: 159-261. AGUIRRE, A. C. & E. . Arquitectura foliar de las especies australianas y neocelandesas de ч (Еарасеае). Bol. Soc. Argent. Bot. 20: 227-240 Ames, H. & W. RIEGEL. 1962. Palynological inves- tigation of coals from the Chickatoon ÉL ation, Alaska. Pollen & Spores 4: 328 [Abstra ANDREÁNSKY, G. 1966. On the Upper ie M flora ungary. Stud. Biol. Hung. 5: 1-151. & Kovacs-Sonkopy (editors). 1955. Glei- derung und Okologie der jungeren Tertiárfloren Ungarns. Magyar Allami Fóldt. Intéz. Evk. 44: 1- 326 ARCHANGELSKY, S. & E. RoMERO. 1974. Losregistros mas antiquos del polen de Nothofagus (Fagaceae) de Patagonia CUM & Chile). Bol. Soc. Bot. México 33: 13-30 ARMSTRONG, J. M. & A. P. WYLIE. 1965. A new basic chromosome number in the family Fagaceae. Na- ture 205: 1340-1341. ЭР D.I. 1944. The Sonoma flora (California). R. W. Chaney (editor), Pliocene Floras of Cal- jui and Oregon. Publ. Carnegie Inst. Wash. 553: 167- 1956. Mio- Pliocene floras from west central . 33: 22 southern Idaho. Univ. Calif. Publ. Geol. Sci. 51: 1 1- 1966. The Eocene Copper Basin flora of northeastern Nevada. Univ. Calif. Publ. Geol. Sci. 59: 1-125 Р. КАУЕМ. А Late Cretaceous апа Tertiary vegetation histo f Africa. Pp. 79-130 in M. J. Werner (e с oscar and Ecol- ogy of Souther Africa. Junk Publ., The Hague. Baas, P. pian Fa leaf anatomy of Tri- pibe Forman Fagaceae. Blumea 28: 171- 17 BANDULSKA, H. 1923. A preliminary paper on the cuticular structure of certain dicotyledonous and coniferous leaves from the Middle appe ek of Bournemouth. J. Linn. Soc., Bot. 46: 2 On the cuticles of some site and fossil Fagaceae. J. Linn. Soc., Bot. 46: 427-441. BARNETT, E. C. 44. Keys to the species groups of Quercus, Lithocarpus and Castanopsis of eastern Asia, with notes on their oe Trans. Bot. Soc. Edinburgh 34: 159- Benson. L. 1962. Plant о Methods and Principles. Ronald Press Co., New York. JONES— FAGACEAE EVOLUTION 253 BENTHAM, G. & J. D. Hooker. 1880. Genera Plan- tarum, Volume 3, Part 1. L. Reeve & Co., London. Berry, E. W. 1916. Lower Eocene floras of south- eastern North America. Profess. Pap. U.S. Geol. Surv. 91: 1-481. 1923. = Ancestors. Williams & Wilkins Co., Baltimor 1 а flora from the Rio гора Argentina. Special Pap. Geol. Soc. Amer. 12: 194 5. The Lower Eocene flora of southeast- ern North America. J. Wash. Acad. Sci. 35: 89. Beyer, A. F. 1954. Some petrified wood from the Specimen Ridge Area of Yellowstone National Midl. Naturalist 51: 553-567. . Die AN eis geformter Pflan- u enprofil des uw Leop. » s. 8: 377 . 1938. A is эш, oak . Bot. 25: 307 -3 m north central Oregon. Oregon Mus. Sci. In- dustr. Occas Nat. Sci. - шесе Е. ALARD. 1960. ntribution a Co etude paleoxylologique de la Patagonie (I). "dici оны Leth. 41: 297-31 Bray, J. В. 60. A note on hybridization between Quercus macrocarpa Michx. and Quercus ple td Willd. in Wisconsin. Canad. J. Bot. 38: 701-7 BRENNER, W. Klima und Blatt bei der cubus Quercus. Flora 90: 114-160. Brett, D. W. Fossil oak wood from the British Eocene. Palaeontology 3: Я 1964. Тһе еке ce of Fagus and Cas- tanea, and the evolution of the cupules of the Fa- А 96- 118. . The species concept in Quercus. 5- Camus, A. 19 21929. Encyclopédie Economique de RM add. Volume 3, Les Chátaigniers. Mono- aphie des Castanea et Castanopsis. Académie e Sciences, Paris. 1934-1954. дао Economique de Sylviculture, Volumes 6-8, Les Chénes. Mono- graphie du Genre Bn: (et Lithocarpus). Aca- démie des Sciences, Paris. 1951. Le genre м Rev. Int. Bot. App l. Agric. Trop. 31: CANDOLLE, A. DE. 1868. а Systematis Na- turalis Regni Vegetabilis. Cupuliferae, Part 16(2). Victoris Masson & Filii, Paris. CHANDLER, M. E. 1957. The Oligocene flora of the Bovay Tracy Lake Basin, poe Bull. Brit. Mus. (Nat. Hist.), Geol. 3: The Lower го о E as of Southe England IV. British Museum (Natural History), London CHANEY, R. W 944, The Dalles flora. /n R. W — — (edito ae Association to Commemorate the 80th 254 Anniversary of the Geological Survey of Japan, E. I. SANBORN. 1933. The Goshen flora of west central Oregon. Publ. Carnegie Inst. Wash. 39: 1-1 E PLANT EDITORIAL GROUP. 1978. CHMURA, C. 1 Ор ра i Macsichtan) angiosperm pollen from the Western San Joaquin Valley, California. U.S.A Mese эми ч Abt. В, Palaophytol. 141: 89— 171. е Н. 1886. Die Flora des Bernsteins. Zwei- Band: Die Angiospermen des Bernsteins. En- elm mann, Danzig. Cookson, I. 1958. Fossil pollen grains of Nothofagus from Australia. Proc. Roy. Soc. Victoria 71: 25- 30. К. M. PikE. 1955. The pollen morpholog of Nothofagus Bl. subsection Bipartitae Steen. Austral. J. Bot. 3: 197-206. Cooper, A. W. & E. P. MERCER. 1977. Morphological variation in Fagus grandifolia Ehrh. in North Car- olina. J. Elisha Mitchell Sci. Soc. 93: 136-149. Corner, E. J. 1939. Notes on the systematy and dis- tribution of Malayan phanerogams. Gard. Bull. Straits Settlem. 10: 239-329. Couper, R. A. 1953. Distribution of Proteaceae, Fa- gaceae and Podocarpaceae in some Southern Hemisphere Cretaceous and Tertiary Beds. New Zealand J. Sci. Technol. 35: 247-250. 1960a. Southern Hemisphere Mesozoic and Tertiary Podocarpaceae and Fagaceae and their paleogeographic significance. Proc. Roy. Soc. Lon- don, Ser. B, Biol. Sci. 152: 491—500 1960b. New Zealand Mesozoic and Caino- zoic plant microfossils. New Zealand Geol. Surv. Bull. 32: 1-87. Cozzo, D. 1950. Estudio del leño fosil de una Di- cotiledonea de la Argentina Nothofagoxylon neu- quenense. Comun. Inst. Nac. Invest. Ci. Nat., Ser. Ci. Bot. 1: 3-12 8. Angiosperm leaves from the Low- nd. Cour. Forsch.- 32. . Betulaceous leaves and fruits from the British Upper Paleocene. Bot. J. Linn. Soc. 83 103-136. CRANWELL, L. M. 19 Auckland Inst. Mus 9 Fossil pollen from ‘Seymore Island, Antarctica. stabs 184: 1782-1 othofagus: living e fossil. In J. K. Gressitt (editor), Pacific Basin Biogeography Sym- Museum Press, Honolulu. 964. Antarctica: cradle or grave for its Noth- ofagus. Pp. 87-93 in L. M. Cranwell (editor), An- cient Pacific Floras. Univ. of Hawaii Press, Ho- . Southern beech pollens. Rec. : 175-1 CREPET, W. L. & C. P. DAGHLIAN. 1980. Castaneoid inflorescences from the Middle Eocene of Ten- nessee and the diagnostic value of pollen (at the subfamily level) in the Fagaceae. Amer. J. Bot. 67: -757. ——— & M.S. ZavADA. 1983. Quercoid catkins from ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 the Middle Eocene. Amer. J. Bot. 70(5, 2): 70. [Abstract.] — L. 1952. ae of Phytogeography. W. , The Hagu Cronous A. 1981. An Integrated System of Clas- n of Flowering Plants. Columbia Univ Numerical approaches to the Evol. 125: 179-187. natomy of vegetative organs of Trigonobalanus Forman (Fagaceae). Kew Bull. 9, 17: 401-4 DaDsweELL, H. E. & Н. D. INGLE. 1954. The wood же of the ie Guinea Nothofagus Bl. Aus- ral. J. Bot. 2: 141-153. пысы C. E L. CREPET. 1983. Oak catkins, leaves and fruits from the Oligocene Catahoula о and their evolutionary significance. r. J. Bot. 70: 639-649. бай ы. Р. J. 1965. Biogeography of the South- End of the World. Harvard Univ. Press, Cam- bridge. DETTMANN, M. E. & G. PLAYFORD. 1969. Taxonomy of some Cretaceous spores and pollen grains from eastern Australia. Proc. Roy. Soc. Victoria, n.s. 9-93. DILCHER, D. L. 1973. A revision of the Eocene flora of southeastern North America. Palaeobotanist 20: 7-18. 1974. Approaches to the identification of an- giosperm leaf remains. Bot. Rev. (Lancaster) 40: 1-157. Доре, 1. А. 1908. Notes dendrologiques sur les cha- 1 тепса. Publ. Carnegie Inst. Wash. 476: 73-124. 1960. ем forests of Yellowstone Ма- tional Park Wyoming. Pp. 253-260 in Billings Geological Society Guidebook, 11th Annual Field Conference, Billings Geological Society, Billings. e petrified аа of Yellowstone Park. Amer. Sci. 210: 105- Dusén, P. 1899-1902. Uber "A tertiüre Flora der Magellanslánder. Svenska Exped. Megellanslán- derna 1895-1897, 1: 87-107, 241-248. . Uber die tertiáre Flora der r Seymor- Insel. Wiss. Ergebn. Schwed. Südpolarexped. 1901- 1903, 3: 1-27. Dvar, S. C. 1936. A key to the species of oaks of eastern North America based on foliage and twig characters. Rhodora 38: 53-63. 910. On the origin of the broad ray in b ES ons s (Ligna). Part i "The genera of Fagaceae in the southeastern United States. J. Arnold Arbor. 52: 159-195 74. Pe Sop in North America. An ыы to Pollen Anal- ysis. Ronald A New York. On the pollen morphology of Trigo- nobalanus (Fagaceae). Bot. Not. 120: 324-333, 1986] ETTINGSHAUSEN, C. F. von. 1887a. Beitráge zur Ke nntniss der Tertiürflora Australiens. Denkschr. serl. Akad. Wiss., Math.-Naturwiss. Kl. 54: 81- 7b. Beitráge zuer Kenntniss der Tertiár- ds. Denkschr. Kaiserl. Akad. Wiss. Math.-Naturwiss. Kl. 54: 143-192 1888. Contributions to the Tertiary flora of Australia. Mem. rv. New South Wales, Palaeontol. 2: 1-189. 1891. Uber tertiare Fagus-arten der südlich- en Hemispháre. Sitzungsber. Kaiserl. Akad. Wiss., Math.-Naturwiss. Cl., Abt., 1, 100: 114-137. . 1896. Über die Nervation der Blátter bei der Gattung Quercus, mit besonderer Berücksichti- gung ihrer vorweltlichen Arten. Denkschr. Kaiserl. Akad. Wiss., Math.-Naturwiss. Kl. 63: 117-180. FERGUSON, D. K. 1971. The Miocene flora of Krezau, western Germany. 1. The leaf-remains. Verh. Kon Ned. Akad. Wetensch., Afd. Natuurk., Tweede Sect. 60: 1-297. Fey, B. S. & P. К. ENpress. 1983. Development and morphological interpretation of the cupule in Fa- aceae. Flora 173: 451—468. FoRMAN, L. L. 1962. A new genus in the Fagaceae. Taxon 11:139-140 964a. Trigonobalanus a new genus of Fa- gaceae, with notes on the classification of the fam- ily. Kew Bull. 17: 381-396. 1964b. Trigonobalanus and its importance in the taxonomy of the acr s Roy. Soc. London, Ser. B, Biol. Sci. 161: 1966a. On the evolution Ks ie in the Fagaceae. Kew Bull. 18: 385-419. 196 Generic delimitation in the Casta- neoideae. Kew Bull. 18: 421-426. UTLER. 1967. Additional notes on Trigonobalanus (Fagaceae). Kew Bull. 21: 331- 337. FRENGUELLI, J. 1941. Nuevos elemento floristicos del Megellaniano de Patagonia austral. Notas Prelim. Mus. La Plata 6: 173-202. GAZEAU, F. & J. C. KOENIGUER. 1968. Les bois het- eroxyles du Mio-Pliocene de Touraine. Com. Trav. Hist. Sci., Mem. Sect. Sci. Paleobot. 2: 56-69. GOEPPERT, H. R. 1855. Die Tertiáre Flora von Schossnitz in Schlesien. Górlitz GOTHAN, W. 1908. Die E Hólzer von der Sey- w Hill-Insel. Wiss. Ergebn. Schwed. е eau u bois fossile de Г Eggenbur- rcoxylon furwaldense n.sp. “80, GRUAS-CAVAGNETTO, C. 1978. Etude ag ata m Eocene du Bassin Parisien. Mem. Soc ol. France, n.s. 56: 1 80, Late а and Eocene floral Ac ademia Sinica (Inst. Geol. Palaeontol.), Nan- jing. HANKS, S. . E. FAIRBROTHERS. 1976. Paly- is dne viis epus of Fagus L. and Noth- JONES —FAGACEAE EVOLUTION in Querc itr J. Arnold Arbor ——— 197 ofagus ligt croscopy and computer analyses. Bot. pu 1: 1- 142. HARDIN, J.W. 1975. Hybridization and introgression . 56: 336-363. 6. Terminol and classification of Quercus trichomes. J. Elisha Mitchell Sci. Soc. 92: 151- 1979. Patterns of n in foliar tri- chomes of eastern North American Quercus. Amer. J. Bot. 66: 576-585. HEER, О. 1869. Miocene baltische Flora. Beitr. Na- turk. Preussens 2: 1-104. . Die Miocene flora and fauna Spitz- bergens. Kongl. Svenska Vetensk. Acad. Handl. 8: 1-98. ———. 1883. Flora fossilis arctica. Die кш flora Grénlands, Part 2. J. Wurster and Co., HERNANDEZ-COMACHO, J., G. LOZANO-C. Hee E. Не. NAO-S. 1980. nibus del genero Trigonobala- nus Forman, e (Fagaceae) en el neotropico-II. Caldasia 13: m Hickey, L. J. ee of bw rato of ене leaves. Amer. J. . 60: 17- 33. 1977. Stratigraphy and paleobotany of the Golden Valley Formation (early Tertiary) of pa rn North Dakota. Mem. Geol. Soc. Amer. 1-181. 79. A revised —— of the archi- tecture of dicotyledonous leaves. Jn C. R. Metcalfe & L. Chalk (editors), y ping of the Dicotyle- dons, 2nd edition. Clarendon Press, Oxfor WoLFE. 1975 [1976]. The bases of angiosperm phylogeny: icr morphology Ann. Missouri Bot. Gar 8—589. , R. M. WES T & M. R. Te WSON. 1984. Arctic y. Science 224: 175- 176. [Letter.] . К. Сно 1983. Arctic terrestrial biota: Eu evidence of age disparity with mid-northern latitudes during the late Cretaceous and early Tertiary. Science 221: 56 . 1983a. Nothofagus macrofossils from the Tertiary of Tasmania. Alcheringa 7: 169-183. 83b. Evolution of Nothofagus cunning- hamii and its relationship to Nothofagus moorei: as inferred from Tasmanian macrofossils. Austral. J. Bot. 31: 453-465 HijELMQvisT, Н. 1948. Studies on the floral mor- phology and PE of the Amentiferae. Bot. Not. Suppl. 2: : iioi A. 1936. SE Tertiary floras of Alaska. Pro- s. Pap. U.S. Geol. Surv. 182: 1-185. HR J. 1954. Forests and climates in the South Island of New Zealand. Trans. & Proc. Roy. Soc. New Zealand 82: 329-410. Hopkins, W. S. 1969. phat e d i the Eocene Kit- ilano Formation, sout tish Columbia. Canad. J. Bot. 47: 110 Hou, D. 1971. Chromosome s of Trigono- balanus verticillata Forman jeu, Acta Bot Neerl. 20: 543-549. Hsu, Y.-C. & H.-W. JEN. 1979. New species and new combinations of Fagaceae from China. Acta Bot. 256 Yunn. 1: 146-148. [In Chinese. Examined in ab- stract form, Biol. Abstr. 71: 6015. Hu, H. H. & R. W. CHANEY. 1940. A Miocene flora dep antung Province, China, Part I, Introduc- io = systematic considerations. Palaeonto l. Ho A. 1983. The Pliocene leaf flora from Ru- SZOW near Zary i in еы Silesia, SW Poland. Ргасе 2 "B eor methods and the so pg к Braces ee 5). Pp. 177-207 in V. A. Brooks (editors), 2d ane in Cladistics. New York Botanical Gar- The Genera of Flowering JENSEN, R. J. & W. H. EsHBAUGH. 1976. merical taxonomic studies of boda in Do. us. Syst. Bot. 1: 1-19. „а Ph.D. thesis. Indiana Univ., Blooming- jme. W., E. Кмовгосн, Z. KvACEK & A. SELMEIER. 1971. Makrofloristische Untersuchungen in Staatssamml. Palaontol. Hist. Geol. 11: 2 Jussieu, A. L. DE. 1789. Genera poene Wheldon i. New York. [1964 Re КАШ, В. B E. C. ABBE. ‚1984. кее аг- һиесї d luti old the Fagaceae. J. Arn Arbor. 65: 375-401. Kepves, M. 1964. Sporomorphs nouveaux des couches Eocenes de Hongrie. Pollen & Spores 6: 195-201. 1967. Sur quelques problémes de stratigra- phie palynologique appliqué au Tertiare inférieur en Europe. Pollen & Spores 9: 321-334 1978. Palynological investigations into sed- iments of the lower Paleogene period in Bulgaria. Acta pU (Szeged 1955+) 24: 23-30. KENT, D. PDYKE, J. J Ma B. J. Mac 1984. peer bus stratigraphic n uu Science 224: 173- 174. [Lette KIRCHHEIMER, F. 1957. Der Laubgewachse der Braunkohlenzeit. Veb Wilhelm Knapp, Halle. KNoBLocH, E. 1969. Te Las Floren von Máhren. Moravskbe Muse m, Brn KocH, B. E. 1963. Fossil und from the Lower Pa- leocene of the Agatdalen (Angmartussut) area, central Nugssuaq Peninsula, We hwest Green- land. Meddel. Grenland 172: Kovacs, E. E. 196 unde ates und ungaris- chen Eichen des Tertiárs. I. Samartische Eichen. Acta Bot. (Szeged 1955+) 8: 283-302. KRASSER, F. 03. Konstantin von Ettingshausen' S Studien über die fossile Flora von Ouricango in Brasilien. Sitzungsber. Kaiserl. Akad. Wiss., Math.- Naturwiss. Cl., Abt. 1, 112: 852-860. KRAUSEL, R. 1924. Beitrage zur Kenntnis der fossilen Flora Südamerikas. I. Fossile Hólzer aus Patagon- — . ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 ien und benachbarten Gebeiten. Ark. Bot. 19: 1- 939. Ergebnisse der смео Prof. Stromer’s in den Wusten Agyptens, IV. Die ose ч ч дни a Bayer. Akad. Wiss h.-Nat s. Abt. -140. Ts UTR Untersuch- Pa- 1. Wevtano ur Kutikular-analyse tertiárer Blatter I. che . Geo KUPRIANOVA, L. 1 . Palyno y . Гр Soviet Rep From the First International Pa- ia Conference U.S.S.R. Acad. Sci., Mos- cow. [Exam in abstract form in Ref. Zurn. Biol. 18: B139. 1963.] 1967. Palynological data for the history of the Сы ай Pollen & Spores 9: 95- KvACEK, Z. & Н. WALTHER. 1981. Studium über Quercus cruciata Al. Braun und analoge Blattfor- men aus den Tertiár Europas. Acta Palaeobot. 21: 77-100. мы R.S. 1952. Catalogue of Cenozoic oni North America through 1950. Mem. Geol. Soc dee 51: 1-381. LANGDON, L. D. 1947. The comparative арени of the Fagaceae I. The genus о Bot. Gaz. (Crawfordsville) 108: 350-37 LAURENT, L 12. Flore fossile x Schistes de Me- nat (Puy-de-Dome). Ann. Mus. Natl. Nat., Geol. deret 1 1-246. & P. RT 1909. Note sur le Castanea avernensis rh Menat. Compt. Rend. Assoc. Franc. Advancem. Sci. (Congr. du Lille, 1909): 1 LAWRENCE, G. H 1951. ге of scd Plants. MacMillan Co., New LEE, C.-S. 1968. Comparative Wood SNC of the eap of Taiwan. M.S. Thesis. Oregon State ., Corvallis. ESAE Г. 1859. On some fossil plants of recent formations. Amer. J. Sci. Arts, Ser. 2, 27: 359- 366. — : © > 1874. о E the Fossil Flora of the Western Territori e Cretaceous Flora. United States Geological rey of the Ter- ritories, Washington 1878. I niributions to the ese ded B the Western Territories, Part II, The Tertiary United ay Geologia Survey of he Territories, Washing ЖОР Е- 1979. Hallazgo del género Us bala- nus Forman, 1962 b E en el neotrópico— I. Caldasia 12: 517-537. Luong, N.-T. 1965. The boundary between the gen- era Castanopsis and Lithocarpus and some data on their taxonomy. Bot. Zurn. (Moscow & Len- ingrad) 50: 997-1001. [In Russian.] MACGINITTIE, Н. D. 37. The flora of the Weaver- ville Beds of Trinity Co., oe Publ. Car- negie Inst. Wash. 465: 83-1 1941. A Middle Eocene a from the central Sierra Nevada. Publ. Carnegie Inst. Wash. 534: 1- 178. 1986] . 1962. The Kilgore flora: a late Miocene flora from Nebraska. Univ. Calif. Publ. Geol. Sci. 35: 67-158. . 1969. The Eocene Green River flora of north- western Colorado and northeastern Utah. Univ. Calif. Publ. Geol. Sci. 83: 1-203. Mal, D. Н. 1964. Die Mastixioideen-Floren im Ter- tiar der Oberlausitz. Paláontol. Abh., Abt. B, Pa- láobot. 2: 1- 970. 1 Пе tertiüren Arten von Trigonobala- us Forman (Fagaceae) in Europa. Jahrb. Geol. 3(1967 ): 381-409. MANCHESTER, S. R. 1981a. Fossil plants of ee Eocene Clarno Geol. 43: 75- Univ., Bloomington e fruits, wood and leaves of the Clarno Formation of Oregon Fagaceae from de 70(5, p 7а рыч ct.] RP. E NE. 1 aid leaves, inflo- rescences e fru n res an extinct genus of fagaceous о from the Oligocene Florissant Flora of Colorado, U.S.A. Ze] у, . Bot. 70: 1147- 11 MASAMUNE, С. & T. TomiYa. 1948. Limlia, a new genu us of еи from Formosa. Acta Bot. Tai- A study of the Neogene plants in apan. II: on the cano-lake deposits. Ann. Sci. (Kanazowa Univ.) 5: 29-77. 1971. Paleogene mega-plant remains of the eters Islands, Japan. Bull. Natl. Sci. Mus. 14: —710. uc H. 1964. Reihe Fagales. Pp. 44-51 in A. Engler, Syllabus der Pflanzenfamilien, 12th edi- erli . А. 1. Floras Tertiarias de la Ar- na. Ameghiniana 8: 357-371. ILI 1975. Las especies de Noth- ofagidites del sedimentos Tertiarios y Cretacicos ia la sara, norte rte de Tierra del Fuego, Argen- niana a 12: 165-183. Wood structure of Trigo- nobalanus excelsa G. Lozano-C., Hdz-C. & Henao (Fagaceae). Caldasia 13: 97-101. METCALFE, C. R. & L. CHALK. 1950. Anatomy of the Dicotyledons, Volumes 1-2. Clarendon Press, Ox- ford. & L. CHALK. 1979. Anatomy of the Dicoty- ledons, 2nd edition, Volume 1. Clarendon Press, rd. MTCHEDLISHVILI, N. D. 1961. Pollen and spores of stern Siberia. Jurassic-Paleocene. Trudy Vse- soyuzn. Neftyanogo Nauéno-Issl. Geol.-Razve- dochnogo Inst. 177: 329-333 MuLLER, C. H. 1942a. The problem A genera and subgenera in the oaks. Chron. Bot 12-14. 1942b. The Central jee pies of Quercus. Misc. Publ. U.S. Dept. Agric. 477: 1- ——, 1952. Ecological control of hybridization in : a factor in cl mechanism of evolution. ВЕНА 6: 147-1 MULLER, J. 1981. E pollen records of extant an- giosperms. Bot. Rev. (Lancaster) 47: 1-145. JONES— FAGACEAE EVOLUTION 257 MULLER-STOLL, W. R. & E. MADEL. 1957. Uber ter- tiäre Eichenhölz aus n ас ажр Вескеп. Senckenberg. Leth. 38: 8. NAKAI, T. 1939. iy eee (dan NN their allies the Japanese Empire. J. Jap. Bot. 15: 185-244, 257-271. NATHORST, A. G. 1888. Zur fossilen Flora Japan's. Paláontol. Abh. 4: 197-250. NAVALE, G. К. 1964. Castanoxylon gen. nov. from Tertiary beds of the Cuddalore Series near Pon- dicherry, India. Palaeobotanist 11: 131-137. Е Ј. 8. . The later extinct floras of rth America. Monogr. U.S. Geol. Surv. 35: 1- Di NixoN, K. C. 1982. In support of recognition of the jenes Nothofagaceae Kuprianova. Publ. Bot. Soc. , Misc. Ser. 162: 102. [Abstract.] eee © & A. D. MIALL. 1984. Arctic biostrati- graphic heterochroneity. Science 224: 174-175. Letter. Мотор, T. 1961. Fossil Fruchte und Samen aus den Niederlausitzer Braunkohlenrevier. Geologie 10: 231-245. OERSTED, A. S. 1871. Bidrag til kundskab om Ege- familien i Nutid og Fortid. Skr. Vidensk.-Selsk., Math. Prou n ж Kl. 9: 333-5 Oesch, Е. 1969. molecula е carbohydrates and polyols in cabis saps of beech (Fagus syl- vatica L.) and a number of other и e Planta 86: 360-380 OGURA, Y. 1949. A fossil wood of Castanopsis-type from the Tertiary of Nanano Prefecture. J. Jap. Bot. 24: 15-18. OLIVER, УУ. В. 1925. Biogeographical relations of the New Zealand region. J. Linn. Soc., Bot. 47: 9 . 1936. The Tertiary Flora of the Kaikorai Valley, Otago, New Zealand. Trans. R. Soc. New Zeal. 66: 289-293 ORLANDO, H. 1964. The fossil flora of the surround- ings of Ardley Peninsula (Ardley island), 25 de I Symp. on Antarctic Geol. North Holland Pub. Co., Amsterdam PALMER, E. J. 1948. Hybrid oaks of North America. J. Arnold Arbor. 29: 1-48. PENNY, J. S. 1969. Late Lu and early Tertiary palynology. Pp. 331-376 in R. H. Tschudy & R Scott (editors), Aspects of Palynology. Wiley Interscience oe New York. PETRESCU, I. Asupra unor lemne de Stejari (Quescoslon дш эке de la sud de Cim- pulung (Jud. A v. Babes-Bolyai din Cluj- Napoca rei Ко. Universitatea, Cluj. PHILIPSON, W. В. AND М. М. PHILIPSON. 1979. Leaf vernation in Nothofagus. New Zealand J. Bot. 17: 417-421. Piet, К. M. 1971. Palynology of Oligocene sediments from pel British Columbia. Canad. J. Bot. 49: 1885-1920. PLATEN, P. 1908. Untersuchungen — кес ie dem Westen der Vereinigten State rd- america. Sitzungsber. Natur Ges. Leipzig 34: 1- 55, 161-164. PRAGLOWSKI, J. 1982. Fagaceae L. (Fagoideae). World 258 Pollen and Spore Flora 11. Almqvist and Wiksell, kholm. PRAKASH, U. & E. S. BARGHOORN. 1961a. Miocene fossil woods from the Columbia basalts of central Washington. J. Arnold Arbor. 42: 165-193. 1961b. Miocene fossil wood from the Columbia basalts of central Washington. J. Arnold Arbor. 42: 347-358. PRANTL, K. 1894. F In A. Engler & K. Prantl, ie Natürlichen к Wilhelm En- gelmann, Leipzig. PreesT, D. S. 1963. А note on the dispersal charac- teristics of the se eeds o a eches and their biogeographical signifi- cance. Pp . 415-424 in J. L. Gressit (editor), Pacific Basin Biogeography. Bishop Museum Press, Ho- nolulu. Prive, C. 1975. Étude de quelques bois de chênes n ge ria. Pp. oe m J. Snowball ito P pro ical Congress (1963), des Zambia. caia мы Nothofagoxylon menendezii, leñ o petrificado c de Tert rtiario. par al Roca, Rio Negro, Argen Ameghin 4: 75-86. 1981. an foliar. de q especies suda- mericanas de Nothofagus Bl. (Fagaceae). Darwin- iana 23: RAMANUJAM, C. G. 1966. Palynology of the Miocene lignites from South Arcot district, Madras, Spain. Pollen & ом 8: 149- т RANIECKA-BOBROWSKA, J. 19 Trzeciorzedowa flora z Osieczowa i Kwisa on. Slask). Prace Inst. eol. 30: 81-223. х > Q © Za m щл т ‚> RAvEN, Р. Н. & D. I. AxELRop. 1974. Angiosperm | 1682. и cesis Nova. Faithorne Kersey, Lon ~ 03. Methodus Plantarum Emendata et ta. Smith & Walford, London 8. а of the >ле апа е. Ph.D. Thesis. Сог- Romer, А. $. 1966. Vertebrate Paleontology, 3rd edi- tion. Univ. of Chicago Press, Chicago ROMERO, E. J. 1973. Polen fosil de Nothofagus (Noth- ofagidites) del Cretacio y Paleoceno de Patagonia. Revista Mus. La Plata, Secc. Paleontol., n.s 291-303. 1977. Polen de Gimnospermas y Fagaceas de la Formacion Rio Turbio (Eoceno), Santa Cruz, Argentina. FECIC Editorial, Buenos Aires. 1980. а foliar de las especies sud- americanas de Nothofagus Bl. Bol. Soc. Argent. Bot. 19: 289-308. A. C. AGUIRRE. 1982. nota foliar Nueva Guinea y Nueva Caledonia. Bol. Soc. Argent. Bot. 21: 213-245. Rosen, D. E. 1978. Vicariant patterns and historical explanation in biogeography. Syst. Zool. 27: 159- Rouse, G. E., W. S. Hopkins & К. M. PIEL. 1971. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Palynology of some late Cretaceous and early Ter- tiary deposits in British Columbia and adjacent Alberta. Special Pap. Geol. Soc. Amer. 127: 213- 246. RUFF e, L. 1978. Evolutionary and ecological trends in Cour. Forsch.-Inst. Senckenberg. 30: 77-83. 1 Wachstums-Modus und Blatt-Mor- phologie bei altertumlichen Fagales und Hama- melidales der Kreide und der Gegenwart. Pp. 329- 341 in 100 Jahre Arbor, 1879-1979. Berlin. . KNAPPE. 1977. Entwicklungsgeschicht- liche und okologische Aspekte zur Oberkreide- Flora, besonders einiger Fagaceae. (Hamamelidi- dae). 2. Geol. Wiss. (Berlin) 5: 269-303. SALARD, M. 1961. Contribution a l'étude poten lologique de la Patagonie II. Rev. Gén. Bot 270. SAPORTA, G. & А. Е. Marion. 1878. Révision de la = ore Heersiene de Gelinden. Mém. Acad. Roy i. Belgique 4 2. Senare H. J. 1973. The Upper Quaternary of the pe Flats area (Cape Province, South Africa). "iie Geol. 15: 1-57. SCHIMPER, W. P. 1870-1872. Traité de Paleontologie Végétale ou la Flore du e Primitif, Volume 2. J. B. Baillière et Fils, P. ScHorTKY, E. 1912. Die Eichen des extratropischen tasiens und ihre pflanzengeographische Bedeu- ung. Bot. Jahrb. Syst. 47: 617-708. SCHUSTER, J. Uber ein pliocánes Eichenholz Idaho. Neues Jahrb. Mineral. Geol. Palaeon- tol. 1908: 4 SCHUSTER, R. Continental movements, *Wallace's Line" and Indom alayan-Australasian dispersal of land plants: some eclectic concepts. Bot. Rev. (Lancaster) 38: 1-86. 1976. Plate tectonics e angiosperm origin and dispersal. Pp. 48-1 n C. B. Beck pes Origin and Early Evolution of море = lumbia Univ. Press, New SCHWARZ, О. 1936a. Entwurf zu einen natürlichen System der Cupuliferen und der o а Г. Notizbl. Bot. Сагі. Berlin-Dahlem . 1936b. Über die Typologie der Echenblats und ihre Anwendung in der Palaobota epe . Nov. Regni Veg. Beih. 86: 60-7 0. SEIN, М. К. 1961. Nothofagus pollen іп the London Clay. Nature 190: 1030-1031. SELMEIER, A. _1970a. Ein Castanopsis- Hólz aus jung- — — Neues Jahrb. Geol. Palaontol. Monatsh. 1970(4): 235-250 1970b. Castanopsis-Hólzer aus obermiozá- nen Glimmersanden der siidlichen Franckenalb. Mitt. Bayer. Staatssamml. Palaontol. Hist. Geol. 10: 309-320. 1970c. Castanoxylon bavaricum n. sp. aus jungtertiáren Schichten Nordost-Bayerns (Basalt- ш E ыа Geol. Blatt. Nordost-Bayern 20: 17-38. 71. Ein verkieseltes Eichen hólz aus jung- tertiáren Schichten Niederbayerns (Aidenbach). Mitt. Bayer. Staatssamml. Palaontol. Hist. Geol. 11: 205-222. —. | Ein Castanopsis-Hólz aus oberchat tischen Steigbach- schichten des Allgaus. Mitt. 1986] Bayer. Staatssamml. Palaontol. Hist. Geol. 12: 97— 104 SHEFFY, M. J. 1972. A Study of the Myricaceae From Eocene Sediments of Southeastern North Ameri- a. Ph.D. Thesis. Indiana Univ., Bloomington. sáb K. 1962. Anatomical studies on the phy- logenetic interrelationships of the genera in the Fagaceae. Bull. Tokyo Univ. Forest 57: 1-64 Simpson, G. G. 1953. The Major Features of Evo- lution. Simon & Schuster Co., New York. 1974. Die fossile Flora Sarmatischer Sed- tleren Slowakei. Acta Geol. Geogr. Univ. Co- menianae 26: 5-85. SMILEY, C. J. & Г. M. Hucaiws. 1981. Pseudofagus idahoensis N. gen. et sp. (Fagaceae) from the Mio- cene Clarkia Flora of Idaho. Amer. J. Bot. 68: 741- 761 SMITH, А. С. 8: J. C. BRIDEN. 1977. Mesozoic and Cenozoic Paleocontinental Maps. Cambridge Univ. Press, New York. SoEPADMO, E. . Florae Malesianae Praecur- sores XLVII. Census of Malesian Castanopsis (Fagaceae). Reinwardtia 7: 383-410. 1 А revision of the genus Quercus L. Subgen. Cyclobalanopsis (Oersted) Schneider in Malesia. Gard. Bull. Straits Settlem. 22: 355-427. . Florae Malesianae Praecursores XLIX. Malesian dapat y Lithocarpus Bl. (Fagaceae). Reinwardtia 8: 197-308. 72. olla Fl. Males. I, 7: 265-403. SRIVASTAVA, S. К. 1969. Assorted angiosperm pollen . 1965. Cuticular studies as an aid to plant taxonomy. Bull. Brit. Mus. (Nat. Hist.), Bot. 4: 1- 78 STEBBINS, G. L., E. B. MATZKE & C. EPLING. 1947. Hybridization in a populat gus of Q. marilandica and Q. ilicifolia. Evolution 1: STERN, W. L. Development E "às amentiferous concept. Brittonia 25: 316- STOPES, M. С. & К. Еол. 1910. Studies p hg struc- fC Trans., Ser. B, 201: 1-90. Stover, L. E. & P. R. Evans. 1973. Upper Creta- ceous-Eocene spore pollen zonation, offshore Gippsland Basin, Australia. Geol. Soc. Austral. Special Publ. 4: 55-72. SzAFER, W. 1961. “Miocene flora from е Gliwice in upper Silesia. Prace Inst. Geol. 33: 1-205. TAKHTAJAN, A. 69. Flowering Plants: ok and Dispersal. Oliver and Boyd, Edinburgh. . Fagaceae. Pp. 60-120 in Flowering Plants of the USSR, Volume 2, Nauka, Leningrad. TANAI, T. 1961. Neogene floral change in Japan. J. Fac. Sci. Hokkaido Imp. Univ., Ser. 4, Geol. 1: 119-398. . 1967. Tertiary floral changes of Japan. P 317-334 in Jubilee Publ. Commemorating Prof Sasa’s 60th birthday. 1970. The Oligocene floras from the Pee Hok- kaido Imp. Univ., Ser. 4, Ge . 1974. Evolutionary trend of the genus Fagus around the Northern Pacific Basin. /n Symposium on Origin and Phytogeography of Angiosperms. JONES— FAGACEAE EVOLUTION 259 Birbal Sahni Inst. Paleobot. Special Publ. 1: 62- 83. К. HuziokA. 1967. Climatic implications of Tertiary floras in Japan. Pp. 89-94 in K. Hatai (editor), Tertiary Correlations and Climatic s in the Pacific. Eleventh Pacific Science . A Mio-Pliocene flora from the Ningyo-Toge area on the border between Tot- tari and Okayama Prefectures, Japan. Rep. Geol. : 1-63. . 1965. Late Tertiary floras from northeas tern Hokkaido, Japan. Special Pap. Pa- laeontol. Soc. Japan 10: 1-117 & KOYAMA. 1975. On the lobed oak leaves from the Miocene Kobe Group, western р Japan. J. Fac. y Hokkaido Imp. Univ., 4, Geol. 17: 129-1 accent P.M. & В.Н. c PN 1979. Fo- liar trichomes of Quercus subgenus Quercus in the eastern United States. J. Arnold Arbor. 60: 350- 366 TIFFNEY, В.Н. 1984 [1985]. = size, si adie syn- dromes, and the rise of the angiospe evidence and hypothesis. Ann. Missouri E Bot. Gard. 71: 551- 6. 57 TRALAU, H. 1962. Die НЫ Fagus Arten Europas. Bot. Not. 115: —176. TRELEASE, W. 1917. Na эы а hybrid oaks. roc. Amer. Philos. Soc. 56: 44—52. 1918. The ancient oaks of America. Mem. Brooklyn к Сага. 1: 492-501. 1924. e American oaks. Mem. Natl. Acad. Sci. 20: pas Tucker, J. M. 1974. Patterns of parallel evolution of leaf forms in new world oaks. Taxon 23: 129- 154. Tucker, J. M., W. E. SUNDAHL & D. O. HALL. 1969. A mutant Lithocarpus densiflorus. Madrono 24: 221-225. UNGER, F. 1850. Genera et Species Plantarum Fos- silium. Wilhelm Braumuller, Vindobonae. VAN DER BURGH, J. nischen Braunkohlenformation, А Braunkohlengruben “Maria Theresia: zu Herzo- genrath “Zukunft West" zu eschweiler und “У1с- tor" (Zülpich Mitte) zu Zülpich. Nebst einer sys- n Gattung Pinus L. Rev. Paleobot. Palynol. 15: 73-275. 1978a. The Pliocene flora of Fortuna-Gars- dorf I. Fruits and seeds of zm osperms. Rev. Pa- laeobot. Palynol. 26: 173-21 8b. Hólzer aus dem [PS der Nieder- rheinischen Bucht. Fortschr. Geol. Rheinl. West- falen 28: 213-275. VAN STEENIS, C. G. 1953. Results of the Archbold pe Papuan Nothofagus. J. Arnold Arbor. 34: 300-374 1971a. ‘Nothofagus, key genus of plant g ography, in time and space, living and fossil, гон. 76. Ecological species, multispe- cies, and oaks. Taxon 25: 233-239. VATER, H. 1884. Die fossilen Hólzer der Phosphor- 260 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 itlager des Herzogthums Braunschweig. Z. Deutsch. Моге, Ј. А. 1968. Paleogene biostratigraphy of non- Geol. Ges. 36: 783-853. marine rocks in King County, Washington. Pro- -33. WARD, L. F. 1905. Status of the Mesozoic floras of fess. Pap. U.S. Geol. Surv. 571 the United States. U.S. Geol. Surv. Pap. 2: 1-616. . 1973. Fossil forms of Amentiferae. Brittonia WHEELER, E. F., R. A. Scorr & E. S. BARGHOORN. 25: 334-355. 1978. Fossil dicotyledonous woods from Yellow- Paleogene floras from the Gulf of Alas- stone National Park, II. J. Arnold Arbor. 59: ka Region. Profess. Pap. U.S. Geol. Surv. 997: 1- -26. 1 WINKLER, Н. 1904. Betulaceae. In А. Engler, Das ZASTAWNIAK B Has d leaf ч from the Pflanzenreich, Volume 19. Wilhelm Engelmann, t Hen n Group of King George Island Berlin. oi holanda Islands, Antarctica). Preimina ary WiLtis, J. С. 1973. A Dictionary of the Flowering Report. Stud. Geol. Polon. 72: 97- Plants and Ferns, 8th edition. Cambridge Univ. Press, Cambridge. APPENDIX I A FOLIAR “KEY”? FOR CONFIRMING FAGACEOUS AFFINITIES AMONG MEMBERS OF THE HAMAMELIDOPSIDAE (AMENTIFERAE) The following “key” is designed to help confirm the identity of fagaceous leaves among similar forms in the Hamamelidopsidae sensu Cronquist (1981). It is based on broad summary publications (e.g., Lawrence, 1951; Metcalf & iat 1950, 1979; Hutchinson, 1967; Willis, 1973; Cronquist, 1981), familial surveys (e.g., Winkler, 04; Sheffy, 1972), personal communications with those working with amentiferous leaves, and personal observations. -— Stomata anomocytic or cyclocytic Stomata neither anomocytic nor cyclocytic Balanopaceae — Casuarinaceae Daphniphyllaceae Urticaceae pro parte 2. Leaves simple 2. Leaves compound Cannabidaceae pro parte Juglandaceae Moraceae pro parte Rhoipteliaceae 3. Leaves with pinnate venati 3. Leaves with distinctly de or flabellate venation cc Cannabidaceae pro parte Cercido E ue Moraceae part rotha iip руй Platanaceae pro parte 5 4. Leaves, when toothed, possessing craspedodromous secondary veins 4. Leaves toothed without purely craspedodromous secondary veins co. Eucommiaceae Platanaceae a parte Leaves lacking anastomosing, fibrous, fine venation and strong marginal veins; are lation well developed consisting sd Каню orthogonal, nearly isodiametric units ................. ous, fine venation; strong marginal veins and elon- ем arecales Didymelaceae when toothed, — less than two teeth per secondary over more than half of the leaf Mis in some Nothofagus species, with more than two teeth per secondary pom two-thirds of the leaf length and always possessing icis unicellular bos D trichomes and often large glands (type 17) 5-5... 6. te with more than Pe be ied spaced teeth per secondary; trichomes restricted to uniseriate nonglan ular forms Eupteliaceae 7. Leaves иш. bearing mme trical bases; epidermal idioblasts absent; types 1 and 2 trichomes, when nt, lack hooks; type 3 trichomes, when present, possess а bases; ae neither calcified nor silicified .................................. 1986] JONES—FAGACEAE EVOLUTION 261 7. Leaves raceme with markedly asymmetrical bases; idioblasts (e.g., Е ралы common in the upper ның 1S; short — ad trichomes (types 1 and )c oy difta hooked; type 3 trichom en present are ei on кшш ынет: epidermis кошш calcified or riche cifie annabidaceae pro parte Urticaceae pro parte 8. Thick-walled type 1 trichomes, when present unicellular 8. Trichomes resembling type 1 always multicellular with thin cross septae .. Lei itneriaceae 9. Leaves neither oo and pinnatifid with semicordate уш» nor spatulate with promin g present, seeded by кше trichome bas 9. Leaves with glandular scales subtended by bicellular (rarely tri- or tetra- cellular) bases, or when absent, leaves linear-oblong and pinnatifid with semicordate stipules, or leaves spatulate with prominant looping sec- ndaries Myricaceae 10. Leaves rarely doubly serrate; when doubly serrate, serrate over more than two-thirds of the lamina length possessing only types 1, 2 and often type 17 trichomes Fagaceae Leaves nearly always doubly serrate possessing modified and/or uniseriate multicellular trichomes and/or glandular trichomes in addition to types 1 and 2 Betulaceae > APPENDIX II KEY TO THE GENERA OF FAGACEAE (BASED ON FOLIAR FEATURES) 1. Leaves lobed Leaves not lobed 2. Upper cuticles revealing sinuate anticlinal walls; may possess only trichome types 1 and 15, neve Fa gus (some pa E Upp ling ight to slightly curved anticlinal walls; may possess a variety of trichome E pasa tufts of various sorts 3 es nearly always pointed and aristate Quercus (sect. Erythrobalanus pro parte) 3. Lobes rounded and nonaristate ttt Quercus [sects. Erythrobalanus (rare), оо Mesobalanus, and Cerris pro parte] 4. Leaves bearing persistent yellowish, peltate (type 11) trichomes Chrysolepis 4. Leaves not bearing persistent yellowish, peltate (type 1 D в 5 AN p? trichomes forming a dense cover over the lower surface of the leaf and/or type 2 trichomes common over major veins; all Pedes lacking tufts; teeth, when кл compound ог biconvex (type А1) thofagus ws Leaves never bearing type 17 trichomes; teeth nearly always simple; type 2 ни, пої abundant and when present usually restricted to the midrib or petiole; type 12 nga Upper cuticles of mature yes o sinuous pp чу walls; secondary veins strictly craspedodromous an raight e near the len E the leaf where they often recurve; trichomes ‚ы to bo ‘ar and rarely 2... Fa ah pro parte cal except = cultivars) 1 11 ondary > Upper cuticles of veins craspedodromous or campt todromous usually slightly curved, or le ale and pere aac usually include tufts in addition to or instead of trichome Ives | 7. у ре bearing appressed е (type 9) trichomes Lithocarpus pro parte 7. Lower cuticles without type 9 trichom 8. Leaves strictly craspedodromous, lac king intersecondary veins; secondary veins regular and straight to slightly curved; teeth regularly spaced ................... mous or craspedodromous with irregularly spaced sec- oo XE < oO "A > 3 5 ee 9 a Teeth not rounded and ds == 10 9. Teeth rounded and fairly со ONES Quercus (sects. d amu Mesobalanus, and Cerris pro parte 10. Leaves lacking large multiradiate (type 10) trichomes 10. Leaves bearing large type 10 trichomes Lithocarpus (L. densiflorus only) 262 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 11. Leaves with abundant thin-walled peltate (type 13) trichomes, with darkly staining bases, on the lower surface — — anopsis pro parte 2 Leaves lacking type 13 trichomes 12. Leaves lanceolate, subcoriaceous, use glabrous at maturity with only simple trichome 5 bases widely scattered on the epiderm кә E < .9 [^7] [e] 2 m A e [e] O о с o = m = e, < HT © E Е) e O [s] £l [^2] — > uc = go uniseriate (type 15) and/or capitate (type 16) trichomes; tufts (type and often abundant; upper cuticle revealing straight to rounded anticlinal w Quercus (sects. Cerris, Sule Erythrobalanus, Lepid la s 5 and 6) a dti obalanus, and Macrobalanus pro parte Castan 13. Leaves with greater than 11 pairs of secondary veins "T Lithocarpus pro parte (L. cornea) ns 13. Leaves with fewer than 11 pairs of secondary ve rcus (sect. C diens irme pro parte) 14. A pom Leaves holly-like, coriaceous to sclerophyllous secondaries usually irr spaced, often entering large spinose teeth ... L Quercus (sects. ener disini pro part epidobalanus pro parte, and Prorobalanus 14. a entire or partially 1 toothed, not holly- like 15. simpl acute), often forming a саар tip 13, date or attenuate (rarely, 1 Ф Aat =e not bearing a drip ti o 16. Lower cuticle Ed with darkly staining thin-walled dien Toe 13) trichome bases; type 13 trichomes often present in abun astan xis po parte d Cas Lower cuticle lacking type 13 trichomes; darkly staining bases, w present, subtending one or more of trichome types 15, 16, and 17 ........ Quercus (sect. C yclobalanopsis pro parte) = 10) trichomes Leaves not bearing 18 p Leaves poss thin uniseriate E APPENDIX III CLASSIFICATION OF TRICHOMES OF THE FAGACEAE Trichomes have been used taxonomically in parts of the Fagaceae for many years (Dyal, 1936; Hardin, 1975, 1979; Thomson & Mohlenbrock, 1979), yet no family. Camus (1934-1954) provided a truly remark- work is fairly limited. The following i is an attempt to classify and describe the major trichome types found Leaves entire-margined ithocarpus pro parte 1 n eh 1 Li а (entire- -margined L. densiflorus) 18 g type 10 tri essing “glandular” peltate (type 19) trichomes or trichomes with ve darkly , pl. 52D) or se 1 rly in the sinus above the teeth, as in Jones (1984, Trigonobalanus Leaves without any of the above characteristics „u. M Quercus (sects. Erythrobalanus, Lepidoba Mac lanus, and robalanus pro parte) among all extant Fagaceae. This classification is built us, D Ha historically on those of Camus, Dyal, Hardin, and oth- ers. The classification must be a arbitrary in | many re- spec cts ‚ because mo st forms intergrade with at least one ot her o not exist I have attempted to combine forms that cannot be distinguished without sectioning or other time con- suming methods. ‚АП types listed here can be identified dard light microscope ora scanning electron п micro- thors, the terms glandular ае, correlated with the functions they imply. They , rather, terms that roughly describe the apparent function of the e types. Each term is associated yndrome or combination of characteristics un- der he жым holes below A. “Nonglandular” trichomes. Trichomes of this group are generally thick-walled, apparently nonglan- 1986] dular, and unicellular or composed of unicellular The function of these trichomes appears class tecteurs" of Camus (1928-1929, 1934-1954). Type 1 (solitary unicellular). Trichomes of this type are usually thick-walled, fairly is and straight to slightly wavy in course (Fig. 1). In those species in which they occur on fine veins, they are usually erect, bristle-like, and relatively short (Fig. 2). When restrict- ed to the petiole, midrib, and other major veins, these trichomes tend to be longer and somewhat appressed to the vein from which they arise. These long appressed forms are usually thinner walled and frequently occur in young leaves. They are not as resistant as the short erect ones and tend to be lost with time. This rather ragile form conforms to "poils tecteurs unicellules courbés a la base" of Саш (1934-1954, volume III). Trichome type 1 is perhaps the most widely dispersed LAT 4 4 +} г Aso f? A; 11 n A (see Table 2). This trichome type appears to form on end of a continuum from solitary unicellular trichomes tecteurs (unicellulaires) isolés" аа 1934-1954) and "solitary" of Hardin (1976). Type 2 (i unicellular conical). Trichomes of this type walled, erect (Fig. 4). They are meio veins as well. Some, a distinctive, darkly is best d d enus hofagus and is rare outside the Fagoideae sensu la Type 3 (papillas) Papillae are included here, though are relatively rare in the Fagaceae, occurring | only i ina ew species of Quercus and Li o: Nothofagus . When present they adorn nearly all Eng rwise undifferentiated lower epidermal cells (Figs. 6, They are thick-walled and arise from normal-sized epi- dermal cells. Type 4 (appressed ¡rifa attached werde le These trichomes resemble type 1 except tha attached laterally and are conca thin-walled. Кт 15 often difficult to determine the point of ber pn s(1934— 1954) illustrated a number of len from ei leaves (Fig. 8). They are exceedingly rare on mature leaves, where they are confined to the ш veins. Their pres- ence appears to be restricted to the genus Quercus. Th pond ty the "poil UTI ae à FE JONES—FAGACEAE EVOLUTION 263 navette" of Camus A nid and “appressed laterala" of Hardin (19 Type 5 (fascicul wal eri ‘nonglandular” pe chomes are We aisi. а by the possession of two o r, thi (g our to seven) unice 1 + » 4 +1 (Fig. 11). Fasciculate tufts occur in a sessile form tha is usually associated with radially septate bases, and a pedicellate form, in which there is a very short epi- rmal pedestal upon which the rays are borne. Fas- jor veins in most mature leaves but ma e cover, in some species, obscuring the entire lower leaf surface. Like most, trichomes of this type are h This trichome type is restricted to the Castaneoideae and Quercoideae. It reaches its greatest кирд ала in the genus Quercus. This type corresponds of Camus’ (1928-1929, 1934-1954) "poils fascicules en bouqets ou buissons, a articulés libres.” How Camus included types 8, 10, and 14 in this broad di. as well. Hardin's (1976) “fasciculate” included those assigned to this group plus those of type 8, which he considered to be a subt Type 6 (stellate). Stellate trichomes consist of three or more “nonglandular, " unicellular, and generally thi ick of attachment in a fashion parallel, or nearly parallel, to the leaf surface (Fig. 12). These trichomes are usually maller than erect tufts, which may occur on the sa me leaf. ma frequently can з distinguished from * *open" forms of type 5 trichomes shaped wedges produced at the шере of the constituent ele- ments (Figs. 13, 14). Two forms of this trichome have been observed. Elements of the more common form occupy a common plane. The second form is bilayered nd contains a second set of elements in a plane above the first (Fig. 15). Stellate trichomes, when abundant, are scattered evenly over the fine venation of the lower u din's (1976) “stellate” type except that it includes bi- 1 y A г. PY ais gt р s 1 2 FS- would be multiradiate. Camus (1934-1954) used the term i i me of those she They are most commonly found distributed over the lower surface of the leaf. Plants that bear fused stellate trichomes usually inhabit warm coastal environments where windy conditions and sandy soil prevail. This 264 is equivalent to Hardin’s (1976) type of the same name. Camus did no pee рн trichomes of this type from other ds T except that the cells are fused well shal nh [s (Fig. 18). They intergrade with t type 5, but differences in their taxonomic and Conarate type . They are usually found in the angles of the sec- ondary veins on the lower sides of the leaves. However, they are scattered over the fine veins as well in a few specimens (Fig. 19). In some cases the elements are extremely Tuis and intertwine (Fig. 20). They are found only in Quercus and, rarely, Lithocarpus. They reach Erythrobalanus. Sessile forms have been observed but most are pedicellate (Fig. 21). Hardin (1976) consid- ур “nonglan- dular” trichome type consists of two or more tition thick-walled, unicellular elements that are nearl planar and approximately parallel with the leaf Aid 2). The bases of the eleme nts 23). They are restricted, with few exceptions, to Litho- carpus, where they occur in nearly all specimens. They are one of the salient characteristics of Lithocarpus leaves. Camus (1934-1954) classified these trichomes under the category “poils en doigts de gant” or hairs like fingers of a glove ‚ Туре 10 (multiradiate). These * *nonglandular" tri- eight), unicellular, generally thick-walled, elements that emerge in a va- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 This type is generally restricted to the genus Quercus b large and impre f in her * zm M i le (1976) included these in a group of the s B. Intermediate trichomes. Trichomes assigned to this group consistently show a combination of the characteristics of “glandular” and **nonglandular" forms. Type 11 (“thick”-walled лсо These trichomes consist of a uniseriate stalk, whic rs a somewhat circular cap of radially fused, o thick-walled cells. These trichomes are most often yellowish and cover the entire lower epidermis (Figs. 27-29). Al- of ways: they are smaller (average radius ca versus 180 um), have thicker cell walls, contain yel- radially fused (Fig. 27), which is not the case of type 13 (Fig. 33). Camus (1928-1929) included this in her “ро|$ ёсаШеих.” Type 12 (curly thin-walled unicellular). Trichomes of this type are thin-walled and collapse upon drying even in mature leaves. There is no evidence of glan- dular function. These persistent trichomes are coiled and unicellular. They occur in abundance only on type riety O quel ndom directions from a typically IV Nothofagus sacs m they form dense mats over on base (Fig. 24). These are usually the lower surface (Fi ). found distributed over the lower surface of the lamina. Type 13 (“thin”. ud peltate). This trichome type — FIGURES 1-6.— of short erect type 1 125 х.— 3. Photomicrograph of type 1 trichomes 500x.—5. Photomicrograph of type 2 t myrsinaefolia (1183), 640 х. [Note: Nu 1. Photomicrograph of solitary unicellular (type 1) trichomes over a secondary vein in ind ofagus alessandri G1 15); note also the large к (type 17) trichomes over the ан es over fine venation, Castanea n the margin of № Scanning electron micrograph of buen conical (type 2) ous. on Nothofagus pum mes showing broad Оет eros phat papillae (type 3 **trichomes") on the lower epiderm mbers in parentheses are Indiana University Paleobotanical Extant Leaf .] fine venation, 2 basal portions, upper cuticle of Nothofagus rmis of Quercus Collection accession numbers. See Jones (1984) for herbarium and collector information FIGURES 7-12.— 7. Scanning electron micrograph of papillae on the lower surface of Quercus iere ea (1183), 500x. —8. Appressed laterally attached unicellular (type 4) trichomes (redrawn from Camus, 1934), ca 00x .—9. Sca х.— 12. electron micrograph of a stellate (type 6) trichome from Castanea mollissima (110), lower surface, 250 x FIGURES 13-18.— 13. Scanning electron micrograph of кич уре 6 trichomes interspersed among type 16 trichomes, upper leaf surface of Castanea mollissima (110), 250 4. Photomicrograph of type 6 d showing prominent pie- -shaped wedges | at the 2 juncture of the elements lower Cuticle of Castanea ozarkens 1 the lower leaf айры vs 25 ри uercus arkansana (6 10), — х.— 16. Scanning electron micrograph. of a fused stellate (type 7) trichome on is lower "foliar surface of Que vid bak ho nn 500 x ad Phot lower leaf surface of Quercus fusiformis ( 144), 2 tomicrograph of type 7 trichomes on the nning Pe ОНА of stipitate fasciculate 18. Sca (type 8) trichomes on the lower leaf ae of m crassipes (640), 2 JONES — FAGACEAE EVOLUTION YA z ш A ca < © = < E Z < E © ea ра = © YN e z шщ т E LL O 3 < Z Z < Z Q E = = o > ía щ < ni Q < e, < by " шщ 2 © pu; ut od m 268 consists of thin-walled cells. They possess uniseriate nearly impossible to distinguish individually by ex- в ai leaf material. . They Pd the lower sur- face of m hey are no found outside of ne Castaneoideae al- though som e 19 trichomes (Fig. 53) of Trigono- ba etaed this type. When lost they leave round simple bases that give Castanopsis cuticles their char- in the earlier work and “‘poils en bouclier a pedicelle court" in the latter. Type 14 (rosulate). These trichomes consist of uni- cellular, often thin-walled elements, arranged in small bushy tufts (Figs. 36, 37). They are frequently dark i in color and h dular function. They intergrade with tufts of types 5 er the and 1 . The generally oc J lower surface of the leaf but are pc found on the upper surface as well. They are most abundant in the genus Quercus but also occur in some species of Lithocarpus. C. “Glandular” trichomes. Trichomes of this group are generally multicellular, apparently glandular in function, and composed of thin-walled cells. Ca- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 mus (1928-1929, 1934-1954) used the term “poils for trichomes of this type. She did not, n (1976) in- dicated, this is probably oe sue intergrade so thoroughly. Type 15 Kip ee These trichomes d ofa single со n of two or more pedra four ve) thin-w early alos size ee 38, 39). This trichome ty type in- tergrades wit th t type 16, , which possesses an enlarged a multiseriate section. The form of! type 15 ‘trichomes varies considerably. They may be short and consist of nearly spherical cells, to long and composed of elongate cylindrical cells (Figs. 40, 41). The terminal cells of and slightly enlarged. When lost, particularly in the a type, of the same name, is e type poets of a rather hetero ogen iseriate stalks with a single mark- edly enlarged terminal cell, or a group of terminal cells, > FIGURES 19-24.—19. Scanning electron presage of type 8 trichomes scattered over fine veins on the lower leaf surface of Quercus imbricaria (3148), 12 pedicellate base, lower foliar surface of Quercus Vp hile (2324), 250 e low of "о parallel tufts (type 9 trichomes) on 3. Photom crograph of type 9 trichomes dem the n nning electron micrograph er leaf surface of Lithocarpus henryi (1169), 1,000 x.— ar arrangement of elements, lower surface of Litho- ine carpus o (3095), 250x.—24. е А micrograph of a maleate (type 10) trichome on the lower leaf surface of Quercus emoryi (617), 2 FIGURES 25-30.— 25. Scanning electron micrograph of large enira: (type 10) trichomes found on the — 26. Sca lower leaf surface of Lithocarpus densiflorus (492), 125 х. trichomes on the n electron micrograph showing a dense c er foliar surface of a Sei tae (505), 125 x .— 30. Scanning electron m covering o Ead : of d thin walled u unicellular (type 12 125x FIGURES 31-36. — 31. Scanning electron micrograph of type 12 trich tomicrograph of type 12 trichomes on the lower surface of a solanderi (313 ,000 x .— 32. Pho leaf surface of Nothofagus fas te pe (31 1 ES othofagus 3), 1 othofagus ко (3121) leaf, 250x.—33. “Thin” e nae (type 13) trichomes of Castanopsis d Lithoca arpus (redrawn from Camus 1934-1954, 1948), ES © e 50x.— 36. Scanning e э surface of Quercus arkansana (610), 500 — 34. Scanning electron micrograph of ty ec ron micrograph of a rosulate (type 14) trichome on the upper leaf JONES— FAGACEAE EVOLUTION ANNALS OF THE MISSOURI BOTANICAL GARDEN (VoL. 73 1986] JONES—FAGACEAE EVOLUTION А, "y x e e №: Ў = va $ M АТК ev ONO RE Ж 7 $ РА ^am ` ' ENIM E 2779, i pe D. t asp, Y "ec > > Of OF. 949 22 = < A o wd $ $ Kia = y } VI \°ee e 212 r, rarely, an intercalary swollen A ings (Figs. 43-45). They are evenly distributed over all or- ders of venation in young leaves but are often nde with n be more n ge tanea. Camus (1928-1929) used the term “poil ca for он d = type. Hardin (1976) called өй bulbou Type 1 7 (large globular) icell ular and c v о large tri- us. They can more frequently it is ir- regular and produces a large number of oddly shaped ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 forms (Figs. 51, 52). The cells are ч colored and r to bear or contain secretory substan pmen Lithocar, of this type are pp ee m “simple branched” by Har- din Type 9 (' ‘glandular”’ о These о have e walls of moder: evidence т confirm this oughly intergrades with type 1 the same leaf (Figs. 51, 53, 54). Туре 19 trichomes аге found only in Trigonobalanus doichangensis. They lack the regularity of type 11 trichomes. This type can be distinguished from type 13 trichomes by its thicker cell walls and darker color. Camus (1934-1954) classified these trichomes as “poils en e’cusson.’ Septate trichomes very similar to those of types 1 and 5 have been reported by Camus (1928-1929, pl. 14:19, 1934-1954, pl. 61:19). The ““crosswalls” are not nearly as thick as the side walls. There is no indication of glandular function. In many cases, apparent cross- walls, which may represent the “degenerative mem- brane thickenings” of Kuester or simply ad debris, can be pen in paa! Pii Truly tate trichomes very rare an e been dude i by Camus (1928-1929, 1934-1954) aa in the genera Castanopsis and Quercus. They have also been ob- served by the author on floral structures of Chrysolepis. FIGURES 37-42.— 37. Photomicrograph of a type 14 trichome on the upper cuticle of — е а 15 ow (3091), 650 х.— 38. Scanning electron micrograph of surfac uercus atriglans (1 x.—39. Phot of Quercus schottkyana (1186), 250 x .—40. Scanni of spherical cells, lower leaf surface of Quercus adenon p trichome com . Phot a Quercus seutiscima t (1170 leaf, 250x — micrograph of the resistant basal portions of type 15 trichomes on the lower cuticle of ) FIGURES 43-48.— 43. iyi di ros micrograph of rp (type 16) trichomes on the upper n e a 4.P Castanea Pee (A1)leaf, ын x, sativa (114), 250x.—45. Draw collapsed type 17 trichome on the lower micrograp ype trichomes, ‘many w пш intercalary multiseriate sections, De spp. (redra ain Camus, 1934-1954, 1938), ca. 200x.— globular (type 17) trichomes scattered over the lower jeu i — cat surface of a Nothofagus betuloides "n 19) leaf, 250x thofagus sabias (3118) leaf showing Ан ered electron nee of a 48. Scanning ectron micrograph of type 17 trichomes scattered among type 2 trichomes on the lower surface of a Nothofagus antarctica (3117) leaf, 125x. RES 49-54.— 49. Photomicrograph of a type 17 trichome base on the lower cuticle of Nothofagus aequi- ids (3587), 250x .— 50. Scanning electron micrograph of branched uniseriate trichomes on the lower epi- dermis of a Quercus palmeri (3141) leaf, 125 x.—51. Scanning electron micrograph of type 18 trichomes on the pert ate trichom Drawings of type 18 trichomes from Camus, 1934-1954, 1948), ca. 100x.— 53. Scanning electron type 19) e on the upper epidermis of Trigonobalanus prue iei Trigonobalanus doichangensis (redrawn from Camus 2). 1934-1 1954 and Cutler, 1964), c ca. 100x (smaller 2) ded 160x (larger 1986] JONES— FAGACEAE EVOLUTION 273 A» J ra: i E J, 297. ps Pe О 42, L LT LLN i 274 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 JONES— FAGACEAE EVOLUTION 275 1986] FOSSIL EVIDENCE REGARDING THE EVOLUTION OF NOTHOFAGUS BLUME EDGARDO J. ROMERO! ABSTRACT Information on fossil Nothofagus, and related genera, is reviewed based on pollen grains, leaves, flower: in the Cretaceous and Eocene, respectively, the rate of morphological change was slow. More uinea and also was possible after the Miocene in New Gui specia o in other areas during the Plio-Pleistocene. Several lines of evidence support former suggestions of giving Nothofagus a higher taxonomic rank. Nothofagus Blume has frequently captured the interest of botanists and paleobotanists. It plays a dominant role 1 in the i Тешрегие forests of the Southern F rt as a timber source, and has an intriguing distri- bution as a vicariant of the other genera in Fa- gaceae. It comprises 34 species living in South America, Australia, New Zealand, New Cale- donia, and New Guinea. It certainly lived in Ant- arctica, but perhaps never in South Africa. The Fagaceae are usually associated with the Betulaceae in the order Fagales (Thorne, 1983; melales, Casuarinales, Myricales, and Hamamelidales, all included in the subclass Hamamelidae. Although this subclass has been subjected to considerable systematic rearrange- ment (Thorne, 1973), the taxa mentioned above were not very much affected The Fagaceae have been segregated into three subfamilies since De Candolle (1864), and this only objection has been the proposal to segregate Nothofagus as a new family, made by Kupriano- va (1965), on the basis of the pollen grains. More recent support for some kind of segregation was advanced by Crepet and Daghlian (1980), Smiley and Huggins (1981), Thorne (1983), Nixon (1984), and Jones (1984b). The classification of the species of Nothofagus into two sections and four series was suggested by van Steenis (1953) and Soepadmo (1972) on the basis of the female cupule and vegetative morphology. Recent in- formation does not support fully such an ar- rangement (Elias, 1971; Philipson & Philipson, 1979; Hill, 1983), but no alternative system has been proposed. In recent years a growing body of information has accumulated about fossil materials of Notho- fagus and related genera, shedding some light on its history and phylogenetical development. This information has not been fully used, and spec- ulation about the evolution and paleobio- geography of the genera has been based mainly n the evidence of living plants (van Steenis, 1971; Darlington, 1965; Melville, 1973, 1981; Cracraft, 1975; Humphries, 1981, 1985; Heads, 1985). Therefore, it seems reasonable to review information concerning fossil forms. dern and fossil data are reviewed first in this paper, followed by a discussion. The review of fossil data is arranged by type of fossil (pollen, leaves, woods, and flowers and fruits). In each case, a brief statement about their morphology in living plants is given. The discussion starts with the Cretaceous history of the genus, and also deals with the fossil record of related families. Then, the Paleogene account comprises the dif- ferentiation of species within the genera, so the morphology of the living ones is also considered. Finally, the Neogene part deals mainly with bio- geography, and the си сопсегп compiled by Wolfe (1973) a Muller (1981). When some information is given without men- ' Departamento de Ciencias Biológicas, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Intendente Guiraldes 2620, 14 Investigaciones Cientificas (CONICET) ANN. MISSOURI Bor. GARD. 73: 276-283. 1986. Buenos Aires, Republica Argentina. Member of Consejo Nacional de 1986] tioning the source, it may be found in these re- views. However, in important cases the original publication is given. Other more recent papers, or ones not considered by Wolfe or Muller, are quoted as well. The references to paleogeography are based on Smith et al. (1981) and those to age are based on Harland et al. (1982). DATA POLLEN GRAINS Pollen grains of Nothofagus are highly char- acteristic and easily determined. They are small- to medium-sized, peroblate, stephanocolpate, crassimarginate, with a spinulose exine. The liv- ing species have been studied in detail by Hanks and Fairbrothers (1976) and Praglowski (1981, 1982). There are three different types (““brassi,” usca,” and ‘“‘menziesi’’) differing in the shape of the mesocolpi, and the size and thickening of the colpi. These pollen grains are very different from others in related genera within Fagaceae, or in related families, which are mostly tricol- porate, with variable, non-spinose sculpturing. Fossil pollen grains provide the most complete record of Nothofagus through geological time. They appear for the first time in sediments of Santonian age in southern Australia (Dettmann & Playford, 1969; Stover & Evans, 1973) with the “brassi” type. This type was found also in (Archangelsky & Romero, 1973; Romero, 1973). The “fusca” type was present in the Maastricht- ian of New Zealand and South America and the Paleocene of Australia (Cooper, 1960; Romero, 1973; Martin, 1981). The “menziesi” type was recorded for the first time in the Maastrichtian of South America, and then in the Paleocene of New Zealand and the Eocene of Australia (Coo- per, 1960; Romero, 1973; Martin, 1981). Since their first appearance in the Cretaceous the three types of pollen grains show almost no morpho- logical change. During the Eocene and Oligocene the partic- ipation of Nothofagus in pollen assemblages reached its maximum (Romero, 1973, 1977; Martin, 1982). During the Miocene it was gen- erally less frequent but it expanded its area, and is recorded for the first time in New Guinea (“brassi”” type; Khan, 1974). There are no rec- ords of fossil Nothofagus from New Caledonia. Pollen grains of genera related to Nothofagus ROMERO- NOTHOFAGUS EVOLUTION 277 are also present relatively early. “Celtis type" pollen and “intermediates to Ulmus type," as defined by Muller (1981) were described from the Turonian (ca. 90 Ma) of Borneo. Borneo was located close to or at the Equator at that time. ollen grains of Alnus, Betula, and Myrica were found in the Santonian of Japan, Canada, and the United States, respectively (see Muller, 1981 for references in this paragraph). In the Cam- panian Cercidiphyllites is recorded in Canada, Momipites (Juglandaceae) in the United States, and Castanea in Canada and Holland. The last one is the first genus of Fagaceae recorded in Laurasia, about 15 Ma, later than the occurrence trichtian, the “Ulmus type" was described in Canada, the United States, Brazil, and India. In the Paleocene the “Liquidambar type" (Hama- melidaceae), Tricolporopollenites parmularius ucommiaceae), and Carpinuspollenites (Betu- laceae) are all known from central Europe. More importantly two other families of the subclass Hamamelidae with a Gondwanaland distribu- tion appear for the first time during the Paleo- cene. They are Casuarinaceae, represented by Haloragacidites in the Paleocene of New Zea- land, Australia, Argentina, and sediments in the Indian Ocean, and Didymelaceae, represented by Schizocolpus marlinensis in New Zealand and sediments of the Indian Ocean. LEAVES Leaves of living species of Nothofagus are rather variable. Those of extant South American, Australian, and New Zealand species (Romero, 1980; Carrasco Aguirre & Romero, 1982) are persistent or deciduous, craspedrodomous (with one exception), and with well marked, sometimes composite teeth. Those of New Guinea and New Caledonia (Romero & Carrasco Aguirre, 1982) are persistent, brochidodromous or semi- craspedrodomous, entire margined or with small teeth. The small teeth are different from those in the former species and were probably indepen- dently derived. Leaves of living species of Fagus can be differentiated from Nothofagus (Jones, 1984a; Dibbern & Romero, 1984; Romero & Dibbern, 1985) and the same is true for other genera of Fagaceae and related families (Wolfe, 1973). Cuticles of the South American species were studied by Ragonese (1981) while Jones (1984a) also sampled several species from Aus- 278 tralia, New Zealand, New Guinea, and New Cal- edonia. Imprints of leaves of indubitable Nothofagus fossil species are not present until the Eocene, and they are probably more frequent during the Oligocene (Ettingshausen, 1886, 1888; Dusen, 1907, 1916; Romero, 1978; Holden, 1983; Hill, 1983). Several show strong similarities with liv- ing species, and thus indicate a very low rate of morphological change (Dibbern & Romero, 1984) Species of U/mus, Betula, and Alnus were also described in the Southern Hemisphere (Ettings- hausen, 1886, 1888; Dusen, 1907, 1916) and they will probably prove to be Nothofagus. How- ever, it is noteworthy that a revision (Romero & Dibbern, 1985) of the original types of Dusen showed that some South American fossils de- scribed as Fagus could not be classified among living Nothofagus and fall within the limits of variation of Fagus. Upper Cretaceous and Paleogene sediments both in the Gondwanaland and Laurasian con- tinents also yielded some extinct genera of sup- posed Fagaceae, such as Debeya, Dryophyllum, Fagophyllum, and Fagopsis (Ettingshausen, 1886, 1888; Dusen, 1916; Berry, 1937). However, the delimitation and definition of these genera was done also many years ago in the Northern Hemi- sphere, and was based only on gross morphology. Its use, especially in Gondwanaland, should be Crane (1983) and Jones (1984a) show the trend to be followed. Among related families, important records of fossil leaves are those of Alnus from the Maas- trichtian of British Columbia; Corylopsis, Cory- lus, and Sinowilsonia from the Paleocene; and Carya and Tetracentron from the Eocene, from several localities in the United States (see ref- erences in Wolfe, 1973). Without ud or ordinal assignment, the earliest fossil leaves considered “їо be me some “Platanoids” from the Albian of the Po- tomac Group, eastern United States (Hickey & Doyle, 1977; Doyle & Hickey, 1976). Another early leaf that shared some Hamamelidoid and Magnolioid characters is an unnamed one, from the Baqueró Formation in Patagonia, Argentina and is late Barremian to early Aptian in age (Romero & Archangelsky, 1985). Cuticles of fossil Nothofagus have been studied ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 in a few Eocene and Oligocene species by Hill (1983) and Jones (1984a). WOODS Nothofagus timber has commercial value and its anatomy has been well known for many years (Dadswell & Eckersley, 1935; Dadswell & Ingle, 1954; Hinds & Reid, 1957; Tortorelli, 1956; Wagemann, 1948). Practical keys for use in each country were published, allowing identification of every species. Fossil woods of Nothofagus have been de- scribed from strata of different ages in South America. The oldest is Nothofagoxylon pichas- quensis (Torres & Rallo, 1981) from the Upper Cretaceous of northern Chile. It is associated with dinosaurs and turtles, at about 25°S of paleolat- itude. Seven other species were found in Eocene, Oligocene, or undifferentiated Tertiary rocks (see review in Ragonese, 1977; and Torres, 1984). In Oligocene (?) sediments of Rio Turbio, south- ernmost Argentina, Salard (1961) described a species of Fagoxylon. A doubtful Nothofagoxylon was studied from PEN ced (Evans, 193 ar As far as I am aware, here aren ibed as Nothofagus in recent ee Amone related families, there is an important record of woods of Platanaceae in the Campanian of the United States, which is the first record of woods of the subclass (Wolfe, 1973). FLOWERS AND FRUITS The cupule of Nothofagus is interpreted as a reduced dichasium (Forman, 1966). Its relation- ships with other genera have been established by Forman (1966) and Endress (1977). One of the species (N. alessandri) has retained seven flow- ers, which is probably the most primitive num- ber of flowers in the family. Nothofagus nitida occasionally has five (van Steenis 1953: 312). The only flower reported from the Southern Hemisphere is a four-parted cupule from the Oli- gocene of Tasmania (Hill, 1983) whose affinities to living species could not be established. In the Northern Hemisphere several flowers were at- tributed to Fagaceae and related families (Crepet, 1981; Smiley & Huggins, 1981). Fagopsis, from Eocene and Oligocene beds in the United States, is especially important (Manchester & Crane, 1983). It is a representative of the Fagaceae, but it is not readily accommodated in any recent subfamily. It has some features similar to every genus in the family, except Nothofagus. 1986] ROMERO— DISCUSSION: A PHYLOGENETIC AND PALEOGEOGRAPHICAL MODEL On the basis of the preceding data, a scenario may be constructed for the history of Nothofagus and related taxa through geological time. The current paradigm for the origin of the angio- sperms is that they derived from unknown gym- nosperms some place in West Gondwanaland towards the end of the Early Cretaceous (Raven & Axelrod, 1974; Brenner, 1976; Doyle & Hick- ey, 1976; Schuster, 1976). The subclass Hama- i META would differentiate in the Northern sphere. Raven and Axelrod (1974) stated 6 Меш like Hamamelidaceae, Fagaceae, and Betulaceae would have originated in Laurasia, and particularly Nothofagus migrated from Asia through Gondwanaland, via the mountain heights of Africa and India, in the Middle Cretaceous. Wolfe (1973) also believed that most of the Ha- mamelidae originated in the аео province (western North America and Asia the Late Cretaceous, and that families such as Fagaceae and Betulaceae evolved there prior to the Maastrichtian, diversifying into extant gen- era during the Maastrichtian and Paleocene. The fossil record does not fully support these ideas. As mentioned in the previous chapter, there are leaves of Aptian-Albian age with Hama- melidoid features in relatively high latitudes of both hemispheres. The oldest fossil record de- termined to family are pollen grains of Ulmaceae of Turonian age in the equatorial belt, in Borneo. And the oldest determined to genus are pollen grains of Santonian age, also at high latitudes in both hemispheres: Nothofagus in Australia and Alnus, Betula, and Myrica in Japan, Canada, and the United States. Therefore, it seems more plau- sible that the subclass originated in tropical areas of West Gondwanaland, with different families differentiating polewards. Nothofagus was at that time the only representative of Fagaceae being a vicariant of the northern Betulaceae and Myri- caceae. Such an early differentiation of only this genus from related families would support the above mentioned suggestion to segregate Noth- ofagaceae as an independent family. From West Gondwanaland, the ancestors of Nothofagus should have migrated to southern South America and/or southern Africa. South merica was at that time united to west Ant- arctica, which in turn was very close to or at the South Pole. Southern Africa was separated from east Antarctica, but only by about 1,000 km; the NOTHOFAGUS EVOLUTION 279 coast of east Antarctica was probably at about 60*S. From west or east Antarctica the ancestors of Nothofagus should have reached Australia be- fore the Santonian and New Zealand before the Campanian. Campanian and Maastrichtian times wit- nessed the diversification and migration of Noth- ofagus, which occupies every Gondwanaland land mass, except South Africa and India, and is re- corded with the three types of pollen grains and petrified trunks. In the Laurasian continents, in contrast, there is a radiation of different families: Platanaceae, Cercidiphyllaceae, Juglandaceae, and Fagaceae (Castaneoideae), which may be found in ыы outcrops as different kinds of fossils. By the Paleocene and Early Eocene, Nothofa- gus was represented in the Southern Hemisphere, though it was still not dominant. Casuarinaceae and Didymelaceae, the two other southern fam- ilies of the subclass, are recorded for the first time. Also there are some cosmopolitan genera (Dryophyllum, Fagophyllum) usually, yet with growing hesitation, attributed to the Fagaceae. In any event, фе м evidence for a relatively tic stock in the Southern Hemisphere. In the Northern Hemi- sphere the radiation continued, with records of the new families Eucommiaceae and Hamamel- daceae, and more genera in the Platanaceae, Betulaceae, Ulmaceae, and Juglandaceae. The Fagaceae were still only represented by the Cas- taneoideae. During the Eocene the Fagaceae had a rich fossil record comprised of every kind of fossil. In South America, Nothofagus is represented by abundant woods, leaf imprints, and pollen grains, and may be dominant in the floras. Australian authors (Christophel, 1981; Martin, 1981, 1982) remarked that in Australia pollen grains of Noth- ofagus may be very abundant in Eocene sedi- ments, but imprints of leaves are very rare, if even present. This is explained by a mosaic pat- tern of forests with patches of a subordinate Nothofagus, but with favorable conditions for dispersal of the pollen grains. Fagaceous leaves of Eocene deposits are the oldest that can be placed in extinct species of living genera. They belong to Quercus, “ Casta- ea," and Fagus in the Northern Hemisphere and to Nothofagus in the Southern Hemisphere. As noticed above, very low rates of morpho- logical change in Nothofagus are recorded in their pollen grains since the Cretaceous and in their important I —. 280 leaves since the Eocene. These long periods of stasis seem to conform to the “punctuated equi- librium” model of evolution. However, the pro- cess and timing of the radiation of the stable species in both cases is not well established. Most of the South American fossil leaves that resemble those of living species are similar to the deciduous taxa inhabiting the northern part of Nothofagus distribution, in northwest Patagonia (Romero & Dibbern, 1985). Also I restudied fos- sil leaves from Australia, formerly published by Ettingshausen (1888) and found that deciduous species were present there at the same time. Therefore, there is a strong possibility that we are witnessing in these floras the development of deciduousness in Nothofagus, or at least an adap- tive radiation from deciduous ancestors in Ant- arctica. About 35 to 40 Ma, that is, by Late Eocene or Early Oligocene, Australia was separated from Antarctica (Raven & Axelrod, 1972; Kemp, 1978). A more continuous oceanic current de- veloped around Antarctica, substantially chang- ing the climate of the Southern Hemisphere. The development of deciduousness was a probable response to that change (Romero, 1984). Later, in connection with the migration of Australasia to the north (Kemp, 1978; Martin, 1981; Smith et al., 1981), some extinction occurred there leaving the Tasmanian N. gunni as the only de- ciduous species. Leaf taphofloras and palynofloras of Eocene and Oligocene age show repeatedly a mixture of species of supposed cold temperate climate with others of supposed subtropical or tropical cli- mate (Kemp, 1978; Romero, 1978, 1984). They are rather variable through geological time and latitude, and have different percentages of Nothofagus as the most important cold temper- ate element. The same authors suggested that these Paleogene associations are not pum by any of the living plant associatio A problem is imposed by the e in South America of fossils attributed to Fagaceae, which are beyond the range of variation of Nothofagus and within that of Fagus. They are leaves and woods and were reported as southern Fagus. Similar but more doubtful remains were de- scribed from Australia and New Zealand. A ge- netical relationship with Fagus seems unlikely, same genetic stock (no matter if it is о at the familial, ordinal, ог superordinal level) and that have remained isolated since the Creta- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ceous. A similar and consistent case for conver- gence could also be made for supposed Notho- fagus widespread in North America (Elsik, 1974). They are not really Nothofagus (Romero, 1977: 181; Muller, 1981) but share important charac- ters. The oldest records of pollen grains of Quercus and Fagus are from the Oligocene, several mil- lion years after the first record of leaves (Romero, 1984). This contrasts with the long history of the pollen grains of Nothofagus and supports the consideration of the southern genus as a more independent, earlier derived taxon. Also, the Oli- gocene record of Fagopsis in the United States, showing features of different genera of Fagaceae, except Nothofagus, is additional argument for an independent treatment of this genus. uring Late Eocene and Oligocene times, New Guinea rose above sea level (Axelrod & Raven, 1982). The earlier records of Nothofagus are pol- len grains of Miocene age. Raven and Axelrod (1972) suggested that the genus migrated from northwest Australia and then diversified during the Pleistocene. New Caledonia has remained isolated since the Cretaceous, and has no fossil record. Raven and Axelrod (1972) believed that the Nothofagus species also remained isolated and are remnants of the old ipd stock. Therefore, о to these authors, Guinea and New Caledonia с different contingents x species, from different continents and with a gap of about 50 million years. How- ever, the living plants in both archipelagos are extremely similar in their flowers (Soepadmo, 1972), leaves (Romero & Carrasco Aguirre, 1982), pollen grains (Cookson & Pike, 1955), wood (Dadswell & Ingle, 1954), and ecology (Ash, 1982). The botanical evidence seems to indicate that the migration occurred from one of the ar- chipelagos to the other, but it is not clear which one was the source and when it happened. The Pliocene and Pleistocene were times of important climatic changes, related to glaciation and fluctuations of sea level. They produced ex- tinctions in some groups of plants and diversi- fication in others. Extinctions occurred with the species producing pollen of the “brassi” type in South America, New Zealand, and Australia. Ra- diation occurred in New Guinea, where the uplift of the mountains favored the differentiation of 13 still closely related species. There are also other groups of species very similar morphologically, which probably differ- entiated very recently. One group is that of N. 1986] betuloides, N. dombeyi, and N. nitida, with sim- ilar flowers (van Steenis, 1953), leaves (Romero, 1980), cuticles (Ragonese, 1981), and pollen (Heusser, 1971; Markgraf & D'Antoni, 1978), which may produce hybrids in mixed popula- tions (Donoso & Atienza, 1984). Another group of this kind, in New Zealand, is that formed by N. fusca and N. truncata, which have the same similarities and frequently hybridize (Cockayne & Atkinson, 1926). Finally, N. antarctica has marked ecotypes, and N. solandri has subspecies, which are evidences of very recent pressure of selection towards speciation. CONCLUSIONS The К 5511 а potheses regarding the evolution of ovx ed and related taxa. They are: 1) Ancestors of Hamamelidales and related or- ders differentiated in tropical areas, in West grated toward Laurasia, where Ulmaceae, ulaceae, and Myricaceae radiated, and towards southern Gondwanaland, where only Nothofagus ancestors evolved. This would support the sug- gestion of giving a higher rank to the southern enus. 2) Nothofagus ancestors migrated from trop- ical West Gondwanaland to southern South America-west Antarctica, and/or to southern Af- rica-east Antarctica. From Antarctica they reached Australia in the Santonian and New Zea- land in the Campanian. 3) During the Eocene living genera of Fagaceae differentiated in the Northern Hemisphere and direct ancestors of living species of Nothofagus differentiated in the Southern Hemisphere. These changes are registered by fossil leaves that rep- resent the onset of deciduousness in response to global climatic changes. Deciduous species of Nothofagus were present in Australia and New Zealand, but were subsequently lost. 4) Long periods of stasis in the morphological change of fossil species seem consistent with the punctuated equilibrium model, but the early, rapid differentiation of the group is not yet well documented. 5) During Eocene times, Nothofagus was a member of mixed forests, comprised of subtrop- ical and temperate elements 6) Nothofagus reached New Guinea by Oli- gocene or Miocene times. The circumstances of migration to New Caledonia are not clear. Bo- ROMERO- NOTHOFAGUS EVOLUTION 281 tanical evidence is consistent with exchanges be- tween the two archipelagos. 7) Climatic changes during the Pliocene and Pleistocene were responsible for the extinction of the Nothofagus “brassi” group in Australia, New Zealand, and South America, as well as for speciation between several closely related mod- ern species. LITERATURE CITED ARCHANGELSKY, S. & E. J. ROMERO. 1973. Los re- istros más 'antiguos del polen de Nothofagus (Fa- gaceae) de Patagonia nce y Chile). Bol. Soc. Bot. México 33: 13-3 AsH, J. 1982. The ro AA Bl (Fagaceae) of New Guinea. Pp. 355-380 in J. L. Gressit (editor), Bio- geography and Ecology of New Guinea. W. Junk Publ., The Hague AXELROD, D. I. & P. H. RAVEN. 1982. m raphy and origin ofthe New Guinea flora 9- 94] in J. L. Gressit (editor), Bopeowaphy and Ecology of New Guinea. W. Junk Publ, The ague BERRY, E. 1937. A Paleocene flora s Patagonia. Johns Hopkins Univ. Stud. Geol. 12: 33-50. 1976. Middle on floral prov- perms. Pp. 23- lution of Angiosperms. Columbia Univ. Press, New CARRASCO AGUIRRE, A. & E. J. RoMERO. 1982. Ar- quitectura foliar de las especies neocelandesas y australianas de Nothofagus Bl. Bol. Soc. Argent. Bot. 20: 227-240 CHRISTOPHEL, D. 1981. Tertiary megafossil M kel Australia as indicators of floristic associations paleoclimate. Pp. 380-390 in A. Keast (editor) Ecological Biogeography of Australia. W. Jun Publ., The Hague COCKAYNE E. AT KINSON. 1926. On the New Zealand wey hybrids of Nothofagus. Gus 8: бы I. С. & К. M. Pike. 1955. The pollen mor- phology of Nothofagus Bl. subsection Bipartitae Steen. Austral. J. Bot. 3: 197-206. Couper, R. A. 1960. New Zealand Mesozoic and Cainozoic plant microfossils. New Zealand Geol. Surv. Paleontol. Bull. 32: 1-82. CRACRAFT, J. 1975. Historical biogeography and earth history: perspectives for a future synthesis. Ann. Missouri Bot. Gard. 62: 250-277. CREPET, W The status of certain families of the Amentiferae during the Middle Eocene and some hypot 1 erus i Evolution, Volume 2. Praeger Publ., New York. . DAGHLIAN. 1980. Castaneoid inflo- rescences from the Middle Eocene of Tennessee level) in the ` Fagaceae. Amer. J. Bot. 67: 739-757. CRONQUIST, A. 1968. The Evolution and Classifica- tion of Flowering Plants. H. Mifflin Co., Boston. 282 DapsweLL, H. E. & A. M. na 1935. The identification of the princip cial Austra- lian timbers other than Eucalypus Tech. Pap. Forest. Prod. a Australia. 16: 1-102. & H. D. | e wood anatomy of the New Guinea Nothofagus Bl. Austral. J. Bot. 2: 141-153. DARLINGTON, P. J. 1965. Biogeography of the South- ern End of the World. Harvard Univ. Press, Cam- bridge. ОЕ CANDOLLE, A. P. 1864. Prodromus Sa e Naturalis Regni Vegetabilis, Volume 16, 2. Pari DETTMANN, M. E. & G. PLAYFORD. 1969. Hed pied of the Australian Cretaceous: a review. Pp. 174- 210 in K. S. W. Campbell (editor), Stratigraphy and ci Australian National Univ. Press, Canberra. E. J. ROMERO. 1984. Revisión de Argentino de oe ogia y Bioestratigrafia. Corrientes, Argen Donoso, C. & J. ATIENZA. ~ 1984, Hibradación natural entre Nothofagus betuloides (Mirb.) Oerst. y Nothofagus nitida (Phil.) Krasser. Medio Am- biente 7: 9- DoYLE, J. & L. J. HICKEY. 1976. Pollen and leaves from the Mid-Cretaceous Potomac group and their bearing on early Angiosperm evolution. Pp. 139- 206 in C. B. Beck (editor), Origin and Early Evo- lution of Angiosperms. Columbia Univ. Press, New York ork. Dusen, P. 1907. Uber die Tertiáre Flora der Magel- "rera а Exped. asus lindas 1(4): 8 on V Über die Tertiáre Flora der Seymour Insel. Wiss. Ergebn. Schwed. Südpolarexped. 1901- 1903, 3: 1-27. Euas, T. $. 1971. The genera of Fagaceae in the Southeastern United States. J. Arnold Arbor. 52: LSIK, W. C. 1974. о: in North America. Pollen & Spores 16: 285-2 Enpress, P. К. 1977. E trends in the Hamamelidales-Fagales Group. Pl. Syst. Evol., Suppl. 1: 321-347. о К. von. 1886. Beitráge zur Kennt- s der Fossilen Flora Neusselands. Denkschr. Kai iserl. Akad. Wiss., Math.-Naturwiss. Kl. 53: 43- . 1888. Contributions to the Tertiary Flora of Australia. Mem. ol. Surv. New South Wales, Palaeontol. 2: 1- Evans, W. P. A fossil Nothofagus (Notho- fagoxylon?) from the Central Otago coal measures. Trans. & Proc. New Zealand Inst. 42: 98-1 FORMAN, L. L. 1966. On the evolution of cupules i in the Fagaceae. Kew Bull. 18: 385-419 — S. L. & D. E. FAIRBROTHERS. 1976. Paly- axonomic investigations of Fagus L. an cue Bl.: light microscopy, scanning elec- * 2 A 1 $ B ot. Syst 1: 1-141. HARLAND, W. B., G. PICKTON, A. G. SMITH & R. WALTERS. V. Coc, P. G. LLEWELLYN, C. A. 1982. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 A Geological Time Scale Series. Cambridge Univ. Press, Cambridge. Eis M. 1985. Biogeographic analysis of Notho- gus (Fagaceae). Taxon 34: 478-480. Hes C. 1971. Pollen a Spores of Chile. Univ. “1977, Early Cretaceous fossil evidence for angiosperm evolution. Bot. Rev 1 . Nothofa agus als up. from the Tertiary of Tasmania. Alcheringa 7: 169-183. . V. . S. REI 957. Forest trees and timbers y New Zealand. Bull. New Zealand Forest S 1-211. erv. HJELMQVIST, н. 1948. Studies оп the floral тог- в and phylogeny of the Amentiferae. Bot. 2: 1-171. . 1983. Fossil Nothofagus from the Longford Formation, Murchison, New Zealand. J. Roy. Soc. New Zealand 12: 65-77. HUMPHRIES, C. J. 1981. Biogeographical methods and D. R. Brooks (editors), Advances in Cladistics, ше 1. New York Bo- tanical Garden, New Temperate DOS HS and an in- temperate botanist. Taxon 34: 480-492. . 1984a. Leaf Architectural and Cuticular ica. Unpublished Ph.D. dissertation. Indiana Univ., Bloomin 1984b. Foliar features of the Fagaceae and their implications on evolution of and within the Fagales. Amer. J. Bot. 71(5, 2): 110. [Abstract.] Kemp, E. M. 1978. Tertiary climatic evolution and vegetation history in the southeast Indian Ocean Region. Palaeogeogr. Palaeoclimatol. Palaeoecol. 24: 169-208. KHAN, A. M. 1974. Palynology of neogene sediments from Papua (New Guinea) stratigraphic bound- aries. Pollen & Spores 16: 265-283. KUPRIANOVA, L. A. 1965. The е of Amen- tiferae. Nauka, Moscow. [In Russian.] MANCHESTER, S. R. & P. R. CRANE. 1087 Attached cene Florissant Flora of Colorado, USA. Amer. J. Bot. 70: 1147-1161. MARKGRAF, V. & H. L. D’ANTONI. 1978. foni duda of Argentina. Univ. of Arizona Press, Tuc ша Н. 1981. The Tertiary Flora p. 393-406 n A. Keast pues Ecological Biogeography of шша W. Junk Publ., The Hague ——_. 1982 [1983]. ca cenozoic barriers and the Australian paleobotanical record. Ann. Mis- souri Bot. Gard. 69: 625-667. MELVILLE, R. 1973. Continental drift and plant dis- tribution. Pp. 439-446 in D. H. Tarling & S. K. Runcorn (editors), Implications of Continental Drift to the Earth Sciences, Volume 1. Academic Press, New York. 1981. Vicarious plant distributions and pa- leogeography of the Pacific Region. Pp. 341-358 nG. Nelson & D. E. Rosen (editors), Vicariance 1986] ие A Critique. Columbia Univ. Press, ork. ew MULLER, ү 1981. a pollen records ofextant An- iosperms. Bot. Rev. (Lancaster) 47: 1-142. NIXON, K. 1984. A sib гасын phylogenetic — of Hamamelidae. Amer. J. Bot. 71(5, 2): 111. [Ab- stract.] i prin W. R. & M. N. PHILIPSON. Leaf rnation in Nothofagus. New PA m 17: 417-421. PRAGLOWSKI, J.C. 1981. Transition within the exine of Not thofagus Blume. Rev. Paleobot. Palynol. 32: B Fagaceae L. Fagoideae. World Pollen and Spore Flora, Volume 2. Almqvist and Wiksell, Stockholm RAGONESE, A. M. 1977. Nothofagoxylon menendezii, lefio petrificado del Terciario de Gen Roca, Río „зы Argentina. Ameghiniana 14: 75—86. 1981. Anatomia foliar de las especies suda- mericanas de Nothofagus Bl. (Fagaceae). Darwin- iana 23: 587-603. RAVEN, P. H. & D. I. AXELROD. and Australasian P: 86. 1379-1 1972. Plate tectonics graphy. Science 176: 1974. Angiosperm biogeography and past continental movements. Ann. Missouri — pow 61: Rom J. E 73. Polen fósil de Nothofagus (Noihofagidies e i es y Paleoceno de Pat- . La Plata, Secc. Paleontol., n.s. T . 1977. "Polen de Gimnospermas y Fagaceas dela Formación Río Turbio (Eoceno), а Cruz, Argentina. FECIC Editorial, Buenos Air 978. Paleoecologia y paleofitogeografí ía de las tafofloras del Cenofitico de Argentina y áreas vecinas. Ameghiniana 15: 209-227. 80. Arquitectura foliar de las especies su- damericanas de Nothofagus Bl. Bol. Soc. Argent. Bot. 19: 289-308. 1984. Historia y evolución de Nothofagus deraciones sobre el origen de . 209-216 in Actas e ongreso Arge entin no de лын y иен Corrientes, Argen & S. ARCHANGELSKY. 1985. Fay Cretaceous Angi perm al leaf from Patagonia: the first record. Poe 227: (in press). A. ASCO AGUIRRE. 1982. Arquitectura foliar de us Mus x wide bn Bl. d de Nueva Guinea y Nueva Caledonia. Bol. Soc dii Bot. 21: 213-24 s M. C. DiBBERN. 1985. Areview ofthe species described as Fagus and Nothofagus by Dusen. Pa- ROMERO- NOTHOFAGUS EVOLUTION 283 leontographica, Abt. B, Paláophytol. 197: 123- 137. SALARD, M. 1961. Contribution a l'etude paleoxy- lologique de la Patagonia. II. Rev. Gén. Bot. 68: 234-270. SCHUSTER, R. M. 1976. Plate tectonicas and its bear- angiosperms. Pp. 48-138 Origin and Early Evolution of Angiosperms. Co- lumbia Univ. Press, New York. SMILEY, С. J. & Г. M. Hucains. 1981. Pseudofagus idahoensis, n. gen. et sp. (Fagaceae) from the Mio- 2 Clarkia flora of Idaho. Amer. J. Bot. 68: 741- 7 SMITH, е G., А. M. HURLEY & J. C. BRIDEN. 1981. Phenerocrie pisei poeta World Maps. Cam- e Univ. Press, Cambridge. NE ind E. 1972. Fagaceae. Fl. Males. I, 7: 277- STEENIS, C. G. G. J. vAN 1953. Results of the Arch- bold Expedition. Papuan Nothofagus. J. Arnold Arbor. 34: E -374. Nothofagus, key genus of plant geog- raphy i in time and space, living and fossil, ecology and phylogeny. Blumea 19: 65-98. Stover, L. E. & P. R. Evans. 1973. Upper Creta- ceous- -Eocene spore-pollen zonation, offshore Gippsland Basin, Australia. Geol. Soc. Australia, . L. 1980. Ой tline of the classification wering Suid (Magnoliophyta). Bot. Rev бз; > 225- THORNE, R. F. 1973. The "Amentiferac" or Hama- melidae as an artificial group: a summary state- ment. Brittonia 25: 395- 983. Proposed new realignments in the an- iosperms. Nordic J. Bot. 3: 85-117. 4 Identificación de madera fósil del р. Congreso Latinoamericano de Paleontologia. México. & M LO. 1981. Anatomía de troncos s siles del Cretácico superier de Pichasca, en el nort de Chile. Pp. 385-398 in Anais II Congreso Pe mae reer de Paleontología, Volume 1. Por- rea L. A. 1956. Maderas y m Argen- tinos. Acme Editorial, Buenos Air WAGEMANN, G. 1948. Maderas chilenas Contribu- id tomi Lilloa 16: 263— 375. WOLFE, J. A. 1973. Fossil forms of Amentiferae. Brit- tonia 25: 334-355. EVOLUTION AND REPRODUCTIVE BIOLOGY OF INFLORESCENCES IN LITHOCARPUS, CASTANOPSIS, CASTANEA, AND QUERCUS (FAGACEAE)! ROBERT B. KAUL? ABSTRACT Of Lithocarpus, Castanopsis, Castanea, and Quercus only M SUARUM Eu branched spikes suggestive of the primitive condition in the Fagaceae; the other g e them in some individuals. Simple and branched spikes can occur on n the same bandi stages from perfect, entomophilous to imperfect, anemoph rcus, which is also the only anemophilous member of these four genera. Lithocarpus is the least incid while Quercus is the most specialized in inflorescence structure of the four genera. Pathways of inflorescence evolution in Cas- tanea, Castanopsis, Lithocarpus, and Quercus have been proposed that identify Lithocarpus and Quercus as possessing the least and most spe- cialized inflorescences, respectively (Kaul & Abbe, 1984). The specialized inflorescence char- acter states of Quercus are associated with ane- mophily and the less specialized states in Litho- carpus, Castanea, and Castanopsis relate to their entomophily (Kaul, 1985). Kaul et al. (1986) sug- gested that the evolution of anemophily in Quer- cus [and probably also in another fagaceous species, Trigonobalanus daichangensis (Camus) orman] was an adaptation for successful pol- lination in seasons of low pollinator activity in seasonal climates, for example, spring in the tem- perate latitudes. In such latitudes the anemoph- ilous Fagaceae — Quercus and Fagus —flower briefly in the spring whereas the entomophilous Castanea and northerly species of Castanopsis and Lithocarpus flower over a longer period in early summer. Likewise, flowering peaks in en- tomophilous paleotropical Fagaceae (Castanop- sis, Lithocarpus) coincide with maximum insect activity (Kaul et al., 1986; cf. Fogden, 1972). Quercus, which is anemophilous throughout its range, flowers at various times in the tropics but not in the wettest seasons. Further evidence of the functional aspects of these inflorescences is scanty, and therefore definitive answers to ques- tions about the ancestry and subsequent func- tional evolution of fagaceous inflorescences must be sought through comparative studies of living and fossil taxa In these genera the spikes and catkins usually bear numerous sessile, subsessile, or, in a few instances, pedunculate flower clusters that are variously called dichasia, cymules, or partial in- florescences. They are the only structures that presumably terminate in a flower in these four genera; all other structures— branches, rachises, spikes, and catkins—apparently terminate in a vegetative bud that may or may not be capable of further growth. The dichasia are surely derived by phyloge- netic condensation and the ancestral inflores- cences were probably thyrse-like or panicle-like. Some modern Fagaceae, most notably some species of Lithocarpus, produce branched spikes that are themselves aggregated into many- branched inflorescences of higher orders, result- ing in the most complex floral displays in the family. The ancestral fagaceous inflorescence was no doubt more complex than any modern inflores- cence in the family. Fey and Endress (1983) have advanced credible evidence that the cupule of the fruit is itself a product of condensation of dichasial branching systems. Further, Hjelm- qvist (1948) interpreted some staminate flowers as pseudanthia. Kaul and Abbe (1984) suggested evolutionary changes in fagaceous inflorescences that included loss of syllepsis in both vegetative and reproductive shoots and separation of sta- ! Research supported by National Science Foundation grants DEB-7921641 and DEB-8206937. ? School of Biological Sciences, University of Nebraska, Lincoln, Nebraska 68588-0118. ANN. MISSOURI Bor. GARD. 73: 284-296. 1986. 1986] minate from pistillate flowers in the total inflo- rescence. We also suggested that the catkins of anemophilous Fagaceae were derived from spikes of entomophilous ancestors and that there is a rough gradient of decreasing inflorescence com- plexity with increasing latitude. In our previous papers (Kaul & Abbe, 1984; Kaul, 1985) some complex fagaceous inflores- cences were described and illustrated. Here I elaborate upon our work, illustrate additional examples of complex inflorescences, and provide further evidence for their interpretation as the least specialized forms among extant Fagaceae. It must be emphasized that even the least spe- cialized extant forms are advanced for angio- sperms as a whole, as evidenced by the presence of such extreme reductions as dichasia and cu- pules in every species. MATERIALS I have examined thousands of specimens of paleotropical, eastern Asiatic, and North Amer- ican species of Castanea, Castanopsis, Lithocar- pus, and Quercus that were gathered by my col- leagues and me with this and related studies in mind. All the specimens illustrated here are in my personal collection but various sets of du- plicates are deposited at A, GH, K, L, and US. Nomenclature herein follows Soepadmo (1972) for the paleotropical species and for generic de- limitation. Figures 4-16 show actual specimens portrayed indicate staminate dichasia; quantities of each are only approximated. TERMINOLOGY Figures 1-3 illustrate many of the structures and terms used here. A shoot system (a major indeterminate axis with a leader that periodically dergoes extension growth, and bearing branches) can include proleptic or sylleptic shoots (produced with or without a resting period of the parent bud, respectively) or both, all or nearly all of them floriferous. Most shoots are long- shoots (with normally-extended internodes) but some, especially in Quercus, are short-shoots (with reduced internodes). The total inflorescence in- cludes all the floriferous shoots on a shoot sys- tem. Sz foliose; they к KAUL—FAGACEAE INFLORESCENCES 285 may be simple spikes without obvious branches or branched spikes that bear one or more florif- erous branches. Catkins are the unbranched, pendent, staminate axes of Quercus and Trigo- nobalanus daichangensis. For simplicity, bract is used here for any non-foliar leaf homologue ir- respective of its size or placement. Except for the dichasia (a term used here gen- erally for the flower clusters, whether or not they are strictly dichasial), all axes terminate in a veg- etative bud that is often abortive. Some of these axes are capable of further growth (indetermi- nate) but some are not (determinate). These gen- era are pantotonic (cf. Briggs & Johnson, 1979) in that nearly every leaf- or bract-bearing node is floriferous, although not necessarily synchro- nously in a flush. OBSERVATIONS PHENOLOGY AND GROWTH HABITS The temperate and tropical species of these genera exhibit rhythmic growth, that is, they flush at intervals. More northerly species flush more or less synchronously at a locality at least once a year in spring or early summer; some have succeeding flushes in the same growing season but only the first flush bears flowers and all but the final flush produce branches. Paleotropical species of low and middle elevations often flush and flower repeatedly in a year but there is wide intra- and interspecific variation in local flushing patterns. Sometimes the flushes are so closely spaced that anthesis in one is finishing when the next flowering begins. In some tropical lowland species flushing is non-synchronous even on a single tree (Kaul et al., 1986). Certain principles apply to the sequence of events in the growth of the trees and some of these are illustrated in Figures 1-3. All species produce flowers on spikes or catkins borne singly or on reproductive shoots and short-shoots in the leaf axils of the current flush and often also in the axils of the previous flush. Most leaves and some bracts subtend one of these structures. In species that branch sylleptically (some species of tropical Castanopsis, Lithocarpus, and Quer- cus) the indeterminate shoots are also floriferous (Fig. 6 and cf. Kaul & Abbe, 1984). Fertile pro- leptic shoots are produced synchronously with the extension growth of the leader in a branch system (Figs. 5, 15, and numerous figures in Kaul & Abbe, 1984) but sometimes they precede it ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 ; e P e Y^ 0 e Y Pistillate spike р оо°° `O, Staminate spike >, оо Stem of current flush Maturing fruit FIGURES 1 a tr Development of a single flush.—1. Before flush begins. Maturing fruit of previous flush present. а d (triangle) still dormant.—2. End of flush; anthesis. Determinate (without apical bud) and indeterminate s apical bud) shoots indicated, along with simple and branched spikes. Uppermost spike is andr dics gs D dichasia shown in black, staminate as open circles. — 3. Post-anthesis. Staminate spikes ко fallen $ eterminate shoot. Fruits of previous flush nearing maturity but those of current flush are immature. “Potential succeeding flush stippled. —4. Lithocarpus elegans (Bl.) Hatus. ex Soep., from Thailand. ). Apical bud of the main axis abortive. — 5. L. leptogyne (Korth.) Soep., from Borneo. Branched spikes are borne on the leader and on the shoots. 1986] KAUL—FAGACEAE INFLORESCENCES 0000000000: 000000000000- o 00000- 00000000- 0000000000000- 00000008 0000000000000- ©, 9 000000 %, 00000000- e000000000000000 FIGURES 6-8. Lithocarpus.—6. L. sundaica (Bl. Rehd., from the Malay peninsula. Maturing о with nate add Pistillate spikes have not yet appeared.—7. L. brev- he . L. lampadaria о А. Camus, from the drogynous branches. The simple spike at the lower инк is eximio eria. and bears one бени with both staminate and alae ng нов (arrow 288 (Figs. 4, 16). These proleptic shoots arise from dormant axillary buds at the nodes of the pre- vious flush, whether or not the leaves of that flush are still present. All determinate axillary reproductive struc- tures, including entire spike- or catkin-bearing shoots, are shed after anthesis except the spikes with pistillate dichasia. Immature fruits of the revious flush are often present at the time of subsequent flushes (Figs. 1-3, 14, 15, 24), a char- acteristic of tropical species and of some tem- perate-zone species. The first nodes beyond the bud scales typically produce only bracts, but these often subtend fer- tile axes as large as, or even larger than, those axillary to the fully-developed foliage leaves be- yond (Figs. 5, 14, 15). Sylleptic shoots usually produce normal foliage at all their nodes. The floral displays of Lithocarpus, Castanop- sis, and Castanea develop with the new flushes at the branch tips and are therefore easily seen massive, light in color, and distinctly fragrant and thus ин clouds ar insects. The displays ofQ ibstantial but are drab, odorless, and inconspicuous, especially when hidden among the leaves in evergreen species. In all four genera the pistillate spikes are borne dis- tally in the total inflorescence and are thus ex- posed. SPIKES, CATKINS, AND OTHER REPRODUCTIVE BRANCHES Many species produce spikes that bear only staminate, only pistillate, or various combina- tions of staminate and pistillate flowers; Quercus has only staminate catkins and pistillate spikes. Furthermore, some individuals of some species (except Quercus) regularly produce simple spikes whereas others produce more complex struc- tures. For example, only one tree each of Litho- carpus dealbata (Fig. 10), L. elegans (Fig. 4), L. harmandii (Fig. 12), and L. lampadaria (Fig. 8) bore branched spikes out of eight, 11, five, and nine trees respectively examined from widely separated areas. Such inconsistency is not con- fined to inflorescences but is well-known in leaf and even fruit morphology throughout the fam- ily. In contrast, most trees of L. gracilis, L. coo- perta (Fig. 9), L. brevicaudata (Figs. 7, 24) and L. leptogyne (Fig. 5) had branched spikes. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 It is obvious from inspection of Figures 4-27 that the structure of reproductive axes varies even on a single shoot. It is common for the extension- growth axes to produce a variety of axillary shoots that are more complex acropetally (Figs. 10, 12, 13) or basipetally (Figs. 5, 6) or medially (Figs. 7, 8, 11). When proleptic reproductive shoots appear synchronously with a new flush, as they usually do, they too are included in these gra- dients (Figs. 4, 5). The leader and some reproductive shoots are indeterminate (e.g., Fig. 2) but their apical buds remain dormant until the next flush, by which time the spikes or catkins have fallen. Other re- productive shoots are determinate and produce only spikes (Fig. 2); these entire shoots are often deciduous after anthesis unless they bear pistil- late flowers. Determinate shoots must be distin- guished from branched spikes; the latter have whereas the former clearly terminate in an abor- tive bud and lack flowers anywhere on the pri- mary axis. In some instances determinate and indeterminate shoots occur, serially, with simple and branched spikes on a branch and the patterns can vary from branch to branch on a tree. Thus the total inflorescence defies classification be- yond panicle, sensu lato. Lithocarpus. Of the four genera, Lithocarpus shows the greatest variety of inflorescence mod- ifications. Whereas some species regularly pro- duce branched spikes mixed with simple spikes and various reproductive shoots (Figs. 2-12, 23- 25, 27), many, perhaps most, species have simple spikes only, although these may be grouped on shoots that themselves produce massive floral dr (Figs. 25, 27). The patterns approach ose of Quercus in some species. pus of the inflorescence complexity is found in the staminate shoots and spikes but in a few igs. and L. lampadaria (Fig. 8), the pistillate spikes are branched too. Of the species studied, padaria has the most complex branching of iiki that bear pistillate flowers. In it, large branched spikes numerous pistillate dichasia and even more numerous staminate dichasia. Some di- chasia include both staminate and pistillate flow- ers (Fig. 8, arrow) and some spikes have the sta- minate and pistillate dichasia mixed for short distances (Fig. 20). The spikes bearing pistillate dichasia in Г. 1986] KAUL—FAGACEAE INFLORESCENCES 289 FIGURES 9-12. Lithocarpus. —9. L. cooperta (Blanco) Rehd., from Borneo. Of the four determinate (without apical buds indicated) shoots, one bears a single branched spike.— 10. L. dealbata (Hook. f. & Thoms.) Rehd., from Thailand. Most spikes are simple but two are branched. The uppermost quie 1 is androgynecandrous and ои идет.) Rehd., from the Malay peninsula. Simple and branched spikes mixed along t the axis. An rogynous spikes are distal. — 12. Г. harmandii A. Camus, from Cambodia. Most spikes are simple but four е ѕрікеѕ еасһ bear a single branch 290 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 se ids >) preme P00, оо оооооо ниче Әә 5 a 3 -00000000000000 QR an о QAR RASA -000000 UJ mu 4 Ao р е ool TEE. O 2928 9998 e0029% 50 38 Е оо eo оо Sm 3:00 AA 99959 0:09 Ф, от oto SS бо oyi S 8 о о MEN JQ 258 5 5 очо o A Пос 9-40 g o Pd TM O 27 © фо осо $98 $ © “сос $9) © goo оо ох Шоо 69 0 © o: $$ 88 ko. Le e 2$ a o Ў oo fo o, 0:0:0 оо 8 оо 0:00 oo оо 0:0 0? отоо, ооо: 9D 00; 0:00 9 8292 233 mtg 38 ae THF ро 98 со, hom HE 900 aay o 089 $6 099 о о Ш” 909 о о50 e 908 foo§ 909 оо обо & of ovo QV. ü Sim, ¿E of: 08 о о ого ^ © оо n о o8 о o9 о о 9.8 о оо о o8 о (бо о фо D 9 FIGURES 13-16.— 13. Castanea crenata Sieb. & Zucc., from Taiwan (cultivated). The uppermost dichasia of one spike are replaced by staminate branches.— 14. Quercus marilandica Muenchh., from Nebraska. Branch from. a tree that annually. bears O catkins and branched catkins (right arrow). Left arrow indicates a d of the previous flush. — 15. О. pachyloma O. Seem., from Taiwan. Typical inflorescence for many aes hires Proleptic indeterminate long-shoots are basally floriferous and the o ost one, as well as the leader, bears pistillate spikes. One staminate short-shoot (arrow) is shown. — 16. um aB om Taiwan. Although the new flush has barely begun to et extension growth, each leaf of fhe dcus flush has produced a proleptic, staminate short-shoot in its axi 1986] KAUL—FAGACEAE INFLORESCENCES 291 FIGURES 17-21.—17. Ein cambodiensis Hick. & A. Camus, from Cambodia. Pistillate spike in anthesis. Some r sta s. Abortive vegetative apex indicated by the arrow. — 18. Q. borealis — ^ iin Priorum Pistillate spike with only: two one-flowered dichasia. Abortive 19 ike just before anthesis. — 20. L L. la mpadaria, from the Malay. po ba cus chapmanii, from Florida. Stem q of spike showing mixed staminate and pistillate dichasia.— 21. Quer segment showing m (left) and two short-shoots, one of which bears a single catkin. Pistillate spikes are forming in two leaf ax 292 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 22-25.— 22. Castanea crenata Sieb. & Zucc. Arrow indicates one of two pistillate dichasia on that spike. Two other spikes have a single basal pistillate dichasium and four spikes are entirely staminate.—23. ble. Some leaves have fallen.—24. L. brevicaudata Hayata showing prominent branched spikes.—25. L. neorobinsonii A us. Arrow indicates an indeterminate reproductive branch whose apical bud is barely visible. Below it, on the right side of the stem, is a determinate reproductive branch. Maturing fruits of the previous flush are shown. 1986] brevicaudata are simpler; only the distal one or two spikes have pistillate flowers and are some- times branched (Fig. 7). Each branch of the spike these species, the pistillate dichasia are most abundant distally in the total inflorescence. A variety of branching patterns in the stami- nate spikes and shoots is shown in Figures 4-7, 9-12, 19, 23-25, where there is a seemingly ran- dom interchangeability of simple with branched spikes on both the leader and lateral axes. Again the gradient of complexity is neither acropetal nor basipetal in the genus as a whole but appar- ently varies, even on the individual tree. In some cases illustrated, the branched spikes are more abundant acropetally (Figs. 9, 10, 12) or basip- etally (Figs. 6, 11) but elsewhere they are borne throughout (Fig. 7). ome spikes are unbranched, some have one branch (Figs. 6, 9, 12), and some have many (Figs. 7, 19, 24). Sometimes staminate dichasia occur directly on the primary axis among or be- low the branches (Figs. 6, arrow; 19). The branches of the spikes occur at dichasial nodes and are subtended by the same bract pattern as the dichasia (Fig. 19). The single New World species of Lithocarpus, gesting that this long-isolated species has re- tained this primitive character state. Castanopsis. There is less variability in in- florescence structure in Castanopsis than in Lithocarpus. Of the 41 species examined, only the single New World representative, C. chrys- ophylla (Dougl.) A. DC., has even an occasional branched spike (Kaul & Abbe, 1984, fig. 38). All others produce determinate and indeterminate shoots bearing simple spikes (e.g., Fig. 26) in patterns like those of Lithocarpus. astanea. Of the five species of Castanea ex- amined (C. crenata, C. dentata, C. mollissima, C. pumila, and C. sativa) only C. crenata bears branched spikes (Fig. 13), which are only occa- sional even in this species. However, a given tree sometimes produces them abundantly. Most spikes are simple and staminate (Fig. 13) but some of the more distal spikes often bear a few proximal pistillate dichasia (Fig. 22, arrow). Quercus. The pistillate spikes are always simple and borne singly in the axils of the most distal leaves (Fig. 15). They have one to ten or KAUL—FAGACEAE INFLORESCENCES 293 more dichasia (Figs. 15, 17, 18), each reduced to a single functional flower. Generally, the lowland paleotropical species have more pistillate di- chasia per spike than do the montane and north- ern species. Occasionally a few staminate flowers are present on the spike, and often there are per- fect flowers (Fig. 17). The vestigial vegetative apex is frequently prominent (Figs. 17, 18, ar- rows). The catkins are always staminate (except in distinctly aberrant instances) and are borne on oth long- and short-shoots. On the long-shoots they are commonly produced singly in the axils of the lower bracts (Figs. 15, 21), but they some- times occur in leaf axils as well. Short-shoots appear singly and proleptically at the nodes of the stem of the previous flush (Figs. 14-16, ar- rows; 21). They carry one to ten catkins and are invested at the base with the bracts of the bud that enclosed them (Fig. 21). Most short-shoots abscise soon after anthesis but a few eventually elongate and produce leaves. Figure 21 shows catkin whereas another has three, as does the long-shoot. Figure 15 illustrates the typical condition in Quercus. Pistillate spikes are borne in leaf axils of the leader and (often) one or a few of the most distal proleptic branches. Catkins appear on all the long-shoots of the current flush as well as on the short-shoots. Because Figure 15 illustrates an evergreen species, leaves of the previous flush st leaves are sterile at the time they appear, except for those that subtend pis- tillate spikes, but many leaves will subtend ax- illary long- or short-shoots in the next flush. Figure 16 illustrates a rather common phe- nomenon in paleotropical Quercus: the appear- ance of proleptic, reproductive short-shoots be- fore significant extension growth occurs. The extension growth also bears some catkins and all the pistillate spikes. However, some such pro- tandry (at the inflorescence level) occurs in more northerly species too, insofar as they display at least some catkins in anthesis before the pistillate flowers are receptive. Branched catkins (equivalent to the branched spikes of the other three genera) are occasional ern United States regularly produces them on an 294 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 28 | FIGURES 26-28.— 26. Castanopsis cuspidata Schottky. Simple staminate spikes are borne singly in the axils h of most new (darker) leaves.— 27. Lithocarpus dealbata (Hook. f. & Thomson) Rehder showing simple staminate spikes, some of them on indeterminate branches (obscured). The upper two spikes bear some pistillate and 1986] occasional tree. Figure 14 illustrates a branched catkin in the latter species, from a tree that was observed to produce them regularly for nine years. Most of the catkins are simple, whether or not they are borne on the current flush’s long- or short-shoots. Some of the catkins, although un- branched, are somewhat indeterminate in that they proliferate beyond the floriferous region and develop a few small leaves; after anthesis the dichasia drop from these proliferative catkins but the axis and leaves persist. An occasional pro- liferative catkin is also branched (Fig. 14, right rrow) and sometimes bears one or two pistillate dichasia. DISCUSSION The multiplicity of branching patterns of sta- minate spikes and shoots, particularly in Litho- carpus, shows that at least in some species flex- ible and variable patterns are the normal condition. The simple, unisexual spike or catkin is the advanced state in the family and occurs in most species, but a few species have retained branched spikes that suggest more complexity in ancestral forms of the genus. Of the four genera considered here, Lithocarpus has not only the most primitive inflorescences but also, according ( structure and have only occasional inflorescences suggestive of more complex ancestry. does Quercus have complete separation of pis- tillate from staminate function in most species, but it also bears the determinate reproductive shoots (short-shoots in this case) only prolepti- cally, in contrast with Lithocarpus, in which such shoots occur both sylleptically and proleptically. In Quercus, the spikes bearing pistillate flowers are morphologically very distinct from the cat- kins and well removed from them. In Lithocar- us, Castanea, and Castanopsis, by contrast, the pistillate flowers usually share a spike with at least a few, and often with very many, staminate flowers. Therefore, I conclude on morphological grounds, by employing the concepts of assessing evolutionary advancement by, among other KAUL—FAGACEAE INFLORESCENCES 295 things, the degree of spatial separation of sta- minate from pistillate functions, that Quercus is the most advanced of the four genera. Separation of staminate from pistillate func- tions has not yet progressed to dioecy in these genera. In fact, many species have morphologi- cally (ifnot a few Quercus species, and mixed-sex dichasia are frequent in Lithocarpus and Castanopsis (Kaul & Abbe, 1984). This contrasts with two puta- tively related families, Betulaceae and Juglan- о in wae nionoecy is clearly established. ilous but only р is, of the four fagaceous genera studied here, and it shows the closest approach to mon- oecy. This adds to the extensive evidence that monoecy (and also dioecy) is characteristic of anemophilous trees. Thus the Fagaceae can be interpreted as having some transitional stages rom perfect, entomophilous flowers to imper- fect, anemophilous flowers and from mixed-sex spikes to unisexual spikes. Homologies of staminate and pistillate spikes are indicated by numerous and frequent transi- tional forms, especially in Lithocarpus and Cas- tanopsis. The relationships of simple to branched spikes and of the spikes to reproductive shoots are also clarified by the presence of intermedi- ates, even on a single shoot system. Furthermore, every transitional stage between strictly stami- nate and strictly pistillate dichasia can be found, especially in Lithocarpus, and both of those ex- t flowers, including an obvious cupule like that of the pistillate di- chasia, there are suggestions of a cupular homo- logue in the abundant bracteation that charac- terizes some of them (Kaul, in prep.). The branches of the branched spikes occur at the sites of dichasia and, in fact, even have the same basal bracteation as the dichasia. Evolution of the simple spike can be interpreted as loss of branching capacities in the branched spikes. (The opposite change is ot И reactivation of branched from simple spikes.) Such a loss seems consistent with the losses of flowers and internodal elon- <— staminate dichasia, the point of separation between the distal staminate and proximal pistillate dichasia being L. indicated by the arrow.— 28 pistillate flowers develop fruit. suffruticosa (Ridl.) Soep. with two mature pistillate spikes. Typically, not all 296 gation that occurred during evolution of the in- dividual dichasium. The interchangeability of branches with dichasia suggests that the dichasia (both staminate and pistillate) are the result of evolutionary condensation of branches. Most of the critical work on the reproductive biology of inflorescences has been done on her- baceous plants [see Wyatt (1982) for a review] but there is no reason to suspect that the prin- ciples emerging from those studies would not apply to trees. Such research in the Fagaceae might help to explain such things as miniaturization of flowers and the accompanying evolution of the cupule, the failure of the family to advance to dioecy despite its antiquity, and the instability of inflorescence patterns in some species. Also of interest is the adaptive value of the formation mophilous monoecious dicotyledons (but not in the monocotyledons, where the opposite condi- tion prevails). Furthermore, such studies could help to explain the factors involved in the change from entomophily to anemophily in the Faga- ceae, a topic reviewed for plants in general by Regal (1982) and Whitehead (1983) and broached for the Fagaceae by Kaul et al. (1986) but in need of substantially more research. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 LITERATURE CITED BRIGGS, В. G. & L. A. S. JOHNSON. 1979. Evolution in the Myrtaceae—evidence from inflorescence structure. Proc. Linn. Soc. New South Wales 102: 157-256. Fey, B. S. & P. К. Емовеѕѕ. 1983. Development and morphological interpretation of the cupule in Fa- gaceae. Flora 173: 451—468. FoGDEN, M. P. L. 1972. The seasonality and popu- lation dynamics of equatorial forest birds in Sa- rawak. Ibis 114: 307-343. HijELMQVvisr, Н. 1948. Studies on the floral mor- phology and payiga of the Amentiferae. Bot. Not. Suppl. 2: 1-17 KAuL, R. B. 1985. gl reproductive morphology of Quercus (Fagaceae). Amer. J. Bot. 72: 1962-1977. ABBE. 1984. Inflorescence architecture and evolution in the Fagaceae. J. Arnold Arbor. 65: 375-401 L. В. ABBE. 1986. Reproductive phenology ofthe oak family (Fagaceae) in the low- and rain forests of Borneo. Biotropica 18 (in press). REGAL, P. J. 1982. Pollination by wind and animals: ecology of geographic patterns. Annual Rev. Ecol. Syst. 13: 497-524. SOEPADMO, E. 1972. Fagaceae. Fl. Males. I, 7: 265- 403 WHITEHEAD, D. R. 1983. Wind pollination: some ecological and evolutionary perspectives. Pp. 97- 108 in L. а Pollination Biology. Academic inflorescence structure: how flower num rrangem and phenology affect pol- Кы. pee нн pres J. Bot. 69: 585-594. FLORAL STRUCTURE, SYSTEMATICS, AND PHYLOGENY IN TROCHODENDRALES! PETER K. ENDRESS? ABSTRACT Study of the floral structure a Cercidiphyllum, Euptelea, Tetracentron, and Trochodendron, at anthesis and in earlier developm ean revealed new features of macrosystematic interest. They s in о and valvate anther dehiscence, similar sal c rpel bulges in Trochodendron and Tetracentron. 3 [o шаг] m [^7 m 3 B о © E e O ч а 1 © e = о Q O 8 a E E о LZ homogeneity of the four genera, which can be seen as S The position of this order is uil qu between the Magnoliidae and Hamamelidae so that the formal inclusion of the order in either group can be justified. The four unusual eastern Asiatic genera Cer- cidiphyllum, Euptelea, Tetracentron, and Troch- odendron, with only six species among them are systematically isolated from other dicots, but they orm a homogeneous group. This is more or less generally accepted today. In recent classifications they appear in the same order Trochodendrales or at least in the same subclass Hamamelidae. This was not always the case, and their taxo- nomic history sheds some light on their system- atic and phylogenetic significance. The outstanding combination of characters such as vesselless wood, lack of a perianth, and a special kind of triaperturate pollen make them intermediate between the Magnoliidae and the Rosidae-Hamamelidae. Therefore, the views on the systematic position of j poy have ‘nell lated between affinities. Siebold and Zuccarini (1839) described the first genus Trochodendron and placed it in the Win- teraneae. Oliver (1889) described the last of the four genera, Tetracentron, and put all of them in the tribe Trochodendreae of the Magnoliaceae. But Baillon (1871) had already proposed that Cercidiphyllum might have affinities with the Hamamelidaceae. Solereder (1899) corroborated this suspicion by extensive morphological work, and Hallier (1901, 1903a, 1903b, 1904) very viv- e idly put forward that Trochodendraceae also are very near the Hamamelidaceae and should even e wung back with I. W. Bailey and his coworkers, who carefully studied all four genera and firmly established the relationship with the Ranales (Magnoliideae) (Bailey & Nast, 1945; Nast & Bailey, 1945, 1946; Swamy & Bai- ley, 1949). Pervukhina (1963) and Endress (1969) were of the same opinion after comparative de- velopmental studies of Trochodendron and Eup- telea, respectively. t the present tendency is to include all four genera in Hamamelidae (Takhtajan, 1964, 1983; Smith, 1972; Cronquist, 1981) (or Hamameli- dales sensu Thorne, 1983); Trochodendron and Tetracentron in Trochodendrales, and Euptelea meli quist, placed the four genera in the Trochodendrales. Every genus is in a separate family or even in a separate order (as in Takhtajan, 1983). Thus, the gross systematic history from the beginning until now shows a change in the em- phasis from magnolialian to hamamelidalian af- finities. Both are somewhat one-sided positions, because there are clearly affinities to both groups, and the new results presented here corroborate this view. However, for purposes of classification ! I thank B. L. Burtt, Royal Botanic Garden, Edinburgh, for кощ жа material of Tetracentron sinense, and U. Jauch, Institut fiir Pflanzenbiologie, University of Z ürich, R. Siegrist, and A. Zuppiger for technical collaboration. I am grateful for valuable comments on the manuscript ida d R. Morin, Rudolf Schmid, and Shirley C. Tuck ? [nstitut für S male Botanik der Universitat Zürich, Zollikerstrasse 107, 8008 Zürich, Switzerland. ANN. Missouni Bor. GARD. 73: 297-324. 1986. 298 it is necessary to place them in only one of the groups, but placement either way can be about equally justified. MATERIALS AND METHODS The following collections, all cultivated ma- terial, were used for this investigation: Cercidiphyllum japonicum Hoffmann et Schultes, male, Endress 6687 (dat. div., 1984); female, Endress 6688 (dat. div., 1984), old Bo- tanical Garden, University of Ziirich. Euptelea polyandra Sieb. et Zucc., Endress 516 (dat. div., 1967-1968), Eidgenóssische For- schungsanstalt für Obst-, Wein- und Gartenbau, Wädenswil (Figs. 3, 32, 40, 44, 58, 59); Endress 6686 (dat. div., 1984), Botanical Garden, Uni- versity of Zürich (all other figures). Tetracentron sinense Oliver (8 June 1976, 10 Aug. 1976), Royal Botanic Garden, Edinburgh. Trochodendron aralioides Sieb. et Zucc., En- dress 6684 (dat. div., 1984), Botanical Garden, University of Zürich. AII material is deposited in Zürich (Z). Figures 32 and 40, where only the generic name is men- tioned, are both based on these species and spec- imens. Flowers at anthesis and in various develop- mental stages were observed in the living state (except Tetracentron). Material fixed in FAA of all four genera was studied with serial microtome sections stained with safranin and astra-blue and with the scanning electron microscope after treatment with the critical-point drying and Au/ Pd-sputtering methods. RESULTS THE FLOWERING SHOOT AT ANTHESIS There is a differentiation in long and short shoots with inflorescences produced only on short shoots, except in Trochodendron. This differen- tiation is most extreme in Cercidiphyllum and Tetracentron, both having short shoots with a single foliage leaf. Trochodendron. An upright cluster of 10-30 yellowish green bisexual flowers (ca. 1 cm diam.) (Fig. 1) with long pedicels appears in late spring while the shoots are expanding. Flowers ob- served in the Botanical Garden at Zürich in 1980 became visible between the bud scales and leaves in early May. On 12 May the flowers were all exposed, on 22 May open nectar was present between the stamens and styles on the ovary sur- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 face, but all anthers were still closed. On 28 May the first stamens were open and contained sticky pollen; dry remains of nectar were still present. During the period when they were fully exposed, the open flowers enlarged from about 0.8 to 1.2 cm diam. Since mature fruits regularly developed from that plant, the flowers seem to be bisexual (or the development may be apomictic). The flowers eem to be proterogynous. There is no sign of androdioecism in contrast to the observation of Keng (1959) on plants in Taiwan. The floral char- acter syndrome points to myophily (or unspe- cialized entomophily). Keng (1959) observed bees and butterflies as visitors of the sweet-scented flowers in mid-April 1957 in Taiwan. Tetracentron. About a hundred very small yellowish bisexual flowers (2 mm diam.) (Fig. 2) are sessile on a long stiff axis. The studied ma- terial, cultivated in Edinburgh, flowered in late summer (August 1976). A secretory surface that apparently also secretes nectar is at the same site as in Trochodendron. Presumably the flowers are jan 1969; Cronquist $). A cluster of 6-12 (Smith, 1946) pendent inconspicuous bisexual pedicelled flow- ers with many stamens and carpels (Fig. 3) is produced by each flowering shoot. In the closed state the long anthers are red, but brownish when open. Pollen is dry and powdery. The minute carpels expose the white stigmatic papillae be- tween the filament bases. No nectar is produced. In contrast to Trochodendron, anthesis is short and takes place in early spring (March or April at Zurich) before the young leaves expand. Thus, Euptelea exhibits an anemophilous character syndrome. Euptelea polyandra is presumably at least partly self-incompatible. Two neighboring specimens cultivated in the Botanical Garden at Zürich regularly set copious fruit, but other plants cultivated in isolation produce few or no fruits. Cercidiphyllum. This is the only genus of Trochodendrales with unisexual flowers (in dioe- cious distribution). The inconspicuous female inflorescences consist of two to seven (Spong- berg, 1979) unicarpellate flowers with big red stigmas (Fig. 4a). The likewise inconspicuous male inflorescences produce a cluster of about 25-32 (up to 40 according to Spongberg, 1979) long pink anthers on slender, pendulous fila- ments (Fig. 4b). The flower number is difficult to determine (see section on floral ontogeny). The opening anthers turn brownish and present co- 1986] ENDRESS— TROCHODENDRALES FIGURES 1-4. Flowers at anthesis.—1. Trochodendron aralioides, nectary stippled.—2. Tetracentron si- nense. —3. Euptelea polyandra. —4. Cercidiphyllum japonicum: a, female flower, stigma stippled; b, entire male inflorescence. pious powdery dry pollen. Nectar is lacking. As in Euptelea, anthesis is short in early spring be- fore shoot expansion (March or April at Zürich). The anemophily syndrome is obvious. PERIANTH One of the oddities of the Trochodendrales is that the flowers lack a perianth. The only excep- tion is Tetracentron with a perianth of four tepals. This was the main reason why Tetracentron has been, at times, removed from the other three genera, such as by Harms (1897), who put Tet- racentron in Magnoliaceae but the other genera in Trochodendraceae. However, the situation is more complicated. Developmental studies have allowed new in- sights. For a better lerstanding ofthe perianth an analysis of the entire inflorescence is neces- Tetracentron. In morphological terms the in- florescence is a spike (at times with a terminal flower, according to Nast & Bailey, 1945, but not present in my material). Each flower sits in the axil of a subtending bract (7 pherophyll in the terminology of Briggs & Johnson, 1979). The four tepals cover the inner floral organs in bud. They are not arranged in a whorl, but in two alternate pairs. The first pair has a transversal position similar to that ofthe two floral prophylls in many dicots (Fig. 5). In Tetracentron, there are no prophylls outside the tepals, but the two outer tepals correspond to prophylls in respect of their position. In correlation with their di- and thin texture, the tepals often lack well-dif- ferentiated vascular bundles (cf. also Nast & Bai- ley, 1945). An interesting detail hi i small, but radially elongated thin-walled epider- +1 300 mal cells at the inner base of the tepals (Fig. 6). Presumably they play a role in the opening movement of the tepals. Trochodendron. The inflorescence is a ra- сете, usually with a terminal flower (“botryoid” in the terminology of Troll, 1964; see also Briggs & Johnson, 1979). In rich inflorescences the low- ermost lateral branch may bear one or more sec- ondary lateral flowers; they are, then, incipient panicles. The lateral flowers of the raceme have two tiny prophylls in a transversal position (Figs. 7-9, 11-13); they are somewhat larger on the lowermost lateral flower (Figs. 14-16). Between the uppermost lateral flower and the terminal flower there are a few scales (Wagner, 1903; Nast & Bailey, 1945; Melville, 1969; Cronquist, 1981) (Fig. 17). They are serially homologous to the subtending bracts (pherophylls) of the lateral flowers on the main inflorescence axis. This is a feature that often occurs in botryoids and has nothing to do with a perianth (*“Zwischenblát- ter,” Troll, 1964; “metaxyphylls,” Briggs & Johnson, 1979). Thus, there are no obvious tepals, as recorded in the literature. However, in studying young stages of lateral flowers, I found very small scales between the two transverse floral prophylls and the androe- cium (Figs. 9, 10). That they are different from the **metaxyphylls" below the terminal flowers and more intimately connected with the flower is shown by the following features: (1) they are smaller and nearer to the androecium than the metaxyphylls, (2) there is a temporal gap between the early appearing two lateral prophylls (how- ever reduced they are) and the additional scales, and (3) if the scales would correspond to the metaxyphylls one would expect the highest num- ber in the lowermost lateral flower (which has, at times, secondary lateral flowers). In the inflo- rescence shown in Figures 14-19 the reverse is the case: the lowermost lateral flower (Fig. 18) does not even have a single scale on the abaxial side, whereas an upper lateral flower (Fig. 19) shows a scale. In the material examined, the number (and apparently also the position) of these scales var- ies from zero to five. In contrast to the tepals of Tetracentron, these scales do not protect the flo- ral organs in any phase of ontogeny. The pro- tective function i is provided strictly by the cata- phy lls andy shoot. In flowers at anthesis these scales are ob- scured by the general thickening of the floral base (Fig. 7). ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Euptelea. The inflorescences are also ra- cemes, but the main axis of the raceme trans- forms into a vegetative shoot after production of the reproductive part (proliferation, Troll, 1964; auxotelic raceme, Briggs & Johnson, 1979). The flowers have no perianth (Fig. 22), but the low- ermost flowers of a raceme often have one or two tiny transverse prophylls (Endress, 1969) (Figs. 20, 21). After production of the prophylls, the floral primordia of Euptelea and Trochodendron look similar in their peculiar zygomorphic or bi- lateral symmetry (see section on floral ontogeny). However, later, the floral axis elongates above the prophylls in Euptelea (Fig. 20), but below in Trochodendron (Fig. 12). Therefore, at anthesis, the prophylls are at the base of the pedicel in Euptelea, but on the top in Trochodendron. Cercidiphyllum. Both perianth and floral prophylls are lacking. The inflorescences are con- gested spikes (heads). General. How do these new results influence the morphological and phylogenetic interpreta- tion of the perianth in the Trochodendrales? The generally held view is that Tetracentron has a perianth of ida tepals, but the other genera have naked flow But if we an instead, *How many phyllomes are between the base of the floral axis and the androecium?” the result is very different. All gen- era except Cercidiphyllum (Euptelea at least in some ofthe flowers) have a first pair oftransverse phyllomes that can be called prophylls. Tetra- centron has two more phyllomes, Trochodendron up to five more. In both genera they are small and not or only weakly vascularized (even in Tetracentron, see also Nast & Bailey, 1945). Con- sidered this way, there is no sharp contrast be- tween Tetracentron and Trochodendron. There is rather a gradual reduction from Tetracentron to Trochodendron and Euptelea to Cercidiphyl- lum (Fig. 23). Whether these rudimentary phyllomes in Trochodendron should be regarded as tepals or as bracts remains somewhat ambiguous. We can state that there are phyllomes between prophylls and stamens, which precede the series of floral organs (sporophylls). At this evolutionary level of flowers exhibiting spiral phyllotaxis and lack- ing differentiation in calyx and corolla, it is al- ways difficult to delimit a perianth from the pre- ceding series of bracts (e.g, Endress, 1980a, 1980b, for Austrobaileya and Hortonia; Endress & Sampson, 1983, for Trimeniaceae). However, the long plastochron after formation of the two 1986] ENDRESS— TROCHODENDRALES 301 . Old flow wer bud with pherophyll (below) and four tepals, mall-celled inner epidermis, х 140. — 1 flower from bud (February) from abaxial si — e same, enlarged, showing the t ИВ and five rudimentary “tepals” (arrows!), x 50.— 10. Lateral flower bud with the pedicel removed, showing one rudimentary tepal on adaxial side, x50. zl Young inflorescence (July); in lower part floral pherophylls removed to show floral primordia, x 65.— 12. Floral primordium from adaxial aie with two (distal) prophylls, x 140.—13. Same, from above, to show the floral zygomorphy, х 130. 302 [Vor. 73 ANNALS OF THE MISSOURI BOTANICAL GARDEN / ч] f ! ^1 a: к= ESOS E ame, from the opposite side, о е young inflorescence (August). — 14. Entire ея from the x25.—15.S ely large prophylls, x 55. — 17. Terminal flower, after removal M all lateral me as Fig. 16, enlarged, tepal rudiments lacking, two large prophylls, FIGURES 14-19. side; on right side vegetative bu wermost lateral welt with us metaxyphylls" (cf. page 300).—18. — 19. Lateral flower from middle region, with smaller prophylls and one rudimentary tepal (arrow!), x 90. 1986] FIGURES 20-22. Euptelea MN E — 20. Floral AE (July) from abaxial side, with one (basal) You 0.—21. Younger flor tepals, x45. prophylls may be seen as a sign that the subse- quently formed scales are to some extent inte- grated into the floral organization. In contrast, within the Hamamelidaceae there are obvious reduction series from flowers with well-differ- entiated protecting sepals in addition to petals amamelis) through tas with reduced с opsis) to реп- anthless flowers (e.g., Distyliopsis) a 1977, 1978) I conclude, based on this, that it is highly im- probable that the Trochodendrales are primi- tively completely devoid of a perianth. ANDROECIUM Tetracentron and Trochodendron. The an- droecium is composed of four stamens in Tetra- centron, but of a variable number of 39 to 46 for lateral flowers and around 70 for terminal flowers in Trochodendron (Liao, 1973: 34-57). In Tetra- centron stamens are much smaller and stouter than in Trochodendron. Tetracentron Trochodendron al primordium (July) from above, to show the bilateral (or zygomorphic) .—22. Base of old floral bud (December), rim around stamens by thickening of floral base, no A newly found feature of systematic impor- tance in both Trochodendron and Tetracentron is that each theca does not open simply by a longitudinal slit but by two valves (Figs. 24-28, 32). The longitudinal dehiscence line extends into two transversal lines at both ends, which results in two window wing-like valves. This feature underlines at the same time the cl between Tetracentron and Trochodendron and their affinities with Hamamelidales (Fig. 32), es- pecially since this feature is otherwise very rare in the angiosperms. Although **valvate" anthers have been mentioned in the literature for various higher groups of the angiosperms, it seems that most taxa actually have simple longitudinal slits without transverse extensions. At present, real valvate anthers seem to be restricted to a few Magnoliidae (Laurales) and Hamamelidae (Fig. 32), whereby those of the Laurales are of a slight- ly different type. However, the search has to be extended to other angiosperms. Further similar- ity with Hamamelidales is provided by the ba- Euptelea ө Se Cercidiphyllum Floral diagrams showing floral pherophylls (6:00 position), prophylls (3:00 and 9:00), and д Кш шы (stippled) and gynoecium (black) not differentiated. Cercidiphyllum: diagram for female flow only. 304 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 SW TUO A FIGURES 24-31. 24-26. ia aralioides, ш anthers.— 24. From the side, the two valves of a theca, x 35.— 25. At opening, from connec ctive tip, short lateral extensions of longitudinal slits, x 85.— 26. Same anther, lateral к Ор оп i mal end of slit, x 85. 27, 28. Tetracentron sinense, open flower, anthers still closed. —27. Flower from above, showing dehiscence lines of anthers on distal side, x 35.— 28. Dehiscence lines on proximal side of anther, from adaxial side, x 55. 29, 30. E era polyandra, undehisced anther, showing dehiscence lines.—29. Distal part, x110.—30. Proximal part, x110.—31. Cercidiphyllum ja- ponicum, open anther from the side, with simple longitudinal slit, x 10. 1986] ENDRESS— TROC ALES 305 Entomophilous Anemophilous Trochodendron Tetracentron Euptelea Cercidiphyllum 1mm | Corylopsis Corylopsis spicata platypetala Sycopsis Parrotia FIGURE 32. Similarity and parallel pee of anther shape, size and dehiscence in and anemophily in Trochodendral dH melidaceae. All stamens from the side, adaxial ub left. (Cor- ylopsis spicata, C. platypetala, Sycopsis sinensis, vat Parrotia persica, all from cultivated material, old Botanical arden, University of Züric h). 306 sifixed, more or less latrorse anthers with a ru- dimentary connective tip (Fig. 32). Euptelea and Cercidiphyllum. In both g era the stamen number is variable: Euptelea with six to 19 per flower (Endress, 1969) and Cerci- diphyllum with at least seven in the lowermost flowers. As part of the anemophily syndrome, the anthers are long and contain more pollen than in Tetracentron and Trochodendron. In Cer- fa yllum they open by simple longitudinal slits (Fi , 32). However, in Euptelea, there are es hon horizontal extensions of the dehis- cence line, mainly at its lower end, which results in two narrow valves on each theca (Figs. 29, 30, 32). As in Trochodendron and Tetracentron, the anthers are basifixed, more or less latrorse, and phyllum than in Trochodendron and Tetracen- tron (Fig. 32), presumably as a result of the early longitudinal growth of the entire anther. General. It is striking that the range of anther shape in the four genera of Trochodendrales is repeated in Hamamelidaceae (Fig. 32). Also in Hamamelidaceae, long anthers with mass pro- duction of pollen and dehiscence by longitudinal slits occur in predominantly wind-pollinated groups, whereas short anthers with less pollen and dehiscence by valves, on which the sticky pollen is presented, are typical for the mainly insect (often fly!)-pollinated groups (Endress, 1977). POLLEN In all four genera the pollen grains are triap- erturate and more or less spheroidal, but in Eup- telea also often pluriaperturate (Praglowski, 1974; Walker, 1976a, 1976b). The apertures are well delintited and coarsely granulate. Trochodendron and Tetracentron. In pollen structure the two genera are again strikingly sim- ilar, also in details not previously mentioned in the literature. Both are tricolpate. The tectum consists of distinct rods that are irregularly in- terwoven and form a loose network between the apertures but are crowded into parallel bundles along the aperture borders (Figs. 33, 34, 37). In short, the exine is rugulate-reticulate between the apertures but striate near the apertures in both genera. In both genera, especially in Tetracen- tron, the pollen is very small (10-15 um diam.). On the whole, the similarity in exine structure between the two genera is even more accentuated ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 than stated by Praglowski (1974) and Walker (1976a, 1976b). Euptelea and Cercidiphyllum. In both gen- era the exine is finely reticulate, with minute tec- tal perforations, and scabrate (with supratectal verrucae). The apertures are shorter, more roundish in outline than in Trochodendron and Tetracentron (Figs. 35, 36, 38, 39). In both Eup- telea and Cercidiphyllum, the apertural exine is coarsely structured. Euptelea pollen often has more than three apertures, in contrast to Cerci- diphyllum. In both genera the pollen is consid- erably larger (20-30 um diam.) than in Trocho- dendron and Tetracentron and fits in the normal range of wind-dispersed pollen (cf. Whitehead, 69 — No Again, the parallel between the pollen of Eup- telea and Cercidiphyllum and the anemophilous genera of the Hamamelidaceae is striking in the trend to shorter apertures and more than three apertures, finer reticulation of the extra-apertural exine, and supratectal verrucae (Endress, 1977; Bogle & Philbrick, 1980). GYNOECIUM AND NECTARY Trochodendron. Leinfellner (1969b) gave a brief survey ofthe gynoecial morphology of Mag- noliales, Laurales, and Trochodendrales. The elaborate diagram of a gynoecium of Trocho- dendron in Nast and Bailey (1945) does not show all the basic gross morphological features. The extension of the inner and outer surface in some critical regions remains unclear. In Trochodendron the carpel number of the gynoecium varies around eight in lateral flowers and around ten in terminal flowers. Smith (1945) found a range of (4—)6-11, Liao (1973) a range of 7-11, and Cronquist (1981) a range of 4-11 (217). In the stylar region the carpels are free and slightly recurved (Figs. 1, 40, 41а-с, 45). In the ovary, the carpel flanks are congenitally fused (Fig. 41e-i). Above the ovary there is a short zone where the neighboring, contiguous carpels are postgenitally fused (Fig. 41d). The ovary is partially inferior (Fig. 40). In the upper part of the ovary a compitum is formed by postgenital fusion of the carpels in the center, while in the lower portion of the ovary the carpels do not meet in the center, but form a ring around a central hole (Figs. 40, 41e-g). The compitum, however, is rather diffuse, because the ventral slits that are lined with a pollen tube transmitting epidermis are very long in horizontal extension. 1986] ENDRESS—TROC ALES 307 FIGURES 33-39. 33-36. Pollen grains.—33. Trochodendron aralioides, х 2,700.— 34. Tetracentron sinense, x 3,000.—35. Euptelea polyandra, x1,700.— 36. Cercidiphyllum japonicum, х 2,000. 37-39. Details of pollen surface. — 37. Tetracentron sinense, same as Fig. 34, x6,000.—38. Euptelea polyandra, x 6,000.— 39. Cercidi- phyllum japonicum, х 9,000. 308 Cercidiphyllum ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Euptelea Trochodendron FIGURE 40. Median longitudinal sections of carpels, with ventral side on the left (Cercidiphyllum, Euptelea), and gynoecia (Trochodendron, Tetracentron) at anthesis, with outer and inner morphological surfaces (stippled, region of postgenital fusion; black, nectariferous region) The compitum is, therefore, not as elaborated as the centralized types in the higher advanced an- giosperm groups (Carr & Carr, 1961). Again, the similarity with the Hamamelidaceae is obvious (cf. Endress, 1967). The stigmatic epidermis differentiates unicel- lular papillae of a dry type (category DPU in the classification of Heslop-Harrison, 1981) (Fig. 46). Ontogenetically, the hole in the center of the gynoecium base arises because the fairly high number of carpels is arranged more or less in a whorl and therefore leaves a free field in the cen- ter of the floral apex. Gynoecial development is slow and retarded compared with that of the an- droecium. In mid-September, when the outer stamens were morphologically already well dif- ferentiated, the carpels were still shallow, hip- pocrepiform mounds (Figs. 56, 57). In mid-Feb- ruary, the postgenital fusion within and between the carpels had not yet started. The carpels have a conspicuous dorsal bulge. This bulge is differentiated as a nectary (see Tieghem, 1900!) (Figs. 1, 40, 45, 50) and is vas- cularized by numerous phloic strands that are connected with the dorsal carpellary bundles (Nast & Bailey, 1945; Pervukhina, 1962) (Fig. 1986] 41d, e). It forms an open landing platform ex- posed to visitors. The epidermis of the nectary is covered with numerous slightly sunken sto- mata. Each stoma is surrounded by a ring of several epidermal cells with conspicuous cutic- ular folds (Figs. 52, 54). All or part of the 15-30 ovules (15-20, my data; 15-28, Liao, 1973; 25-30, Cronquist, 1981) in each carpel develop into many small dust-like seeds, suited for wind-dispersal. They are re- leased from capsules that open by ventral and short dorsal slits. The mature ovules have long integuments and a chalazal protrusion, which both contribute to seed appendages. In the cha- lazal appendage the ovular vascular bundle forms a hair-pin loop (Nast & Bailey, 1945; Mohana Rao, 1983). The testa differentiates five cell lay- ers, the middle layer being sclerified (Melikian, 1973). Tetracentron. The gynoecium of Tetracen- tron although strikingly similar to that of Troch- odendron (Fig. 40), differs in that it is much smaller and contains only four carpels (Figs. 42c, 51), each with five or six ovules. The very similar differentiation of a nectary should be especially emphasized, since it has not been reported pre- viously, although its presence was mentioned by Cronquist (1981). The numerous stomata on the dorsal carpellary bulges are also slightly sunken below the general surface (by the smaller size of the guard cells compared with the other epider- mis cells) and surrounded by a ring of epidermal cells with prominent cuticular folds (Figs. 53, 55). Striking are also the facts that in both genera the stigma is presumably dry and has unicellular papillae (Fig. 47), that the ovary is syncarpous and semi-inferior (Fig. 40), that the ovary locules are filled with copious secretion around the ovules, and that dust-like seeds (testa five-lay- ered with middle layer sclerified, Melikian, 1973) with a chalazal and micropylar appendage are released from small capsules. The ovules with the long integuments and chalazal protrusion with a hair-pin-like vascular bundle are very similar (Nast & Bailey, 1945) in Tetracentron and Tro- chodendron. Also in both genera the carpels have five main vascular bundles, three dorsal ones and one lateral one on each side (Fig. 42). Trochod- endron has more secondary lateral veins appro- priate to its larger size. In Tetracentron the nec- tary does not have a separate vascular supply, probably due to the much smaller size of the gynoecium (Fig. 42b, c). Cercidiphyllum. The female flower consists ENDRESS—TROCHODENDRALES 309 of nothing more than a slightly stipitate single el on a rudimentary floral axis (Fig. 4a) (So- lereder, 1899). The carpel is plicate throughout (in the terminology of Leinfellner, 1950) without any ascidiate basal portion (Fig. 43). Leinfellner’s (1969b) observation of a slight indication of an ascidiate base in Cercidiphyllum could not be confirmed with my material (Fig. 43k). The ovary contains about 17-24 lateral ovules in two rows (17-20, my data; 20-24, Harms, 1916). The ovary is relatively small at the time of pollination and the ovules immature, and the stigma occupies about a third of the length of the carpel (Fig. 4a); it is extended over almost the entire circumfer- ence of the carpel apically and tapers towards the base (Figs. 4a, 43a, b). The stigmatic surface bears unicellular papillae and is of a dry type (Fig. 49). The carpel has three main vascular bundles (Fig. 43), a median and two laterals with a network of secondary lateral bundles in between, which are only weakly differentiated at anthesis. The seeds have a conspicuous micropylar and a chalazal appendage with a vascular bundle de- scribing a hair-pin loop. However, the integu- ments protrude less over the micropyle than in Trochodendron and Tetracentron (Swamy & Bai- ley, 1949). Euptelea. The gynoecium consists of about eight to 31 free carpels (Smith, 1946; Leinfellner, 1969a; Endress, 1969). The carpels are conspic- uously stipitate, but lack a style. The ovary is strongly ascidiate (Fig. 44) and contains 1-3(-4) lateral ovules that are, as in Cercidiphyllum, im- mature at anthesis. The stigmatic surface sits im- mediately on the ovary; it bears long, unicellular papillae and is of a dry type (Fig. 48). In the carpel three vascular bundles differentiate (Fig. 44), a (dorsal) median and two laterals. The two laterals fuse in the ascidiate region in proximal direction and the resulting ventral median bun- dle fuses with the dorsal median in the stipe (Fig. 44; Leinfellner 1969a). The carpel shape of Eup- telea is very distinctive and resembles certain agnoliidae, such as Winteraceae, Schisandra- ceae, and Ranunculaceae, much more than any Hamamelidae. FLORAL ONTOGENY, PHYLLOTAXIS, AND ORAL SYMMETRY Floral symmetry and phyllotaxis are intimate- ly correlated, at least in early developmental stages. Examples were recently shown by Tucker 1984). It seems that there is also a connection between the early floral symmetry and the ar- 310 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURE 41а-е. Trochodendron aralioides. Series oft ti i t anthesis (dashed lines, regions of postgenital fusion; bius k, xylem i in vascular bundles). –а. ое region. — b. Upper stylar region. — c. Lower stylar region. —d. nsition region between style and ovary (stippled, nectariferous tissue). — e. Up pper symplicate region of ovary w rg the carpellary flanks meet in the center; note numerous peripheral phloic шкан supplying nectaries FIGURE 41f-i. Trochodendron aralioides. Series of t d lin regions of postgenital fusion; black, xylem in vascular bundles).—f. Middle isse d region of Ovary vtl 1986] ENDRESS—TROC ALES 311 minimal extension of postgenitally des region of carpellary flanks meeting in the center кошш —g. Low symplicate region of ovary with carpellary flanks retracted from the center, leaving a hole; at the perip n tamen bases and stamen traces eu —h. ee region between symplicate and ө н region (only center of ovary drawn).—i. Synascidiate region of o 312 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 FIGURES 42-44. Transverse sections of a flower (Tetracentron, Cercidiphyllum) or of a carpel (E uptelea) at anthesis mens spi dian of "postgenital fusion; black, xylem in vascular bundles). — 42. Tetracentron sinense, abaxial side below (stipp a, stylar region; b. transition region between styles and ovary; Lx с, transition ar pen apocarpous and. symp plica te region; d, symplicate region, the two median stamens used with floral base, le two lateral stamens (partially) free; e, transition region between symplicate and on. А (in h-k): a, b, upper (а) and lower (b) stigmatic Bes (wavy outline, receptive rini c, middle stylar region; d, transition region between style an dieti ovary; f, ovary below placentar region; g, ovary base; h, transition region between ovary and carpella , mas sive carpellary base, ventral slit Dy dashed line) still pen k, carpellary base united with floral, и on = ae lowermost extension of ventral slit still present. 44. Euptelea polyandra, ventral side of carpel facing upwards: a, b, plicate apex of carpel; c, upper ascidiate peni of carpel; d, lower end of stigma, lateral bundles fus a hen ventral median bundle; e, ovary with lateral ovule; f, lower end of ovary locule; g, h, carpellary base, ventral and dorsal median bundle fusing; i, carpellary base partially united with floral base 1986] ENDRESS— TROC dupla 45—49.—45. Trochodendron aralioides. Gynoecium at early anthesis, x 10. 46-49. Stigmatic surfaces at anthesis. — 46. Trochodendron aralioides, x1 x 30. 49. Cercidiphyllum japonicum, x 130. T of the entire shoot bud (Endress, 1969), w is obvious in Trochodendron, Euptelea, she yas di ula Trochodendron. The young flowers have been shown to pass through a zygomorphic phase with lateral flowers of the inflorescence (Fig. 56), and the terminal flower is radial throughout its on- togeny (Fig. 57). The androecium has usually been described as being spirally arranged. This seems to be true from the earliest ontogenetic stages and it can be extended to the gynoecium. This is even implied eight carpels, which is a Fibonacci number. As in Ийсит (Robertson & Tucker, 1979), the broad residual floral apex the frequent 20.— 47. Tetracentron sinense, x170.—48. Euptelea polyandra, favors a secondary more or less whorl-like ar- rangement of the spiral carpels. It is interesting that the spiral arrangement is also present in the lateral flowers. Therefore, it can be supposed that in the youngest stages when the stamens and car- pels are initiated, the floral apex is (at least) not (strongly) zygomorphic. This problem and the exact phyllotaxis have to be studied by detailed ontogenetic investigations. Tetracentron. The flower is described as hav- ing three tetramerous whorls, with the stamens opposite the tepals, but the carpels alternating with the stamens (Bailey & Nast, 1945). How- ever, the flowers are more exactly described as dimerous in the perianth and androecium with four alternate pairs of organs, but tetramerous in the gynoecium (Fig. 2). The two tepal pairs are distinct by their aestivation, the two stamen pairs by the position of the anthers. Only the two me- 314 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 LIE 50-55. 50, 51. Gynoecium at anthesis, from above, Pone nectariferous dorsal bulges of the 1s.— 50. Trochodendron aralioides, sunken stomata visible as black dots, x 25.— 51. Tetracentron sinense, i ан removed, nectaries covered with secretion, x 30. 52, 53. Sunken stoma of nectary, encircled by several epidermal cells. — 52. Trochodendron aralioides, x 600.— 53. Tetracentron sinense, х 960. 54, 55. Longit al section of nectariferous surface with sunken stomata.— 54. Trochodendron aralioides, x 350. —55. Tetracentron x 350. sinense, dian stamens are contiguous in bud, while the two lateral ones are more remote. Further, the lateral tepals and stamens are attached at a lower level than the median ones, as seen from trans- verse section series (Fig. 42d). Therefore, the flowers are not exactly radial, but slightly bilat- erally symmetric. Unfortunately, the early floral ontogeny is still unknown in Tetracentron. 1986] Euptelea. The flower has a radial symmetry at first glance. However, the floral base is broader in transverse than in median direction, and is, therefore, bilaterally symmetric as in Tetracen- tron. Interestingly, floral primordia show a sim- ilar pattern of zygomorphy as in Trochodendron. However, it is just reversed in that mostly the abaxial side is retarded (Figs. 58, 59). Cercidiphyllum. The female flowers are nec- essarily “zygomorphic,” since they consist of a single carpel, which is directed towards the sub- tending bract (pherophyll of the flower). Leroy (1980) ascribed a special morphological signifi- cance to this position. However, it is the position that would be expected for the first phyllome of a lateral shoot with an adaxial prophyll in the same way as it occurs in the vegetative region of Cercidiphyllum. Ontogeny shows that not only the flowers but also the entire inflorescence is zygomorphic (Fig. 60). The inflorescence pri- mordium is directed towards the adaxial side. The flowers appear in one or two decussate pairs The first (transversal) pair of subtending bracts (pherophylls of the flowers) is not exactly op- posite, but both are positioned somewhat to- wards the adaxial side. The second (median) pair of pherophylls is exactly opposite, since it is sit- uated in the symmetry plane of the inflorescence (Fig. 60). One or both pherophylls of the second pair may be lacking, but its ““carpel” is present (Fig. 61). Other authors have found up to three (to four) pairs of flowers (Swamy & Bailey, 1949; Spongberg, alius Yan Heel (pers. comm.) ob- serveda tary floral apex between вя um and the pherophyll in another speci- The male flowers of Cercidiphyllum are diffi- cult to delimit. The inflorescence contains about 25-32 stamens (Harms, 1916: 16-35) (Figs. 68, 69). Since the symmetry of the young inflores- cence and the position of the bracts in the floral region is the same as for the female, a similar position for the flowers has to be expected: two more or less opposite flowers or several flowers in decussate pairs (Figs. 62-67). However, it is difficult to find floral boundaries. All stamens have the thecae in a lateral position relative to the inflorescence axis and all have collateral bun- dles or at least the xylem more towards the center of the inflorescence axis (Figs. 68, 69). From this one can judge that all turn their ventral side to- wards the inflorescence axis. According to this interpretation of the inflorescence, all flowers would, then, be highly zygomorphic, with sta- ENDRESS—TROC ALES 315 mens only on the abaxial side of the flower. In early ontogeny the median stamen of each flower would be the largest, the lateral ones successively retarded (Figs. 62, 65, 67). Thereby, the lower- most flower on the adaxial side of the inflores- cence axis (frontal view in Fig. 65) would contain the highest stamen number, whereas the lower- most flower on the abaxial side (frontal view in Fig. 64), the lowermost lateral flowers (frontal view in Figs. 66, 67), and the more apical flowers of the inflorescence would contain fewer sta- mens. Another tentative, but perhaps less convinc- ing, interpretation would be that the first pair of individual flowers consists of the row of about seven stamens adjoining each of the two lateral bracts. Each bract of the second (i.e., the median) pair (if present at all) has only about one to three stamens in its axil. The following stamens replace bracts in some sense and form a terminal flower (as in some Hamamelidaceae, cf. Endress, 1978; Wisniewski & Bogle, 1982). The residual apex on top of the inflorescence would be the apex of the inflorescence in the first interpretation, but the apex of the terminal flower in the second (Fig. 68a). More information could be expected if the search is extended to other specimens. Swamy and Bailey (1949) depicted a male inflorescence that is easier to interpret since groups of several stamens are situated in the axils of two successive pairs of bracts. Here all flowers seem to consist of eight to 13 stamens. SOME HISTOLOGICAL FLORAL FEATURES OF SYSTEMATIC INTEREST In Tetracentron all floral organs contain scat- tered enlarged cells with the contents dissolved in the fixed and sectioned material (Fig. 70). They much resemble the so-called “oil” cells that are generally present in many Magnoliidae, but ab- sent in Hamamelidae. They are often more or less spherical as in Magnoliidae, apparently in contrast to the vegetative region where they have been described as elongate or branching by Bailey and Nast (1945), or termed “Sekret- schláuche” by Harms (1897). In ере асе), branched idioblasts. However, in the gynoecium, at least at anthesis, many of these cells are not sclerified and ег also cor- respond to “oil” cells (Fig. In addition, the tissues d the anthetic floral organs of both Tetracentron and Trochodendron are heavily tanniferous (Figs. 54, 55, 70, 71), but 316 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Ficures 56-61. 56, 57. Trochodendron aralioides. Flowers of a youn (August) from above. — 56. Lateral zygomorphic flower, x 55.—57. Terminal radially symmetric л х40. 58, 59. Euptelea p (July). — 58. Young inflorescence from above, floral pherophylls removed to show floral primordia, x Same, enlarged; one zygomorphic flower with stamen primordia (larger on adaxial side), x 110. 60, 61. i diphyllum japonicum. Young female inflorescence (July) with three unicarpellate flowers, the carpel clefts oriented towards their pherophylls. – 60. From above, x120.—61. From the abaxial side, median carpel (flower) without pherophyll, x110 1986] ENDRESS—TROC ALES 317 FIGURES 62-67. Cercidiphyllum japonicum. Young male inflorescences (July).—62. Very young stage, from er ik and first stamen primordia visible, abaxial side towards top, x 140. 63-67. Somewhat older stage, viewed from different кук —63. From above, abaxial side towards top, x90.—64. From the abaxial side, x 85.—65. From a adaxial side, x 90.—66. From lateral (right side of Fig. 63), x 80.—67. From lateral (left side of Fig. 63), x9 318 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 68, 69. Cercidiphyllum japonicum. Young male inflorescences, series of transverse sections. — 68 Inflorescence with pherophylls on adaxial joining vascular stele of inflorescence (arrows). — 69. Inflorescence without pherophylls on median side: and abaxial sides: a, level with uppermost stamens oming free ; с, vascular arrow); d, vascular traces of lateral pherophylls a, leve with uppermost stamens becoming free from floral base; b, level of free lateral pherophylls; c, vascular traces of lateral pherophylls joining vascular stele of inflorescence (arrows). less so in Euptelea and Cercidiphyllum. A high tannin content is also typical for the flowers of Hamamelidales and many Magnoliales. CONCLUSIONS CIRCUMSCRIPTION OF TROCHODENDRALES From our present knowledge it is evident that Trochodendrales sensu lato (including Trocho- endron, Tetracentron, Euptelea, and Cercidi- phyllum) are clearly related and form a coherent group because in none of the four genera can a relative be found in another order that would be more close than any of the three other genera. It is true that the extremes are relatively far apart. There is a relatively large step in floral structure between Tetracentron and Cercidiphyllum. How- ever, in vegetative morphology the two genera resemble each other so much that the relation- ship is instantaneously evident. Furthermore, the meristic variation in floral organs is a typical constitutive feature of primitive angiosperms. The range in floral structure within the four gen- era is of about the same level as within the Ham- amelidales sensu stricto (Hamamelidaceae, Pla- tanaceae, possibly Myrothamnaceae). Therefore, 1986] ENDRESS—TROCHODENDRALES 319 FIGURES 70, 71. “oil” cells present in all organs, x55 x 140. the four genera are formally best treated as an order Trochodendrales (like Dahlgren, 1980, but not CLASSIFICATION OF THE TROCHODENDRALES Of the four extant genera, Trochodendron and Tetracentron are the two most closely related. The present investigation shows that they are much more similar than has been pointed out earlier. This is reflected in Table 1. The most distinctive resemblances include: occurrence of a perianth (although rudimentary in Trocho- dendron), markedly valvate anther dehiscence, presence of a nectary on the carpellary dorsal bulges containing many sunken stomata sur- rounded by a ring of epidermal cells with heavily sculptured cuticle (all new characters); addition- ally, the very similar and distinctive pollen, ova- ries and ovules, fruits and seeds, and also the similar vesselless wood and similar stomata in the vegetative body (Bondeson, 1952; Baranowa, 1983 All this strongly points to the inclusion of both genera, Trochodendron and Tetracentron, in the same family Trochodendraceae (as suggested earlier by Gundersen, 1950, or—before the in- flation of taxonomic group numbers on all hier- archic levels—by Hallier, 1903b: Trocho- dendreae of Hamamelidaceae). The remaining two genera of the order, Euptelea and Cercidi- phyllum, may remain in separate families. Therefore, for the moment, a fairly balanced classification of the group would be as follows: Trochodendrales Trochodendraceae Trochodendron Tetracentron “Oil” cells in floral tissue. — 70. Tetracentron sinense; longitudinal section of old floral bud .—71. Trochodendron aralioides; transverse section of ovary at anthesis, Eupteleaceae Euptelea Cercidiphyllaceae Cercidiphyllum FOSSIL RECORD OF THE TROCHODENDRALES Today, all four genera of the Trochodendrales are restricted to more or less small areas in tem- perate or subtropical Eastern Asia from Nepal (Tetracentron) to Taiwan (Trochodendron) and Japan (Cercidiphyllum, Euptelea, Trochoden- ron). Cercidiphyllum was widely distributed in the Northern Hemisphere in the Tertiary (newer finds and reviews, e.g., Brown, 1962; Hummel, 1971; Becker, 1973; Chandrasekharan, 1974; Iljinska- ja, 1974; Jáhnichen et al., 1980; Scott & Wheeler, 1982; Basinger & Dilcher, 1983; Hickey et al., 1983; Stockey & Crane, 1983) and perhaps back to the Upper Cretaceous (Maestrichtian) (Hickey et al., 1983). In the Paleocene other genera were affiliated with Cercidiphyllum, such as Joffrea (Crane & Stockey, 1985) and with less certainty Jenkinsella (Chandler, 1964; Crane, 1978). Re- tallack and Dilcher (1981) even pointed to sim- ilarities between early ee and the mid-Cretaceous genus Prisca. Fossils of the three а genera сап Бе iden- tified with less certainty. Fossil leaves have been compared (with doubts) with Trochodendron and Tetracentron as far back as to the mid-Creta- ceous (Iljinskaja, 1972, 1974). Tetracentron-like wood has been described from the Upper Cre- taceous (Page, 1968). Wood ascribed to Euptelea was found in the early Tertiary of North Amer- ica (survey in Wolfe, 1973). mbined presence of leaves, fruits, and seeds of Cercidiphyllum in various fossil beds 320 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE 1. Character states occurring in more than one but not in all genera of the Trochodendrales to show their relationships. Tetracentron Trochodendron Euptelea Cercidiphyllum entomophilous nectaries on carpel dorsal surface short anthers, short connective tips anther dehiscence markedly valvate anemophilous nectaries lacking slightly valvate pollen 10-15 wm diam. exine striate-rugulate apertures (colpi) long ovules mature at anthesis apertures short carpels sessile carpels stipitate gynoecium syncarpous ovary semi-inferior ovary superior seed coat 5-layered, middle layer sclerified phyllomes between prophylls and stamens present (tepals) floral prophylls regularly present “oil” cells present in floral organs “oil” cells lacking wood without vessels wood with vessels long anthers, long connective tips anther dehiscence longitudinal or exine finely reticulate and scabrate ovules immature at anthesis gynoecium apocarpous or unicarpellate seed coat multilayered phyllomes between prophylls and stamens lacking floral prophylls lacking or present in basal flowers only . Б weakly peltate = carpels strongly peltate pollen triaperturate pollen pluriaperturate fruits dehiscent fruits indehiscent seeds with appendages seeds without appendages | carpels epeltate | pollen triaperturate fruits dehiscent seeds with appendages flowers bisexual leaves alternate flowers unisexual leaves opposite stamens 4 stamens numerous flowers enclosed flowers enclosed in bud carpels 4 by cataphylls in bud flowers sessile stipules present 1-leaved short shoots present carpels numerous flowers pedicelled stipules absent 1-leaved short shoots lacking | carpel 1 flowers sessile | stipules present 1-leaved short shoots present leaves deciduous | | leaves evergreen | | leaves deciduous 1986] ENDRESS— TROC ALES 321 TABLE2. Character states of the Trochodendrales shared with certain Magnoliidae or Hamamelidae to show their relationships. Some of the character states occur in both subclasses but are more im portant on the side where they are mentioned. The initials of the genera of Trochodendrales are mentioned in brackets if a feature does not occur in the other ones (C, Cercidiphyllum; E, Euptelea; Te, Tetracentron; Tr, Trochodendron). Magnoliidae Trochodendrales Hamamelidae carpel number with wide range carpel stipitate, style lacking (E) nectary on carpel dorsal surface (Te, Tr) stigma with unicellular papillae wood vesselless (Te, Tr) **oil" cells present (Te, Tr) single adaxial prophyll (C) chloranthoid leaf teeth (Te, Tr, C)! plants more or less glabrous lack of myricetin? filaments long anther dehiscence valvate (Te, Tr, E) pollen tri- or pluriaperturate with marked apertural exine structure ovary semi-inferior (Te, Tr) ovules immature at anthesis (E, C) fruits dehiscing, seeds edged and winged (Te, Tr, C) presence of an unelaborated compitum in syncarpous taxa (Te, Tr) stipules present (Te, C) ! Hickey and Wolfe (1975). ? Kubitzki and Reznik (1966). facilitates the identification of the material. That the fruits and seeds of Cercidiphyllum are better suited to fossil preservation than those of the other three genera is probably due to their more robust texture and flat shape limiting deforma- tion. All these many finds, especially those of Cer- cidiphyllum and extinct genera related to it with an obviously much wider distribution and great- er diversification in the early Tertiary corrobo- rate the impression that the Trochodendrales are a relic group with a long history, and are now in the state approaching extinction. Present modest diversity on the generic level (Euptelea and Cer- cidiphyllum with two species each) points to rel- atively recent differentiations. POSITION OF THE TROCHODENDRALES BETWEEN THE MAGNOLIIDAE AND HAMAMELIDAE The Trochodendrales are closely related to Magnoliales and at the same time to Hamamel- idales. The Trochodendrales are intermediate 1 4L AA KIA AD i | /На amelidae (Table 2): Some have retained vessel- less wood and “oil” cells, but they have already acquired tricolpate pollen and valvate anthers of a hamamelidalian type. The odd female flowers of Cercidiphyllum consisting of a single carpel all rather in the range of Magnoliidae than Ham- amelidae. Tas distance 19 the соге Magnoliales and I t the same for all four genera, but for each from different angles. Ad- ditional embryological features not mentioned in the text, sieve tube plastid differentiation or chromosome numbers do not contribute muc to this special question because they are too uni- form in the critical groups under consideration here (Yakovlev, 1981; Ly-thi-Ba, 1981; Behnke, 1981) or too diverse (Ratter & Milne, 1973, 1976). The perianth may be seen as a marker trait. In the Magnoliales the perianth is not yet differ- entiated into typical sepals and petals (cf. Hiep- ko, 1965). In the Trochodendrales the perianth is reduced, whereas in the Rosales/Hamameli- dales it is often differentiated into sepals and petals, sometimes also reduced (at least partly 322 from a double perianth, cf. Endress, 1977; see also Ehrendorfer, 1977). It seems reasonable that the Trochodendrales evolved from an ancestral group whic a perianth, but not yet differentiated into sepals and petals, as in Magnoliales. The Hamameli- dales, in contrast, originated from a group where a well-differentiated perianth with sepals and petals already occurred, Е жш 1sepal БЕ lost in man whole, the hee шы are not m to the Hamamelidales. They are a con- servative, isolated group. However, they have common roots with Hamamelidales and have retained more magnolialian traits than have the Hamamelidales. Therefore, the formal inclusion of Trocho- dendrales in either Magnoliidae or Hamameli- dae can be justified. An inclusion in the Hama- melidae seems reasonable (if this subclass is retained as such at all) because it is a smaller subclass than Magnoliidae. The urgent need now is a detailed comparative study of all families of Magnoliidae and Ham- amelidae, and not a premature phenetic or cla- distic classification solely from the available in- formation in the literature. Our ultimate goal is better knowledge of the living plants and better understanding of their phylogeny. WIL LITERATURE CITED ВАШУ, I. W. & С. С. Nast. 1945. Morphology and relationships of Trochodendron and Tetracentron. . Stem, root and leaf. J. Arnold Arbor. 26: 143- 153. a H. 1871. Nouvelles notes sur les Hama- ées. Adansonia 10: 120-137. . On the laterocytic stomatotype 102. R,J.F.& 1983. Fruits of Cercidiphyllum from the early Tertiary of Elles- Island, arctic Canada. Amer. J. Bot. 70(5, 1973. The York nee flora of the U х by River Basin, southwestern Montana. PCR Abt. B, M 143: 18- BECKER, H. BEHNKE, H.-D. 1981. Sieve-element characters. bwai Bot. 1: 381—400. ВосгЕ, А. L. & C. T. PHILBRICK. 1980. A generic atlas 95 hamamelidaceous pollens. Contr. Gra Herb. 210: 29-103. BONDESON, W. 1952. pa eo und Bau der Spaltóffnungen bei den Trocho- dendron Sieb. et Zucc., Tetracentron Oliv. und Drimys J.R. et G. Forst. "Acta Horti Berg. 16: 169- 217. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 BRN B. G. & L. A. S. Јонмѕом. 1979. Evolution the Myrtaceae—evidence from inflorescence тыш Proc. Linn. Soc. New South Wales 102: 157-272. Brown, R. W. 1962. Paleocene flora of the Rocky Mountains and Great Plains. Profess. Pap. U.S. Geol. Surv. 375. Carr, S. G. M. & D. J. CARR. 1961. The functional significance of syncarpy. Phytomorphology 11: GARDAR. M. E.J. 1964. The Lower соо ian out Summary and Surv a servations. British. Museum (Natural History), London. CHANDRASEKHARAN, А. 1974. Megafossil flora from the Genesee locality, Alberta, Canada. Palaeon- tographica, Abt. B, Paláophytol. 147: 1-41. CRANE, P. R. 1978. Angiosperm leaves from the low- er Tertiary of Southern England. Cour Forsch.- Inst. Senckenberg 30: 126-136. & R. A. Stockey. 1985. Growth and repro- ductive biology of Joffrea speirsii gen. et sp. nov., a Cercidiphyllum-like plant from the Late Paleo- cene of Alberta, Canada. Canad. J. Bot. 63: 340- 364. po en A. 1981. An Integrated System of Clas- sification of Flowering Plants. Columbia Univ. Press, geh York. DAHLGREN, R. 1980. A revised system of classifica- tion of the angiosperms. Bot. J. Linn. Soc. 80: 91- 124. 1983. General aspects of angiosperm evo- lution and macrosystematics. Nordic J. Bot. 3 119-149. EHRENDORFER New ideas about the early differentiation of angiosperms. Pl. Syst. Evol. Suppl. 1: 227-234, ENDRESS, P. K. 67. Systematische Studie über die verwandtschaftlichen Beziehungen zwischen den Hamamelidaceen und Betulaceen. Bot. Jahrb. Syst. 87: 431- 1969. Gesichtspunkte zur systematischen Stellung der Eupteleaceen (Magnoliales). Ber. Schweiz. = Ges. 79: 229-278. 1977. Evolutionary trends in the Hamamel- idales-Fagales-group. Pl. Syst. Evol. Suppl. 1: 321- 347. Blütenontogenese, Blütenabgrenzung und systematische Stellung der perianthlosen Hamamelidoideae. Bot. Jahrb. Syst. 100: 249-317. 80a. The reproductive structures and sys- tematic position of the Austrobaileyaceae. Bot. Jahrb. Syst. 101: 393-433. Floral structure and relationships of Hortonia (Monimiaceae). Pl. Syst. Evol. 133: 199— 221. & F. B. SAMPSON. 1983. Floral structure and relationships of the Trimeniaceae (Laurales). J. Arbor. 64: 447-473. Families of Dicotyledons. lyphyletischen Ursprung der Sympetalen and 1986] Apetalen und die Anordnung der Angiospermen ber Abh. Geb. Naturwiss. Hamburg 16. Über die Verwandtschaftsverhált- nisse bei Engler' s Rosalen, Parietalen, Myrtifloren und in anderen Ordnungen der Dikotylen. Abh. Geb. Naturwiss. Hamburg 18. 1 г den Umfang, die Gliederung und die Verwandtschaft der Familie der Hama- се Beih. Bot. Centralbl. 14: 247-260. 1904. Uber die оя Daphniphyllum, ein Uberga ngsglied von den Magnoliaceen und Ha- mandi zu den Kátzchenblüthlern. Bot. Mag. e 18: 55-69. Eo die Stellung der Gattung Tet- milie der Trochodendraceen. 60. 916. e Blütenverháltnisse und die systematische Siling der Gattung Cercidiphyl- lum Sieb. et Zucc. Ber. Deutsch. Bot. Ges. 34: 272- 283. i pcd yond ct Y. 1981. Stigma characteristics angiosperm taxonomy. Nordic J. Bot. 1: 401- Hickey, L. J. & J. A. WoLFE. 1975 [1976]. The 2 of angiosperm phylogeny: vegetative morphology. Ann. ppm gs Gard. 62: 538-589. , M. R. Dawson & D. K. hes 1983. yeh eal biota: paleomagnetic e dence of age disparity with mid-Northern жөр during the late Cretaceous and early Tertiary. Sci- : 1153-1156. 65. Vergleichend-morphologische und entwicklungsgeschichtliche Untersuchungen ü das Perianth bei den Polycarpicae. Bot. Jahrb. Syst. . The genus Cercidiphyllum in the Tertiary flora of Poland and of the neighbouring regions. Roczn. Sekc. Dendrol. Polsk. Towarz. Bot. 25: 63-75. ILJINSKAJA, I. А. 1972. Correction of the volume of the genus Trochodendroides and new fossil species Zurn. (Moscow & Leningrad) 57: 1974. T Trochodendrales, Cercidiphyllales. Pp. 1 12-128 in A. Takhtajan (editor), oe | Fossilia URSS I. Nauka, Leningrad. [In Russian.] JAHNICHEN, H., D. H. Mar & H. WALTHER. 1980. Blatter und Friichte von Cercidiphyllum Siebold & Zuccarini im mitteleuropáischen Tertiár. Schrif- tenreihe Geol. Wiss. 16: 357-399. KenG, Н. 1959. Androdioecism in the flowers of Trochodendron aralioides. J. Arnold Arbor. 40: 158-160. Kusitzki, К. & Н. REZNIK. 1966. Flavonoid-Muster der Polycarpicae als systematisches Merkmal. I. Übersicht über die Familien. Beitr. Biol. Pflanzen 42: 445-470. LEINFELLNER, W. 1950. Der Bauplan des synkarpen zeums. Oesterr. Bot. Z. 97: 403-436. —. 1969a. Uber die Karpelle verschiedener Magnoliales. VII. Euptelea (Eupteleaceae). Oes- terr. Bot. Z. 116: 159-166. . 1969b. r die Karpelle verschiedener Magnoliales. VIII. "Überblick über alle Familien der Ordnung. Oesterr. Bot. Z. 117: 107-127. ENDRESS— TROC ALES 323 Leroy, J.-F. 1980. Développement et organogenése chez le Cercidiphyllum japonicum: un cas sem- génie des Ranales et de a nales arborescentes а, Rev. Gén. Bot. 88: 43-82. MELIKIAN, A. P Seed-coat types of Hama- melidaceae and allied iiie in relation to their systematics. Bot Zurn cow & Leningrad) 58: 350-359. (In Russian.] MELVILLE, R. . Studies in floral structure and evolution. I. The Magnoliales. Kew Bull. 23: 133- 180. MOHANA Rao, Р. В. 1983. Seed and fruit anatomy of Trochodendron aralioides. Phytomorphology 31: Nast, С. G. & 1. W. ВАпЕҮ. 1945. Morphology and relationships of Trochodendron and Tetracentron. II. Inflorescence, flower, and fruit. J. Arnold Ar- bor, 26: 267-276 Morphology of Euptelea and comparison with Trochodendron. J. Arnold Arbor. : 192. 1889. Tetracentron sinense Oliv. Hook- 1892. 1968. Angiosperm wood from the upper Cretaceous of Central California: part II. Amer. J. ot. 55: 168-172. PERVUKHINA, N. V. 1962. Interesting peculiarity of the ovary of Trochodendron aralioides. Bot. Zurn. pec & Leningrad) 47: 993-995. [In Russian.] 196 i On the position of dades dei. t in Fen & Leningrad) 48: 939- 948. е те si ] OFFE. 1962. The morphology of Trochodendron: flower (A RN to the phy- logeny of a rn. (Moscow & ningrad) 47. 1709-1730. [In oa 1974. The pollen morphology of the onomy. Pollen & Spores 16: 449-467 RATTER, J. A. & C. MILNE. 1973. Chromosome num- bers of some primitive angiosperms. Notes Roy. Bot. Gard. Edinburgh 32: 423-428. & Chromosome counts in primitive angiosperms: II. Notes Roy. Bot. Gard. Edinburgh 35: 143-145. RETALLACK, G. & D. L. DILCHER 81. Early angio- a © B LESS © > deposits in Kansas, U. SA. ранган СА. Abt. B, Paláo- spain 179: 103-137. ROBERTSON, R $. C. TUCKER. 1979. Floral on- togeny of Illicium floridan num, with emphasis on stamen and carpe | development. Amer. J. Bot. 66: 2617 Scorr, R. A. & E. A. WHEELER. 1982. Fossil woods from the Eocene Clarno formation of Oregon. In- ternatl. Assoc. Wood Anat. Bull., n.s. 3: 135-154. 324 SIEBOLD, P. Е. & J. С. ZUCCARINI. 1839. Flora Ja- nomic review of Trocho- dendron and As J. Arnold Arbor. 26: 123-142. ———, 1946. A taxonomic review of Euptelea. J. Arnold Arbor. 27: 215-226. 1972. An appraisal of the orders and families of primitive extant angiosperms. J. Indian Bot. Soc. 50A: 215-226 SOLEREDER, H. Zur Morphologie und Syste- matik der Gattung Cercidiphyllum Sieb. et Zucc., mit Berücksichtigung der Gattung Eucommia Oliv. Ber. Deutsch. Bot. Ges. 17: 387-406. . 1979. Cercidiphyllaceae hardy in temperate North America. J. Arnold Arbor. 60: 367-376. STOCKEY, К. A. & P. К. CRANE. 1983. In situ Cer- cidiphyllum- like seedlings from the Paleocene of rta, Canada. Amer. J. Bot. 70: dera Swamy, B. G. L. & I. W. BAiLEv. 19 mor- phology and relationships of rn J. Arnold Arbor. 30: 187-210. TAKHTAJAN, A. 64. The taxa of the ecd plants dido the rank of order. Taxon 13: 164. 1969. Flowering Plants; Origin d Dispers- al. ‘Oliver & Boyd, Edinburgh. 3. The systematic arrangement of dicot- yledonous iode Pp. 180-201 in C. R. Metcalfe & L. Chalk (editors), Anatomy of the Dicotyle- dons, a ume 2, 2nd edition. Clarendon Press, THORNE, R. Е, 1983. Proposed new realignments іп the angiosperms. Nordic J. Bot. 3: 85-117. TIEGHEM, P. УАМ. 1900. Sur les Dicotylédones du ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 oupe des Homoxylées. J. Bot. (Morot) 14: 259- 361. . 1964. Die Infloreszenzen. Typologie und Stellung im Aufbau des Vegetationskórpers. 1. Fischer, Stuttgart. TUCKER, SHIRLEY C. 1984. Origin of symmetry in flowers. Pp. 351-395 in R. a ru W Dickison (editors), d Plant Eu Academ Rr dim Or lan WAGNER, R Beiträgez zur jede ue der сае Frochodendron Sieb. et Zucc. Ann. Naturhist. Hofm 409- 422. WALKER, Y Comparative pollen mor- 9 in C. B. Beck (editor), Origin and Early а ч Angiosperms. Columbia Univ. Press, Yor 197 6b. Sener significance à the exine in the pollen of primitive angiosperm Ferguson & J. Muller (editors), The а Significance of the Exine. Linn. n . Symp. Ser. 1: 251-308. Academic Press, Lon WHITEHEAD, D. R. 1969. Wind bailas in the angiosperms: evolutionary and environmental considerations. Evolution 23: 28-35. WisNIEWSKI, M. & A. L. BoGLE. 1982. The ontogeny of the inflorescence and flower of Liguidambar styraciflua L. (Hamamelidaceae). Amer. J. Bot. 69: 1612-1624. Wo rt, J. A. 1973. Fossil forms of Amentiferae. Brit- tonia 25: 334-355, YAKOVLEV, M. S. (editor). 1981. Comparative Em- bryology of Flowering Plants. Winteraceae— Ju- glandaceae. Nauka, Leningrad. [In Russian.] THE FLORAL MORPHOLOGY AND VASCULAR ANATOMY OF THE HAMAMELIDACEAE: SUBFAMILY LIQUIDAMBAROIDEAE' A. LINN BOGLE? ABSTRACT florescence and floral morphology and floral e anatomy of the subfamily Liquidambaroideae In (Liquidambar L., Altingia Nor.) are described. The i partially deve - functionally pistillate flowers a cycle of sterile phyllomes of uncertai y condensed bes?) is inserted on the hypanthium between the stamens and carpels of the partly inferior ovary. b Lines flowers the stamens derive a single trace and the phyllomes a ramifying system of bundles m hypanthial trunk bundles. Each carpel contains a dorsal and two ventral bundles. Numerous margin receive а е as the most primitive within the Нат from the ventral bundles. The gynoecium is interpreted melidaceae, and subfamilial status of Liquidambaroideae, rather «bon segregation as the family iia is supported. There has been an increase in interest and dis- cussion in recent years concerning the nature and origins of the *Amentiferae" —a group now ac- cepted by a majority of modern systematists as being made up of highly advanced and special- ized taxa —and how they relate to more primitive groups of angiosperms. Central to this problem are the nature and position of the family Ham- amelidaceae, which occupies an intermediate po- sition in the various theories and phylogenetic schemes that have been put forth. However, any considerations of the phylogenetic position of the Hamamelidaceae, and of the trends of special- ization both within and without the family, must begin with a reasonably clear understanding of which members of the family are primitive, or most nearly so, and which are advanced. At pres- ent, opinion is divided on this point and on cir- cumscription of the family. This is due to the remarkable morphological diversity evident among the 28-30 genera that are now known to comprise the five distinct subfamilies tradition- ally included in the family. Such diversity is par- ticularly evident among the seven genera that make up the four small subfamilies Liquidam- Bucklandioideae or Symingtonioideae authors, including Chunioideae of Takhtajan, 1980, and Mytilarioideae of H. T. Chang, 1973), and Disanthoideae. Each of these subfamilies, and the Liquidam- baroideae in particular, is characterized by in- teresting combinations of primitive and ad- vanced morphological and anatomical features that have led to varying interpretations of their Thus, A/tingia and Liquidambar, comprising the subfamily Liquidambaroideae, have been re- moved by some workers (e.g., Blume, 1828; < 1 Scientific Contribution Number 1370 from the New Hampshire Agricultural E i ion. This paper m the University o D. dissertation submitted to esearch Fellow ard University, and as a . The author is к for ds e support, either direct or indirect, provided by these institutions, and particularly to Ernst C. Abbe, Reed C. i ial ai fNe Rollins, Richard A. еек. 1, Lawrence С. W. Jensen roject no. 216, and the Central University Research Fund, Grant no. Mi is gratefully acknowledged. Field colecions , Ly Ў made оп ту behalf by Ernst all three I peat , Shiu-Ying Hu, H ohs, Rob Demaree are also gratefully acknowledged. Т thank Margaret Bogle, Garrett Crow, James or reading and editing the manuscript; A. most particularly, Shiu-Ying Hu for help with the Chinese of the Arnold Arboretum, Baranov for help with have ui shared ү һап 2 Botany and Plant Pathology Department, Univereity ‘of New Hampshire, Durham, New Hampshire 03824. ANN. MISSOURI Bor. GARD. 73: 325-347. 1986. 326 Hayne, 1830; Lindley, 1836; Wilson, 1905; C. T. Chang, 1959, 1964; Melikian, 1971, 1973a, 1973b; P. R. M. Rao, 1974; T. A. Rao & Bhupal, 1974; Skvortsova, 1960) from the Hamameli- daceae as a пие family Altingiaceae. Others üdambar and Altingia to be relatively арштын taxa within the Ham- amelidaceae (e.g., Makarova, 1957; Schmitt, 1965; Meeuse, 1975), whereas such recent au- thors as Schulze-Menz (1964), Cronquist (1968), and Takhtajan (1969) considered these genera as relatively advanced members of the Hamamel- idaceae. Similarly, Willis (1966) and Wolfe (1973) segregated Rhodoleia as a family Rhodoleiaceae, principally on the basis of leaf architecture. Cronquist (1968), while maintaining the Ham- amelidaceae in the broad sense, stated that “the monotypic genus Disanthus has at least as much claim to family status as the Altingiaceae." The most extreme position is that of Nakai (1943), who went so far as to elevate all four subfamilies to family status, thus restricting the Hamameli- daceae to the genera of subfamily Hamamelidoi- deae. With regard to what constitutes the most prim- itive floral type within the family, Takhtajan (1969) stated that Disanthus cercidifolius Max- im., the sole member of the subfamily Disan- thoideae, “15 the most primitive in floral struc- ture." Cronquist (1968) considered Disanthus as "clearly the most primitive" genus in the family, or (Cronquist, 1981) that it “may be the most archaic surviving genus." This view of the prim- itive nature of Disanthus is concurred in by Wolfe (1973) on the basis of leaf architecture (appar- ently assuming the exclusion of Rhodoleia from the Hamamelidaceae), and by Goldblatt and En- uae е on evidence from chromosome cou Co ANM there is still relatively little ba- sic information available in the literature from comparative morphological and anatomical studies of a comprehensive nature on these rel- atively primitive members of the family. More data are needed before suppositions concerning primitive or advanced conditions, and the mor- phological limits of the Hamamelidaceae, can be made with any assurance and employed in con- siderations of extrafamilial relationships. This paper is intended to present an account of the floral morphology and vascular anatomy of the two genera comprising the subfamily Liq- uidambaroideae (Liquidambar, Altingia) based on original anatomical work and morphological ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 observations, as part of an effort to provide com- parable data for all genera of the family. Some observations concerning the little known segre- gate genus Semiliquidambar are also presented. MATERIALS AND METHODS Inflorescences of Liquidambar and Altingia in various stages of development have been col- lected from both wild and cultivated trees in southeast Asia, and North America either by me, or for me by others. In addition, inflorescences from herbarium specimens in the Arnold Ar- boretum (A)! have also been examined, including The spec- imens examined, collection data, and location of voucher specimens are listed in Table 1. The methods of ү preparation and staining of sections, and clearing of floral parts are standard and have been e previously (Bogle, 1970). I have examined both sectioned and cleared specimens of staminate and pistillate flowers and inflorescences of Liquidambar and Altingia in anthesis and post anthesis stages. Drawings were made with the aid of the draw- ing attachments designed for the Wild M-20 and M-5 microscopes. I am grateful to i н and curators of the d herbaria for permitting me to use their living ине herbarium collections, and for supplying material for the study: Arnold Arboretum of Harvard University; Gray Her- barium of Harvard University; Morris Arbore- tum of the University of Pennsylvania; Scott Foundation of Swarthmore College; University of Washington Arboretum, Seattle; and the Roy- al Botanic Gardens, Kew, Great Britain. RESULTS In the following paragraphs, the descriptions of the vasculature of the functionally pistillate flowers (hereafter referred to simply as pistillate flowers) are based primarily on serial sections, whereas those of the functionally staminate flow- ers (hereafter referred to simply as staminate flowers) are based primarily on cleared material. Emphasis is given here to the functionally pis- tillate flowers because these are, in fact, mor- phologically and sometimes functionally perfect and more useful for comparison with flowers of the other subfamilies. The staminate flowers, in contrast, appear to represent the culmination of a line of reduction from the perfect condition within the subfamily. 1986] TABLE 1. BOGLE—LIQUIDAMBAROIDEAE 327 Collections on which anatomical and morphological observations presented here are directly based (A = Arnold Arboretum, ALB = author’s collection). Taxon Collector Locality Deposited Altingia chinensis (Champ.) Bogle 583 Hong Kong Botanical Garden ALB Bogle 588 Hong Kong, forest above Peel Rise, off Aberdeen ALB Road Bogle 591 Hong Kong, forest above Peel Rise, off Aberdeen ALB Road Hu 6700 Hong Kon ALB Pételot 5944 Tonkin, DO 1,500 m A Altingia chingii Metc. Lau 4356 China, Kiangsi A (paratype) Tso 20760 China, N. Kwangtung, Lok Chong, Yao Shan A Altingia excelsa Nor. Bogle 313 Malaya, Pahang, Cameron Highlands, Mentigi ALB Forest Reserve Abbe 10298 Malaya, Pahang, Cameron Highlands, Mentigi ALB Forest Reserv Liquidambar formosana Bogle 610 Taiwan, Yin-Ping, between Chitow and Chushan ALB Hance Liquidambar orientalis Bogle 973 USA, Minnesota, Minneapolis, cultivated speci- ALB Mill n Univ. of Minnesota greenhouse Liquidambar styraciflua L. Bogle 790 USA, и Montgomery Co., Willow ALB Grove, Red Barn Roa Bogle 823 Mexico, Hidalgo. vic. El Barrio Road, km 322, S ALB of Tamazunchale SUBFAMILY LIQUIDAMBAROIDEAE Two (or three) genera of large trees make up this distinctive subfamily. The most well known of these, and the most widespread in both living and fossil form, is Liquidambar L., which is char- acterized by its palmately-lobed and -veined, de- ciduous leaves. Liquidambar is disjunctly dis- tributed in eastern Asia, Asia Minor, and in North and Central America. Less well known is Altingia Nor., which differs from Liquidambar in its simple, elliptic to ob- long, pinnately-veined, evergreen leaves. Altin- gia is widely distributed in southeast Asia. The segregate genus Semiliquidambar H. T. Chang (including Altingia chingii Metc.) was es- tablished (H. T. Chang, 1962, 1973) to include specimens whose leaf morphology is interme- diate between that of Altingia and Liquidambar. These species are described as having leaf blades that vary from undivided to 2- or 3-lobed, tri- plinervous venation, and with pistillate flowers lacking staminodia, among other morphological features. I have seen specimens of Altingia chin- gii in which the leaves appear to bridge the gap in leaf shape between Altingia and Liquidambar (Tso 20760, paratype, A; Lau 4356, A; see also Metcalf, 1931, pl. 58; and description of A. chin- gii var. parvifolia in Chun, 1934). These genera are in need of monographic revision to better define the number of species and clarify the va- lidity of Semiliquidambar. I have not had access to floral material of Semiliquidambar suitable for detailed morphological and anatomical work. Liquidambar and Altingia share several mor- phological and anatomical characteristics, in- cluding: gum ducts associated with the vascular bundles of the stems, leaves, and floral organs; formation of terminal buds enclosed within nu- merous bud scales; spirally arranged stipulate plex, bisexual inflorescences не питег- ous many-flowered heads; flowers naked, perfect, or imperfect, and пша. pistillate or sta- minate; a cycle of sterile organs surrounding the ovary in pistillate flowers; pollen grains poly- porate (forate); ovules numerous, inserted on both carpel margins, but only the lowermost one or two on each margin fertile; and seeds winged. This combination of characteristics sets the 328 subfamily quite apart from the rest of the family, but specific characteristics (e.g., gum ducts, com- plex inflorescences, sterile floral organs, poly- porate pollen grains) do occur in certain genera of the other subfamilies, so are not unique to the Liquidambaroideae. For descriptive purposes I shall use the generalized term phyllome (cf. Esau, 1965: 424) in reference to the cycle of sterile organs as they appear in Liquidambar and Altin- gia, because they are not clearly either stami- nodial or carpellary in nature. 1. Liquidambar L., Sp. Pl. 2: 999. 1753. Figures ІА, 2, 3, 4А-Е, 6A-D, I, J The genus Liquidambar consists of two to five species of large deciduous trees distributed in the Northern Hemisphere. Liquidambar formosana Hance is widespread in eastern Asia (Tardieu- Blot, 1965; Li, 1977; H. T. Chang, 1962, 1973). Liquidambar acalycina occurs in at least nine provinces of China (H. T. Chang, 1973, 1979). Liquidambar orientalis Mill. is known from Tur- key and some islands of the Aegean Sea such as Rhodes (Rechinger, 1943) and Cyprus (Holm- boe, 1914; Meikle, 1977, as L. styraciflua). Liq- uidambar styraciflua L. is widely distributed in eastern and southeastern North America and southward at high elevations on the mountains of Mexico and Central America to Honduras (Little, 1971, map 135-E, 135-N). The Mexican and Central American plants are recognized by some authors as a distinct species, L. macro- phyllum Oerst. (Sosa, 1978). The three principal species (L. formosana, L. orientalis, L. styra- ciflua) are also widely cultivated as ornamental trees and form the basis for my observations. The stipulate leaves of all three species are palmately-lobed and -veined, but differ in the degree of lobation and pubescence. The leaves of Liquidambar formosana are usually 3-lobed, those of L. styraciflua are usually 5(-7)-lobed (see Duncan, 1959, for a detailed analysis of leaf form in L. styraciflua), whereas those of L. orientalis are also 5(-7)-lobed, but with the lobes further subdivided. Leaf form in the last two species is intergrading, however, and some authors con- sider them conspecific (Rechinger, 1943; Meikle, 1977). Harms (1930, see also H. T. Chang, 1979) rec- ognized two sections within the genus. Section Cathayambar Harms, containing only Liquid- nosana (Fig. 2A), is distinguished pri- h L шоч JOTT ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 marily by the presence in the pistillate inflores- cences of elongate, fleshy, subulate, pubescent setae (““borsten””) of varying length, which he in- terpreted as being “between the female flowers,” whereas in section Euliquidambar Harms, con- taining L. orientalis and L. styraciflua de 3A), such setae are said to be lacking (H. T. Chang, 1979, also placed L. acalycina in this section). Morphology. The flowers of Li are naked and usually functionally unisexual. They are fused in small, spherical to elongate heads that are spirally arranged in a narrowly conical, complex inflorescence. One to three long-pedun- culate, pistillate heads occupy the basal nodes of the inflorescence (Fig. 1A), followed distally by numerous (ca. 15-20) short-pedunculate to ses- sile staminate heads (Figs. 1A, 4A, staminate portion of the inflorescence, bracts, and stamens removed to show axes). In some inflorescences a pistillate head may be lacking. The staminate and pistillate flowers and heads are usually dis- tinct in form and function, but exceptional and intergrading forms are common. Some examples of such forms are discussed below in a brief con- sideration ofthe inflorescence following the mor- phological descriptions of typical pistillate and staminate flowers. Pistillate flower. The functionally pistillate owers are naked but perfect. Six to eight basal or sub-basal, elongate, hyaline bracts subtend the head and initially enclose it but quickly drop off after anthesis. I could find no morphological or anatomical evidence of either a calyx or corolla in any of the primordial or mature flowers ex- amined. The androecium of the pistillate flower con- sists of a cycle of five to eight (four to ten in Harms, 1930) stamens inserted at the angles of the floral periphery. The stamens vary in their degree of development from head to head. The anthers may be sessile or borne on short fila- ments and may range in form from strongly re- Bass and sterile (Figs. 2A, 3A) to fully devel- ped anthers t sterile or even re pollen. When fertile, however, the anthers apparently do not dehisce until the stigmas of the same flower are well past their receptive pe- riod, thus preventing self-fertilization in these proterogynous female inflorescences (Schmitt, 5) The gynoecium consists of two (one to six) involute carpels that are fused below but apo- carpous above, forming a syncarpous, half-in- ferior ovary. In the base of the ovary the fused 1986] 2 -— METRIC 1 BOGLE—LIQUIDAMBAROIDEAE 329 IG 1.—A, B. Total inflorescences of Liquidambar styraciflua L. and altingia chinensis (Champ.) Oliv. The single, basal, pistillate head in each inflorescence is indicated by an arrow. All other heads are staminate. carpel margins form a thick but shallow septum (Figs. 2F, G, 3E) that divides a short distance above the base of the locule to form two ap- pressed, compound, parietal placentae (Figs. 2H, Each carpel margin bears two rows of pendent, anatropous ovules. The number of ovules on each margin in my material ranges from ten to 15 in Liquidambar formosana (Bogle 610, Fig. 2D), and ten to 13 in L. styraciflua (Bogle 790, 823, “L. macrophylla”) and L. orientalis (Bogle 97 3). The ovules are arranged in a double row on the central part of each carpel margin, giving way to single ovules above and below (Fig. 2D, see also Griffith, 1836, pl. XVI, as Sedgewickia). The ovules of Liquidambar styraciflua lack an egg apparatus at the time of pollen release (Schmitt & Perry, 1964). In transverse sections the ovary appears bilocular below but is essentially uni- locular above the top of the septum, even thou the tightly appressed placentae give the appear- ance of two locules. Above the septum the tightly appressed carpel margins are not sealed, and the carpels are thus “open” at pollination. Exter- nally, the dorsal surfaces of the carpels become free from the hypanthium, and adaxially from each other, at a level slightly above the top of the septum (Figs. 2J, К, 3G—I). In the material I studied, the styles of L. orientalis and L. sty- raciflua (Fig. ЗА) w were notably shorter and stout- er, and their , than those of L. formosana, but the possibility of regional variation in style-stigma morphology in these species has not been ruled out. Surrounding the base of the apocarpous por- tions of the carpels is a cycle of fleshy, sterile phyllomes (Fig. 2A) that are of particular interest because of the variety of interpretations that have been placed on them by various workers. In Liq- uidambar formosana these take the form of the long setae (““borsten” of Harms, 1930) that char- acterize section Cathayambar. Oliver (1867) de- scribed these structures as “long spines, pro- duced from the calyx-limb . . . not at first sight readily distinguishable from the persistent in- durated styles." Ek (1902) considered them to 330 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ER a, \ | | | | | | / / | Г и half -inferior —À-—T ovary ГА _ о xe 94 \ 1986] be “isolated perianth parts.” Guillaumin (1920) called them “bracteoles” surrounding the female flower, and Harms (1930) suggested that they are vestigial styles of sterile flowers inserted between the fertile flowers. These styliform bodies appear outwardly very much like the functional styles except for their variable size and the absence of stigmatic surfaces (cf. Fig. 2A). The two setae inserted on either side of the ovary in the median plane are usually the largest, perhaps reflecting a slightly greater amount of space for growth and enlargement in the angles between the two car- pels. One of these setae is usually larger than its opposite number, often approaching the styles in length. In flowers of L. orientalis and L. styra- ciflua these sterile structures are reduced to a cycle of short, blunt, glabrous, fleshy lobes sur- rounding the carpels (“phyl” in Fig. 3A) and scribed by Harms (1930). Schmitt (1965) de- scribed these organs in L. styraciflua as “papillae” that “form on the surface of the capsules.” Staminate flower. The naked staminate flowers (Fig. 4) are highly reduced and fused in spirally arranged, congested, bracteate (B,) spikes, heads, or clusters along the primary inflorescence axis. Each head represents a secondary inflores- cence axis (А,), the larger of which axes, in turn, bear secondary bracts (B,) with axillary flower clusters (Fig. 4, anthers removed from heads; A, Liquidambar styraciflua; B-F, L. orientalis). The staminate flowers may be strictly staminate or contain the merest vestiges of rudimentary ас and stigmas of several: ovaries: 43) and 4F, note tips of rudimentary carpels visible within cycle of stamens]. Each flower contains four to eight (to ten) stamens arranged in a single cycle, although the flowers and stamens are often so BOGLE—LIQUIDAMBAROIDEAE 331 densely crowded that it is sometimes difficult to determine this detail of floral structure. 1 found no evidence in L. orientalis to support Ek's (1902) statement that each staminate flower consists of one stamen. There is no evidence in the typical staminate flower of the cycle of sterile phyllomes that is so prominent in the pistillate flower. Near the base of each staminate head distor- tion of the flowers, due to such factors as reduc- tion of the peduncle, fusion of the peduncle to bu primary inflorescence axis, and differential wth, make it difficult, if not impossible, to а individual flowers. In the middle and distal portions of the head, however, it is often apparent that the stamens are arranged around shallow, pubescent, elongate pits or depressions in the surface of the axis [Fig. 4C(3)]. The pu- bescent tips of two abortive carpel primordia are frequently visible within the ring of stamens at either end of the pit. Tong (1930, fig. 10) noted such vestigial gynoecia in male flowers of Liq- uidambar formosana The mature fertile stamens (Figs. 2C, 3C) con- tain four sporangia and dehisce by means of two simple, lateral, longitudinal slits. At anthesis the pistillate heads are pendent on long peduncles, whereas the main axis of the total inflorescence, bearing the staminate heads, is more or less erect. After pollination the entire distal staminate por- tion of the inflorescence (Fig. 4A) quickly withers and drops away. The peduncles of the pistillate heads remain attached and the heads ripen an enlarge several-fold. Most of this enlargement is due to the growth of the numerous ovaries within the head, but part of it is apparently attributable to the activity of a vascular cambium in the axis of the head (Schmitt, 1965). The seeds are shed in the fall. Inflorescence variation. Although staminate and pistillate heads are usually strongly differ- — FIGURE 2. VERBA i formosana ip edd — А. Pistillate flower at anthesis, excised from a head.— B. Floral of o diagram. — C. en from a staminate flower. — D bundles and two compound ventrals visible.— Through lower part rof locules, showing ovule traces departing from co Arrangement Serial sections UE a a pistillate iue at dcus cut slightly obliquely.—E. Thr Bases of locules; compound ventral bundles rgins of one carpel. E-M. ough the receptacle; diio trunk rmed.—G. mpound ventrals.—H. Through ovary; the compound ventral bundles dividing.—I. Through ovary; approaching top of the inferior arise trunk bundles branching to form traces ч stamens, phyllomes, and carpe stamens and phyllomes; procambial strands of phyllome supply stippled ls free of each Set adaxially; hypanthium nearly free. — L. Insertion of remaining stamens and from the ovary; insertion of som rpe here.— oe els.—J. Partial separation of the hypanthium phyllomes. — M. Through the style bases, anthers, and phyllomes. (Key: As, tertiary axis; d, dorsal carpel bundle; lb. учат branch: loc, locule; ov, ovule; phy], n st, stamen, stamen trace; sty, style; tb, trunk bundle; v, ventral carpel bundle; vv, compound ventral bundle 332 F SOM D 22 ESSEN ANNALS OF THE MISSOURI BOTANICAL GARDEN Y) [VoL. 73 a s T----4 +---- 1986] entiated, unusual variations in inflorescence structure have frequently been observed and are also reported in the literature. On several occa- sions I have seen inflorescences of Liquidambar styraciflua and L. orientalis in which the flowers f the lower staminate heads contain partially developed but sterile gynoecia. Various degrees of development of the gynoecium among sta- minate flowers in a single head of L. orientalis are о in Figure 4B, С. For the sake of these drawings the secondary bracts dicated by hatched areas), as have the anthers. In addition, each axillary group of flowers is il- lustrated in diagrammatic form to the left of the drawings. The secondary bracts are numbered in ascending order on the axis. Each stamen is rep- resented in the diagrams by a small circle, a ru- dimentary ovary by an ellipse, and the absence of an ovary by a cross. Figure 4B(1) and 4E il- lustrate a single staminate flower consisting of seven stamens and lacking any external evidence of a vestigial ovary. This flower is seen in lateral view in Figure 4C(1), and diagrammatically in Figure 4E. Figure 4C(3) shows a secondary bract witha single, axillary, жапше flower that con- tains t ded by nine stamens (cf. diagram in Fig. 4F). The tips of the two abortive carpel primordia are apparent at either end of a minute pit in the center of the flower. Figure 4B(5) illustrates an axillary com- plex of seven staminate flowers of which five contain partially developed ovaries and styles, while two contain no ovaries. Although quite crowded on a short, conical axis, the limits of each flower could be ascertained by studying the placement and orientation ofthe stamens. Where two or three flowers abut, the stamens may occur in tufts of three or four. The composition of the BOGLE—LIQUIDAMBAROIDEAE 333 seven flowers in the complex is shown diagram- matically in Figure 4D. The presence within a single staminate head of several secondary bracts with associated axillary floral clusters indicates clearly that each head of the inflorescence is a reduced branch system containing axes of at least two orders (secondary and tertiary) in addition to the floral pedicels and points up the complex- ity of the inflorescence in the subfamily. Inflorescences of Liquidambar styraciflua in which all heads are developed to the pistillate state were reported by Britton (1887) and illus- trated by Kirchheimer (1947). I have seen inflo- rescences in this condition on trees in the Arnold Arboretum and University of Washington Ar- boretum (Seattle), as well as on trees obtained from commercial sources and on native trees in eastern Pennsylvania. Such inflorescences are unusual but not rare and may be considered as concrete evidence of the bisexual potential of the staminate flowers of Liguidambar. Inflorescence variation of this type has also been observed in Altingia. Although functionally pistillate, the female flowers normally contain stamens that are re- duced and sterile. However, I have observed pis- tillate heads of Liquidambar formosana in which the stamens vary in development from tiny, cla- ios neo as in Figure 2A, to fully de- t pollen. The о of a staining reaction with lactophenol- ted that this pollen ported viable pollen in the anthers of pistillate flowers of L. styraciflua in New York and North Carolina, respectively. Floral vascular anatomy. In the following paragraphs the floral vasculature of Liquidambar formosana is first described in detail. The pat- <— FIGURE 3. Liquidambar styraciflua L.— A. Pistillate о i anthesis; one of the basal flowers in a head ral with subtending secondary bract. — B. Floral diagr of a pistillate papi at a aie trunk bundles on ей: зераг Through bases of styles, papillate anthers of sterile stamens. (Key: Ay, am.—C. Sta sis, cut slightly obliquely.—D. Through the receptacle, showing several large ocules; ventral bundles forming. – Е. Through lower part of the ovary, showing branching ation of placentae; traces to ovules departing from the ventral bundles. – С. Through n from a staminate flower. D—J. Serial sections carpel bundle; Ib, lateral branch; loc, locule; ov, ovule; phyl, чое st, stamen, stamen trace; (06 style; tb, trunk bundle; v, ventral carpel bundle.) 334 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 E м FIGURE 4. Staminate partial inflorescences of Liquidambar styraciflua, L. orientalis, and Altingia chinensis. — A. The staminate portion of the total inflorescence of L. styraciflua after anthesis, with all anthers removed, showing central primary axis with primary bract scars, and secondary axes with secondary bract scars. B, C. Portions of secondary axes of the staminate partial inflorescence of L. orientalis, showing various stages of 1986] terns of the other two species are then described to the extent that they differ from that of L. for- osana. Pistillate flower. A typical pistillate flower of Liquidambar formosana, excised from a head, is illustrated in Figure 2A. At anthesis the major vascular features of the flower are discernible even though some of the vasculature is only pro- cambial in those portions of the flower that later the cycle of phyllomes, or the vascular network of the carpel walls). The vascular supply to the flower is variable. As many as ten bundles enter the base of the sessile flower from the peduncular stele (Fig. 2E). In the receptacle these rapidly give rise to a pe- ripheral system of five to eight trunk bundles (tb), which eventually supply the stamens, the phyl- lomes, and the carpels as an inner series of four ventral (v) and two dorsal (d) carpel bundles (Figs. F-H, 6I). There is also considerable variation in the organization of the dorsal and ventral bun- dles. A dorsal bundle may be formed in the ovary wall by the radial or tangential division ofa trunk bundle; through the fusion of two lateral branch- es from adjacent trunk bundles; or directly by one of the bundles entering the receptacle from the peduncular stele. Except in the last case the formation of the dorsals occurs at varying levels in different carpels. The ventral bundles originate from the recep- tacle below the base of the locules. The ventrals of adjacent fused carpel margins may originate individually as branches from trunk bundles, or they may be united in a compound ventral bun- 2E, F). If come distinct in the septum a short distance above the bases of the locules (Fig. 2G, H). The peripheral trunk bundles (tb) ascend in the wall of the hypanthium to supply the stamens and phyllomes. At the base of the androecium BOGLE—LIQUIDAMBAROIDEAE 335 each trunk bundle branches tangentially. The abaxial branch supplies a single trace (st) to the base of a sterile stamen (Fig. 2H-J). The adaxial branch then divides to form two to three branch traces (phyl), which supply the bases of adjacent phyllomes. If a single large phyllome occupies the position between two adjacent trunk bundles (or sterile stamens), its base may receive traces from each of the adjacent trunk bundles. If two smaller phyllomes occur between two trunk bun- dles, each will receive one or more traces from each of the nearest trunk bundles. The traces а ари in the fleshy base of the phyl- lom ming an ascending system of bundles йлы by stippling in Fig. 2], К). At higher levels in the phyllome only a single prominent central bundle persists (Fig. 2K, L), ascending to the tip of the phyllome, as has been noted also by Schmitt (1965). In transverse sections the bas- es of the phyllomes often appear lobed and di- vided (Fig. 2J, K), and the larger lobes, at least, may be supplied by procambial strands at an- thesis. The secretory ducts that accompany the vascular bundles throughout the plant are also present in the bundle systems of the phyllomes. At the level of separation of the carpels from the phyllomes (Fig. 2J-L), it often appears that the stamen and phyllome bases remain fused for a very short distance, as if in a floral tube. This may or may not represent the vestige of a hy- panthial tube. This condition is occasionally present in Liquidambar styraciflua as well. The dorsal carpel bundles pass upward and, in the apocarpous portion of each carpel, produce at least two pairs of subopposite secondary branches that move upward and toward the car- pel margins, where they fuse with the ventral bundles. Transverse sections may thus show four to seven bundles in the carpel wall (Fig. 2L, M). The ventral carpel bundles (vv, v) ascend through the septum and into the placentae (Fig. 2F, H, I). Traces to the ovules are produced in both the septum (Fig. 2G) and in the free com- — iia s of the flowers illustrated at B n inflorescence of Altingia chinensis. п vestigial carpel primor gynoecial development in staminate flowers.—B. Secondary bract (5), with axillary complex of seven flowers, which contain partially developed gynoecia. Compare with the dia, grammatic presentation at D. ry rdia indicated by arrows. (Key: A,, axis; A,, secondary axis; B,, primary bract; B,, secondary bract; circled numerals designate bracts and > ae associated axillary flowers.) 336 pound placentae (Fig. 2H, I); placentation is thus axile below, but parietal above. In the distal por- tion of the ovary the ventrals produce an occa- sional minor branch to the carpel wall before passing into the style bases. The vascular anatomy of the female flowers of Liquidambar orientalis and L. styraciflua (Fig. 3) is essentially like that of L. formosana. The origins of the dorsal and ventral bundles of the ovaries of these species are fully as variable as in L. formosana (cf. Fig. 3D—F). An interesting deviation, however, is the adnation in these two species of a lower pair of carpel secondary veins to nearby trunk bundles of the hypanthium. Each of these secondary bundles originates below the level of separation of carpel from hypanthium, sie as а branch from a trunk bundle (or one of its major branches ) on either side of the dorsal bun dle Gass Figs. 3G, H, 6D). They move adaxially into the carpel wall and eventually fuse with adjacent ventral bundles. In Liquidambar styraciflua (Bogle 823) the ventral bundles may appear slightly diffuse (Fig. 6A, B), with some of the strands diverging to form ovule traces while others ascend into the style bases. In other cases the ovules may be supplied from a branch of the ventral that dies out after supplying traces to the ovules. he dle system that vascularizes phyllomes, a: as seen in cleared specimens of Liq- uidambar styraciflua, varies from quite complex (Fig. 6C, Bogle 823) to relatively simple (Fig. 6D, Bogle 790). In scanning electron micro- graphs widely scattered stomata can be seen in the epidermis of both the phyllomes and carpels of L. styraciflua. Staminate flower. Typical staminate flowers in all three species are highly modified (Fig. 4B, C), and the vasculature of their receptacles and axes is very diffuse and complex. In cleared sta- minate heads individual flowers are barely or not at all discernible on the basis of vasculature alone. Stamen bundles radiate from the margins of gaps in a loosely expanded peduncular stele. Each gap appears to correspond to a single flower, but gaps may become confluent, forming large open spaces in the stele. A bundle ting from the margin of a gap may: supply a single stamen base di- rectly; divide to form two traces to separate sta- mens; or anastomose with a second bundle to form the supply to one stamen. I found no evi- dence of a vascular supply to any of the abortive carpel primordia mentioned above and illus- trated in Figure 4. But it would not be surprising ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 to find such a supply in staminate flowers with partially developed gynoecia, such as those of Liquidambar orientalis illustrated and dia- grammed in Figure 4B(5) and 4D. 2. Altingia Noronha, Verh. Batav. Genootsch. Kunsten 5(2): 1. 1790. Figures 1B, 4G, 5, 6E-J. About seven to 14 species are currently rec- ognized in the genus Altingia (Vink, 1957; Tar- dieu-Blot, 1965; H. T. Chang, 1962, 1973, 1979). The genus is distributed from Bhutan and Assam eastward to southern China, and southward through Malaya to Java and Sumatra. In their natural state the massive, evergreen trees of A/- tingia may achieve heights of up to 60 meters, prompting their description as the kings of the mountain forests in southeast Asia, particularly in Indonesia. The evergreen leaves of A/tingia are simple to pinnately-lobed, stipulate, with an entire or serrate margin. They contrast sharply in form with the deciduous, palmately-lobed and -veined leaves of Liquidambar. However, intermediate leaf forms, and conditions of persistence, are de- scribed in the species attributed to the segregate genus Semiliquidambar H. T. Chang (1962, in- cluding A. chingii Metc. as Semiliquidambar chingii, and four new species, since reduced to three by H. T. Chang, 1979). Morphology. Inflorescences of Altingia are terminal on vegetative axes, and appear to basically similar to nin of Liquidambar in com- position, with a comparable range of variation. The staminate heads are often more elongate and oblong in appearance than those of Liquidambar (Fig. 1), and the internodes and peduncles are often longer, giving the total inflorescence a long- er, looser appearance. The majority of herbari- um specimens seen bear only older pistillate heads; very few bear anthesis stage or young in- florescences that permit observations of the total inflorescence. This problem is noted by Tardieu- Blot (1965) in her treatment of the six species of Altingia occurring in Cambodia, Laos, and Viet- nam. In fact, flowers are listed as unknown for five of the six species she describes. Vink (1957) described the inflorescences of Malaysian A/tin- gia in part as: “male heads in racemes consisting of masses of stamens, intermingled with some minute ?bracts," and “female heads solitary, in racemes or in the lower part of male racemes.” Young inflorescences containing at least one 1986] pistillate head at the basal node, with numerous distal nodes bearing pedunculate to sessile sta- minate heads, have been observed in herbarium specimens (GH, A) of Altingia chinensis (Champ.) Oliver, A. excelsa Nor., A. gracilipes, and A. obo- vata Merr. & Chun. Some inflorescences of A. chinensis have two or three pistillate heads at the basal nodes. In contrast, unisexual inflorescences of A. excelsa containing 11-16 pistillate heads arranged in a narrowly conical fashion occur in specimens of Abbe 10298 in my collections. In these, the lower heads are pedunculate, the in- termediate and upper heads grade from short- pedunculate to sessile, whereas the two or three most distal heads are fused in a terminal mass. e number of flowers in each pistillate head varies among the species of Altingia from nu- merous (in excess of 25) to as few as six. In fruit this results in heads that range in shape from globose in species such as A. chinensis or A. yun- nanensis, to obovate or obpyramidal in A. takh- tajanensis and A. siamensis (compare illustra- tions in Tardieu-Blot, 1965, pl. IV, V). This contrasts with the many-flowered pistillate heads of Liquidambar, which are typically globose. Pistillate flower. As in Liquidambar, I found no morphological evidence of a calyx or corolla (Fig. 5A, B). On the surface of the pistillate head three to ten stamens define the periphery of each flower. Their anthers are often rudimentary and misshapen, but in some cases they are fully formed and contain abundant pollen. The an- thers are broadest abaxially, and the connective is frequently minutely apiculate adaxially (Fig. 5A). Stamens in which the anther connectives are elaborated into stylodia with stigmatic sur- faces (Fig. 6F-H) have been observed in one specimen of A. chinensis (Pételot 5944, A). A cycle of short, inconspicuous, pubescent phyl- lomes occurs between the stamens and style bas- es (Fig. 5M). In maturing flowers the phyllomes appear broader and more massive, with more complex lobing than those in Liquidambar. The apocarpous portions of the two carpels are short, thick, and pubescent. Within the half-inferior ovary a shallow partial septum divides the base of the locule, but shortly gives way to two ap- о и placentae (Fig. 5E, F). A count of the r of ovules per carpel margin in specimens at hand showed 12-16 (28-31 per car- pel) in A/tingia is (Pételot 5944, A) to 20- 25 (41-47 et carpel) in A. excelsa (Bogle 583). The ovules are anatropous and become angular to weakly winged at maturity. BOGLE—LIQUIDAMBAROIDEAE 337 The form of the sterile phyllomes varies among the species of A/tingia. In most species these or- gans are short, broad, blunt, and pubescent struc- tures (Fig. SA, M). In a specimen of “А. chi- nense" from Tonkin (Pételot 5944, A), faeta. I observed subulate phyllomes (Fig. 6E) com- parable to those of Liguidambar formosana, and similar subulate organs 4—6 mm long are de- scribed T. Chang (1962, as “calyx limb") in Semiliquidambar cathayensis H. T. Chang (A. chingii Metc. pro parte). Staminate flower. The staminate flowers of both Altingia chinensis and A. excelsa are also similar to those of Liquidambar. The peduncu- late secondary axis of a staminate head of А. chinensis, from which all the anthers have been removed, is illustrated in Figure 4G. The indi- vidual flowers are indistinct, having merged with the axis. In undistorted flowers a single cycle of stamens is inserted around the periphery of a shallow depression in the surface of the axis. The anthers are tetrasporangiate, with a broad, fleshy cap formed by the connective (Fig. 5C). The sur- face of the depression is very finely and minutely pubescent, catching many pollen grains. At either end of the depression the tip of an abortive carpel primordium is often evident. Each tip bears a tuft of hairs (Fig. 4G, arrows). The scars of three secondary bracts (B;) are also apparent on the surface of this axis. Each secondary bract sub- tends a group of staminate flowers, the lowest of which are greatly distorted. The axis of this head thus represents at least three axillary floral clus- ters fused into a common mass. This morpho- logical evidence suggests I е partial inflo- rescences of Alt densed branch systems comparable to those Mise for Liq- uidambar styraciflua by Wisniewski and Bogle (1982). However, similar studies in A/tingia are needed. Floral vascular anatomy. The principal basis for comparison with Liquidambar used here is the vascular anatomy of the functionally pistil- late flowers. The vasculature of the staminate flowers is greatly reduced, modified and irregu- lar, as in Liquidambar, and no attempt is made to describe it here. Pistillate flower. Reduction of the peduncu- lar and pedicellar axes in the pistillate heads in Altingia appears more advanced than that in Liq- uidambar, resulting in increased congestion of the flowers. The peduncular stele is expanded into a loose network of branching and anasto- mosing bundles, and the abbreviated, loosely 338 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 1986] formed pedicellar cylinders seen in the recepta- cles of pistillate flowers of Liquidambar are ab- sent here. An indication of the increased reduc- tion and fusion within the heads is the occurrence of basal fusions between the trunk bundles of adjacent flowers. These bundles separate to their appropriate flowers at a relatively low level. Such fusions were not noted in the species of Liquid- ambar. Several (five to seven) vascular bundles di- verge directly from a gap in the peduncular stele into the relatively broad base ofthe sessile flower (Fig. 5D). These bundles may function simply as “disk” lobes of Vink, 1957), or they may divide to supply dorsal or ventral bundles to the gy- noecium as well. The number of trunk bundles corresponds to the number of stamens in most cases (Fig. SE, F). As the trunk bundles ascend in the hypanthium they begin to branch at a level slightly below the separation between the carpels and the phyllomes (Fig. 5G). Each trunk bundle usually produces a trace (st) to a rudimentary stamen, as well as lateral branches (lb) to one or both sides. The latter may function directly in supplying the base of a phyllome or may anas- tomose with a lateral branch from an adjacent trunk bundle to form such a supply. Each major lobe of the cycle of phyllomes is thus supplied with a vascular bundle. Within the lobe the bun- dle ramifies to form a complex system of minor bundles, procambial at anthesis, many of which terminate in small protrusions on the abaxial side of each lobe (Fig. 5H-J). At the level of separation of the carpels from the phyllomes the stamen and phyllome bases may appear fused for a short distance, as if in the rim of a hypanthial tube (Fig. 5I, J). The formation of the carpel ventral bundles is BOGLE—LIQUIDAMBAROIDEAE 339 variable. They are formed immediately in the base of the flower, frequently before either dorsal bundle is evident. One to three vascular strands become associated at either side of the receptacle in the transverse plane to form a compound ven- tral bundle (vv in Figs. 5D, E, 6J). In some serial sections, and in some cleared flowers, the com- pound ventral bundles appear to consist ofa cen- tral strand flanked on either side by minor lateral strands that diverge slightly, giving the impres- sion of a second bundle in each carpel margin. In cleared material these divergent bundles are seen to persist for only about a third ofthe length of the ovary. In serial sections they are seen to dissipate in procambial strands to the lowest in- serted ovules. They appear to represent minor placental strands of the ventral bundles, rather than a second bundle in each margin. A similar situation is seen in Liquidambar styraciflua (Fig. At the level of insertion of the lowest ovules the compound ventral bundles (vv) divide, form- ing individual ventral bundles (v) to each of the adjacent fused carpel margins (Fig. 5G). Above this level the traces to the ovules originate from the free ventral bundles (Fig. 5H), which finally pass unbranched into the styles (Figs. 5J, 6J). The carpel dorsal bundles (d) may be distinct at the base of the locules, but more often they are fused basally with one or more of the pe- ripheral trunk bundles and do not separate until about halfway up through the inferior portion o the ovary (Fig. 5G). Secondary carpel wall bundles appear only as the hypanthium begins to separate from the ovary. Above this level (Fig. 5I-L), lateral branch bun- dles pass directly from dorsal to ventral bundles, often almost horizontally. Each lateral is accom- panied by a gum duct, as are the dorsal and ven- tral bundles. There is no evidence in my material = FIGURE 5. qeu m a staminate flower. D-L. Seri cep tacle; bundles to the phyllomes shown. — I. Partial separation of the hypanthium from the from each other.—J. Insertion of the stamens and phyllome veral гонаг trunk bundles and compound ventral pos: dd evident. — E. Altingia excelsa Nor.— A. Pistillate flower at anthesis. Parts of adjacent flowers shown. — B. Floral Stamen from al sections of a pistillate flower "n anthesis. — D. Re- carpels, and of the carpels s. — K, L. Sections through the style bases, showing phyllomes and anthers of the sterile иаа —M. Drawing made em a aput oe cain hand gem through a ovules, in sert n of phyllomes. (Key: A,, primary axis; pistillate flower in the median plane, sho A,, secondary axis; d, dorsal carpel bunte "Ib. lateral branch; loc, oa ov, ovule; phyl, Sd ай st, stamen, .) stamen trace; sty, style; tb, trunk bundle; v, ventral carpel bundle; vv, compound ventral bundle 340 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURE 6. A-D. Liquidambar styraciflua. — A. Vasculature of gynoecium of the pistillate flower, showing dorsal and ventral carpel bundles, and the ovule traces originating from the ventral bundles on the near side (Bogle 823). —B. Vasculature of one carpel enlarged, showing do Pu bundle and ventral bundle of near margin with ovule traces; ovules pads (Bogle 823).—C. Ramifying vascular supply to phyllomes of the pistillate flower; one stamen trace and insertion scar indicated. Note that one dendritic system supplies branches to more than one lobe of the in of bn in both C and D (Bogle 823). —D. Less ela — vascular supply to phyllomes of a pistillate flower (Bogle 790). E-H. Altingia chinensis (Pételot 5944). —E. Pistillate flower with one stylodeus stamen with half-anther (stl), one scale-like staminodeum with «Уагыз (sto) and numerous 1986] of vascular bundles contributed to the carpel walls by the hypanthial bundles, as in Liquidambar orientalis and L. styraciflua. DISCUSSION The taxa of subfamily Liquidambaroideae ex- hibit an interesting combination of relatively primitive and advanced morphological and an- atomical characteristics. The complex inflores- cences are terminal on long or short vegetative axes. Subsequent vegetative growth develops from axillary buds. Morphological evidence from mature inflo- rescences indicates that the total inflorescence in the subfamily is derived from a compound ra- ceme by reduction, and that a minimum of three orders of axes (or four if the floral pedicels are taken into account) are present. Ontogenetic study of the total inflorescence of Liquidambar styra- ciflua (Wisniewski & Bogle, 1982) supports this interpretation. The primary or central axis of the total inflo- rescence elongates at anthesis, carrying the par- tial inflorescences (“heads”) out of the bud and separating them through internodal elongation. The internodes may become progressively short- er toward the tip of the axis, and the most distal internodes may fail to elongate, resulting in the aggregation of a few distal heads into an irregular common mass. One or two functionally pistillate eads. completely pistillate or staminate total inflores- cences are not uncommon. The secondary axes, often referred to as the peduncles of the heads, originate in the axils of primary bracts of the primary axis, and show varying degrees of reduction. The proximal ones may elongate considerably, producing long- to short-stalked heads bearing secondary bracts with axillary flower clusters, whereas the more distal ones may fail to elongate at all, resulting in heads sessile on the primary axis. Failure of the distal internodes of the secondary axes to elongate re- BOGLE—LIQUIDAMBAROIDEAE 341 sults in terminal compound clusters of eet hs similar to those at the tip of the primary a The tertiary axes that bear the several de of each cluster are generally abortive, as are the floral pedicels (quaternary axes), resulting in the aggregation and fusion of the flowers. An abscission layer forms in the primary axis just above the node bearing the uppermost pis- tillate head, allowing tl portion of the total inflorescence to fall after anthesis and pollination. The pistillate heads enlarge greatly in fruit and fall only after the seeds are shed in autumn. Artificial pollinations have shown the three principal species of Liguidambar to be in- terfertile (Santamour, 1972), but I know of no similar experiments in Altingia, or between Al- tingia and Liquidambar The individual flowers of Liquidambar and cally bisexual, but typi- ta from typically abortive staminodia to rarely fer- tile. When fertile the viable pollen may be she well after stigma receptivity in the same flower, preventing self pollination (Schmitt, 1965). Abortive carpel primordia or rudimentary ova- ries occur in some staminate flowers, but ovules have not been observed. The inflorescences and flowers clearly are adapted to anemophily and outcrossing. There are numerous в observations 5 I aUsU of a perianth in “the Ei üidaniberoideae: AL though there is general agreement that the sta- minate flowers of both genera lack a perianth, there are several reports of a perianth in the pis- tillate flowers of both Liquidambar and Altingia. I believe that these reports are based on misin- terpretation of either: 1) the cycle of sterile sta- mens (as suggested by Horne, 1914); 2) the cycle of elongate sterile phyllomes (disc lobes) that oc- curs between the stamens and carpels in Liquid- ambar formosana and some species of Altingia (as Semiliquidambar); or 3) a layer of scleren- chymatous tissue that forms between and unites h..1 11 2:291 itted. — F-H. Three variants of stamens with stylodeus connectives, bearing. two or four sporangia. dambaroideae. — I. short distance basally (vv). ibas traces ovule, ovule trace; st, stamen, carpel bundle; vv, compound cd bund shown J. Diagrammatic representations of carpe Adjacent ventral weh [a тт of апу fusion.—J. nly one ventral bundle. (Key: d, dorsal carpel bundle; ov, amen trace; stl, ind stamen; sto, staminodium; tb, trunk bundle; v, ventral le.) ] vasculature in Liqui- Adjacent ventral bundles fused for a 342 the inferior portions of the flowers in the pistil- late heads, and becomes protrusive as the heads dry out at maturity. In Liquidambar various workers have de- scribed what they considered to be perianth parts (Oliver, 1867; Sargent, 1890, 1922; Ek, 1902; Sosa, 1978). Oliver (1867) apparently considered the sterile phyllomes of all three species of Liq- uidambar to be calycine. The indurated periph- eral rim between flowers of mature heads of L. styraciflua, considered by Sargent (1890, 1922) to be a rudimentary calyx, is more probably formed by layers of bony tissue (possibly the "epidermal hairs" or “hair-like cells” of Schmitt, 1965) that externally delimit and unite the hy- panthial walls of the fused, half-inferior ovaries that form the female partial inflorescences. The distal margins of these bony layers become pro- trusive in the surface of the head as the paren- chymatous tissues of the fruits dry out and col- la pse. Endress (1977) stated that “a perianth is al- most totally lacking" in subfamily Liquidam- baroideae, while Niedenzu (1891), Harms (1930), Samorodova-Bianki (1957), and Schmitt (1965) all found the pistillate flowers to lack a perianth in Liquidambar In Altingia, references to a perianth in the pis- tillate flowers are common. Blume (1828) ap- parently interpreted the sterile stamens as calyx lobes. Several workers described calyces united between flowers, but lacking a limb (e.g., Guil- laumin, 1920; Hutchinson, 1967), or immersed in a disc with connate calyces fused with the stamens (Nakai, 1943). Several authors appear to have interpreted the sterile phyllomes (disc lobes) as part of the calyx limb (Griffith, 1836; Hooker & Thomson, 1858; Clarke, 1858; Nie- denzu, 1891; Noronha, 1790), whereas the re- mainder believed that a perianth is almost totally lacking (Endress, 1977), or absent (Hemsley, 1906; Schulze-Menz, 1964; Tong, 1930; Wilson, 1905; Vink, 1957) In Semiliquidambar H. T. Chang, Chang (1962) described the pistillate flowers as having con- fluent calyces with subulate calyx limbs ranging from 1-6 mm long. In a comparative chart he listed the calyx lobes as persistent or obscure in Liquidambar, persistent in Semiliquidambar, and lacking in A/tingia. However, in fruits of A/tingia chingii Metc. [7 Semiliquidambar chingii (Metc.) Chang] that I have seen (Lau 4356, Tso 20760, both A) the only structures of the dimensions described are sterile phyllomes. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 In the many specimens of Liquidambar and Altingia that I have examined I have found no evidence of calyx lobes or limb in any of the staminate or pistillate flowers at anthesis or be- yond and have to conclude that none exist. Nor is there any evidence of a perianth in the earliest as have ES (Озан, 1920; р. 1967), that a calyx or calycine tissue was present in the ancestral line, forming a hypanthium that is now incorporated into the head as confluent tissue between the pistillate flowers. It ma this tissue that forms the interlocking “hair-like cells uniting the adjacent capsules” described by Schmitt (1965, fig. 2), and which eventually pro- duces the sclerotic tissue uniting the ovaries of the flowers. I base my assumption that a perianth limb has been lost, rather than that the flowers were primitively naked, on comparative evi- dence from closely related subfamilies that also have their flowers fused in heads or spikes to some degree. In the dichlamydeous flowers of Rhodoleia (Rhodoleioideae) a very reduced, membranous, inconspicuous calyx is present. In the Exbucklandioideae (including Mytilarioi- deae H. T. Chang, 1973), Mytilaria is dichla- mydeous with distinct calyx lobes, Exbucklandia is also dichlamydeous but shows evidence of a calyx only in the early stages of floral ontogeny, and Chunia lacks both calyx and corolla. Disan- thus (Disanthoideae) is dichlamydeous, with dis- tinct sepals and petals. It appears that reduction or loss of perianth in Liquidambaroideae, Rho- doleioideae, and Exbucklandioideae is related to condensation of inflorescences and increasing degrees of fusion between flowers in the heads. In this process, reduction and loss of the calyx limb may precede that of the corolla (Bogle, pers. observ.). Loss of the perianth has apparently oc- curred independently in both Liquidambaro- ideae and Exbucklandioideae, as well as in the more advanced members of the subfamily Ham- amelidoideae (Bogle, 1970; Endress, 1970, 1978). As is apparent from the foregoing discussion of the perianth, the interpretation of the cycle of sterile phyllomes (disc lobes) has been problem- atical. I have used the term phyllome (sensu Esau, 1965: 424, 540) to distinguish these organs from sterile and reduced каше (staminodes), which , because they show no direct relationship in structure or in transitional forms to stamens. Just as some au- thors considered them calycine, others have in- OMMMOMLY OCCUI ir 1p 1986] terpreted them as bracteoles surrounding the pis- tillate flowers (Guillaumin, 1920); as vestigial styles of sterile flowers interspersed among the fertile flowers (Harms, 1930); as staminodia (Tong, 1930); as “disk” lobes (Vink, 1957); as papillae on the surfaces of the capsules (Schmitt, 1965); or as the sterile setae (“borsten”) of Liq- uidambar formosana on which Harms (1930) based his section Cathayambar. Similar elongate sterile bodies, comparable to those seen in L formosana, also occur in some species of Altin- gia, where I believe they have been interpreted as calyx lobes (H. T. Chang, 1962, 1973,as Semi- liquidambar). Croizat (1947) considered the lobes or sterile phyllomes to be “scales” arising “from the body of the head, or torus," and that the staminodia surrounding each flower "either belong to the scales that immediately surround the carpels, or to a row of subsidiary undeveloped scales abax- ially located from these." He wrote that Liquid- ambar is “in active evolution away from the amentiferous аде, still retaining the head- or ament- sex- ual expression, which leads on occasions to pure- ly male or female heads or flower However, the vascular supply to n phyllomes arises from the same trunk bundles as that of the stamens and carpels, and at a relatively high level within the flower. This does not support an in- a. of the phyllomes as extra-floral scales ising de novo from the torus. Neither does it а жиы? an interpretation of them as extra-floral organs of any kind. Vink (1957: 364) emphasized the position of the sterile phyllomes (as “‘disk lobes”) in A/tingia and Exbucklandia (as Symingtonia) between the stamens and ovary. This position might suggest that they are staminodia, glandular tissue, or re- duced sterile carpels. Tong (1930: 9, 13, 14, fig. 2C, D) interpreted the phyllomes as staminodia, noted their spine- mosana, and cited staminodia with functionless anthers as transition forms from stamens to staminodia in pistillate flowers of both Liquid- ambar and Altingia. His diagram (1930, fig. 2D) of the staminate flower of Liquidambar indicated two cycles of stamens. I have not observed this condition, nor have I seen any evidence of sterile phyllomes with abortive anthers. I have seen sta- mens with stylodeus elaborations of their con- nectives (Fig. 6E-H) external to the cycle of phyl- BOGLE—LIQUIDAMBAROIDEAE 343 lomes, but the adjacent phyllomes show no such modification. The resolution of the proliferated vascular supply in the base of the phyllomes to a single, central vascular bundle in their distal portions could suggest the single bundle that nor- mally supplies each stamen. The short, blunt, glabrous phyllomes in Liq- uidambar styraciflua and some Altingia species look somewhat like similar organs in Rhodoleia and Exbucklandia, however. In Rhodoleia these organs form a cycle of very small, glabrous, ir- regular lobes inserted between the stamens and the ovary and are reported to be glandular (Leeu- wen, 1938). In Exbucklandia, however, a cycle of blunt, glabrous, fleshy lobes is elevated on the inner rim of the hypanthium, and is clearly sep- arated and distinct from the gynoecium. These lobes are not reported to function as glands, and their position on the rim of the hypanthium sug- gests a relationship to the androecium. The proliferated system of procambial strands seen in the base of the sterile phyllomes or the stomata in the phyllome epidermis of L. styra- ciflua (Wisniewski & Bogle, 1982) might also suggest a glandular or secretory function, but there are no reports in the literature, nor any field ob- servations of mine, that indicate such activity in these organs. Nor do the pubescent surfaces of the phyllomes of Liquidambar formosana and Altingia chinensis suggest a glandular function. In these species the epidermal trichomes of the phyllomes and the styles are similar and are not glandular. A gynoecial origin could be inferred from the styliform phyllomes of Liquidambar formosana, and to a lesser extent from some specimens of Altingia (as Semiliquidambar) in which these or- gans approach the functional styles in length, and bear the same type of pubescence. Although they look very much like abortive carpels both at an- thesis and in fruit, and have been interpreted as such (Harms, 1930), I have seen no evidence of stigmatic surfaces or ovule production in any of them. Many specimens of other species of Liq- uidambar and Altingia have relatively short, blunt, glabrous phyllomes, which sometimes ap- pear to be inserted on the vestige of a hypanthial tube, apart from the ovary. The late appearance of the phyllomes in on- togeny (Wisniewski & Bogle, 1982), after the sta- mens and functional carpels are initiated (at least in Liquidambar styraciflua) neither establishes nor eliminates a gynoecial relationship, but the ramifying vascular supply to the base of the phyl- 344 lomes does not appear comparable to that of the functional carpels. A secondary origin of the sterile bodies might be hypothesized if some function relating to pol- lination or the reproductive к could be confirmed for them, as in Rho a. On the oth- er hand, the presence of aie Nu of sterile organs in such divergent genera as Liquidambar, Altingia, Rhodoleia, and Exbucklandia suggests an origin in a common ancestry, rather than a secondary origin related to reproductive pro- cesses. It seems probable that the interpretation of the sterile phyllomes in Liquidambaroideae may re- late directly to that of the similar organs in Rho- doleioideae, Exbucklandioideae, and Hamamel- idoideae. If such fleshy, indeterminate organs were brought into close association with the de- veloping gynoecium as a result of the conden- sation of the inflorescence, crowding of flowers, and eventual eii Maced fusion of hypan- thium and gynoecium rphogenetic factors might influence s pen toward a sty- lodial morphology. That is not to suggest that Exbucklandioideae or Rhodoleioideae are in any way ancestral to the Liquidambaroideae, but rather that these En probably shared a rela- tively close, comm ncestral stock. The flow- ers of Mytilaria (Exbucklandioidene). with a di- cyclic androecium in which the inner cycle of stamens appears to be sterile, and in which all stamens have remarkable fleshy, horned exten- sions ofthe anther connective, may provide clues as to the origin of the sterile phyllomes of the Liquidambaroideae. Additional ontogenetic studies in Liquidambaroideae, Rhodoleioideae, and Exbucklandioideae are needed. The gynoecium in the Liquidambaroideae is the most primitive in the family. The carpels are follicular and unsealed. The shallow partial sep- members of the family. The extensive stigmatic surfaces are clearly adapted for wind pollination. However, anemophily is probably a derived con- dition that has evolved several times within the il carpel vascular supply of two marginal ventral bundles and a median dorsal bundle is comparable to that in the Rhodoleioideae, Ex- bucklandioideae, and Disanthoideae, and differs from the five bundle carpels of the Hamameli- doideae (Bogle, 1967, 1970). The ventral bundles ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 of adjacent fused carpel margins may be distinct or fused basally for a short distance. The latter condition is shared with Rhodoleioideae and also comes more pronounced in other subfamilies. The large number of ovules per carpel, and their insertion in more than one row on both margins, is in keeping with the primitive con- dition of the carpels. Only the basal or sub-basal ovules on each margin are fertile, the remaining ovules are presumably sterile by reduction. The number of seeds produced per flower is very low in relation to the total number of ovules initiated in each flower. The tendency toward ovule abor- tion and reduction in number that characterizes the family, culminating in one ovule per carpel in some genera of Hamamelidoideae (Bogle, 1970), appears to be well established in the Liq- uidambaroideae Placentation has been described as either axile or parietal in the Hamamelidaceae. In the Liq- uidambaroideae the lowermost ovules may be inserted on the partial septum in the base of the ovary. This condition is sometimes termed axile placentation. Most of the ovules are inserted above the partial septum on the two lateral com- pound placentae, hence placentation is parietal. At a still higher level the uppermost ovules may be inserted on free carpel margins. Consequently, I use the term parietal as defined by Lawrence (1951: 763): “borne on the walls within a simple or compound ovary, or on intrusions of the wall that form incomplete partitions or false septa within the ovary." There has been some disagreement concerning the taxonomic relationships of A/tingia (Noron- 1828; de Candolle, 1830; Lindley, 1836, 1853; Oken, 1841; Bentham, 1861), and more recently by Leroy (1982). Bentham and Hooker (1865) and other subsequent workers have recognized them as distinct genera. The major differences between the two genera are the more or less elliptic to oblong, pinnately- veined, persistent leaves of A/tingia, as opposed to the palmately-lobed and -veined, deciduous being intermediate between A/tingia and Liquid- ambar in this respect, with leaves polymorphous (trilobate, simple, or lobed on one side), tripli- nerved, and deciduous or not (H. T. Chang, 1962). Semiliquidambar occupies an extensive area of 1986] overlap between Altingia and Liquidambar for- mosana in southeastern China and on the island of Hainan (distribution map in H. T. Chang, 1992). This distribution Medi that some of have. originated as hybrids between Altingia and Liquidambar. The appearance of the subulate phyllomes seen in the pistillate flowers of these specimens may thus be inherited from a L. for- mosana gm nt. Further morphological investi- miliquidambar, and attempts at Liquidambar such as those conducted in ambar by Santamour (1972), would be highly desirable. A curious parallel can be seen between the Hamamelidaceae and Platanaceae in that similar contrasting leaf forms occur in Platanus. The tropical, southeast Asian Platanus kerrii has per- sistent, elliptic to oblong, penninerved leaves, whereas the remaining species of the genus have deciduous, palmately-lobed and -veined leaves. On the basis of these and other morphological characteristics, Leroy (1982) recently proposed the establishment of two subgenera in P/atanus to separate these two divergent forms. Should the intermediate form attributed to Semiliquid- ambar eventually prove to originate in hybrid- ization between Altingia and Liquidambar, these two genera might warrant similar taxonomic treatment. CONCLUSIONS It now seems clear that the total inflorescence in the Liquidambaroideae is more complex than was formerly thought. Four orders of axes (in- cluding the floral pedicels) are present, but only those of the first and second order are obvious. Primary and secondary inflorescence bracts are present, but tertiary bracts subtending individual flowers have not been observed. The flowers are potentially bisexual, becoming functionally pistillate or staminate through fail- ure of the stamens or carpels, respectively, to develop. Unisexuality is not completely estab- lished. There is no morphological or anatomical evi- dence of a perianth in early ontogeny or in ma- ture flowers. The perianth limb is assumed to have been lost phylogenetically as a function of inflorescence condensation leading to crowding and fusion of flowers. However, there is the sug- gestion of a brief hypanthial tube in the fused bases of the stamens and phyllomes. BOGLE—LIQUIDAMBAROIDEAE 345 The stamens are few and variable in number in each flower, dehiscing latrorsely by means of longitudinal slits. The connective often forms a fleshy cap above the sporangia in staminate flow- ers. Each stamen receives a single vascular trace. The cycle of sterile phyllomes develops be- tween the stamens and carpels of pistillate flow- ers. The phyllomes are not clearly staminodial or carpellary in form, origin, or vascularization and are not known to have glandular or secretory function or a role in pollination. They may relate directly to similar cycles of sterile organs in Rho- doleioideae and Exbucklandioideae. The partly inferior ovary is usually bicarpel- late. The carpels are involute and follicular, with numerous ovules on each margin, but usually have only one fertile ovule at the base of each margin. The carpel margins are not sealed in their apocarpous portions. Each carpel receives a dorsal and two ventral bundles. The ovules are supplied from the ventral bundles. Although seemingly specialized and advanced within the family in such characteristics as inflo- rescence structure, tendency toward unisexual owers, distinctive pollen morphology, and ane- mophily, the Liguicgenparoneae, оной vea have a more p h follicular carpels, three distinct “vascular bundles in each carpel, and numerous ovules on each margin) than any other genus in the family. The gynoe- cium of the Rhodoleioideae most closely ap- proaches that of Liquidambaroideae, and that of the Disanthoideae shows considerable modifi- cation from both of those groups. The gynoecial structure, in particular, of Liquidambaroideae represents the least specialized condition in a family in which the general trend is toward re- duction in the gynoecium, dee in the num- ber of ovules to one per c l, and increased connation of carpels and ed bundles. When thisi 1S taken 1 in conj unction with such specialized resence ofresin ducts, con- densed inflorescences, unisexual flowers, rugate to porate pollen associated with anemophily, and sterile floral organs (phyllomes) of uncertain na- ture, each characteristic occurring in at least two other subfamilies, the Liquidambaroideae can be seen as part of a continuum or reticulum within the Hamamelidaceae, rather than as an indepen- dent family, the Altingiaceae. LITERATURE CITED BENTHAM, G. 1861. Flora Hongkongensis. L. Reeve, London 346 & J. D. Hooker. 1865. Genera Plantarum, Volumes 1, 2. L. Reeve, London. vea Flora Javae. Bru 7. Floral Vascular joa and the e of the Hamamelidaceous Flower. Ph.D. aul. ogy and vascular anat- omy of the Hamamelidaceae: с Mec qun gen- era of th old Arbor. 51: 310-366. C. T. PHILBRICK. 1980. A generic atlas of hamamelidaceous pollens. Contr. Gray Herb. 210: 29- BRITTON, E. G. 1887. Elongation of the Vues of Liquidambar. Bull. Torrey Bot. Club 14: 95 CANDOLLE, A. P. pr. 1830. Prodromus ан Naturalis Regni Vegetabilis, Part 4. Pari CHANG, C. T. The pollen morphology of Liq- uidambar L. and Altingia Nor. Bot. Mos- cow & Leningrad) 44: 1375-1380. pls. | 1-5. [In Russian with English summary.] 1964. The pollen morphology of the families Hamamelidaceae and Altingiaceae. Trudy Bot. Inst. Akad. Nauk SSSR, Ser. 1, Fl. Sist. Vysš. Rast. 13: 173-232. pls. 1-21. [In Russian.] CHANG,H. T. 1962. Semiliquidambar, Novum Ham- amelidacearum Genus Sinicum. Sunyatsen Univ. Bull., Nat. Sci. 1: 34-44. 73. A Revision of the Hamamelidaceous Flora of China. Bull. Sunyatsen Univ. 1: 54-71. 1 Hamamelidaceae. Jn Flora Reipub- licae Popularis Sinicae 35(2): 36-116. Сним, W. У. 19 Contributions to the flora of Kwangtung and South-eastern China. Sunyatsenia 1: 209-317. [Hamamelidaceae, pp. 36, 241-247.] CLARKE, B. 1858. On the Structure and Affinities of Myricaceae, Platanaceae, Altingiaceae, and Chlo- ranthaceae. Ann. Mag. Nat. Hist., Ser. 3, 1: 100- 109, 112, 113. pl. VI CroizaT, L. 1947. Trochodendron, Tetracentron, and their meaning in phylogeny. Bull. Torrey Bot. Club 60-76. uy sk A. 1968. The Evolution and Classifica- of Flowering Plants. Houghton Mifflin Co., Bos ton. Я ie An Integrated System of Mp cu : Flowering Plants. Columbia Univ. Press, New York. [Li iquidambar m Altingia here Ae as subfamily Altingioidea Duncan, W. H. 1959. es variation in Liquidambar styraciflua L. Mie 24: Ek, B. 902. quidambar dentalis Miller. A и ы Histologo-Pharmacognostical Study in Connection with the Secretion p = Plant of a Liquid Styrax (Styrax pun Dissertation. Moscow Uni ENDRESS, P. К. 1970. Die оол дег apetalen Hamamelidaceen, ihre grundsatzliche morpholo- gische und systematische Bedeutung. Bot. Jahrb. Syst. 90: 1-54. . 1977. Evolutionary trends in the Hamamel- idales-Fagales group. Pl. Syst. Evol., Suppl. 1: 321- 347 1978. Blütenontogenese, Blütenabgrenzung ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 und systematische Stellung der perianthlosen Hamamelidoideae. Bot. Jahrb. Syst. 100: 249-317. Esau, K. 1965. Plant Anatomy, 2nd edition. John iley & Sons, Inc., New GOLDBLATT, P. & . ENDRESS. 1977. Cytology and evolution in Hamamelidaceae. J. Arnold Arbor. 7-71. GRIFFITH, W. 1836. Description of two genera of the fami f e es (Calcutta) 19: 94-114. pls. XIII-XVIII. [A/tin- gia here described as Sedgewickia.] 1838. Description de deux genres de ee a stemon et d'une ерке 4е Kaulfussia. и "Sci. Nat. Bot., Sér. : 176. misery er A. 1920. Hamamelidacées. Pp. 709- n M. H. Lecomte (editor), Flore Générale de l'Indochine, Volume 2. Masson et Cie, Paris. е Н. 30. Hamamelidaceae. Jn A. Engler $ K. Prantl, Die Naturlichen Pflanzenfamilien, 2nd edition. 18a: 330-343. . 1830. Abbildungen von Storaxbáumen. 2. ; 906. Altingia gracilipes. Hemsl. Hooker’s Icon. Pl. 29: tab. 2837. HOLMBOE, J. 1914. Studies on the Vegetation of Cy- prus, based upon researches during the spring and summer of 1905. Bergens Mus. Skr. 1(2): 1-344. Hooker, J. D T. THOMSON. 1858. Praecursores ad Floram Indicam. J. Proc. Linn. Soc., Bot. 2: 54-103. vaga: pp. 84-86.] HORNE, A. L. ontribution to the study of the evolution of the с with special reference naceae. Trans. Linn. Soc. London, Bot., Ser. 2, 8: 239-309. pls. 28-30. HuTCHINSON, J. 1967. The Genera of Flowering Plants, Angiospe rmae. Dicotyledons, Volume 2. Clarendon Press, Oxford. [Hamamelidaceae, pp. 93-103.] KIRCHHEIMER, F. Über verzweigte Fruchstande der us da L. Planta 35: 106- 109. LAWRENCE, G. H. M. 1951. Taxonomy of Vascular Plants. Macmillan Co., New York. LEEUWEN, W. M. D. vAN. 1938. Observations about the biology of tropical flowers. Ann. Jard. Bot. o 48: 27-68. pl. X. [Rhodoleia, pp. 48- l. X. LEROY, rà EJ 982. o anus rer Compt. Hebd. Séances , Sér. 3, 295: 251 -254. 1977 Hamamelidaceae. Pp. 1-9 in H.-l. Li, u, T.-c. Huang, Т. Koyama & С. E. DeVol dtr Flora of Taiwan, M eni III, Angio- permae. Epoch Publ. Co., Tai киш, 1 1836. С piss S nen of Botany, 2nd e n. Longm —— 1853. The ab Kingdom, 3rd edition. Bradbury and Evans, London. LINNAEUS, C. 1753. Species Plantarum, Volumes 1, 2. kholm. LITTLE, Е. L. . Atlas of United States Trees, Volume 1, Conifers and Important Hardwoods. 1986] c. Publ. United = Dept. Agric. 1146. [Lig- са Maps 1353 Е. MAKAROVA, Z. I. 1957. i the history of the genus Liquidambar L. Bot. Zurn. (Moscow & Leningrad) 42: _ . Floral Evolution i in the Ham- les and associated groups including Urticales, and final conclusions. Acta Bot. Neerl. 24: 181-191 MEIKLE, В. D. 1977. Flora of Cyprus, Volume 1. Bentham-Moxon Trust, Royal Botanic Garden, Kew MELIKIAN, A. P. 1971. ae structure of the spermodermis in the members of the genera Liq- uidambar L. and Altingia Nor. in шй to their taxonomy. Izv. Biol. Nauki 24: 50-55. [In Rus- sian. 1973a. Seed coat types of Hamamelidaceae and allied families in relation to their systematics. Bot. Zurn (Moscow & Leningrad) 58: 350-359. [In Russian. ]} . 1973b. The anatomy of seed coat and sys- tematics of the family oe Izv. Biol. Nauki 26: 104—105. [In Russ METCALF, F. P. 1931. Botanical ae on Fukien and Southeast China, I. Lingnan Sci. J. 10: 413-414. 59 1943. Ordines, Apnd: tribi, genera, sec- et combinationes novae. ls. 58, NAKAI, T. гуы z 1891. Hamamelidaceae. Pp. 115-130 er & K. Prantl, Die Naturlichen Pflan- See a Teil 3, Abteilung 2a. W. Engelmann, Leipzig. NORONHA, F. 1790. A. 6 malaice еї јауапісе Rasamala; Lignum papuanum Rumphii Herbar. Verh. Batav. Genootsch. nist 5: 1- 20 OKEN, L. 1841. Allgemeine Naturgeschichte für alle ánde. Band 2 und 3 oder Botanik, Band 1 und 2. Hoffmann'sche Verlags-Buchhandlung, Stutt- gart. [Fide E. D. Merrill, J. Arnold Arbor. 31: 280. 1950.] OLIVER, D. 1867. "rhe es risen Hance. Hooker's Icon. Pl. 14. pl. 1020. Rao, P. R. M. 1974. 2 iis eet in some Ham amelidaceae and phylogeny. Phytomorphology 24: 113-139. Rao, T. A. & O. P. BHUPAL. 1974. кие, x Кылы sclereids in various taxa of Hamamelidace d. Wiss. Wien, Math.-Naturwiss. Kl., Denkscht. 105: 1-924. ни С. В. De genera Liq- ui ar L. notulae systematicae. Mater. Gerb. Glavn. Bot. Sada SSSR 18: 77-89. [In Rus- sian.] BOGLE—LIQUIDAMBAROIDEAE 347 SANTAMOUR, F. S., JR. 1972. Interspecific rae Sci. 18: 23-2 rth America 2nd edition, Volumes 1, 2. Dover Publications nc., New York $снмітт, D. The pistillate inflorescence of sweet-gum (L. styraciflua). Silvae Genet. 15(2): 3-35. . PERRv. 1964. Self-sterility in sweet- gum. Forest Sci. 10: 302-305. SCHULZE-MENz, G. К. 1964. Hamamelidaceae. Pp. lien, 12th edition, Volume 2. Borntraeger Bros., Berlin. SHOEMAKER, D. N. 1905. On the development Hamamelis че Bot. Gaz. alia an Fa 39: 248-266 SKVORTSOVA, N. T. 1960. The pr structure of the conducting system of leaf pe resentatives of families Eb a melidaceas and AI- tingeaceae. Dokl. Akad. Nauk SSSR 133: 1231- 1234. Sosa, V. 1978. Hamamelidaceae. Pp. 1-6 in A. Gó- mez- -Pompa (editor), Flora de Veracruz, uber" in 1.1 ticos, A.C., Xalapa, Mexic TAKHTAJAN, A. 1969. еа Plants: Origin апа Dispersal. Oliver and Boyd, Edinburgh. [Trans- lated from Russian by C. Jeffre З 80. Outline of the classification of flow- ering ES Noe Bot. Rev. (Lancas- ter) 46: 225 TARDIEU-BLOT, "a x 1965. Hamamelidaceae. Pp. 75-116 in A. Aubréville & M. L. Tardieu-Blot (editors), Flore du Cambodge du Laos et du Viet- nam, Fascicule 4. Muséum National d'Histoire Naturelle, Paria, Томс, K. 1930. Studien ueber die familie der Ham- amelidaceae, mit besonderer berucksichtigung der systematik und entwicklungsgeschichte von Cor- i Uni ylopsis. Bull. Dept. Biol. Sun Yats Vink, W. 195 amamelidaceae Steenis (editor), Flora Malesiana 5(1): 363-379 Wis, J. С. 1 ictio f the Flowering Plants and Ferns, 7th edition. Cambridge Univ Press, Cambridge. [Revised by H. K. Airy Shaw.] WILSON, P. 1905. Altingiaceae. Jn М. L > tton & L Underwood (editors), North rican Flora Am 22(2): 189. New York Botanical an. New York. WISNIEWSKI, M. & A. L. BoGLE. 1982. The ontogeny of the _inflorescence and flower of vip. del amamelidaceae). Amer. J. WOLFE, J. A. an Fossil forms of Amentiferae. Brit- tonia 25: 334-35 COMPARATIVE POLLEN MORPHOLOGY AND ITS RELATIONSHIP TO PHYLOGENY OF POLLEN IN THE HAMAMELIDAE! MICHAEL S. ZAVADA?? AND DAVID L. DILCHER? ABSTRACT Data on pollen morphological features from 200 species in 20 families commonly included in the Hamamelidae and particular species The basic descriptive analyses presented are snows d some variability, and each of the species was sco in the Anacardiaceae and Salicaceae are presented i in this paper. derived fr hirty pollen characters red for these characters. These da linkage strategy а у these analyses. С . Three mmiaceae, and aceae. The Liq yricaceae. The Balanopaceae and Nothofagus are some- what isolated and peripheral ane but hold dann in both linkage strategies. Thirty pollen char- acters of 78 taxa were analyzed using o produce a cladistic tree. The outgroup used was Tetracentron. Three phylogenetically related died sorted out, which are the same as those already recognized in the Groups I, II, and III mentioned above. Group I occurs at the base of the tree (primitive), and Group II occurs as intermediate between Groups I and III (derived). In general, these data support the relationships suggested by к for the Hamamelidae, based upon vegetative and floral features and the о of Cronqu This survey of pollen in the Hamamelidae was initiated with three primary goals in mind: (1) to provide comparative morphological data for assessing the relationships of fossil-dispersed pollen with possible hamamelidaceous affinity, (2) to assess at which taxonomic level pollen characters of extant hamamelidaceous taxa are useful in determining taxonomic position, and (3) to assess the phylogenetic relationships of taxa within the Hamamelidae as elucidated by pollen morphology and ultrastructur e y (SEM)] from pub- ed pollen data from 42 previously uninvestigated taxa. We have amassed pollen data from 20 of the 24 families recognized by Cronquist (1981), P over _ species (Didymelaceae, Urticaceae, Мог and Cecropiaceae are excluded о, the cladistic and phenetic analyses). In addition, pollen from the Anacardiaceae and Salicaceae are in- cluded in the analysis In a survey as broad as that presented here, it is often difficult to decide which taxa to include or exclude. Members of the Hamamelidae have been placed in a number of subclasses, and a comprehensive pollen survey of all the families of the different classifications was not attempted here. We decided to use the Cronquist (1981) system as a starting point and introduced into our analysis selected taxa from other subclasses that have been suggested to be phylogenetically related (e.g., Thorne, 1973) 1 ti dhi Itati wish to give special thanks to Robert Scl on the cladistic and phenetic analyses. We also thank Karl Longstreth, Linn Bogle, , and Greg Anderson for their e manuscript. We thank Thomas Delendick of the to D. L. Dilcher artment of Biology, Indiana University, Bloomington, ad 3 Current address: rea of Botany, University of the Witwatersrand, 1 Van Smuts Avenue, Johan- nesburg, 2001 South Africa ANN. MissouRi Bor. GARD. 73: 348-381. 1986. 1986] ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 349 METHODS All pollen was removed from dry herbarium material. A list of specimens used in this study and the herbarium data are presented in Appen- dix I. The pollen was first acetolyzed and pre- pared for TEM by dehydration in an ethanol series and embedded in Spurr’s Low Viscosity Resin. Sectioning was done on an MT-2 ultra- microtome; the sections were post-stained in uranyl acetate and lead citrate for 15 minutes, respectively, and viewed on a Phillips EM-300. Acetolyzed pollen was prepared for SEM by mounting the pollen on stubs with the high vac- uum wax Apiezon W-100 and coated with gold- palladium. A Cambridge Stereoscan scanning electron microscope was used for viewing. e terminology used to describe aperture type, tectal and supratectal ornamentation, and pollen wall structure follows Faegri and Iversen (1964). Zavada (1984) discussed the criteria used to identify wall layers. Identification of wall layers relies not only on staining properties but also on some developmental data. However, very few developmental data on hamamelidaceous taxa are available; thus, pollen descriptions in this paper are based on the morphology and staining properties with the electron micrograph stains uranyl acetate and lead citrate. DESCRIPTIVE PALYNOLOGY TROCHODENDRALES Trochodendraceae (1 genus, 1 species). This taxon is restricted to Japan and Formosa. Pollen of Trochodendron aralioides Sieb. & Zucc. is tri- colpate, oblate to slightly prolate. Pollen is about 24 um in polar diameter and 20 um in equatorial diameter (Erdtman, 1952). Exine sculpturing is reticulate. Wall structure is tectate-columellate with a relatively thick footlayer. A thin endexine Tetracentron sinense Oliv. ex Hook. is tricolpate, spherical to slightly prolate. Pollen averages about 15 um in diameter. Exine sculpturing is reticulate (Walker, 1976). Pollen wall structure is tectate- columellate with a footlayer that is underlain by an endexine, which thickens slightly in the aper- tural region (Praglowski, 1974). CERCIDIPHYLLALES Cercidiphyllaceae (1 genus, 1—2 species). Cer- cidiphyllum japonicum Sieb. & Zucc. is endemic to Japan. Pollen is tricolpate (Walker, 1976; this study) or pericolpate (Walker, 1976; Fig. 1). Pol- len averages 36 um in polar diameter and 31 um in equatorial diameter (Erdtman, 1952). Pollen shape is spherical to prolate. Exine sculpturing is reticulate (Fig. 1). Pollen wall structure is tec- tate-columellate with a footlayer of varying thickness (Fig. 2). A thin endexine that does not thicken in the apertural region is evident (Prag- lowski, 1974; Fig. 2). EUPTELEALES Eupteleaceae (1 genus, 2 species). Euptelea pleiosperma Hook. f. & Th. is found in China and India, and Е. polyandra Sieb. & Zucc. is endemic to Japan. Pollen of E. polyandra was studied and is predominately tricolpate (Walker, 1976; this study); however, pericolpate grains have been reported (Praglowski, 1974; Walker, 1976). Shape varies from oblate to prolate (Fig. 3). Pollen ranges between 29 and 39 um in polar diameter (Erdtman, 1952). The exine sculpturing is reticulate (Fig. 3). Pollen wall structure is tec- tate-columellate with a thin footlayer (Fig. 4). Endexine underlies the footlayer and is thin in nonapertural regions but thickens somewhat in apertural regions (Praglowski, 1974; this study). HAMAMELIDALES Platanaceae (1 genus, 7-12 species). Platanus is widely distributed in temperate and subtrop- ical regions of the Northern Hemisphere. Pollen of seven species was studied with SEM and TEM (Appendix I) and vas studied by Hesse (1978) using TEM. А (1982) presented а methodological study of the pollen of some re- cent and fossil species of Platanus. Platanaceae are a stenopalynous family. All species studied are treated together due to their morphological and ultrastructural similarity. ollen is tricolpate (Figs. 5, 6, 8, 11), the colpi are lenticular (Fig. 5) to slightly ovoid (Fig. 11). The grains are spherical to slightly prolate in equatorial view and circular to semi-angular in polar view and 19-24 um in polar diameter and 17-22 um in equatorial diameter. The exine sculpturing is reticulate (Figs. 5, 6, 8, 9, 11), the wall structure is tectate-columellate with a foot- layer the same thickness as the tectum (Figs. 7, 10, 12-14). A thin endexine may be present (Figs. 7, 10, 13, 14); however, this may be an artifact of staining in some taxa (Fig. 12). Myrothamnaceae (1 genus, 2 species). This 350 ANNALS OF THE MISSOURI BOTANICAL GARDEN (VoL. 73 1986] family of xerophytic shrubs consists of Myro- thamnus moschatus Baill. in а апа М. flabellifolia Welw. in South Afric he pollen of Myrothamnus HN Bail. and M. flabellifolia Welw. is usually shed in tet- rahedral tetrads (occasionally in tetragonal tet- rads) that range between 23 and 25 um in di- ameter for M. hatus and 24 and 26 um for M. flabellifolia (Figs. 15, 19, 21). Single pollen grains of M. moschatus range between 11 and 14 um in polar diameter and 12 and 15 um in equa- torial diameter. The pollen of M. flabellifolia ranges between 12 and 15 um in polar diameter and 13 and 16 um in equatorial diameter. Pollen in both taxa is spherical and triporate (Fig. 21). The three somewhat circular, ill-defined pores are symmetrically positioned and distally offset from the equator. The exine sculpturing consists of clavate processes that are ornamented with minute papillae (Figs. 16, 18, 20, 22). The wall averages 0.7 um in thickness and is intectate. The pollen wall structure consists of an outer colu- mellate layer. The columellae are the clavate pro- cesses easily observed with SEM (Figs. 16-20) and the bases of the clavate columellae are fused to a footlayer (Figs. 18-20) that is at times dis- continuous (Fig. 18). Beneath the footlayer is a thin endexine that is not thickened adjacent to the apertural region (Fig. 18). In the region where pollen grains are joined in the tetrad, the colu- mellae are short, stout pegs, and their apices are joined and hold adjacent pollen together (Figs. 17, 19). The only interspecific differences are that the clavate processes in Myrothamnus moschatus are more widely spaced than in M. flabellifolia (com- pare Figs. 16 and 22). In addition, the footlayer in M. moschatus is usually thicker than the foot- layer of M. flabellifolia. However, due to the lack of fertile herbarium material, intraspecific vari- ation is unknown and we do not know how con- sistent these differences are within wide-ranging populations. ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 351 Hamamelidaceae (28 genera, 100 species). This family has a cosmopolitan distribution. Pollen of 12 genera (out of 28) and 16 species (out of 100) was studied ultrastructurally. Bogle and Philbrick (1980) examined 28 genera using SEM. Hesse (1978) described the ultrastructural char- acteristics of three genera and four species, and Endress (1977) illustrated eight taxa using SEM. The taxa investigated are treated here on the subfamilial level. Liquidambaroideae. Pollen of Liquidambar styraciflua L. and Altingia obovata Merr. & Chun were studied with TEM and SEM. Bogle and Philbrick (1980) examined L. orientalis Mill. and A. chinensis Oliv. ex Hance using SEM. The pol- len is spherical, periporate, the pores circular, often with islands of ektexine on the pore mem- brane (Fig. 31), and 32-58 um in diameter. Exine sculpturing is finely reticulate with small supra- tectal rugulae, the wall structure is tectate-colu- mellate, with a footlayer that is the same thick- ness as the tectum (Figs. 31, 32). The columellae are thin and short in Liquidambar (Fig. 31) but somewhat thicker in A/tingia (Fig. 32). The foot- layer is often underlain by a thin, less electron- dense wall layer (presumably endexine) that does not considerably thicken in the apertural region (Figs. 31, 32). Rhodoleiodeae. Pollen of this monotypic subfamily (Rhodoleia championii Hook. f.) was studied with TEM (this study) and SEM (Bogle & Philbrick, 1980). Pollen is tricolpate, shape is subprolate, and 16-26 um in equatorial diameter and 20-29 um in polar diameter (Bogle & Phil- brick, 1980). Exine sculpturing is finely reticulate (= scrobiculae of Bogle & Philbrick, 1980), and the wall structure is tectate-columellate with a footlayer as thick as the tectum (Fig. 29). The endexine appears as a thickened layer in the aper- tural region only (Fig. 29). UR Pollen of Exbucklandia populnea (R. Br. ex Griff.) R. W. Brown was studied with SEM mae & Philbrick, 1980; this = FIGURES 1-7.—1. Scanning electron micrograph of Cercidiphyllum japonicum sowing reticulate exine -— showing reticulate exine sculpturing and pericolp icana; x Transmission о е footlayer, ‘and possible endexine (En); x1 ate apertures; х 1,300.—4. Tr of E. polyandra showing tectate-columellate wall structure, thin footlayer, and less dense e 5. Scanning electron micrograph of Platanus cda x 3,100.— 6. —7. aph of P. acerifolia showing tectate-c UOLUTTICIUAL vv 11 of Euptelea тобаа Scanning electron microgra лишена 352 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 8-14.— 8. Scanning electron micrograph of Platanus occidentalis, x 3,200.— 9. 09 Meis micrograph of P. not an uncommon occurrence in this taxon; x 5,800. — 1 ransmission electron micrograph of P. occidentalis showing tectate-columellate wall structure, footlayer, and pos: ssible endexine; x12,600.— 11. Scanning electron micrograph of P. kerrii; x 3,100.—12. T mission eh : е : ry evident .— 14. Transmission electron шол of Р. racemosa showing identical wall structure to other Platanus species; х 16,200. 1986] study) and TEM (this study). Pollen of Mytilaria laosensis Lecomte and Chunia bucklandioides H. T. Chang was studied with SEM by Bogle and Philbrick (1980). Pollen is tricolpate and spherical to prolate in all three taxa and 26-32 um in equatorial di- ameter and 23-37 um in polar diameter in Е. populnea, and 26—37 um in equatorial — and 29-38 um in polar diameter in C. buckla dioides. Exe кишш 1 is ишиае in Ex- finely reticulate in Chunia А Pol- len wall structure of E. populnea is tectate col- umellate with a thin footlayer (Fig. 23). No en- dexine is evident. Disanthoideae. Pollen of Disanthus cercidi- folius Maxim. is tricolpate, spherical to prolate, ranging between 22 and 23 um in equatorial di- ameter and 22 and 33 um in polar diameter, exine sculpturing is reticulate. Wall structure is tectate-columellate with a thin footlayer (Fig. 26). The endexine is thin in nonapertural regions but thickens in the apertural region. Hamamelidoideae. This is the largest sub- era. All genera were studied with SEM (Bogle & Philbrick, 1980) and six gen- era were investigated with TEM [Hamamelis (3 spp.), Lorapetalum (1 spp.), Corylopsis (2 spp.), Fothergilla (2 spp.), Matudaea (1 spp.), this study; and Parrotia (1 sp.), Hesse, 1978]. In most taxa, pollen is tricolpate but varies from tricolpate to pericolpate in Fothergilla, and daea. Pollen varies from slightly oblate to mostly spherical and prolate and is 16-53 um in di- ameter. Exine sculpturing is finely reticulate to coarsely reticulate, often with supratectal ver- rucae. In all taxa studied with TEM, pollen wall structure is tectate-columellate with well-devel- oped footlayers (Figs. 25, 27, 28, 30). Endexine is evident in Fothergilla monticola Ashe (Fig. 28), Corylopsis pauciflora Sieb. et Zucc. (Fig. 27), and Matudaea hirsuta Lundell (Fig. 30), but ab- sent in Parrotia persica (DC.) C. A. Mey. (Hesse, 1978), Hamamelis japonica Sieb. et Zucc. (Fig. 24), and Lorapetalum chinense Oliv. (Fig. 25). DAPHNIPHYLLALES Daphniphyllaceae (1 genus, 35 species). This monotypic order is distributed in eastern Asia and the Malay Archipelago. Five species were investigated with SEM and TEM (Appendix I). ZAVADA & DILCHER C HAMAMELIDAE POLLEN MORPHOLOGY 353 Pollen is tricolpate (Figs. 33, 35, 39, 41), spher- ical to oblate in equatorial view (Figs. 33, 39) and circular in polar view (Figs. 35, 37), 17 um in polar diameter and 21 um in equatorial di- ameter (Erdtman, 1952). Exine sculpturing is somewhat psilate to verrucate (Figs. 34, 36, 38, 40, 42), the wall structure is tectate-columellate with a thin footlayer, and the tectum is occa- sionally perforated with small channels (Figs. 36, 40, 41). Endexine is present and is thin in non- apertural regions (Figs. 36, 40) but thickens con- siderably in the apertural region (Figs. 34, 38, 42) DIDYMELALES Didymelaceae (1 genus, 2 species). This fam- ily is restricted to Madagascar. Pollen is tricol- pate with two pores per colpus: one in the distal hemisphere and one in the proximal hemisphere, oblate to spheroidal, about 21 um in equatorial diameter and 23 um in polar diameter, exine sculpturing is reticulate (Erdtman, 1952). Little is known of pollen wall structure but it appears to be tectate-columellate with a footlayer. Oc- currence of endexine is unknown (Erdtman, 1952) EUCOMMIALES Eucommiaceae (1 genus, 1 species). Eucom- mia ulmoides Oliv. is found in China. Pollen is prolate and tricolpate: one colpus narrows near the equator but expands in the polar region (Fig. 43), the two other colpi appear straight (Fig. 44). The pollen is about 42 um in polar diameter and 31 um in equatorial diameter (Erdtman, 1952). Exine sculpturing is minutely spinulose to ver- thicken in the apertural regions (Figs. 45-47). LEITNERIALES Leitneriaceae (1 genus, 1 species). This monotypic family occurs in the southeastern United States (Leitneria floridana Chapm.). Pol- len was studied with SEM and TEM. Pollen is tricolporate (occasionally tetracolporate), oblate to spheroidal (Fig. 48), 26-28 um in size (Erdt- man, 1952). Exine sculpturing is minutely ver- rucate (Fig. 49); the tectum is microperforate (Fig. 50), the wall structure is tectate-columellate with 354 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 19 : | | po A ox ; 20 FIGURES 15-20.— 15. Scanning electron micrograph of Myrothamnus moschatus showing tetrahedral tetrad with d a exine sculpturing; x1,300.— 16. Scanning electron micrograph of M. moschatus showing clavae with minute papillae; x15,500.— 17. Transmission electron micrograph of M. flabellifolia showing eras of pollen in a tetrad. Note thin endexine (arrow); x8,094.— 18. Transmission electron micrograph of M. fla- bellifolia through the distal face of a pollen grain adjacent to the apertural region (AR). Note iin. а (arrow), which does not thicken near the aperture and discontinuous footlayer (е.р., above arrow); х 6,840. — 19. Transmission electron micrograph of M. moschatus showing attachment of pollen in a tetrad. Мор inter- 1986] ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY FIGUR clavate sculpturing, and porate aperture (A); x RES 21, 22.— 21. Scanning electron micrograph of M. flabellifolia showing a tetrahedral pollen tetrad, x 2,0 —22. Scanning electron micrograph showing clavae with minute papillae. Note that the clavae are more crowded than in M. moschatus (compare with Fig. 16); x15,500. a very thin footlayer (Fig. 50). An endexine is present in nonapertural areas and thickens con- siderably in apertural regions. URTICALES Barbeyaceae (1 1 genis. 1 d Barbeya rtheast Africa. Pollen was studied with SEM d TEM. Pollen is tricolporate (pores are equatorially elongate and wider than the colpi), oblate to spheroidal (Fig. 52), about 26 um in polar diameter and 22 um in equatorial diameter (Dickison & Sweitzer, 1970; this study). Exine sculpturing is scabrate (Figs. 52, 53). Pollen wall structure is tectate- columellate with a thick tectum traversed by mi- croperforations (Fig. 51), the tectum is underlain y short, stout columellae (Fig. 51). The colu- mellae rest on what appears to be endexine (Fig. 51). The footlayer is essentially absent. The en- dexine thickens somewhat in apertural regions. Ulmaceae (15—17 genera, 150—200 species). The family is distributed throughout temperate and tropical regions of the world and is usually divided into two subfamilies, Ulmoideae and Celtidoideae. Some pollen has been studied with SEM and TEM (Zavada, 1983) Ulmoideae. The pollen is 4—7-stephanopo- rate, the pores circular to elliptical, spheroidal to respond to thick areas of the pollen wall (Fig. 54; except in P/anera, which is psilate). Pollen wall structure consists of a tectum occasionally tra- versed by minute channels, granular infrastruc- uu and a thin continuous footlayer (Fig. 55). o endexine is evident (Zavada, 1983). n is 2—5-stephano- verrucate (Fig. 56), and wall structure consists of an outer tectum that has minute channels. The infrastructure consists of a middle layer of anas- tomosing rods and an inner layer of irregular- shaped columellae, which rest on a thin basal layer (Fig. 57; Zavada, 1983). Cannabaceae (2 genera, 2 species). This family occurs in north temperate regions. Pollen of Cannabis sativa L. was studied with SEM and TEM. Pollen is triporate with annulate pores; the <— cytoplasmic channels (arrow) and the thin endexinal apertural membrane (AR); x3,876.— 20. Transmission electron micro moschatus of the distal face of a pollen grain showing clavae with minute papillae (P) and thick basal layer. In this particular section, endexine does not stain differently than ektexine; х 15,960. 356 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 23-32. t h ollen of th lid gI of the tectate-columellate wall structure. — 23. E ee x 8,040.—24. Hamamelis japonica, x12,310.—25. Loropetalum chinense, x16,540.— 26. Disanthus cercidifolius; x 8,040. — 27. Corylopsis pauci- endexine (En, arrows); x .—29. Rhodoleia championii, note endexine (En) adjacent to apertural region (A); x 16,540.—30. Ма tudaea hirsuta; x16,540.—31. Liquidambar styraciflua (A = apertural area); x 5,600.— 32. Altingia obovata, note thin endexine (En); x12,310. 1986] annulus around the pore is formed by an elab- oration of the infrastructural layer in this region (Text-Fig. 7F), (Fig. 58), oblate, 30 um in equa- torial diameter. Exine sculpturing is finely spi- structures) with a thin footlayer (Fig. 59) and what appears to be a very thin endexine. Moraceae (including the Cecropiaceae) (40 genera, 1,000 species). This family is wide- spread in subtropical and tropical regions of the world. For the size an of this family, little is known palynologica ally. However, a few taxa have been studied ultrastructurally (Niez- goda & Nowaczyk, 1976; Hamilton, 1976; Punt, 1978; Punt & Eetgerink, 1981). Pollen is variable and, in the taxa studied, it is di-stephano- or peri- porate, spherical to oblate, and averages about 30 um in diameter. Exine sculpturing can be ru- gulate to scabrate with small supratectal spinules. Pollen wall structure (only studied in Dorstenia sp.; this study) is tectate-granular with a bilay- ered basal layer (Fig. 73). The tectum is occa- sionally traversed by minute chann Urticaceae (45 genera, 700 iind This family is widely distributed in subtropical and tropical regions of the world. Pollen wall ultra- structure is unknown. Pollen is stephano- or peri- porate, predominantly oblate, and 10-20 um in To (Erdtman, 1952). Exine sculpturing is cabrate to rugulate with supratectal spinules did. 1976). rt JUGLANDALES Rhoipteleaceae (1 genus, 1 species). Rhoiptelea chiliantha Diels et Hand is native to southeast Asia. Pollen has been studied with SEM and TEM by Stone and Broome (1971). The pollen is ob- late, tricolporate with very short colpi appearing to approximate the triporate condition, averag- ing 27 um in equatorial diameter. Exine sculp- turing is scabrate, pollen wall structure is tectate- granular, and the infrastructural area is underlain by a footlayer. The tectum is traversed by minute pollen was studied with SEM and TEM by Stone and Broome (1975). They recognized the four basic pollen types described below ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 357 Engelhardtia-type — Pollen is triporate (varies from 2 to 8), oblate, 14—26 um in diameter, and lacks pseudocolpi. Platycarya-type— Pollen is triporate (varies from 2 to 5), oblate, 15-16 um in equatorial o with pseudocolpi Carya-type — Pollen is —Q Neia from 1 to 6), obi. and 33-66 um in diam Pterocarya-type— Pollen is acq NN oblate to spheroidal, and 28-50 um in equatorial diameter. Pollen wall structure in all of the above taxa is tectate-granular with a relatively thick foot- layer. The tectum is occasionally traversed by minute channels. A thin endexines 18 оћеп present that d thicken in th & Broome, 1975). rY MYRICALES Myricaceae (3 genera, 50 species). This fam- ily is distributed in temperate and subtropical regions. Pollen is triporate, oblate, about 26 um in equatorial diameter. Exine sculpturing con- sists of small scabrae (Praglowski, 1962; Lieux, 1980). Pollen wall structure of Myrica aspleni- folia L. was studied with TEM (Fig. 64), and Coetzee and Praglowski (1984) investigated 11 species of Myrica with SEM and TEM and found that the pollen in those species is 21-25 um in polar diameter, 25-35 um in equatorial diame- ter. The pollen wall structure is tectate-granular; columellae are often discernable atively thin footlayer is present; no endexine is evident. FAGALES Balanopaceae (1 genus, 9 species). This fam- ily is distributed in regions of the southwest Pa- cific (e.g., New Caledonia). Pollen of Balanops vitiense (A. C. Smith) Hjedruqvist is 3-5-colpate, oblate, averaging 35 um in diameter (Erdtman, 1952). The exine sculpturing consists of small spinules (Fig. 68), the wall structure is tectate- granular to columellate (Fig. 69), and the tectum is traversed by microperforations. The infra- structure consists of both irregular-shaped gran- ules and robust columellae om rest on an uneven tlayer (Fig. 69). and The footlayer is underlain by an endexine that, 358 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 33-38.— 33. Scanning electron micrograph of Daphniphyllum calcinum; x 2,200.— 34. Transmission electron micrograph of D. calycinum showing tectate-columellate wall structure, footlayer, and endexine that thickens in apertural region (En); also note tectal perforations (small arrow); x 7,000. — 35. Scanning electron 1986] ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 359 FIGURES 39-42. mission electron micrograph o perforation (small x 6,000.—42. Transmission tro erture region (A) with dd endexine (En); x15 arrow) ae endex at times, is discontinuous and thickens some- what in the apertural region (Fig. 69). a enera, 800 species). The family has a cosmopolitan distribution. It 1s usu- ally separated into three subfamilies. —39. Scanning electron micrograph of Pere himalayense, x3,200.— . hi 40. Tra ima alayense showing tectate-columellate wall structure, thin footlayer, cia x 25,500.—41. Sca nning electron micrograph of D. /aurinu iu of D. laurinum showing tectal perforation (small arrow), = ‚540. Castaneoideae. The pollen is tricolporate, prolate, and averages 17 um in and 11 um in equatorial diameter. Exine sculp- turing is striate. The wall structure is tectate- columellate with a well-developed footlayer, the _— T of D. gracile; x 2,900.— D. neilgherrense, x 3,200. — (A) with endexine (En); x10,000. 36. Transmission electron micrograph of D. gracile showing thin endexine non-apertural region and a tectal perforation (small arrow); x8,5 500.— 37. Scanning electron micrograph of 8. Transmission electron micrograph of D. neilgherrense showing aperture region 360 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 —43. Scanning rg micrograph of Eucommia ulmoides showing the colpus that narrows 4. Scanning electron micrograph of E. ulmoides showing straight, slit-like FiGURES 43-47. in the equatorial region; x 3,000. ана brie micrograph of E. ulmoides showing atectate ektexine underlain by ,000. —46. ulmoides in the hemisphere with slit-like apertures showing atectate wall, endexine (En), and rugulate (Ru) sculpturing (compare to Fig. 44); x 15,000.—47. Trans- mission electron micrograph of E. ulmoides in the region of the colpus that narrows equatorially showing small spinules (Sp) and wall structure; x 10, — FIGURES i I — 48. Scanning electron micrograph of Leitneria floridana; х 2,400.—49. Scanning electron micrograph of L. floridana showing exine sculpturing; х 6,400.— 50. Transmission electron micrograph of L. floridana к: tectate-columell late wall structure, thin footlayer, endexine (En), and tectal perforation (small ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY micrograph of B. oleoides; x 3,500.— : showing trirudiate thick region in the polar region; this characteristic is reminiscent of some fossil normapolles dispersed pollen types; x 3,200. 362 tectum is traversed by small perforations. A thin endexine underlies the ektexine in nonapertural regions but thickens considerably in apertural regions (Crepet & Daghlian, 1980; Miyoshi, 1983). Fagoideae. The pollen is tricolporate in Fa- d stephanocolpate in Nothofagus, and is spherical-oblate (Fagus) to oblate (Nothofagus). agus pollen is about 37 um in polar diameter and 38 um in equatorial diameter; that of Noth- ofagus is about 20 um in polar diameter and 32 um in equatorial diameter. Exine sculpturing in Fagus pollen consists of minute striations. Pollen of Nothofagus has died spaced spines. Pollen wall structure of Fa tecta a well-developed iss the tectum is tra- versed by small perforations. A well-developed endexine, which thickens in the apertural re- gions, is also evident. Pollen wall structure of Nothofagus is tectate-granular with a relatively thick footlayer that is underlain by a well-de- veloped endexine. The а о reni in the apertural regions. The is im forate ек Daghlian, У та 1981, duisi The pollen is tricolporate, pro- late to spheroidal, about 28 um in polar diam- eter. The exine sculpturing is scabrate to rugu- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 late, the wall structure is tectate-columellate with a thick footlayer that is underlain by an endexine. The endexine thickens in the apertural regions (Smit, 1973; Crepet & Daghlian, 1980; Solomon, 1983a, 1983b). Betulaceae (6 genera, 120 species). This fam- ily is distributed primarily in temperate regions of the Northern Hemisphere. Pollen is triporate (Betula, Ostrya, Corylus, Ostryopsis, Carpinus) to stephanoporate (A/nus), d о ит in equatorial diameter. Exine sculpturing is mi- nutely scabrate to slightly rapit (Fig. 66; Erdt- man, 1952; Adams & Morton, 1972; Kedves, 1979; Crane, pers. comm.), Fus wall structure is tectate-granular (Fig. 67; Erdtman, 1969; Row- ley, 1981; Rowley & Prijanto, 1977; this study); however, columellate-like structures are often present in the infrastructural layer. (Dunbar & Rowley, 1984). The tectum is traversed by mi- nute channels (Fig. 67). The footlayer is under- lain by a very thin endexine that does not thicken in the sides regions in oe however, in Betula s separated from the кошл to form an atrium. CASUARINALES Casuarinaceae (1—2 genera, 50 species). This family is native to some southeast Pacific islands, FIGURES 54-59.— . Scanning electron micrograph of Ulmus glabra, х1 ‚840.— > 5. Transmission electron 0 aph of C. sativa showing tectum, granular infrastructure, and thin 60. x2, бо шуш... А thin, less dense layer is Me evident in this taxon (endexine); x10,2 FIGURES 60-67.—60. Scanning electron micrograph of d stricta; x1, us —61. Transmission electron ck tectu micrograph of C. stricta showing thi thin footlayer; x 10,094. — x15,540.— 66. Scanning electron micrograph of Betula alba; x ransmissio of B. alba showing minute tectal channels (small arrow) and granular infrastructure: x18, FIGURES 68-73. els (small arro n electron micrograph 960. x 3,400. — 69. Transmission electron granular to columellate See озшш footlaye Pistacia terebinthus showing p orate (ulceroid-like) ap- ,500.— grain); x15,540.— 70. Scanning electron microgra ertures, finely reticulate tectum with small со — 68. Scanning кашса micrograph of р vitiense showing spinulose exine sculpturing; горгар , and dark- -staining endexine (unacetolyzed pollen ransmission electron micrograph of P. terebinthus showing tectate- columellate wall ‘Structure and thick footlayer; x 12, 500. —72. Pollen of d deltoides. (a) T small, ulceroid-like pores (arrows); x4, 500. (b) Transmission electron micrograph showing tectate- abun ellate wall structure and very thin footlayer; х 18,500.— . Transmission electron micrograph of Dorstenia showing tectal perforations (arrows) and granular infrastructure; x 21,000. 1986] ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 363 ^ nye A E e H КАРЫ \х a? B LY. ыл Дд “д; | Ав P СА Та ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 1986] ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY т 79%, ey Wt i "EUIS MEE оды о. 4 w Чазаа 365 366 New Guinea, Sumatra, and Australia. The pollen is triporate, oblate, averaging 29 um in equatorial diamet о 1970) and 23 um in polar dia nnd Figs. 62). Exine sculpturing con- sists of small м or scabrae and sometimes appear rugulate, the wall structure of Casuarina is tectate-granular (Figs. 61, 63), the thick tectum is traversed by minute channels (Fig. 61). A thin footlayer is present that separates from the gran- d layer in the apertural region to form an o endexine is evident, but Coetzee and P (1984) reported a thin endexine in the eight species they investigated. Pollen wall structure in Gymnostoma is identical to that in Casuarina; however, the footlayer remains in contact with the granular layer in the apertural region (this study). POLLEN CHARACTERS USED IN THE ANALYSES The pollen characters chosen for the phylo- genetic and similarity cluster analysis of Ham- amelidae are those that vary among the taxa in- vestigated. Characters that are present but do not vary are excluded. The characters are of two types: (a) those that are present (1) or absent (0) and no other variation of the character exists, and (b) those that are present (1) or absent (0) and a dependent associated character state also exists that can thus be determined to be present or absent, for example, pollen shed in monads; if pollen is shed in some other unit, the monad character is scored absent (0) and the character state it possesses (e.g., tetrads) is scored present (1). Thirty character states that vary are recog- nized for pollen of the Hamamelidae (see legend Appendix II). The character states are discussed below, in addition to their association with taxo- nomic units above the generic level. POLLEN UNIT (TWO CHARACTER STATES) All species of Hamamelidae shed their pollen in monads except Myrothamnaceae, which shed pollen in tetrads. POLLEN SHAPE (THREE CHARACTER STATES) Pollen shape (prolate, spherical, oblate) varies e equatorially positioned pores are midi Piin ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 (e.g., Juglandales, Casuarinales), and taxa with peri-porate or -colpate pollen are generally spherical (e.g., Liquidambaroideae). If any species varies between two or three of the shape classes, both or all shape characters are scored as present (1). APERTURE TYPE (NINE CHARACTER STATES) he predominant aperture type is tricolpate (e.g., Trochodendrales, Hamamelidales). The tri- porate and stephanoporate (greater than three equatorial-positioned pores) are common in the Juglandales, Myricales, Casuarinales, Ulmaceae, and Betulaceae. The pericolpate and periporate condition occurs in the Liquidambaroideae, Moraceae, and Urticaceae. Stephanocolpate pol- len occurs in two taxa: Balanops vitiensis (A. C. Smith) Hjed tand spp. The tri- colporate type i is restricted to the Fagales, Leit- neriales, and possibly Rhoipteleaceae (Stone & Broome, 1971), and the Barbeyaceae. The di- porate type is found in some Moraceae and the Celtidoideae. SUPRATECTAL SCULPTURING (SIX CHARACTER STATES) Exine sculpturing is highly variable in many plant groups, including the Hamamelidae. In the semitectate (reticulate) pollen of the Trocho- dendrales and Hamamelidales, the tectum is often predominantly in the Juglandales, Casuarinales, Myricales, and Betulaceae. Scabrae occur in the Urticales and the rugulate-verrucate sculpturing type in Urticales, Juglandales, and Fagales. Cla- vate sculpturing occurs only in the Myrotham- naceae, and the striate type only in the Fagaceae (e.g., Fagus, Castanea). WALL STRUCTURE TECTUM (FOUR CHARACTER STATES) Pollen of all taxa have a tectum except for Myrothamnus, which is intectate. The pollen wall structure of Eucommia is unique in having atec- tate pollen (cf. Degeneria). Semitectate (reticu- late) pollen is easily recognizable with light mi- croscopy and occurs in the Trochodendrales and Hamamelidales. In Daphniphyllales, Leitneri- ales, Barbeyaceae, and Fagaceae the tectum is more continuous and is traversed by microper- forations discernable only by SEM. In some Ur- ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 1986] ticales, Juglandales, Myricales, and Casuarinales delia т, the tectum appears imperforate with SEM; how- UNWEIGHTED PAIR GROUP (UPGMA) STRATEGY ever, TEM reveals that the tectum is traversed оо е е т AN by microchannels. f L . | TETRACENTRON SINENSE ULM CERCIOIFOLIUS RHODOLEIA CHAMPIONII INFRASTRUCTURE (TWO CHARACTER STATES) EXBUCKLANDIA POPULNEA "i AMAT ЭВР r ‚ LATANU The infrastructural layer of pollen walls in an- FOTHERGILLA MONT ICOLA giosperms is usually columellate or granular. The 08:09515 spp columellate type, which consists of cylindrical Eten deg “APONTE : - TUDAE columns supporting the tectum, is found gener- LTQUIDANBAR STYRACTFLUA ally associated with the semitectate (reticulate) DORSTENIA SP нЕ анна і i ы DAP YLLUM 1 AYENSE or microperforate tectal condition (i.e., Troch- BAI EM и odendrales, Hamamelidales, Daphniphyllales, DAPHNIPHYLLUR CA CA URINUM : : F LTNERIA FLOR itneriales, and Fagales). The granular infra- FAGUS SPP es 1 S structure (loosely defined here), which can be QUERCUS SPP S VERTICILLATA composed of anastomosing rods, spherical gran- CASTANOPSIS SPP ules, or irregular-shaped columns, is generally ОСАО SPP Р s . А POPULUS DELTOIDES associated with taxa that have microchannels(i.e., PISTACIA TEREBINTHUS — . . AAA A Urticales, Juglandales, Myricales, and Casuar- AMPEL OERA CUBENSIS inales). APHANANTHE ASPER on PHIL IPPINENSIS ETACHME ARISTA EQ IS FOOTLAYER (ONE CHARACTER STATE) SIROMMTERA,SUBEQUAL IS А А : : . MIRANDACELTIS E A footlayer is present in all taxa investigated CANNABIS SAT die » А . r * PARASPONIA ANDERSONII except Barbeya oleiodes, in which it is repre- АР Pisis sented by a very thin discontinuous layer, апд "ITE EA DAVIDI у, much of the infrastructural layer rests directly ом ҮЗ ОМ RHAMNOIDES : T the endexine. PLANERA AQUATICA Р А SOB LL ACEA A MYRICA ASPLENIFOLIA AAA ENDEXINE (THREE CHARACTER STATES) P TEROCARYA | CARYA SPP ! RI GHIAN ! The presence, absence, and thickness i in aper- di ALFAROA QUANACASTENSIS 1 tural and nonapertural regions varies ALFAROA GUATEMALENSIS among taxa. There appears to be no apparent СБ AMERICANA y relationship among taxonomic units and the oc- EIOS Ora SPP RIGIDA | h ENGELHARDT LA l currence of endexine. Three character states are ALFAROA COSTARICENSIS - 1 1 : FAROA MEXICANA recognized for this wall layer: (1) presence Ог ALEAROA MEXICANA | | absence, (2) endexine is thicker in the apertural Шр VITIENSIS — regions than nonapertural regions, and (3) en- '©"О 55 $PP | | 1.00 75 50 25 .00 dexine does not thicken in the apertural region relative to nonapertural region. If endexine is xT-FIGURE 1. absent, all three character states are scored (0). © strategy. If endexine is present only in apertural region, character states one and three are (0) and two is (1) (this occurs only in Rhodoleia). If endexine ing the ри pair sr bic method e is present but does not thicken in apertural re- Text-Fig. 1) a gion, character states one and three are (1) and Text-Fig. 2). Both produced similar EAN sults. Three major groups are discernable, designat- ed Group I, II, and III. Group I in both analyses NEA wh Carmi a consists of Trochodendraceae, Cercidiphylla- ceae, Eupteleaceae, Platanaceae, Hamamelida- ceae (including Liquidambaroideae), Eucom miaceae, and Myrothamnaceae. Group I taxa are prolate to spherical in equatorial view and cir- cular in polar view. The tricolpate aperture pre- Similarity cluster analysis, two is (0). The similarity cluster analysis used the Baroni- Urbani-Buser Coefficient. Using the Cluster Analysis Package af Archer, Hohn, and Horo- witz (1984), two d g g 368 о ANALYSIS OF HAMAMELIDAE NI-URBANI-BUSER rl COMPLETE LINKAGE STRATEG 1.00 7S .50 м л © e TETRACENTRON SINENSE DISANTHUS CERCIDIFOLIUS EID POPULNEA HAMAMELIS SPP RNODOLELA CHAMP TONI I LOROPETALUM CHINENSE SALIX SPP PLATANUS SPP FOTHERGILLA MONTICOLA а wn CORYLOPSIS ОНА Lum JAPONICUM UPTELEA SPP ECON IA U ge MYROTHAMNUS SPP LIQUIDAMBAR 1b ALTINGIA OBOVA p= DOR АА ADEL TOLOE 5 PISTACIA TEREBINTHUS DAPHNIPHYLLUM GRACILE DAPHNIPHYLLUM HIMALAYENSE DAPHNIPHYLLUM МЕ ILGHERRENSE DAPHNIPHYLLUM CALYCINUM DAPHNIPHYLLUM LAURINUM BARBEYA OLEIODES Lapi FLORIDANA M. CAS CASTANDPSIS SPP FRICONOBAL ANUS VERTICILLATA AMPELOCERA CUBENSIS AMPELOCERA RUIZII PARASPONIA ANDERSONI I TREMA SPP APHANANTHE ASPERA APHANANTHE PHILIPPINENSIS CELTIS SPP ME ARISTATA GIRONNIERA SUBEQUAL IS Е al ORTU EA TIS MONOICA PTEROLELTIS TATORINOWI I rez dE SATIVA CHAET HENIPTELEA DAVIDII gi INTEGRIFOL [A qc ld LON RHAMNOIDES КОХ, SERRATA PLANERA AQUATICA RHOIPTELEA CHILIANTHA PLATYCARYA STROBILACEA СА — ЛЕ CORYLUS POP CASUAR 1НА Bs GYMNOST Ra ALNUS RI BALNPS Y 11116619 NOTHOF AGUS — 1.00 .75 .50 .25 $ TExT-FIGURE 2. Similarity cluster analysis, CLS. dominates; however, the pericolpate (Euptelea, Matudaea), periporate (Liquidambaroideae), and triporate (with ulceroid apertures, Myrotham- naceae) apertures are also present. Pollen is usu- ally shed in monads (except in Myrothamna- ceae). The tectum is usually reticulate, and the size of the perforations vary (except Eucommi- aceae are atectate and Myrothamnaceae have clavate sculpturing). The infrastructure is pre- dominantly columellate and all taxa have a foot- ayer. Endexine is usually present. Supratectal sculpturing is variable. In both linkage strategies, ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Trochodendron and Tetracentron exhibit close phenetic relationships with Disanthus of the Hamamelidaceae. The genera Cercidiphyllum and Euptelea also exhibit a high degree of similarity. Taxa of the Platanaceae and Hamamelidaceae (excluding the Liquidambaroideae) are also phe- poen similar. problematic taxa of Group I are the Liq- а. Eucommiaceae, and Myro- thamnaceae. Members of these three groups oc- cupy an intermediate position between Groups I and II, using the complete linkage strategy (Text- ig. 2). However, using the unweighted pair group method, all three families appear isolated be- tween groups II and III and may be considered groups themselves (Text-Fig. 1). Group II is small and exhibits yd аа relationships with Group I a II. It consists of Daphniphyllacea el arbeyaceae, and Fagaceae pim ng Nothofa- gus). Pollen is generally spherical but can be pro- late or oblate in equatorial view and is tricolpate or tricolporate. Pollen is invariably shed in mo- nads. The tectum is perforate, the perforations are usually not discernable with light microscopy but evident with SEM. The infrastructure of all taxa is columellate. The footlayer is present in all taxa except Barbeya. Endexine is occasionally present but best developed in Barbeya. Exine sculpturing is variable. The clustering patterns ofthe taxa investigated in this group generally reflect familial relation- ships. The Daphniphyllaceae and Leitneriaceae are closely linked. Taxa of the Fagaceae form a group, except that Fagus exhibits a closer rela- are somewhat isolated from the other taxa in both analyses. Group III is the largest of the three groups, consisting of the Ulmaceae, Cannabaceae, tonal View and circular to angular in polar view. ^ or + short colpi, numbering two to many. Pollen EM small channels traverse the tectum. Infrastruc- ture varies somewhat but is generally granular to columellate-granular. The infrastructural layer 1986] can consist of irregular-shaped rods, anastomos- ing rods, or spherical granules. The footlayer is usually thin, but always present. The endexine is relatively rare in this group and when present is usually a very thin layer and exine sculpturing can be rugulate, scabrate, or spinulose. Branching patterns in this group generally re- flect familial relationships. Using the unweighted pair group strategy (Text-Fig. 1), the Ulmaceae appear as three distinct groups: (1) members of the Celtidoideae, (2) Trema and Parasponia, and (3) Ulmoideae. Using the complete linkage strat- egy, two groups are evident: Ulmoideae and Cel- tidoideae (sensu Grudzinskaya, 1967). Canna- baceae link closely with the Celtidoideae in both strategies. Juglandaceae (sensu Manning, 1978) are sep- arated into four groups based on linkage patterns with the complete linkage strategy (Text-Fig. 2). Platycarya (Platycaryeae), isolated from other taxa of the Juglandaceae, exhibits similarities to Rhoiptelea and Myrica. Pterocarya and Juglans (Juglandeae) show close phenetic relationship woh Carya (Eiicoreae). Members of the Engel- ardt y linked with the Betulaceae and Casuarinaceae. The un- weighted pair group strategy produced similar results, however, taxa of the Betulaceae and Ca- suarinaceae fall within the Engelhardtieae. The Betulaceae and Casuarinaceae form a closely linked group in both strategies that is near or within the Engelhardtieae (Juglandaceae; Text- Figs. 1, Dalanons and Nothofagus of the Fagaceae ap- pear as a close group, isolated, but within Group III in both linkage strategies. In addition to the taxa already discussed, four other taxa were introduced into the analysis: Salix, Populus (Salicaceae, Dillenidae), Pistacia (Rosideae), and Dorstenia (Moraceae). The Salicaceae were placed with amentiferous taxa, which are now members of the Hamamel- idae by Engler (1926). Salix has tricolpate, spher- ical to ied pollen, prolate, reticulate exine sculpturi and a columellate infrastructure леон ere 1969). Erdtman (1952) noted the similarity of Sa/ix pollen to that of the Plat- anaceae. In our analysis, Salix is most similar to modification of the exine similar to pollen with ulcerate apertures (Fig. 72a). The tectum is per- forate and the columellae rest on a very thin ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 369 footlayer (Fig. 72b). With the complete linkage strategy Populus appears as an isolated taxon be- tween Groups I and II (Text-Fig. 2), and between II and III in the unweighted pair group strategy (Text-Fig. 1). In both cases, neither Salix or Pop- ulus exhibits close phenetic relationships with amentiferous taxa of Group III. Pollen of Pis- tacia is 6—7-porate. The pores are distributed irregularly about the equator and are ulceroid in nature (Fig. 70). Pollen is spherical (Fig. 70) and shed in monads, and the tectum is perforate. The infrastructure is columellate; the columellae are fused to a relatively thick footlayer (Fig. 71), and no endexine is evident. Thorne ( 1973) со onsid- ered the J III taxa, this study) to be closely allied with the Anacardiaceae (Rosideae). Ноу, пе роп Grou p of P with that of the J uglandaceae or Rhoipteleaceae. Pis- tacia, along with the salicaceous taxa, form an isolated group between Groups I and II with the complete linkage strategy (Text-Fig. 2), and be- tween II and III with the UPGMA strategy (Text- Fig. 1). Pollen of Dorstenia (Moraceae), also included in the analysis, is periporate, spherical in shape, and its wall structure is tectate-granular with a well-developed footlayer (Fig. 73). The tectum is occasionally traversed by minute channels (Fig. 73). This taxon shows greatest phenetic similar- ity with members of the Liquidambaroideae (Hamamelidaceae) (Text-Figs. 1, 2). Itisunlikely that the introduction into the analysis of a single taxon from such a large family as the Moraceae with about 1,000 species will accurately reflect familial relationships. Light and SEM studies of a few moraceous taxa (Niezgoda & Nowaczyk, 1976; Hamilton, 1976; Nair & Sharma, 1975; Zamora, 1977) indicate that the pollen is very diverse in the Moraceae, and a much broader survey is necessary | can be used to suggest phenetic relationships with taxa of the Hamamelidae or any other subclass. THE PHYLOGENETIC ANALYSIS The data set used for the similarity cluster analysis was also analyzed cladistically (Appen- dix II). The computer program used, Phyloge- netic Analysis Uing Parsimony (PAUP), was de- veloped and installed in the Indiana University computer system by David L. Swofford of the Illinois Natural History Survey (Swofford, 1984). Seventy-eight operational taxonomic units 370 TEXT-FIGURE 3. Cladistic analysis of the Hama- melidae based on pollen (Adams, 1972, consensus tree). AMP = Ampelocera, BAL = Balanopaceae, BARB = Barbeyaceae, BET = Betulaceae, CAN = Cannaba- ceae, CAS = Casuarinaceae, CELT = Celtidoideae, CER = Cercidiphyllaceae, DAP = Daphniphyllaceae, Eupteleaceae, FA- = Myrica- TROCH = Trochodendrales, ULMOID = Ulmoi- deae. (OTUs) and 30 pollen characters were analyzed. The character states are unordered; however, an outgroup was chosen. The consensus tree gen- erated rooted the outgroup (Text-Fig. 3). Tetracentron of the Trochodendrales was cho- ndrales (1981) and Endress (1986) to be the basal or the plesiomorphic group in the Hamamelidae, and (b) the pollen characters of the taxa in the Trochodendrales are common in many of the alleged primitive families and orders of other subclasses. The commonality of these characters see the descriptive palynology section for pollen characters). From the consensus tree generated, three ma- jor phylogenetically related groups are evident. These groups are comparable to the groups rec- ognized in the similarity cluster analysis (Text- ig. 3; Groups I, II, and III). In addition to the consensus tree, the program generated 250 equal- ly parsimonious trees. The branch swapping in all of the equally parsimonious trees takes place within these three groups, at all times maintain- ing the phylogenetic relationship of the three groups. Group I taxa always occur at the base of the tree, showing a close phylogenetic relation- ship with the outgroup. Group II taxa always 11 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 OCcur as an intermediate group between I and III (Text-Fig. 3). Group III always occurs as the most derived. All groups exhibit a number of equally parsimonious possibilities within each group. This suggests mat it is impossible to resolve the i between mimber of each group based on pollen alone. Not surprisingly, in order to reach a finer reso- lution of the relationships at the generic or spe- cific levels, characters other than pollen will have to be used. The taxa from the Salicaceae and Anacardi- aceae introduced into the analysis bear no phy- logenetic relationship to the amentiferous taxa of Group III, again supporting the results of the similarity cluster analysis. Salix exhibits a close cladistic relationship to a member of the Ham- amelidaceae, and Populus and Pistacia appear as between Group II and Group Ш taxa. It can be concluded from this analysis that Group I taxa are primitive, Group II taxa are intermediate between I and III, and Group III taxa are derived. DISCUSSION The taxonomic composition (at the family level) of the Hamamelidae has been a subject of controversy for some time. However, there is some agreement on the families that comprise the “соге” taxa in this group. Both Thorne (1973) and Cronquist (1981) included the Trochoden- draceae, Tetracentraceae, Cercidiphyllaceae, Eupteleaceae, Hamamelidaceae, and Platana- ceae in the Hamamelidae (Hamamelidiflorae of Thorne, 1973). These families comprise a well- defined phenetic group based on vegetative (Group I of Barabe et al., 1982) and pollen char- acters (Group I, this study; Text-Figs. 1, 2). In айншопй, our SAIS. analysis also suggests a g these taxa (Text- -Fig. 3) The Eucommiaceae, a family Cronquist (1981) placed in its own order and Tippo (1938, 1940) considered closely related to the Urticales, are most similar to Group I taxa in our analysis. Based on a number of vegetative and floral characteristics, Thorne (1973) consid- ered the Eucommiaceae closely allied with the Hamamelidales, as our analyses also suggest. The y I group. Base on pollen morphology, this family exhibits phe- netic similarity to Group I taxa (also see Barabe et al., 1982) but has many unique pollen char- 1986] acters not found in any other taxa in the Ham- amelidae (also see Thorne, 1973). Cronquist (1981) placed this taxon in the Hamamelidales without question. Although our phenetic anal- ysis does not show a strong relationship between Myrothamnus and other Group I taxa, our cla- distic analysis demonstrates a close phylogenetic relationship between Myrothamnus and cerci- T go and hamamelidaceous taxa (Text- g. 3; cf. Cronquist, 1981). mia II consists of a number of small, iso- lated families (Text-Figs. 1-3), which exhibit many pollen characters that can be considered intermediate between Groups I and III. Of the four families included in this group, Thorne (1973) considered only the Fagaceae as a mem- ber of the Hamamelidae. The Barbeyaceae (tax- on incertae sedis of Thorne, 1973) are considered by Dickison and Sweitzer (1970) and Cronquist (1981) as closely allied with the Ulmaceae of the Urticales. However, Barabe et al. (1982) placed the family within their Group II (similar in com- position to our Group II). On the other hand, our cladistic analysis clearly indicates a close re- lationship between Barbeya and the Celtidoideae of the Ulmaceae. The Daphniphyllaceae have been variously treated. Thorne (1973) consid- ered this family allied with the Buxaceae and Brunelliaceae of the Rosidae. A recent cladistic analysis by Barabe (1984), using vegetative and floral features, lends support to Thorne's treat- ment. However, pollen data do not support this alliance; pollen ofthe Buxaceae is predominantly polyporate (Erdtman, 1952), and pollen of the Brunelliaceae, although tricolporate, is prolate with reticulate exine sculpturing (Cuatrecasas, 1970; similar to Group I pollen, this study). Leit- neria (Leitneriaceae) is interesting because it has many features intermediate between Groups I and III. Its oblate shape and microperforate ex- ine is characteristic of many Group III taxa and its tricolporate apertures are similar to some Group I taxa. In the compressed state (fossil pol- len), it has a triradiate polar thickening (Fig. 53) of some fossil normapollis types (e.g., dispersed pollen believed to be (in part) related to some Hamamelidae. Group III taxa form the largest and closest knit phenetic group of the three. Results of the cla- distic analysis show that the branching patterns among Nothofagus, Balanops, Rhoipteleaceae, Betulaceae, Myricaceae, Casuarinaceae, and Ju- glandaceae cannot be resolved based on pollen. ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 371 The remaining taxa, all members of the Urti- cales, exhibit a high degree of paraphylesis (Text- Fig. 3). The phenetic and cladistic analyses based on pollen support Cronquist's (1981) treatment of these families. Thorne (1973) has argued for the placement of the Rhoipteleaceae, Juglanda- ceae, and Myricaceae in his Rutiflorae and sug- gested a close relationship of these families to the Anacardiaceae taxa. Our phenetic and cla- distic analyses included the genus Pistacia, which Thorne believes to exhibit many characteristics suggestive of this alliance, a relationship our pol- len data fails to support (Figs. 70, 71; Text-Figs. 1-3). The pollen characters that define Group III taxa (2-3 to many equatorially placed pores or short colpi, microchanneled tectum, columel- late-granular infrastructure, thin footlayer, and thin endexine) represent a unique array of fea- tures in the dicots. The Poaceae with wind-pol- linated porate pollen are the only family that show some convergence with Group III taxa, but this family has major differences in the place- ment and structure of the aperture; the annulate, distally located pores of pollen of the Poaceae are constructed by a thickening of the basal layer (Skvarla & Larson, 1966). The protruding pores in Group III taxa (aside from being equatorial apertures) are constructed by an elaboration of the tectum and infrastructure (Text-Fig. 7F-D. Thorne suggested that these taxa, as defined by Cronquist (1981), be separated into two different subclasses. Given the unique combination of characters in Group III, this suggestion finds lit- tle support in our analyses. The pollen data pre- sented here lend support to Cronquist's system (1981). The following discussion of phylogenetic relationships of the Hamamelideae relies on the acceptance of these groups as monophyletic (sen- su Cronquist, 1981). The results of the phenetic and cladistic anal- yses are especially interesting when compared with the recent analysis by Barabe et al. (1982) and what is currently known about the fossil re- cord of the Hamamelidae. The similarity cluster analysis ofthe Hamamelidae based on vegetative and floral characteristics by Barabe et al. (1982) distinguishes three groups. Group I consists of the same taxa that make up our Group I. Recall that Group I is made up of families that Thorne (1973) and Cronquist (1981) consider “core” taxa. The discrepancy between the analysis of Barabe et al. (1982) and our phenetic analysis concerns the relationship of Group II taxa to Groups I and III and the composition of Groups Il and III. 372 +Euc +Cas +В; al *Euc ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 TEXT-FiGuRE 4.— А. Phenetic de of the major poup recognized by pea et ne Ao се for the taxa listed above the axa within that relationships of the major groups aes t group are the same as in our gro —B. Phene ermin ned in our analysis. — C. =з рун жаюга relationships oft e groups iir on the fossil record. — D. Phylogenetic relationships of the e major groups bas n a cladistic analysis of the pollen characters and first occurrence of the pollen types represented by each маг Group II of Barabe et al. (1982) consists of Leit- neria, Barbeya, members of our Group II, Di- Es elaceae (not considered in our analysis), Balanops (a rather isolated member of our Group III), Casuarinaceae (member of our Group III), and Eucommiaceae (member of our Group I). However, in our study, Eucommia occurs as ап isolated member between Groups I and II in the complete linkage strategy and between Groups II and III in the unweighted pair group strategy. Group III of Barabe et al. (1982) is very similar to our Group III except for their inclusion of the Fagaceae (Text-Fig. 4). The Fagaceae are mem- bers of our Group II. The general picture that emerges from their analysis is that Groups II and III are phenetically closer to each other than to Group I (Text-Fig. 4A). In our phenetic analysis of the pollen groups, I and II are closer to each other than either is to Group III (Text-Fig. 4B). The relatively close agreement between the phenetic analysis of Barabe et al. (1982) and ours, especially in terms of the members of each group and the phenetic relationship between the three groups using different sets of characters, underscores the gradational nature of pollen, flo- ral, and vegetative features among members of the three groups. The results of the cladistic analysis compare well with the phenetic analyses of Barabe et al. (1982) and that presented in this study (Text- Fig. 4D). For the most part, the members of each phenetic group are maintained in the same rel- ative position in the cladogram (Text-Fig. 4). The cladistic analyses indicate that Group I taxa are primitive. Group II taxa show features of Group I and III (and are intermediate between them), and Group III taxa are derived. However, the branching patterns within each of these groups are difficult if not impossible to resolve based on pollen data alone and will require an analysis that includes a wider range of characteristics than available in pollen. The choice of Tetracentron (Trochodendrales) as our outgroup (see above), and the use of par- mony to derive the other taxa, orders the pollen d (Text-Fig. 5). The pollen unit is pre- ominantly a monad in all three groups. The occurrence of tetrads in the Myrothamnaceae is considered a derived condition. Pollen shape is primarily prolate in the primitive groups and oblate in the derived taxa. Likewise, shape in polar view ranges from circular in primitive taxa to angular in the derived groups. The tectum varies from reticulate in Group I to microper- forate in Group II to microchanneled in Group III. Pollen wall structure is columellate in Groups POLLEN UNIT Monads—— — —ee Tetrads (Myrothamnaceae) POLLEN SHAPE Prolate ———m Spherical —m Oblate Circular —e Angular TECTUM Reticulate E rope Intectate rforate —™ Microchanneled (Myrothamnaceae INFRASTRUCTURE Columellate = Granula FOOTLAYER Present ——— Thin-Absent (Barbeyaceae) ENDEXINE + APERTURE TYPE Colporate (Tri-, Tricolhex-) Group II Tricolpate < Stephanoaperturate (Di-, Tri-, Porate, Colpate) Group III Periaperturate (Colpate, Porate) Group I SCULPTURING te, Rugulate, Papillate TEXT-FIGURE 5. Evolutionary trends of pollen characters in the Hamamelidae, based on a phenetic and cladistic analysis of extant pollen. 1986] ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY o x он < O < > = < . t © FF Oa со à E б > Е © = Е a u x ш сє 2 « 2 = ш 5 3 ш > oc a о I кюк E ш ш ш о J 2 5 ш> 2 >) "b E Hm EE te umm - y Н Н 1 i Н У D 10 ¡0 - zi: b E TEM HEINE ME ME IP «|l. f: Д Ц о | В te fie Bie Elfi f: ! : ie Yi! io DAE НЫ | cil] n іе ME НИ pie LU e ' i І H 19 je е E MN |: EE Н us = Н ; [ E o ' ' 3 Ш * 2 * E a * О === MEGAFOSSIL RECORD DISPERSED POLLEN IN SITU POLLEN * EARLIEST POLLEN RECORD TEXT-FIGURE 6. earliest record of pollen similar to The known = record of hamamelidaceous taxa. The earliest — od (*) 1s the at found in the Hamamelidae group (not necessari melid affinity). mpiled from Wolfe (1973), Magie (1981), Jones (1984), Zavada and Crepet (1981), qure (1984), and Crane (pers. comm.). I and II (the plesiomorphic condition) and col- umellate to granular in Group III, with some taxa having an infrastructure composed of columel- lae-like elements and granules (e.g., Balanops, Betula, Celtis). The footlayer is present in all taxa except Barbeya, and this is considered a derived state. Pollen aperture type is predominantly tri- colpate in Group I taxa and is considered the plesiomorphic condition. The periaperturate type may occur in Group I taxa (e.g., A/tingia) but is considered a derived type in this group. Group II taxa are primarily tricolporate and Group III porate. Although it is convenient to consider Group II as an intermediate aperture type be- tween I and III, it is just as likely that the porate pe of Group III is derived directly from the tricolpate type. Thus, the aperture type of Group III can conceivably be derived from the Group II type (tricolporate) or Group I type (tricolpate). However, the most parsimonious transition is from the tricolpate type to porate condition be- cause such a transition does not first require the evolution of the pore and then the reduction of the colpus, but only a reduction of the colpus. The polyporate pollen ofthe Liquidambaroideae (Group I) is also believed to be derived directly from the colpate condition based on its close adum cds relationship to the colpate pollen of me H melidaceae. poss ls record is also important in inter- preting the neontological data. As indicated by the cladistic analysis, we might expect Group I taxa to appear first in the fossil record. The Plat- anaceae and Cercidiphyllaceae both appear be- fore any of the other hamamelidaceous taxa, cor- roborating to some extent the neontological data (Text-Fig. 6). Group II taxa, however, appear as megafossils later than Group III taxa (Text-Fig. 6). This situation taken at face value supports our earlier suggestion that Group III taxa are as likely to be derived from Group I as Group II taxa. Thus, the order in which Groups II and III are derived would be reversed (Text-Fig. 3C). If the first occurrence in the fossil record of the pollen type for each of the three groups is con- sidered, we find that Group I pollen appears first, 374 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 y. TEXT-FIGURE 7. А-Е. Representative Group I and II taxa n ae thickenings in the apertural region (redrawn) (a = a —A. Rhodoleia championii; x12,540.— B. Leitneria floridana ae 540.—C. Barbeya oleoides; x 8,094.—D. Fagus crenata; х 10,500 (redrawn from cole & Daghlian, 1980).— E. Quercus spinosa; x13,250 (redrawn from Crepet & Daghlian, 1980). F-I. Representative Group III taxa with exo- DUM in the apertural region. — Е. Cannabis sativa; x 8,094. —G. Rhoiptelea chiliantha; х13,500 (redrawn from Sto Broome, 1971).—H. Pterocarya delavayi; x 3,700 (redrawn from Stone & Broome, 1975).—I. Juglans mallia x 8,200 (redrawn from Stone & Broome, 1975). 1986] Group II shortly thereafter, and then Group III (Text-Fig. 6). However, unless these pollen types can be taxonomically rud m the Ham- amelidae, this cannot b trong port for the intermediate position Group II seems to occupy based on the neontological data. How- ever, the elucidation of the taxonomic affinities of many of the dispersed pollen taxa of the Nor- mapollis group may have an influence on our interpretation of Hamamelid phylogeny. Many of these dispersed pollen types are believed to be related to amentiferous taxa (Kedves & Dinitz, 1981). Kedve’s (1981, 1983) recent treatment of this fossil-dispersed pollen group is relevant; he recognized three major groups: probrevaxones, and p ll rains ofthe probrevaxones group are tricolporate with short colpi, have tectate-columellate wall, and reticu- late to psilate to slightly verrucate-scabrate exine sculpturing. The germinal aperture lacks an an- ul s (1981) subdivided the norma- polles group into three subgroups: pronorma- polles, eunormapolles, and paranormapolles. The pronormapolles are similar in many respects to the probrevaxones types except that they possess endannuli (a thickening in the inner part of the apertural region (e.g., Complexiopollis, Limai- pollenites, Kedves & Pardutz, 1982). Members of the pronormapolles have, in addition to the tricolpate aperture, a microperforate exine and tectate columellate wall; shape is oblate in equa- torial view and angular in polar view. The char- acters observed in both the probrevaxones and pronormapolles are similar to those in Groups I and II in our analysis. These pollen types are the y doit epi earlier types and are considered e primitive types by Kedves (1983). The taxa d. e subgroups cunormapolles 3nd paranor- four Group III taxa (columellate-granular infrastructure, thin basal layer, oblate equatorial shape, and circular to slightly angular shape in polar view; Stanley & Kedves, 1975; Kedves & Stanley, 1976). The presence of the annulate porate aperture is also common. In these fossil taxa the annulus is con- structed of ektexinal material; this is also ob- served in the extant Group III taxa (Text-Fig. 7; compare Groups I and II taxa A-E with Group III taxa F-D. In addition to sharing many char- acteristics with Group III taxa, the Eunorma- polles and paranormapolles taxa occur strati- graphically higher. Although the taxonomic affinities of the probrevaxones and pronorma- ZAVADA & DILCHER — HAMAMELIDAE POLLEN MORPHOLOGY 375 polles groups are unknown, their chronistic re- lationship to the other normapolles —along with a suite of characters suggestive of Group II (or rma polles are hamamelidaceous taxa possibly relat- ed to some of the relictual Group II taxa of the extant Hamamelidae. However, elucidation of the taxonomic affinities of these dispersed pollen types will be much better understood when their attachment to megafossils is known. Until this is accomplished, the importance of this fossil material in determining the phylogenetic rela- tionships of the Hamamelidae will remain spec- ulative. In summary, the phenetic and cladistic anal- yses based on pollen morphological data suggest the Hamamelidae as defined by Cronquist (1981) are a reasonably circumscribed subclass, and are in general agreement with other phenetic anal- yses (e.g., Barabe et al., 198 LITERATURE CITED ADAMS, E. N. 1972. Concensus techniques me the comparison of taxonomic trees. Syst. Zool. 21: 390-397. ADAMS, R. J K. Morton. 1972. An Atlas E and the Adjacent United States, y . Univ. of Waterloo, Ontario. ARCHER, A., . S. Horowitz. 1984. Cluster Analysis Package. Computer Program. In- diana University Department of Geology, Bloom- ington, Indiana BARABE, D. 1984. Application. ias 5 а la sys- tematique de An — s des Hamameli- dales. Candollea 39: 51- . BERGERON & G. A. pO 1982. Etude quantitative de la classification des Hamamelidi- BocLE, А. 1. & C. T. Рниввіск. 1980. A generic atlas of Up OW edd pollens. Contr. Gray Herb. 210: 29-103. COETZEE, J. A. & J. o AR 1984. Pollen evi- dence for the occurrence of Casuarina and Myrica in the Tertiary of South Africa. Grana 23: 23-41. CREPET, W. L. & C. P. DAGHLIAN. 1 . Castaneoid inflorescences from the Middle pelas of Ten nesssee and the diagnostic value of nn a ilie subfamily level) i in i Fagaceae. Amer. J. Bot. 67: 739-757. CRONQUIST, A. 81. An Integrated System of Clas- sification of Flowering Plants. Columbia Univ. , New York. CUATRECASAS, J. 1970. Flora Neotropica, 2. Brunel- liaceae. Hafner Publ. Co., Connecticut DickisoN, W. C. & E. he mor- phology and relationships of Barbeya oleoides. Amer. J. Bot. 57: 468-476. 376 DUNBAR, А. & J. В. Rowley. 1984. Betula pollen 977. Evolutionary trends in the — -Fagales group. Pl. Syst. Evol., Suppl. 1: 321-347. Ў 1986. Floral structure, systematics, and phy- logeny in Trochodendrales. Ann. Missouri Bot. . 1926. Die Naturlichen Pflanzenfamilien, Volume 14A, Angiospermae. zig. ERDTMAN, G. 1952. Pollen Morphology and Plant Taxonomy. Angiosperms. Almqvist ca Witsell, Stockholm. . 1969. Handbook of Palynology. Hafner Publ. Co., New Yo je FAEGRI, K. & J. I SEN. 1964. LM of Pollen pe " . The TEPAT and rea- sons for distinguishing the ETA ua as a sep- arate family, ne TE rae rn. (Mos- cow & Leningrad) 52: жч Ruaan, ] HAMILTON, A. C. 19 6. a nication of east African Urticales poen. Pollen & Spo c2 e, Ha eli Platanaceae, und Fagaceae. Pi. [m Evol. 130: 1 4. Leaf sur gn and Cuticular and 'Fagaceous' JONES, J. H. esis. Indiana Univ., Bloo- mington. Kepves, M. 1979 some selected recent Amentiflorae pollen I. Acta Bot. Acad. Sci. Hung. 25: 75-82. —. 981 Definitions of, evolutionary trends, within d len. Rev. Paleobot. beue 35: 149-1 1983. nt ofthe сд brevax- ones pollen grains and the main stages of their evolution during the lower and middle Senonian. Pollen & ады 25: 487—498. NITz. 1981. Probrevaxones, а new pollen diets for the first pe coi form-genera from the Upper Cenonian of P gal. Acta Bot. Acad. Sci. Hung. 27: 383-402. A. PARDUTZ. 1982. Ultrastructure inves- tigations of the early normapolles taxa Canplex- iopollis and Limaipollenites. Palynology 6: 149— 159. & E. A. STANLEY. 1976. оо ч investigations of the normapolle p and me other selected European and North panes can angiosperm pollen II. Pollen & Spores 18: 105- KERSHAW, А. D. 1970. Pollen morphological varia- tion within the Casuarinaceae. Pollen & Spores 12: 145-159. Lieux, М.Н. 1980. Anatlas d aede of trees, shrubs, and woody vines of Lou and other south- eastern states, Part II, Peu aie to hind Pollen & Spores 22: 192-231. ANNALS OF THE MISSOURI BOTANICAL GARDEN . Scanning electron microscopy of [Vor. 73 MANCHESTER, S. R. 1981. Fossil History of the Ju- glandaceae. Ph.D. Thesis. Indiana Univ., Bloo- mington. MANNING, W. E. 1978 [1979]. The classification within the Juglandaceae. Ann. Missouri Bot. Gard. 65: 1058-1087. MivosHi, М. 1983. Pollen morphology of the genus Castanopsis (Fagaceae) in Japan. Grana 22: 19- 23. Nair, P. K. K. & M. SHARMA. 1975. Pollen mor- phological studies in Indian Urticales. Bot. Not. 118: 177-186. NIEZGODA, C. J. & J. NOWACZYK. 1976. шш studies in Acanthinophyllum, Clarisia, Soroc and Trophis (Moraceae). Pollen & Spores 18: $13. 522. PACLTOVA, B. 1982. Some pollen of recent and fossil species of the genus Platanus L. Acta Univ. Caro- linae Geol. Pokorny 87- PRAGLOWSKI, J. s on the ollen mor- phology of Swedish trees and shrubs. Grana Pal- ynol. 3: 45-65. 1974. The pollen morphology of the Troch- odendraceae, Tetracentraceae, ор у. Pol- T E Spores 16: 449-467. 981. Transition within the exine of Noth- m Blume. Rev. Paleobot. Palynol. 32: 369- 9 Fagaceae L. Fagoideae. World Pollen and Spore Flora 11. Almqvist and Wiksell, Stock- holm PuNT, W. On the pollen morphology of Scy- кн апа Dorstenia (Moraceae). Grana 17: 17- T: E. EETGERINK. 1981. On the pollen mor- а ology of some genera of the tribe Могеае (Мо- raceae). Grana 21: 15-19. ROWLEY, 1. Pollen wall characters n em- phasis upon applicability. Nordic J. Bot. 1: 357- 380. B. PRIJANTO. 1977. Selective destruction of the exine of pollen grains. Geophytology 70J: SKVARLA, J. J. & D. A. LARSON. 1966. Fine structure of Zea mays pollen. I. Cell membranes and exine t. 53: 1112-112 . 1973. A — electron microscopical сна of the pollen morphology in the genus Quer- us. Acta Bot. Neerl. 22: 655-665. ue A.M. taxonomy o Amer. J. Bot. 70: 481- 1983b. Pollen Р and plant tax- onomy of red oaks in eastern North America. Amer. J. Bot. 70: 495-507. STANLEY, E. A. & M. KEDVES. 1983a. Pollen old mg раді Ғ _ 1975. Electronmi- and some other ‘selected о апа N rth American angiosperm pollen. I. Pollen & els 17: 233-272. STONE, D. E. & C. R. BROOME. 1971. Pollen ultra- structure: evidence for relationship of the Juglan- oer nd the Rhoipteleaceae. Pollen & Spores 13: 5-1 1986] ——. . Juglandaceae A. Rich. ex Kunth. World Pollen and Spore Flora 4. Almqvist and Wiksell, Stockholm SwtrTZER, E. M. 1971. Comparative anatomy of UI- maceae. J. Arnold Arbor. 52: 523-585. SWOFFORD, D. 1984. PAUP; Phylogenetic Analysis Using Parsimony, Version 2.2. d Natural History Survey, Champaign, Illino THORNE, R. F. 1973. The ii” or Hama- melidae as an artficial group: a summary state- ment. Brittonia 25: 395-405. TiPPo, О. 1938. The comparative anatomy of the secondary xylem and phylogeny of the Eucom- miaceae. Amer. J. Bot. 27: 832-838. 40. Comparative anatomy ofthe Moraceae and their presumed allies. Bot. Gaz. (Crawfords- Я 1984. Cuticular anatomy of an- rm leaves from the Lower Cretaceous Po- tomac Group. I. Zone I leaves. Amer. J. Bot. 71: 192-202. m. o ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 377 WALKER, J. W. 1976. Comparative pollen morphol- Evolution of Angiosperms. Columbia Univ. Press, New Yor WOLFE, J. A. 1973. Fossil forms of Amentiferae. Brit- 7. Morfológia de los granos de polen de la familia Moraceae en México. Bol. Soc. Bot. México 36: 71-93. ZAVADA, M. S. 1983. Pollen morphology of Ulma- ceae. is 22: 23-30. 4 [1985]. Angiosperm origins and evo- шо. ae on dispersed fossil Жуу ultrastruc- ture. Ann. Missouri Bot. Gard. 71: & W. L. CREPET. 1981. Investigation of an- giosperms from the Middle Eoc of North America: flowers of the Celtidoideae. pond J. Bot. 68: 924—933 378 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 APPENDIX I Taxon Collection Locality Herbarium Altingia obovata e Liang 64734 China US Balanops vitiens A. C. pd Hjedruavist Degener 15519 Fiji US Barbeya oleoid Schweinf. Burger 1926 Ethiopia US Betula alba L. Knowlton, 5/13/1933 Vermont IU Cannabis sativa L. Chase, 8/23/1899 Illinois IU ‚шн stricta Christophel 3531 Australia — Cercidiphyllum japonicum Sieb. e Ya Tokuburchi s.n., 1891 Sapporo, Japan US Cor pauciflora Sieb. et Zuc Blaney 57613 Japan (Cult.) NYBG Corps olas ol. et Wils. Rock 8127 China US антре ав calycinum Benth. Chun 6428 China US Daphniphyllum gracile Guff. Brass s.n. New Guinea US Daphniphyllum himalayense var. chartac (Rozenth.) Uuan Datla s.n. Siwalik US Ее laurinum (Benth.) B Boeea 7381 Sumatra US Daphniphyllum neilgherrense (Wight) Thw Bernardi 15811 Ceylon US prm ii Max Maelsawa 409 Japan US lo sp. Coll. Ig. nom. s.n. Connecticut (Cult.) CONN Eucommia ulmoides Oliv. Coll. Ig. nom., 4/27/1932 Brooklyn Botanical Garden BKL Euptelea хеши Sieb. et Kobayashi & Oonisi 34579 Japan US некин pe (R. Br. ex Griff.) R. W. кл Rock 7574 China US Fothergillia major Lodd. Hurbison 5930 Georgia US Fothergillia monticola Ashe Walker 9816 Pennsylvania US Gymnostoma deplancheanum (Miq) L J. Christophel 3488 New Caledonia — Hamamelis japonica Sieb. et Zucc. Konta 4923 Japan US Hamamelis vernalis rg. Sally 1018 Arkansas US Hamamelis virginiana L. Fosberg 45888 Virginia US Leitneria floridana apm. Demaree 31693 — IU Liquidambar styraciflua L. Burbank 2012 Georgia US Lorapetalum chinense Oliv. Tsung 23400 China US Matudaea hirsuta Lundell Bogle 848 Mexico US 1986] ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY APPENDIX I. Continued. 379 Taxon Collection Locality Herbarium Myrica asplenifolia L. Deam 27402 Indiana IU Myrothamnus flabellifolia w. Bayliss BS/1762 Swaziland US Myrothamnus moschatus Bail. Fosberg 52576 Madagascar US Pistacia terebinthus L. Young 9105 Italy IU Platanus gentryi Nixon et Poole Gentry 5862 Sinaloa, Mexico MICH Platanus kerrii Gagnep. Langson 8745 Vietnam — Platanus mexicana Moric. Ventura 596 Veracruz, Mexico ENCB Platanus mexicana Moric. Vela, 2/22/1962 Puebla, Mexico ENCB Platanus occidentalis L. var. occidentali Poole & Watson 2522 Texas TEX Platanus racemosa utt. Ferris 8505 Baja, California, Mexico DS Platanus racemosa Nutt. Elmer 3831 California NY Platanus rzedowski Nixon et Poole King 3900 San Luis Potosí, Mexico MICH Platanus wrightii Wats. Frye & Frye 2271 Sonora, Mexico DS Platanus wrightii Wats. McVaugh 8105 Hidalgo, Mexico MICH Populus deltoides Marsh Welch 5436 Indiana IU Rhodoleia championii Hu, 1/14/1972 Hong Kong US Hook. F. 380 ANNALS OF THE MISSOURI BOTANICAL GARDEN APPENDIX II Appendix II lists the presence (1) or absence (0) of the 30 character states for The * indicates the outgroup used in the cladistic analysis. [VoL. 73 each of the 78 taxa (OTUs). N єє e e e e e ee —— ы оооло ол Б шо м ороо чоло t = NN м Wn — N > Tricolpate iporate Pericolpate Triporate Atectate Stephanocolpate Tricolporate Stephanoporate iporate Intectate Striate TET SIN CER JAP EUP SPP DAP HIM DAP NEI DAP CAL EUC ULM BAR OLE AMP CUB AMP RUI APH ASP APH PHI CEL SPP CHA MEX Tetracentron sinense Cercidiphyllum icum Liquidambar styraciflua Altingia obov Rhodoleia Panem Exbucklandia populnea Disanthus cercidifolius Hamamelis spp Loropetalum chinense Fothergilla monticola Matudaea spp. Daphniphyllum gracile Daphniphyllum laurinum Daphniphyllum hi ense Daphniphyllum neilgherrense Daphniphyllum calycinum Eucommia ulmoides Barbeya oleiodes Ampelocera cubensis Ampelocera ruizii Aphananthe aspera Aphananthe philip- аа Celtis sp е aristata Gironniera subequalis Lozanella enantiophylla Mirandaceltis monoica Pteroceltis tatorinowii Parasponia andersonii Trema s Chaetoptelea mexicana HEM DAV HOL INT PHY RHA PLA AQU ULM SPP ENG ROX ENG RIG TRI VER ALN RUB BET LUT COR AME CAS SPP GYM SPP POP DEL PIS JER SAL SPP Hemiptelea davidii Holoptelea integrifolia Phyllostylon rhamnoides Planera aquatica Ulmus spp. Zelkova serrata Pii sativa Dorsten Poen qm Rhoiptelea chiliantha Platycarya strobilacea lhardtia Engelhardtia rigida Engelhardtia spicata Oreomunnea spp Alfaroa costaricensis Alfaroa quanacastensis Alfaroa guatamalensis Alfaroa mexicana Alfaroa manningii Alfaroa williamsii Pterocarya spp. Juglans spp Carya spp. Myrica asplenifolia Balanops vitiensis Castanea s Lithocarpus spp. agus spp. Nothofagus spp. Quercus spp Trigonobalanus verticillata Alnus rubra Betula lutea lus americana Piin 2m Gymnostoma spp. Populus deis Pistacia terebinthus Salix spp 1986] ZAVADA & DILCHER—HAMAMELIDAE POLLEN MORPHOLOGY 381 APPENDIX II. Continued. |2 |+ o n o j= oa aal |= m m |+ = = о | | со = o N кю — кю кә кә ы 24 25 26 27 28 29 30 1 0 0 0 0 0 Rh KK о – о c= H © un ч < aba аасы ыык] RRR eee сес оо о ОООО ооооо о = о H = > rs Bn a dede IA > = "d e a w == 2 OO OH OF pm m e = > = m 24 ооннонн ннн ннн A A A A A A a ooooo o m 2 un ro ru Ri a е ooooco e = = n "d "ч ooococooo оооо MORPHOLOGY AND DEVELOPMENT OF PISTILLATE INFLORESCENCES IN EXTANT AND FOSSIL CERCIDIPHYLLACEAE! PETER R. CRANE? AND RUTH A. STOCKEY? ABSTRACT Comparison of the shoots, diit inflorescences, and infructescences of Joffrea speirsii Crane & Stockey from the upper Paleoc of Can ada, Nyssidium arcticum (Heer) Iljinskaya m the upper t a mur, eastern U.S.S.R., reveals о diversity in phyllotaxy, т! € position and the number and crowding of follicles in each infructescence. Despite this diversity, "thc pist stillate шке of all these fossil Cercidiphyllum- like na are directly com- parable in basic organization, - ко of extant Cercidiphyllu to the evidence that the fossil and e a arpel, fossil taxa flow carpels. The arrangement of carpels in bicarpellate flowers resembles that in extant Hamamelidaceae. The Cercidiphyllaceae contain a single genus, Cercidiphyllum, with two very similar species of dioecious trees, C. japonicum Siebold & Zuc- carini native to Japan and central and western China, and C. magnificum (Nakai) Nakai native to Honshu (Spongberg, 1979). The Cercidiphyl- laceae have been placed in the Hamamelidales S h (Takhtajan, 1969). Cercidiphyllum is thought to be “related on the one hand to the Hamameli- daceae (especially Disanthus) and on the other to the Trochodendrales and Magnoliales” (Cron- quist, 1981: 167), although most authors agree that Cercidiphyllum is an isolated E sepa- rated by | ts closest living relatives. The pistillate ı кыы struc- tures of Cercidiphyllum are unusual: they super- ficially resemble an apocarpous flower with two to eight carpels, but the carpel orientation is ap- parently anomalous, with the ventral suture di- rected abaxially. Swamy and Bailey (1949) sup- ported the earlier hypotheses of Brown (1939), Harms (1916), and Solereder (1899), and inter- preted the pistillate reproductive structure as a reduced inflorescence. As part of the evidence for this hypothesis Brown (1939) and Swamy and Bailey (1949) cited the elongated inflorescences of fossil Cercidiphyllum-like plants, which were ceous and lower Tertiary (Brown, 1939, Chandrasekharam, 1974; Schloemer-Jager, 1958). Although this hypothesis has been dis- cussed by subsequent authors, the precise simi- larities between the fossil and extant inflores- cences have remained obscure due to inadequate years knowledge of these fossil plants has in- creased considerably (Crane, 1984; Crane Stockey, 1985; Krassilov, 1976; Stockey & Crane, 1983), They are now known from seedlings, leaves, shoots, pistillate inflorescences, infruc- tescences, seeds, and possible staminate inflo- rescences, and it is clear that these extinct Cer- cidiphyllum-like plants are more closely related to Cercidiphyllum than to any other genus (Crane & Stockey, 1985). In this paper we compare the ! This work was supported in part by a United States National Science Foundation em BSR-8314592 to RAS. PRC, and a NSERCC (National Sciences and Engineering Research Council of Canada) gr. nt A6908 to We thank Dr. V. A. Krassilov, Institute of Biology and Pedology, Far-eastern Scientific сес Vladivostok, U.S.S.R., for the use of Figures 16-18 and for helpful discussions of the Amur material. We are also ко to Paul and Chris Rechten for directing our attention to the material from Decker, Montana; to Mrs pei for collecting most of the specimens of Joffrea used in this study; aa * Р. К. Endress, E. М. Friis, pu J. A. Wolfe for very helpful comments on an earlier draft of this manuscri ? Department of Geology, Field Museum of Natural History, pat Road at Lake Shore Drive, Chicago, Illinois 60605. 3 Department of Botany, University of Alberta, Edmonton, Alta, Canada T6G 2E9. ANN. MISSOURI Bor. GARD. 73: 382-393. 1986. 1986] © FIGURES 1, 2. Cercidiphyllum japonicum Sieb. & Zucc.—1. , and three bud-scales (BS) cences and leaves: note the long styles and stigmas leaf on the right: inflorescence on left shows a bract (B) subte CRANE & STOCKEY —CERCIDIPHYLLACEAE 383 Two short shoots with young pistillate inflores- and surrounding the inflorescence nding one of the four carpels, x 3.5.—2. Tw short shoots showing leaf scars, add bud (AB), infructescence with four follicles, and bract scars (B) at base of follicles, x 3.3. Scale bars, 2 mm shoots and pistillate reproductive structures of extant Cercidiphyllum with early Tertiary Cer- cidiphyllum-like plants from western Canada (Joffrea speirsii Crane & Stockey, 1985), south- ern England [Nyssidium arcticum (Heer) Iljin- skaya; Crane, 1984], and eastern U.S.S.R [Trochodendrocarpus arcticus (Heer) Kryshto- fovich; Krassilov, 1973, 1976]. The nomencla- ture for fossil taxa follows that adopted in Crane (1984), Crane and Stockey (1985), and Krassilov (1976). Discussion of the complex nomenclature and systematics of fossil Cercidiphyllum-like in- fructescences is outside the scope of this paper. Comparisons among fossil taxa demonstrate considerable diversity in phyllotaxy, shoot growth, inflorescence position, and the number and crowding of follicles in each infructescence. However, there are fundamental similarities in construction of the pistillate reproductive struc- tures in the extant and fossil plants. These sim- ilarities add to the evidence that the fossil and d en taxa are closely related, explain some of ual morphological features of extant v rune dus and facilitate more meaningful comparison of the genus with other extant taxa in the Hamamelidae. EXTANT CERCIDIPHYLLUM Extant Cercidiphyllum has branches differen- tiated into long and short shoots. In trees that we have examined, long shoot leaves are oppo- site, subopposite, or occasionally in irregular whorls of three. Short shoots develop from the axillary buds of long shoot leaves. Each axillary bud has three bud-scales (Figs. 1, 19); the outer and inner are positioned opposite to the second bud-scale and the leaf of the preceding season. The new leaf develops opposite the inner bud- scale (Swamy & Bailey, 1949), and each short shoot bears a single leaf in each growing season (Fig. 1). Short shoot growth is sympodial, con- tinuing each season through the activity of a new axillary bud. The resulting short shoots are dis- tinctly curved pigs the long shoot on which they are borne (Fig. In fertile short Ao the pistillate inflores- cence is terminal and consists of a short axis that has a cluster of two to eight carpels at the apex 384 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 1986] (Figs. 1, 2). The carpels are arranged in opposite and decussate pairs, and the pairs are borne a slightly different levels (Swamy & Bailey, 1949). Each carpel is subtended by a membranous bract (Fig. 1) that is shed early in carpel development, leaving only a scar and sometimes a small flap of tissue at t the base of the mature follicle (Fig. 2). TI , including follicles, — is up to 25 mm long. Attachment of pairs of carpels at different levels and the single bract associated with each carpel (Swamy & Bailey, 1949) are not consistent with the suggestion that the cluster of carpels is a simpl er (Hutchinson, 1964). Each bract and: the single carpel that it subtends are more readily inter- preted as a single flower that lacks a perianth. The cluster of carpels is therefore an inflores- cence of two to eight very simple flowers (En- dress, 1986; Swamy & Bailey, 1949). The ventral suture of each carpel is oriented abaxially with respect to the inflorescence axis. JOFFREA SPEIRSII CRANE & STOCKEY (PALEOCENE, CANADA) Joffrea speirsii Crane & Stockey, from the up- per Paleocene, Paskapoo Formation of central Alberta, Canada (Crane & Stockey, 1985), is the most completely understood of all fossil Cer- cidiphyllum-like plants. Joffrea exhibits long and on the short shoots (Fig. 5) indicate that the leaves were borne in opposite and decussate pairs; an arrangement also clearly seen in the seedlings of Joffrea (Crane & Stockey, 1985; Stockey & Crane, 1983). The number of leaves borne on each short shoot in a single growing season is unknown. In Joffrea at least some short shoots bore two or more pistillate inflorescences at the apex (Fig. 3). Neither this nor the short shoot leaf arrangement CRANE & STOCKEY —CERCIDIPHYLLACEAE 385 can be accounted for by sympodial growth. To- gether these features strongly suggest that short oot gro was monopodial, with pistillate in- florescences developing from axillary buds o short shoot leaves. The pistillate inflorescence is elongated, up to 65 mm long, with bud-scales at the base (Figs. 3, 8). The inflorescence bears approximately 40 carpels spaced at intervals of 1-5 mm on short side branches of the inflorescence axis (Fig. 3). Each side branch bears one or two short-stalked carpels. The junction between the carpel stalk and the side branch of the inflorescence axis is marked by a distinct joint (Fig. 9), which may mark the attachment of a caducous bract as in extant Cercidiphyllum. Single carpels or pairs of carpels frequently have broken away from the inflorescence axis at this joint (Fig. 7). At ma- turity, the inflorescence elongated to approxi- mately twice its original length, and the follicles were more widely spaced at intervals of 6-11 mm (Fig. 4). At this stage the joint at the base of the follicles is often difficult to detect; however, well- preserved impressions of similar (perhaps con- specific) infructescences from the upper Paleo- cene of Montana clearly illustrate this joint and the tendency of the follicles to break away at that point (Figs. 13, 15). Joffrea carpels that are borne singly have the ventral suture oriented adaxially with respect to the inflorescence axis, and carpels borne in pairs have the two sutures opposite each other (Figs. 6, 7). At maturity at least some of the single follicles twisted so that the ventral su- ture was oriented abaxially with respect to the inflorescence axis (Fig. 10) e = NYSSIDIUM ARCTICUM (HEER) ILJINSKAYA (PALEOCENE, ENGLAND Cercidiphyllum-like infructescences and folli- cles from the upper Paleocene Reading Beds — FIGURES 3-10. Joffrea speirsii ei & Stockey, from Red Deer, Alberta (Paskapoo Formation, upper Pa- leocene). All specimens (prefix attached to a sin Вх x2.—8. Tw a distinct joint (arrow) at the base of a pair of follicles (513323), and in pairs (S13357A), х0. 95. —5. Short shoot showing dia gual indicated with arrows (S13333), x3.7.—6. Ridged ao axis ye bscissed: note sutures facing one another (S12358), x1.75.— a orientation of sutures with respect to floral axis, 2n remains of styles o bud- nre (arrows) at the base of a broken inflorescence (S12387), x 2.— 9. Inflorescence the University of Alberta Paleobotanical Collections (UAPC-ALTA).— 3. . Pair of carpels x 2.8.— 10. Mature follicle showing showing a A ads em stalk and abaxial orientation of suture with respect to infructescence axis (S12410B), x2. Scale s4m 386 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 a 11-15. Cercidiphyllum-like plants from the Paleocene of southern England and western Nort a. Specimens prefixed v. are from Berkshire, southern England (Reading Beds, Woolwich and Reading Formation. upper Paleocene) and are in the collections of the Department of Palaeontology, British Museum 1986] (Woolwich and Reading Formation) of southern England were originally described as Carpolithus gardneri by Chandler (1961). The original spec- imens and additional material have recently been reexamined in detail (Crane, 1984) and assigned to Nyssidium arcticum (Heer) Iljinskaya. There is no evidence to suggest that the infructescences from the Reading Beds were borne on short shoots. The most informative specimens (Figs. 11, 12) suggest instead that the infructescences were borne as part of a long shoot system. These specimens exhibit whorls of oval scars that we interpret as leaf scars (Figs. 11, 12; see also Chan- dler, 1961). Growth in N. arcticum from the Reading Beds apparently resulted in the periodic production of whorls or pseudowhorls of leaves. Some leaf whorls are not followed by the pro- duction of an inflorescence (e.g., lower whorls, Figs. 11, 12). Other whorls of scars apparently delimit the base of infructescences (e.g., upper whorl, Fig. 11), and suggest that long shoot growth in this species was frequently terminated by the production of an inflorescence from the apical bud. Occasional branching occurred between the whorls of leaves, but the precise relationship of TI to leaf and axillary bud production is not ea dd pistillate inflorescences of N. arcticum from the Reading Beds are unknown, but the most complete infructescences are up to 130 mm long with about 15 widely spaced follicles ap- proximately 5-6 mm apart. The follicles were borne singly (Fig. 11), or in pairs (Fig. 14), on short side branches of the infructescence axis (Crane, 1984). The best preserved specimen from the Reading Beds (Fig. 11) shows a distinct joint between the follicle stalk and the side branch of the inflorescence axis. Frequently this is not vis- ible in poorly preserved material. However, oth- er specimens indicate that follicles broke away from the infructescence axis at this point of CRANE & STOCKEY —CERCIDIPHYLLACEAE 387 weakness (Fig. 14). As in Joffrea, the ventral su- tures of pairs of follicles face each other. Suture orientation in single follicles is often unclear, but at least some were probably oriented abaxially with respect to th t maturity (Crane, 1984). TROCHODENDROCARPUS ARCTICUS (HEER) KRYSHTOFOVICH (PALEOCENE, U.S.S.R.) Trochodendrocarpus arcticus (Heer) Kryshto- fovich from the lower Paleocene (Danian) of Amur, eastern U.S.S.R. (Krassilov, 1973, 1976, 1977) is known from large specimens showing shoots that have attached infructescences. The infructescences are widely spaced and borne al- ternately, perhaps helically, on long shoots (Figs. 16, 17). A scar is often visible where each in- fructescence is attached to the long shoot (Figs. 16, 17). These scars are large and in the size range of petiole bases of associated Cercidiphyllum- like leaves. We suggest that the scars are those of alternate leaves and that each infructescence developed from an axillary bud. An alternative interpretation would be that each “‘infructes- cence" is a side branch of a large paniculate in- fructescence (Krassilov, 1977). This interpreta- tion is more d lttor ith t of other Paleocene Coconut like plants that are currently known. e infructescences are up to 40 mm long and bear eight to 14 closely spaced follicles at dis- tances of approximately 2-4 mm (Fig. 16). The follicles were borne singly (Krassilov, 1976, pl. 23: fig. 5) or in pairs (Fig. 18). None of the Amur material shows a distinct joint between the in- fructescence axis and the follicle stalk, although some of the dispersed follicles appear to have become Pn at this point (Krassilov, 1976, pl. 24: fig. 9). The orientation of ventral sutures in the follicles i is unclear. ICUVIILUU ile w <— (Natural History). Specimens prefixed PP are from Decker, Montana (Tongue River Member, Fort Uni Cercidiphyllum-like plant from Decker, M scars (arrows) where follicles have abscissed, and the abaxial orientation of the follicle sepe with respe . N. arcticum showing a pair of follicles attached to common 208 the infructescence axis (PP34177), х3.— 14 mation, = Paleocene) and аге in the paleobotanical collections of the Field Museum of Natural History ssidium arcticum (Heer) Iljinskaya, long shoot with attached infructescences: note whorls of d E N. follicle suture with respect to el уена: 1 axis (PP34178), х 2.7. Scale 0 5 тт 388 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES Pepe Trochodendrocarpus arcticus (Heer) oe 3 pe lower Paleocene of Amur U.S.S.R.— 16. Long shoot with Mor d infructescences: note x ].— 17. Long shoot with attached infructescenoes note scars at arrows, х 0.6.— 18. Pair of follicles ae pau о attached to а common stalk, . Scale bars, 10 mm. шш. courtesy of V. Krassilov DISCUSSION es and extant Cercidiphyllum, but these short shoots Shoot growth and phyllotaxy. Cercidiphyl- аге of quite different types. Those of Joffrea are lum and its fossil relatives exhibit considerable similar to those found in some gymnosperms variation in shoot growth. Long and short shoot (е.р., Ginkgo) and many flowering plants (e.g., differentiation is known in both Joffrea speirsii Prunus), which involve small growth increments 1986] of an otherwise normal monopodial shoot. The sympodial short shoots of Cercidiphyllum are more unusual and less common in flowering plants as a whole. The extant vesselless dicoty- ledon Tetracentron sinense, often interpreted as closely related to Cercidiphyllum (Takhatajan, 1969), also exhibits this unusual sympodial short shoot growth (Bailey & Nast, 1945; Nast & Bai- ley, 1945). In Nyssidium arcticum from southern England and Trochodendrocarpus arcticus from Amur, short shoots are unknown Phyllotaxy in the fossil and Recent plants is as diverse as shoot growth, ranging from alter- nate to opposite to whorled. Leaf morphology in extant Cercidiphyllum is extremely variable (Brown, 1939; Swamy & Bailey, 1949; Chandra- sekharam, 1974) and is further complicated by pronounced differences between long and short shoot leaves. Latest Cretaceous and early Ter- tiary Cercidiphyllum-like leaves exhibit extreme variation in shape and venation (Hickey, 1977; Wolfe, 1966). Some of these leaves may hav been produced by genera only distantly related to Cercidiphyllum, but much of the variation may also be accounted for by the wide range of shoot growth patterns and D in fossil Cerci- diphyllum-like plant Inflorescence miM and organiza- tion. Inflorescence production in Cercidiphyl- lum and its fossil relatives is highly variable (Ta- ble 1), but there is now a clear basis for directly interpreting the organization of the inflorescence in several fossil taxa in terms of the inflorescence of extant Cercidiphyllum (Figs. 19-24). In both the fossil and extant plants the inflorescence is a raceme. The joint between the branch of the in- florescence or infructescence axis and the stalk of the carpel or follicle is comparable to the joint at the base of each follicle in extant Cercidi- phyllum. In extant Cercidiphyllum this joint marks the attachment point of a bract, and a bract may also have been borne at this point in the fossil plants (B, Figs. 19, 20). The joint is clear in Joffrea and Nyssidium arcticum but in- ferred in Trochodendrocarpus arcticus. The joint is also clearly visible in other early Tertiary ma- terial from North America (Figs. 13, 15), and we suggest that it was a general feature of Upper Cretaceous and early Tertiary Cercidiphyllum- like plants. In Joffrea speirsii the base of the inflorescence has associated bud-scales (BS, Figs. 19, 20) that are comparable to those that surround the inflo- rescence, leaf and new axillary bud in extant Cer- CRANE & STOCKEY —CERCIDIPHYLLACEAE 389 cidiphyllum. These bud-scales have not been seen in the fossil material from southern England or Amur. Under the interpretation given above, the in- florescences of the fossil plants are directly com- parable to the inflorescences of extant Cercidi- phyllum, and the homologies hypothesized by Brown (1939) and Swamy and Bailey (1949) are supported and considerably clarified. The differ- ences between the fossil and Recent inflores- cences are principally quantitative and concern the interrelated factors of inflorescence length, follicle number, and follicle crowdin Suture orientation. The айп of the $и- ture in fossil material has frequently been diffi- cult to determine (Crane, 1984; Hickey, 1977), and discussions of this feature are complicated by the need to distinguish between suture ori- entation with respect to the inflorescence axis and suture orientation with respect to the floral axis. In Joffrea, suture orientation can be com- pared in both inflorescences and infructescences. At carpel stage all the sutures of single carpels are clearly adaxially oriented with respect to the inflorescence axis, but in paired carpels the su- tures face each other and their orientation with respect to the inflorescence axis is unknown. At maturity the stalk of single follicles often twists to orient the suture abaxially with respect to the infructescence axis. This may be related to dis- persal of seeds from the follicles (Crane & Stock- ey, 1985), and the same phenomenon may have occurred in other North American early Tertiary material (Brown, 1939, pl. 55: figs. 3, 10). In upper Paleocene material from Decker, Mon- tana, single follicles also have the suture oriented adaxially with respect to the infructescence axis (Figs. 13, 15), but in these cases the lack of ob- at carpel stage may also have been adaxial. The corded in the literature (Basinger & Duces 1983; Crane, 1984; Hickey, 1977) Several explanations have been offered to ac- count for the suture orientation in extant Cer- cidiphyllum (Harms, 1916; Hutchinson, 1964; Leróy, 1980; Solereder, 1899; Swamy & Bailey, 949). Based an the interpretation of the Cer- an apocarpous flow- er, Hutchinson (1964) suggested that the appar- ently abaxial suture orientation was anomalous 390 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 19, 20. Interpretative drawings showing homologous structures in the infructescences EX Cercidi- phyllum and Joffrea. — 19. Cercidiphyllum japonicum Sieb. & Zucc. Short shoot with infructescence. — 20. Joffrea speirsii Crane & Stockey. Short shoot with infructescence. AB = axillary bud, B = bract, BS = bud zi F= follicle, LS = leaf scar, P- petiole, TB = terminal bud. 1986] CRANE & STOCKEY —CERCIDIPHYLLACEAE 391 TABLE 1. Comparison of extant Cercidiphyllum with fossil Cercidiphyllum-like plants. See text for details of interpretations Trochodendro- carpus arcticus Cercidiphyllum Joffrea speirsii Nyssidium arcticum (Paleocene, (Extant) (Paleocene, Canada) (Paleocene, England) U.S.S.R Shoots Long ara dl Long shoots, mono- Long shoots, short Long shoots, short e podial short shoots absent? shoots absent? oots shoots Phyllotaxy Opposite (occasion- — Whorls or pseudo- Alternate Position of inflo- B subop- posite, or whorled on long shoots, single on short shoots Terminal on sym- podial short ally alternate?) on long shoots, oppo- site on short shoots In leaf axils on monopodial short whorls Terminal on long shoots In leaf axils on long shoots shoots shoo Inflorescence type Raceme Raceme Raceme Raceme Maximum length 25 mm 130 mm 130 mm 40 mm of infructes- Follicle number 2-8 ca. 40 ca. 15 8-14 per infructes- cence Follicle orienta- Always abaxial Single follicles, ie follicles, Unknown tion with re- adaxial initially, daxial, some spect to infruc- some twisted and d at maturi- tescence axis abaxial at maturi- ty; unknown in ty; unknown in paired follicles paired follicles Follicle number Always 1 l or 2 l or 2 l or 2 er flower Follicle orienta- ?Adaxial ?Adaxial, certainly ?Adaxial, certainly Unknown tion with re- adaxial when 7 to floral paired axis? adaxial when paired @ Follicle orientation with respect to floral axis is difficult to assess in unicarpellate flowers. and due to twisting of the carpels. This is not supported by detailed developmental studies or by carpel and follicle vasculature (Swamy & Bai- ley, 1949). The complex explanations of Harms (1916) and Leréy (1980) both involve transfor- mation of a bud-scale on a vegetative shoot into a carpel, the single leaf of the shoot being trans- formed into a bract (Harms, 1916) or being lost (Leróy, 1980). Solereder’s (1899) view is more straightforward, simply involving the loss of one of a pair of carpels that originally had opposite ventral sutures (adaxial with respect to the floral axis) and were oriented perpendicular to the in- florescence axis. This interpretation receives some support from the variable vascular supply to Cer- cidiphyllum carpels: some carpels have the dorsal vein supplied by a single vascular bundle from the stele, whereas others are supplied by two vas- cular bundles which fuse in the base of the carpel (Swamy & Bailey, 1949: 192, fig. 12). Equally significant however is the regular occurrence of pairs of follicles that have opposite oue su- tures in the inflorescences of Joffrea arc cum, and T. arcticus (Figs. 6, 7, 14, 18). Fuel it is unknown whether the paired carpels and follicles are perpendicular or transverse with re- spect to the inflorescence axis, in all other re- spects the fossil plants conform to the archetype 392 Ny ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Y 2. у Lo WS MAN FIGURES 21-24. phyllotaxy d infi position in extant and fossil poene uen dnd —21. pile жалы japonicum yee short shoot with a single leaf, axillary bud, and terminal infructescence. — 22. monopodial short shoot with opposite and decussate leaves, terminal bud, and axillary and showing long shoot with bone leaves and terminal carpus arcticus (Paleocene) from Amur showing long shoot with alternate Massil eiusd е sides southern Engla infructes 4. Trochodendro leaves Anda an dp ct e ofextant Cercidiphyllum envisaged by Solereder. Fossil infructescences more similar to those of extant Ci t recorded from the Oligocene (Jáhnichen et al., 1980). CONCLUSIONS Extinct latest Cretaceous and early Tertiary Cercidiphyllum-like plants are very diverse in phyllotaxy, shoot growth, pistillate inflorescence production, and the number and crowding of follicles per infructescence (Figs. 21-24). This extreme morphological diversity emphasizes that : 0, showing sympodial Joffrea speirsi (Paleocene) showing infructescences. — 23. the complex of plants that produced Cercidi- phyllum-like leaves during the latest Cretaceous and early Tertiary was highly heterogeneous, and further variations other than those considered here remain to be described in detail (Basinger & Dilcher, 1983). Extreme caution will therefore necessary in assigning isolated and poorly understood organs of these Cercidiphyllum-like plants such as Ji relationships among these fossil plants remain to 1986] be resolved by detailed studies of well-preserved inflorescences and shoots. Nevertheless, despite the structural diversity of Cercidiphyllum-like d e the direct comparability of the inflores- cences of extant Cercidiphyllum to those of the fossil Cercidiphyllum-like plants is now estab- lished. The fossils also provide support for the hypothesis that the suture orientation in extant Cercidiphyllum is due to loss of one of a pair of opposite carpels. The suture in extant Cercidi- phyllum is therefore abaxial with respect to the inflorescence axis but probably phylogenetically adaxial with respect to the floral axis of the “an- cestral" flower. The recognition that bicarpellate flowers may be the primitive condition in the Cercidiphyl- laceae strengthens the idea that the family may be closely related to the Hamamelidaceae (Cron- quist, 1981). The close resemblance among the infructescences of Joffrea, Nyssidium, and Trochodendrocarpus and the racemes of unisex- ual, often apetalous, bicarpellate flowers seen in Sinowilsonia (Hamamelidaceae, Endress, 1977) is particularly striking. The phylogenetic rela- tionships between the extant and fossil Cerci- diphyllaceae and Hamamelidaceae deserve de- tailed study aimed at determining whether bicarpellate flowers constitute an important de- fining character for this group. LITERATURE CITED ВАНЕУ, I. W. & C. G. Nast. 1945. Morphology and relationships of Trochodendron and Tetracentron, I. Stem, root, and leaf. J. Arnold Arbor. 26: 143- E BasiNGER, J. Е. & D. L. DitcHer. 1983. Fruits of и from the early Tertiary of Elles- mere Island, arctic Canada. Amer. J. Bot. 70(5, 2): 67. [Abstract.] Brown, R. W. 1939. Fossil leaves, fruits and seeds RO of Cercidiphyllum. J. Paleontol. 13: 485-499. 62. Paleocene flora of the Rocky Moun- tains and Great Plains. Profess. Pap. U.S. Geol. Surv. 375: 1-119. CHANDLER, M.E. J. 1961. The Lower Tertiary Floras of Southern England, I. Palaeocene Floras: Lon- don Clay Flora сно & ail British Mu- seum (Natural History), Lo CHANDRASEKHARAM, A. 1974. rem flora from the Genesee locality Alberta, Canada. Palaeon- CRANE, P. R lum-like plant fossils from the British early Ter- tiary. Bot. J. Linn. Soc. 89: 199-230. 85. Growth and repro- ductive biology of Joffrea speirsii gen. et sp. nov., CRANE & STOCKEY — CERCIDIPHYLLACEAE 393 a Mery cles like plant from the Late Paleo- Alberta, Canada. Canad. J. Bot. 63: 340- CRONQUIST, A. 1981. An Integrated System of Clas- sification of Flowering Plants. Columbia Univ. . 1977. Evolutionary trends in the Hamamelidales-Fagales group. Pl. Syst. Evol. Suppl. 1: 321-347. 1986. Floral structure, systematics and phy- logeny i in Trochendrales. Ann. Missouri Bot. Gard. 73: 297-324. Harms, H. 1916. Uber die Blütenverháltnisse und die systematische Stellung der Gattung Cercidi- phyllum Sieb. & Zucc. Ber. Deutsch Bot. Ges. 34: Hickey, L. J. 1977. Stratigraphy and paleobotany of the Golden Valley Formation (early Tertiary) of western North Dakota. Mem. Geol. Soc. Amer. 150: 1-293 HurcHINSON, J. 1964. The Genera of Flowering Plants, Dicotyledons, I. rie = Oxford. JAHNICHEN, H., D. H. MAI 1980. Blatter und Friichte von pene Hen Siebold & Zuccarini im mitteleuropáischen Tertiár. Schrif- tenreihe Geol. Wiss. 16: 357- KRassiLOv, V. A Mesozoic plants and the problem of angiosperm ancestry. Lethaia 6: 163-1 78. . 1976. The Tsagayan Flora of Amur Region. Nauka, Moscow. [In Russian.] 9 The origin of angiosperms. Bot. Rev. (Lancaster) 43: 143-176. LERÓY, J. Е. 80. Développement et organogenése chez le Cercidiphyllum japonicum: un cas sem- ST, Morphology and relationships of Trochodendron and Tetracentron, nflorescence, flower, and fruit. J. Arnold Ar- m 26: 267-2 dro en JAGER, A. 1958. Altertiáre pflanzen aus n der (n NET eur Palae- Du E a, Abt. B, Palàophy 0 9-103. SOLEREDER, H. 99, Zur oe. und Syste- matik der Gattung Cercidiphyllum Sieb. & Zucc., мет Berucksichtigung der Gattung Eucommia Oliv. Ber. Deutsch Bot. Ges. 17: 387-406. SPONGBERG, S. A. 1979. Cercidiphyllaceae hardy in temperate North America. J. Arnold Arbor. 60: 367-376. StockEY, В. A. & P. В. CRANE. 1983. In situ Cer- cidiphyllum-like seedlings from the Paleocene of J. р 70: ^ ин & I. W. ВАпЕҮ. 1949 mor- phology and relationships of E J. Arnold Arbor. 30: 187-210. TAKHTAJAN, A. 1969, pas de a эм Origin and pi а Oliver & Boyd, WoLFE, J. А. 1966. Tertiary m ca the Cook Inlet Region, Alaska. Profess. Pap. U.S. Geol. Surv 398B: 1-32. FRUIT AND SEED DISPERSAL AND THE EVOLUTION OF THE HAMAMELIDAE! BRUCE H. TIFFNEY? ABSTRACT The assumptions (1) that the Hamamelidae attained their zenith in the Cretaceous and, (2) that abiotic dispersal dominated i in Cretaceous ойор perms, Suggest that dispersal mode could be used as ciated with the Hamamelidae. A review of modern a character i in evaluating y) most “lower” н amamelids are а dispersed, but that кок sal. In ma putatively ч families Aa g., Fagaceae, cases the dispe abiotic to biotic dispersal around the O Circumstan vide suggests a Moraceae/Cecropiaceae/Urticaceae comp orthy evolutionaril ) emphasizes impo ss biotic dispersa yp n sal mechanisms of a family are the same in the fossil record and the present da xi However, in he Juglandaceae and Fagaceae the fossil record indicates fi е Tertia the Mor. witch i ircum oie anal fossil and den aceae/Cecropiaceae/ Urticac eae and possibly the suggest that families dominated by biotic dispersal are more diverse than families dominated by abiotic dispersal; (4) the Cretaceous-Tertiary boundary marks a time of major change in dispersal mode in the angiosperms; (5) the primitive fruit morphology of the Urticales appears to be the achene; and (6) derivation of fleshy structures from extra-ovarial tissues plays an important role in the dispersal of many species of the Moraceae, Cecropiaceae, and Urticaceae BACKGROUND AND HYPOTHESIS The Hamamelidae are increasingly recognized as having been significant in the early history of the angiosperms. Members of the group have been traced to the Early Cretaceous (e.g., Doyle & Hickey, 1976; Hickey & Doyle, 1977) and in the earliest portions of the Late Cretaceous (Schwarzwalder & Dilcher, 1981); it is possible that the ко may have had a separate origin om the gnoliidae (Nixon, unpubl. data). Similarly, eae for the importance and di- versity of Hamamelidae in Cretaceous floras is growing. In particular, Leo Hickey (unpubl. data) has suggested, on the basis of fossil leaves, that the Hamamelidae reached a zenith of diversity in the Cretaceous and, with certain exceptions, have decreased in importance to the present day (see Cronquist, 1981: 153 The fossil record of the angiosperms also re- veals a pattern in relative sizes of diaspores. Cre- taceous fruit and seed floras are dominated by small diaspores (1-3 mm on largest axis). Ter- tiary fruit and seed floras contain similarly small diaspores but also possess many much larger ones (Tiffney, 1984). As a broad generalization, small fruits and seeds may be abiotically or biotically dispersed, whereas large fruits and seeds are more often biotically dispersed. The small size of the Cretaceous angiosperm diaspores, together with the general absence of modern dispersal agents, suggests that Cretaceous angiosperms were large- tant in angiosperm biology today. These animals moved larger seeds than would abiotic means and were contributing factors to the evolution of modern angiosperms forming closed-canopy, late-successional forests such as those found in the warm-temperate and tropical areas of the world today (Tiffney, 1984). In sum, this scenario ' I thank Michael Zavada and David Dilcher (Indiana University) for the opportunity to examine this topic ough their invitation to participate in the Symposium thro the leaf fossil record an led and c nd for his criticisms throughout the gestation of this paper, and three Hamamelidae, Leo Hickey for information on reviewers for their m on the detai onstructive criticisms. Research was partially supported by the United States National Science 306002. Foundation т. Soap ? Peabod Haven, Conosci d 06 ANN. Missouni Bor. GARD. 73: 394-416. 1986. m of la History and Department of Biology, Yale University, P.O. Box 6666, New 51 1986] suggests that Cretaceous angiosperm groups should be dominated by abiotic dispersal, and that biotic dispersal should be significant only in groups that either evolved in the Tertiary or al- tered their dispersal biology during that period. If the assumption that the Hamamelidae are primarily a Cretaceous group is correct, then all *true" Hamamelidae should show fossil or re- cent evidence of abiotic dispersal. In theory, this provides one test of hypotheses on the systematic affinities of certain families [e.g., Juglandaceae (Hickey & Wolfe, 1975)] whose association with the Hamamelidae has been questioned. Families dominated by biotic dispersal mechanisms would presumably have affinities outside the Hama- mélidae, whereas those with abiotic dispersal at least not disqualified fr om the sub-class. However, this conclusion rests on the further assumption that the primary dispersal mode within a family has not changed over time. The reader is cautioned that, while focused on dispersal, this paper carries no implication that dispersal biology is more (or less) important than any other character in elucidation of the phylog- eny of the Hamamelidae. METHODS AND MATERIALS In this paper, I follow Cronquist (1981) for disposition of families within the sub-class Ham- amelidae and Willis (1973) for generic com- position of the families. Data on the modern modes of dispersal and associated information (Table 1) were assessed at the generic and occasionally the specific level, and were compiled from the following sources: Engler and Prantl (1894), Prain (1917), Standley (1920-1926, 1928), Engler and Prantl (1930), Ridley (1930), Standley and Steyermark (1946), Lawrence (1951), Martin et al. (1951), Standley and Steyermark (1952), Hutchinson and Dalziel (1954), Vink (1957), Jacobs (1960), Backer (1963), Hutchinson (1964), Melchior (1964), Owhi (1965), Hutchinson (1967), Radford et al. (1968), van der Pijl (1969), Miller (1971), Willis (1973), Walker (1976), Montgomery (1977), Soe- padmo (1977), Croat (1978), Heywood (1978), Chang (1979), Kuang and Lu (1979), Li and Cheng (1979), Anonymous (1980), Elias (1980), Ming (1980), Wiggins (1980), Cronquist (1981), Dassanayake and Fosberg (1981). These sources are cited specifically only where манар. In ] mode the relative contribution of the different modes TIFFNEY —HAMAMELIDAE FRUIT AND SEED DISPERSAL 395 is expressed as a percentage of the total number of genera or species within the family. Cases in which dispersal mode is not recorded in the lit- erature, or where it appeared equivocal or gen- eralized in light of fruit or seed morphology, are noted individuall The fossil record of angiosperm fruits and seeds is good in the Tertiary (Tiffney, 1977) but scanty in the Cretaceous. Only a few fruits and seeds of the Hamamelidae are known from the Creta- ceous, although the record of the group is much better in the Tertiary. The appearance of modern dispersal agents roughly at the Cretaceous-Ter- tiary boundary presumably had a substantial ef- fect on fruit and seed morphology, and may have influenced the appearance of “modern” charac- ters. The absence of modern fruit and seed taxa in the Cretaceous may also partially be an artifact of scientific interest. Only within the last 15 years have researchers seriously examined Cretaceous angiosperm fruit and seed floras (e.g., Knobloch, 1971, 1977; Friis, 1983, pers. comm.; Knobloch & Mai, 1984; Tiffney, unpubl. data). The fossil record of the families of the Ham- amelidae is summarized in Figure 1. Figures 2- 8 summarize the history of families with many genera reported from the fossil fruit and seed record. The data come from the primary litera- ture and are taken from a file on the occurrences of fossil fruits and seeds that I have been assem- bling as a parallel to Muller's (1981) summary of the fossil pollen тооз of modern families. The generic re accepted as giv- en by the primary authors. I have excluded some clearly erroneous reports but have not examined every occurrence in detail. In some cases this may be misleading. For example (Tiffney, in prep.), it may prove impossible to separate the seeds of many modern genera of the Hamamel- idaceae in the fossil record. Thus, the reports of individual genera of Hamamelidaceae in Figure may be misleading. Similarly, biases of pres- ervation, mosaic evolution, and the attitudes of examining scientists may result in the placement of extinct forms in modern genera, or may cloud the recognition of modern genera in the fossil record. Stratigraphic locations are taken as reported by primary authors except where clearly in error. Localities from western North America are dated from Evernden and James (1964) and Wolfe (1981). Period and Epoch durations are based on the time scale of Harland et al. (1982). The summary of dispersal patterns in the fossil 396 TABLE 1. ANNALS OF THE MISSOURI BOTANICAL GARDEN the Hamamelidae sensu Cronquist (1981). See text for sources [Vor. 73 Characters of diversity, dispersal, distribution, habit, ecology, and pollination for the families of Number Genera Genera/ Abiotically Biotically Family Species Dispersed Dispersal Morphology Dispersed Tetracentraceae (1) 1/1 100% winged seed 0 Trochodendraceae (2) 1/1 100% winged seed 0 Cercidiphyllaceae (3) 1/2 100% winged seed 0 Eupteleaceae (4) 1/2 100% winged fruit 0 Platanaceae (5) 1/6, 7 10096 hairy achene or nutlet 0 Hamamelidaceae (6) 28/100* 100% winged seeds (18%) 0 ballistic seeds (82%) Myrothamnaceae (7) 1/2 100% tiny seeds in capsule 0 Daphniphyllaceae (8) 1/35 0 fleshy drupe 100% Didymelaceae (9) 1/2 0 large drupaceous fruit? 100% Eucommiaceae (10) 1/1 100% winged fruit 0 Barbeyaceae (11) 1/1 100% nut with large accrescent se- 0 pals Ulmaceae (12) 18/150 33% winged fruits and fleshy to 67% semi-fleshy drupes Cannabaceae (13) 2/3 unclear—the bracts aid in wind dispersal, but Ridley (1930) records animal dispersal Moraceae (14) 51/1,333 8% genera dry achenes, some drupes, 78% genera (13% spp.) many pseudo-drupes (86% spp.) Cecropiaceae (15) 8/275 0 pseudo-drupes from fleshy flo- 10096 ral parts Urticaceae (16) 46/1,255 3796 genera abiotic via dry am ballis- 47.596 genera (66% spp.) tic, or winged fru (29% spp.) biotic via drupes, - ries, receptacle-fruit, sticky surfaces, and eliasomes Leitneriaceae (17) 1/1 100% (?) dry, possibly floating, drupe 0 Rhoipteleaceae (18) 1/1 100% winged nut 0 Juglandaceae (19) 9/60 56% genera winged nut or drupaceous 44% genera (37% spp.) (63% spp.) Myricaceae (20) 3/50 unclear—no particular adaptations exist, and both abiotic and biotic dispersal observed Balanopaceae (21) 1/9 0 acorn-like dru 100% Fagaceae (22) 8/800 0 nut within cupule; rarely mild- 100% ly winged Betulaceae (23) 6/120 83% nut or samara 17% Casuarinaceae (24) 1/50 100% samaroid 0 397 1986] TIFFNEY —HAMAMELIDAE FRUIT AND SEED DISPERSAL TABLE 1. Continued. Decidu- Ever- Dioecious/ Distribution Habit green Monoecious Dispersal Unit Pollination Mode Nepal, China, Burma tree D perfect seed anemophily Korea, Japan, Taiwan ree E androdioecious seed secondarily ane- mophily China, Jap tree D dioecious seed anemophil China, Pun Assam tree D perfect to prot- fruit anemophilous, androus some ento- mophil Mediterranean to Hi- tree D monoecious fruit anemophily alayas, As à Mexico to Canada subtropical to temper- trees and D&E perfect or uni- seed primarily ento- ate regions, Old and shrubs sexual mophily New Worlds Africa, Madagascar small shrub ? dioecious seed anemophily E Asia, Malaysia trees or shrubs ? dioecious fruit ? Madagascar tree E dioecious fruit ? anemophily ina tree D dioecious fruit anemophily NE Africa, Arabia small tree ? dioecious fruit anemophily tropical and temperate trees, shrubs, D&E monoecious or fruit anemophily emi- and vines perfect sphere М temperate to tropical herbs — monoecious or fruit anemophily ioeciou widespread, largely trees, shrubs, D&E monoecious ог fruit or “fruit” но ropical and vines dioecious of floral some ento- mopni tropical trees, shrubs, ? rarely monoe- fruit or “fruit” Mass 8 and vines ious, com- of floral some ento- monly dioe- mophily cious tropical and subtropi- herbs, pini — monoecious, fruit or “fruit” — anemophily cal, temperate vines, a dioecious, of floral trees Te and polyga- parts ostly mous herbs) SE United States shrub to small D dioecious fruit anemophily tree SW China, Vietnam tree D cunda some- fruit anemophily nisex- ua Northern Hemisphere trees D monoecious fruit anemophily temperate and tropical, shrubs D&E monoecious fruit anemophily Id and New Worlds SW Pacific, New Cale- tree E dioecious fruit anemophily cosmopolitan, except trees and D&E monoecious, fruit anemophilous, Africa seldom dioe- secondarily cious, rarely entomophi- t lous in some species g ei Northern trees and D monoecious fruit anemophilous Hemisphere shrubs SW Pacific, ats trees and E monoecious fruit anemophilous Indomales shrubs 398 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Уог. 73 I T | T T T T т" TROCHO " | Trechodendiabaae | _letracentraceé o Cercidiphyllacea А i I T T7, E E ue } | | № | | | Expteleaceae | < | aries | e G | „Platanaceae | | = PUR " ESA SEO aliud E | Hamamelidaceae 5 тт? i р Ж ш: = | | | т | |Myrothamnaceae e Y DAPH | IDaphniphyllacead в DID. T baies ai I è | | | EUCOM. | [24 + i 4 | | | | Barb ›уасеае e | Лтасеае 1 А | | LI n | | | Cannabaceae | | 5 р) zi | | ————— = C < l | [ ш О | | Могасеае | T Е | | 1 б, re № O = | | Cecropiaceae "YP к Urticaceae ` | | 216 | | os | ч | | | um T LI Т D | | | Leitneriaceae | Ра | = ? 7 Rhoipteliaceae | n ' | | 5 ш ди anne ene l d а 9 7 iam | : mo beds = о | + t MYRIC p----1---1----1--1---- 1 j 9 0] | | | |Ва!апорасе ле е = ‚ ragaceas | UN | | x TS sen = T 2 ! — Betulaceae Fd к —O ИЕ И dibus dans a eos ene | | D I y — LI 1 . нін ыса, : 1 Е 1 < 4 К Р T о | | E | 2.1 | Ep Ерм ра [Еро Е IM|L > ; = | | | | | | CENO. |5[9| $5 | CAMP. |МАА.] РЕ ЕОСЕМЕ OLIGOCENE MIOCENE РОО | | 97.5 91.0 83.0 730 65.0 ! 54.9 | ! 38.0 24.6 S.L 2.9 FIGUR t-occurrence data for families of the Hamamelidae sensu Cronquist (1981) based on fossil fruits, irt rl pollen. Solid line, record of biotic dispersal within a family; dotted line, record of abiotic dispersal within a fam jug к у line, first a for pollen n Muller, 1981). CENO. = Cenomanian, TURON. = Turoni = сола. SANT. = - Santonian, CAMP. = Campanian, MAA. = Maastrichtian, = Paleocene, PLIO = Si Q = Quaternary, E = Early, M = a iddle, L = Late. Absolute dates after Harland et al. (1982). Trocho. = Trochodendrales, Daph, = и Did. = impe sed Eucom. = Eucommiales, Leit. = Leitneriales, Juglan. = Juglandales, M — Myricales, Cas. — Casuarinales. Notes: (1) ala dispersed; (2) Biotically dispersed; (3) Eucommia.. like fruits of the early Eocene. "Pierce and Hickey s. comm.); (4) Abiotically dispersed; (5) Cannabaceae appear to have “generalized” dispersal mechanisms; (6) Muller (1981) suggested possible Oligocene je of e (7) Friis (1983) reported pem that conform most closely to those of the Juglandaceae but also possess similarities with those of the Myricaceae and Rhoi- pteleaceae; (8) Casholdia, Polyptera, and Cyclocarya, Pall wind dispersed, appear in the Upper Paleocene. The first animal-dispersed fruit of the family is Juglans, which appears in the Middle Eocene; (9) Biotically dispersed; om in the Early Eocene; Quercus and possibly Trigonobalanus (animal dispersed) appear in the Middle Eocene. 1986] record is based only on fossil fruit and seed rec- ords. Mosaic evolution (e.g., Manchester, 198 1a; Knoll et al., 1984; Stebbins, 1984) dictates that it is inappropriate to infer a modern dispersal mechanism on the basis of a fossil leaf or pollen grain belonging to a modern genus. In these de- scriptions, when no reference is cited for a spe- cific conclusion (e.g., that the morphology of the seeds of Tetracentron suggests abiotic dispersal), the conclusion is that of the present author. RESULTS Tetracentraceae. The fruit is a follicetum, dehiscing to release small seeds with spongy, wing-like outgrowths (Cronquist, 1981; Law- rence, 1951). The size and wings suggest abiotic dispersal. bli (1974) placed bip que speci- mens of Nordenskioldia Heer in the Trocho- dendrales, but there is no reason to accept these as members of either the Trochodendraceae or Tetracentraceae. The same reference also places Nýssidium Heer in the Trochodendrales, but Crane (198 f many fossils assigned to this genus with the Cercidi- phyllaceae (see below Trochodendraceae. The fruit is a follicetum, dehiscing to release many quite small seeds. The seed size suggests abiotic dispersal. Reid and Chandler (1933) described Trocho- dendron(?) pauciseminum Reid & Chand. from the Early Eocene London Clay flora of England. The identification was given with a question mark, and is based on five- to six-loculed, sep- ticidal fruits containing small seeds with (wing- like) extensions. If the identification is correct, then Trochodendron was abiotically dispersed in the Eocene. Cercidiphyllaceae. The separate follicles bear small, asymmetrically-winged seeds. Wind dis- persal is redonda in ko literature and borne out by personal obse The follicles of ign РРР are fairly com- mon fossils (e.g., Brown, 1939; Jáhnichen et al., 1980; Crane, 1984a). The oldest presently veri- fied fruit record is from the Paleocene (Brown, 1962; Crane, 1984a). However, Crane (1984a) demonstrated that many fruits assigned to the form genus Nyssidium Heer represent the Cer- cidiphyllaceae, and reports of Nyssidium extend back to the Turonian (Takhtajan, 1974; see also Krasilov, 1976). Thus, fruits of the family could be present in the Late Cretaceous. It is likely that TIFFNEY —HAMAMELIDAE FRUIT AND SEED DISPERSAL 399 the Cercidiphyllaceae were more diverse in the Late Cretaceous and early Tertiary. Crane and Stockey (1985) describe an extinct multi-organ assemblage as the genus Joffrea Crane & Stockey from Late Paleocene sediments: while varying in other characters, its seeds are very similar to those of Cercidiphyllum. Seeds found with fossil Cer- cidiphyllum follicles (Reid & Chandler, 1933; Crane, 1984a) resemble the seeds of the extant species, suggesting wind dispersal. teleaceae. The fruits are small samaras or winged nutlets adapted for wind dispersal. No fossil fruits are reported. Platanaceae. The infructescence is a globose head of densely hairy achenes or nutlets that are shed at maturity. The hairs increase surface area, and the fruits may be wind dispersed or float on water. Schwarzwalder and Dilcher (1981) described Cenomanian-age leaves and infructescences from Kansas that they suggest belong in the Platana- ceae. Friis (1984) reported platanaceous inflo- rescences from the Late Cretaceous of North Carolina and Sweden. Hickey (pers. comm.) in- dicated that “Platanus-like” infructescences oc- cur in the Lower Cretaceous sediments of the Potomac Group (see Hickey & Doyle, 1977). These records suggest that the Platanaceae are the oldest family of the Hamamelidae repre- sented by fruiting remains. The morphology of the individual fruits is often similar to that of the fruits of modern P/atanus, suggesting abiotic dispersal, although Manchester (1986) notes an Eocene Platanus with slightly larger fruits lacking the pappus-like hairs ofthe modern genus. These fruits are still quite small (2-3 mm) and Man- chester (pers. comm.) suggests that they are prob- . The majority of genera possess a ballistic dispersal mechanism in which the seed is “squirted” out from the fruit by pressures created in the fruit wall by moisture loss (e.g., Hamamelis L.). However, in the Al- tingioideae (A/tingia Nor., Liquidambar L.) the seeds are very small and occasionally possess wings [although Vink (1957) noted that the oily seeds of Altingia have been observed to attract onkeys, birds, and ants]. Similarly, in the Hamamelidoideae, Exbucklandia R. W. Brown possesses large, winged seeds indicative of wind dispersal, and Rhodoleia Champ. ex Hook. has small, disk-like seeds dispersed by wind Both Liquidambar and the extinct genus Stein- hauera Presl appear in the Paleocene (Chandler, 400 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 [ Dod T pg | Liquidambor | 1 | t Te | ! | | | Соор | | 1 | | . Rhodoleio' - - lik | Rhodoleia | | l | | Disanthus | | T | Fortunearia | | | Distylium | | | | Hamamelis | | | | 1 Bucklandia | | Т | Steinhauerat ! | I 1 | | | | | I | | Fothergilla | | = e— | | | Parrotio | | | Сане | | | | | | z E L E! M L E L E MI L О 12| | СЕМО с SANT] CAMPANIAN | MAA. PALEOCENE | EOCENE OLIGOCENE MIOCHNE PLIOJQ E 1 1 | FIGURE 2. First-occurrence data for genera in the Hamamelidaceae based on fossil fruits and seeds. = ventions as in Figure 1, Chandler (1961b); “Rhodoleia-like” (Klikovispermum Champ., Mai and Walther (1985); Disanthus Maxim + indicates extinct genus. Liquidambar L., Chandler (1961b); Corylopsis Sieb. & Zuc Knob. & Ma i), Knobloch and Mai , Mai and Walther (1978); Fortunearia Rehdr. & Wils., (1984); Rhodoleia Mai and р ие Distylium Sieb. & Zucc., Takhtajan (1974): Hamamelis L., Zablocki (1930); Bucklandia . Br. ex Griff., first possible appearance, Brown (1946), later possible appearances, Reid and Reid (1915), Szafer (1946); Fothergilla L., Szafer (1946, 1954); Parrotia C. A. Mey., Tralau (1963); Steinhauera Presl, Mai (1968). 1961b; Mai, 1968; Collinson, unpubl. data). The shape of the fruits and seeds suggests wind dis- persal. Knobloch and Mai (1984) drew a com- parison between the fossil seed K/ikovispermum waltheri Knobl. & Mai from the Maastrichtian of Czechoslovakia and the seeds of Rhodoleia Champ., but the first clear report of Rhodoleia is provided by Mai and Walther (1985) in the Late Eocene. Ballistically-dispersed members of ported by Mai & Walther, 1978), and Distylium Sieb. & Zucc. (Takhtajan, 1974). There is no evidence in the fossil record that the Hamamel- idaceae were ever anything but wind or bal- listically dispersed. Myrothamnaceae. The fruit is a dehiscent capsule bearing many seeds (Cronquist, 1981), which Willis (1973) states are small. This sug- gests abiotic dispersal, but specific evidence is lacking. There is no fossil record of the fruits of the family. Daphniphyllaceae. The fruit is a one-seeded, fleshy, indehiscent drupe that is black or green at maturity and about 1 cm long (Ridley, 1930; Walker, 1976; Cronquist, 1981). The fruit char- acters suggest biotic dispersal. There is no fossil record of the fruits of this family. Didymelaceae. The fruit is “a large one- seeded drupe, with lateral grooves (as in Pru- nus)" (Willis, 1973). The morphology is sugges- tive of biotic dispersal. There is no fossil record of the fruits of this family. Eucommiaceae. The fruit is a large samara, similar in shape to those of Ailanthus Desf. or North America (see note under Ulmaceae). The 1986] TIFFNEY – HAMAMELIDAE FRUIT AND SEED DISPERSAL 401 р | Gironniera | | Ulmus l Celti I | ! | | Chaetoptelea | | | | | "Banksites" lineatus | i Е | | | "Embothrites" borealis | | | | | Aphananthe | Zelkova | i| ?Pteroceltis | Trema | | | | > E | L E! М L E L E MI L о |2 I | СЕМО. 5 @5АМТ| CAMPANIAN | МАА. PALEOCENE | ЕОСЕМЕ OLIGOCENE MIOCENE PLIO |Q 1 FIGURE First-occurrence data for genera in the Ulmaceae based on fossil fruits and seeds. аа аѕ іп Figure 1. Gironniera Gaudich., unpublished data cited in Mai and Gregor (1982); Ulmus L., first report Brown (1962), next report Givulescu (1980); Celtis L., unpublished specimens кошы. by Leo J. Hickey; Chaetoptelea Liebm., MacGinitie (1941); * Banksites" lineatus Unger, N Unger, Manchester (in prep.); Aphananthe Planch. , Takhtajan (1982); Zelkova Spach., a (1957); ? Ptero- celtis Maxim., first report Weyland (1937), questioned by Kirchheimer (1957); Trema Lour., Holy (1975). latter fruits are shorter and broader than those of modern Eucommia and might constitute an extinct genus. Similarly, Pierce and Hickey (pers. comm.) are investigating a samara from the Pa- leocene of western North America that is similar to the fruits of the modern genus but differs in several specifics. The last two reports hint at a greater taxonomic diversity in the family in the early Tertiary and underscore the importance of wind dispersal in the group Barbeyaceae. The fruit is a “short-beaked nut with associated accrescent, somewhat membra- nous, prominently veined sepals” (Cronquist, 1981). In some species the length of the sepals exceeds that of the nut by 2:1 (Prain, 1917). These morphological characters suggest wind dispersal, although animal predation and dis- persal of the nut could not be excluded. There is no fossil record of the fruits of this family. Ulmaceae (Fig. 3. The Ulmoideae (five gen- era) possess wind-dispersed samaras, with the exception of Planera J. F. Gmel, which has a fleshy fruit. The Celtidoideae (ca. 12 genera) pos- sess fleshy-walled, animal-dispersed, drupes with the exception of Pteroceltis Maxim., which has a winged fruit, and Chaetacme Planch., which has a tiny, hard-surfaced drupe (Hutchinson, 1967). Twenty-six percent of the species of the family are abiotically-dispersed samaras and 74% are drupes presumed to be biotically dispersed. Fruits of the two subfamilies appear in the fossil record at approximately the same time. The Ulmoideae are first represented by U/mus L., which appears in the Paleocene of western North America (Brown, 1962), although this rec- ord may include some material of Eucommia (P. Crane, pers. comm.). Chaetoptelea Liebm. ap- pears in the Early Eocene of western North America (MacGinitie, 1941), and the extinct form * Banksites" lineatus Unger appears in the Mid- dle Eocene of the same area, persisting to the Early Oligocene (Manchester, in prep.). The ex- tinct form “Embothrites borealis" Unger appears in the Upper Eocene of Europe and persists through the Early Miocene (Manchester, in prep.). The Celtidoideae are first represented by en- 402 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 | : У | | РЕН. | | | Becktonia* | | | р | | Ficus | | y | | | | orus M | |! | | NE | | | | Moroidea * | | | | | | | | Broussonetia | | | | | | Chlorophora | | | | Сиагап!а | | oo | | | | z eli [ЕГ m lupe L E MI L a [] 1 CENO. |2 Ы$АМТ| САМРАМАМ | MAA. PALEOCENE | косме! OLIGOCENE MIOCENE PLIO.|Q FIGURE 4. First-occurrence data for genera in the Moraceae based on fossil fruits and seeds. Conventions as e 1, + indicates extinct genus. Ovicarpum Chandler, pied е о Chandler, first report in Figur Chandler (1963), last report Chandler (196 1a); Ficus L., Chandle о d 1961b); Moroidea y inis L Vent., Chandler (1925-1926, 196 1a); Chlorophora Gaudich., Chandler (1925-1926, 196 1a); Cudrania Trécul, Palamarev ( 1968). docarps of Gironniera Gaudich., in the Paleocene of Europe (unpubl. data cited in Mai & Gregor, 1982). This is closely followed by endocarps of Celtis L. from the Early Eocene of western North America (Leo Hickey, pers. comm.). Only one more genus of the Ulmoideae appears, this in the Oligocene, compared to three more of the Cel- tidoideae; possibly two in the Oligocene and one in the Miocene. [Weyland’s (1937) report of Petro- celtis Maxim. has been questioned by Kirch- heimer (1957).] This imbalance of subsequent appearances could be taken to reflect differential diversification of the two subfamilies; the Cel- tidoideae are more numerous in the present day. Cannabaceae. The fruit is an achene invest- ed to a greater or lesser degree in a persistent calyx. If large enough, the calyx wings permit wind dispersal in Humulus L., but often the wings are reduced and wind dispersal is impossible (Ridley, 1930). Animal dispersal is known in Cannabis L. (Ridley, 1930) and may be signifi- cant. Neither dispersal mechanism is dominant in either genus. Dorofeev (in Takhtajan, 1982) reported four species of Humulus L. and two species of the extinct genus Humularia Dorof. [which requires a new name, as Humularia Duvign. (Legumi nosae) has nomenclatural priority] as appearing in the Oligocene of western Siberia. The first re- port of Cannabis L. is in the Miocene of eastern Siberia (Dorofeev, 1969). The morphology ofthe fruits suggests no particular change from their present generalized" dispersal adaptations. Moraceae (Fig. 4). Out of 51 genera of Mo- raceae examined, 7896 had fleshy diaspores and 8% had dry, presumably abiotically-dispersed fruits. I was unable to ascertain fruit type in 1496. At the species level, 86% of the diaspores are fleshy and 1396 are dry. Based on the total sample ofthe fleshy-diaspore genera, cnly 1196 are drupes (true fruits in which flesh is derived from the carpel wall) whereas over 6796 are dispersal structures in which the flesh is provided by en- larged, accrescent floral parts or enlarged recep- tacles surrounding an achene. This pattern is clearer at the species level, where almost all fleshy diaspores are derived from perianth parts. Thus, although biotic dispersal adaptations strongly dominate in the Moraceae, the attractive dia- spore structure does not develop from the true fruit, but from structures external to it. Floral arts also participate in abiotic dispersal. Ridley (1930) noted in Sloetia sideroxylon Teijsm. & Binneud ex Kurz. that the swollen sepals may squeeze the mature achene out of the floral re- mains with enough force to throw the achene a yard. In Dorstenia L., Ridley noted that drying of the receptacle and enclosed flowers creates 1986] TIFFNEY —HAMAMELIDAE FRUIT AND SEED DISPERSAL 403 Jicameria holyii '+ | 1 | | | rticoidea + | | | | Ы | | | polithes schenkii 2+ | | | | | | 2 Boehmeria | ji | | | 2 | | Urticicarpum scutellum turticicarpum oligocenum + | | | * | | | | | ! | l Pilea | | | | | | І | Laportea | | | ] | | | | | | Urtica | | | | Girardinia | | EE | | | | | | z Е | Le M | Е L E M i = р | | П CENO. |? [SANT] CAMPANIAN| МАА. JPALEOCENE [EOCENE | OLIGOCENE MIOCENE PLIOJQ FiGURE 5. First-occurrence data for genera in the Urticaceae based on fossil fruits and seeds. Conventions as in Figure 1, + indicates extinct genus. Notes: (1) "Senonian" age assigned, this encompasses stages ranging from the Coniacian to Maastrichtian; (2) age given as кү to Paleocene; (3) next report of Во Takhtajan (1982); (4) affinities with the Urticaceae uncerta Mai (1984); Urticoidea Knobloch & Mai, Knobloch Маг Mai (1 Knobloch (1971); Boehmeria Jacq., Knobloch (1971); са scutellum R Dorofeev (in Takh Mai and Walther (1978); Urtica г. Dorofeev (in Takhtajan, 1982): Girardinia perap (1933), Urticicarpum oligocenum Dorof., 1973), Laportea Gaudich., Gaudich., Dorofeev (in Takhtajan, 1982). enough pressure to “shoot” the achenes away from the plant. Fig-like objects (Ficus ceratops Knowl.) have been reported from the Late Cretaceous of west- ern North ран but lack the internal structure l necessary for se- cure systematic assignment. Shoemaker (1977) reviewed these fossils and placed them in the form genus Carpites Schimper. The first assured occurrence of this family is in the Early Eocene oras of southern England, where Ficus L. (Chandler, 1962), Morus L. (Chandler, 1961b), and two extinct genera, Ovicarpum Chandl. (Chandler, 1962) and Becktonia Chandl. (Chan- dler, 1963) are reported. This is followed by the appearance of Broussonetia L'Hérit ex Vent., Chlorophora Gaudich., and the extinct genus Moroidea Chandl. (all Chandler, 1961a) in the Late Eocene. These fossils largely are of achenes that could have been borne within a fleshy berry of the involucre or could have been dispersed dry. If one accepts the identifications of the mod- ern genera (Ficus, Morus, Broussonetia, Chlo- rophora) as correct, biotic dispersal was domi- nant in the family from its first appearance. and extern ehmeria, n, Bicameria holyii Knobloch & Mai, Knobloch and 984); Ip schenkii Knobloch, eid & Chandler, Reid and tajan, 1982); Pilea Lindl., Palamarev Cecropiaceae. All eight genera possess fleshy diaspores indicative of animal dispersal. Similar to the Moraceae, the flesh in these diaspores is primarily derived from floral parts, rather than from the carpel wall. There is no fossil record of the fruits or seeds of this family. Urticaceae (Fig. 5). Asin the Moraceae, the basic fruit type is an achene, a small nut or rarely a drupe, and the floral parts play a strong role in dispersal. However, abiotic dispersal is common in this family; out of 46 genera, over 4796 are biotically eT about 3796 abiotically dis- persed, and data could not be obtained for 15%. The pattern is inverted at ше species leves Al- most 66% of the over 29% biotically, and almost 5% unknown. Of the abiotically-dispersed genera, roughly half had mechanisms developed from the true fruit, whereas the other half had mechanisms depen- dent on accrescent floral parts forming wings, hairs, or aiding in ballistic dispersal. Of the biot- ically-dispersed genera, well over half are berry- mimics based on fleshy floral parts and one-third have sticky or hairy floral parts that are assumed to attach the fruit to animals. In one genus, the A disper seu, 404 floral parts appear to form an eliasome-mimic that attracts ants. Only three и of animal- dispersed forms have true drupe The fossil record of the ып = is “bipar- tite," with a cluster of largely extinct genera of fruits from the Cretaceous and early Tertiary, followed by the appearance of modern genera of fruits in the Oligocene. Knobloch and Mai (1984) relate Bicameria holyi Knobl. & Mai of the Se- nonian and Urticoidea cucurbitoides Knobl. & of the Maastrichtian to the family, and Knobloch (1971) described Microcarpolithes schenkii Knobl. and Boehmeria ctyrokyi Knobl. from undifferentiated sediments ranging from Campanian to Paleocene in age. If the last iden- tification is correct, then it is the oldest report of a modern genus in the group. If not, then the first is Pilea Lindl., which appears in the Upper Eocene (Palamarev, 1973), followed by Boeh- meria Jacq. (Takhtajan, 1982), and Laportea Gaudich. (Mai & Walther, 1978) in the mid-Oli- gocene. As with the Moraceae, it is not possible to infer mode of dispersal from the morphology of the fruits. If one follows the dispersal modes of the modern genera in the fossil record (Boeh- meria, Pilea, Laportea, Urtica), they are abioti- cally dispersed with the exception of Boehmeria. Leitneriaceae. The fruit is a dry drupe with a leathery, slightly spongy exocarp. This could facilitate animal dispersal or, in light ofthe moist habitats in which Leitneria Chapm. grows, pos- sibly aid in water dispersal. Dorofeev (1963; recent stratigraphy from Takhtajan, 1982) reported Leitneria venosa (Ludwig) Dorof. from several Oligocene locali- ties in western Siberia. No difference in dispersal mechanisms from the extant L. floridana Chap- man can be inferred from the fossil. Rhoipteleaceae. The two-winged samaroid nut is wind dispersed. See discussion under Ju- glandaceae for fossil record. Juglandaceae (Fig. 6). The fruit is a nut, which may either be samaroid or drupaceous, corresponding to wind or animal dispersal. By species number, the drupaceous forms (Carya Nutt. and Juglans L.) outnumber the wind-dis- persed ones about 2:1. Further details of dis- persal in this family are provided in the section on the alteration of dispersal mode within fam- ilies. The oldest records of fruits with possible ju- glandalean affinity are of the form genera Man- ningia Friis, Antiquocarya Friis, and Caryanthus ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Friis from the Senonian of Sweden. While these flowers and fruits differ from those of extant Ju- glandaceae in some characters and in their size, and while they may be compared to modern flowers and fruits of the Myricaceae and Rhoi- pteleaceae in some respects, Friis (1983: 185- 186) concluded “ће best correlation of the fossil fruits and floral structures described here is with members of the Juglandaceae." Given this evi- dence, it seems best to recognize the fossils as closely associated with the precursors of modern Juglandaceae, if not actually representative ofthe family. Cyclocarya lljinskaja co-occurs in the upper Paleocene with the extinct genera Polyptera Manchester & Dilcher (both Manchester & Dilcher, 1982), Casholdia Crane & Manchester (Crane & Manchester, 1982), and the form genus Juglandicarya Reid & Chandler (Manchester, 198 1а); all but the last-named are winged. This is followed by the appearance of the winged gen- era Englehardtia Leschen. ex Bl. (Jáhnichen et al., 1977; Manchester, 1981a), Platycarya Sieb. & Zucc. (Chandler, 1964; Wing & Hickey, 1984), and Pterocarya Kunth. (Manchester & Dilcher, 1982) in the Early Eocene. The animal dispersed Juglans L. appears in the Middle Eocene (Man- chester, 1981a), and Carya Nutt. appears at the Eocene-Oligocene border (Mai, 1981). The Senonian fruits are quite small and show no sign of fleshy exocarps; features suggestive of па. dispersal. There is a clear transition in numerical dominance from wind-dispersed species to animal-dispersed species through the Tertiary (see Fig. 9 and discussion below). Myricaceae. The fruit is drupaceous or al- most a nutlet, sometimes enclosed by small brac- teoles. As with the Cannabaceae, it is difficult to reach a satisfactory generalization about dis- persal in the family. Ridley (1930) and Martin et al. (1951) noted that the fruits are dispersed by birds, but water and wind dispersal are also possible. In light of these observations, no single dispersal mode is assumed for this family. The earliest report is of Comptonia octocostata (Knobl.) Knobl. from the early Maastrichtian of Europe (Knobloch, 1975; Jung et al., 1978). This is followed by C. goniocarpa Mai & Walther from the Early to mid-Oligocene of East Germany (Mai & Walther, 1978). Myrica boveyana (Heer) Chandl. appears in the Early Eocene of southern England (Chandler, 1961b) and is followed by a host of subsequent reports of other species. The 1986] TIFFNEY —HAMAMELIDAE FRUIT AND SEED DISPERSAL 405 Manningiat | e Antiquocarya + е жей Caryanthust * | C ashotdiot | T dan poo Engelhardti y | Platycarya Pterocarya T Pal F T | Paraengelhardtia * + —[— — hh — .— — — <= — +... e. fj — — — — + i Hooleya t | | | Juglans 1 | | | | | | Carya| | | | | | | | | | | | | | = Е L EI M [et E | L E M L o [ а | | | СЕМО. =: SANT| CAMPANIAN | MAA. ee ¡EOCENE | OLIGOCENE MIOCENE PLIO.|Q FiGURE 6. First-occurrence data for genera in the Juglandaceae based on fossil fruits and seeds. Conventions as in Figure 1, + indicates extinct genus. Although published sources of чч occu Manchester and Dilcher (1982); Casholdia Crane & Manc Iljinskaja, Manchester and Dilcher (1982); Engelhardtia esses note: Manchester placed all fossil Engelhardtioid орн into ү form genus Palaeocarya Saporta, as the (1981a, range of variation seen in the fossils exceeds that in the m rrences are given, much insight ryanthus Friis, Friis (1983); Juglan- x Bl., Jáhnichen et al. (1977), Manchester genus sensu stricto. I retain Engelhardtia here as a more familiar concept, accepting that it may ida several genera); Platycarya Sieb. & Zucc., Chandler (1964), Wing and Hickey (1984); Pterocarya Kunth, Dilcher et al. (1976); Paleooreomunnea Dilcher, Potter & Cre Manchester and Dilcher (1982); Paraengelhardia Berry, pet, Dilcher et al. (1976); Hooleya Reid & Chandler, Wing and Hickey (1984); Juglans L., Brown (1962), Manchester (1981b); Carya Nutt., Mai (1981 morphology of the fossils parallels that of the living рен suggesting a al *generalized" dispersal syndrome in the pas The fruit is a fleshy-walled, acorn-like drupe, sitting in an acorn-like cup (Willis, 1973; соса. 1981). The fruit mor- phology suggests biotic dispersal. There is no fos- sil record of the fruits of this family. Fagaceae (Fig. 7). The fruits are nuts, often of considerable size, sane biotic dispersal. Only in Nothofagus B occasionally с although (1977) noted that, in most species, they are large and disperse very poorly. The family may be considered biotically dispersed. The oldest fruit of the family may be repre- sented by reproductive material from the San- tonian- Campanian of Mossscnuset United States fruits of Lithocarpus Bl. in morphology (Tiffney & Friis, unpubl. data). This record requires val- idation. The extinct genus Fagopsis Hollick ap- pears in the Early Eocene (Manchester & Crane, 1983), followed by Trigonobalanus Forman (Mai & Walther, 1978), Quercus L. (Bones, 1979; Manchester, 1981b), and possibly Castanea Mill. (Crepet & Daghlian, 1980) in the Middle Eocene. Lithocarpus Bl. may appear in the Late Eocene (Axelrod, 1966), although Manchester and Crane (1983) are dubious of this record. Fagus L. ap- pears in both Europe (Chandler, 1957) and North America (Chaney, 1927; supported by Smiley & Huggins, 1981) in the mid-Oligocene. The Cretaceous fossils are about 3 mm in di- ameter, show no signs of a fleshy covering, and occur in prodigious numbers, all suggestive of abiotic dispersal. Fagopsis is the earliest known Tertiary representative of the group and pro- duces small fruits borne within wing-like cu- pules, often aggregated in rings, the whole ap- 406 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 | | "Fagaceae" 1 | { | | | | Fogopsis* | Trigonobalanus | Quercus | | | Castanea | | 1 | | Lithocarpus | 1 | | Fagus ! | Nothofagus | T T 11 Pseudofagust 11 Castanopsis | T | | | [| z E L E! M L E L E MI L O |2] | ceno. |5 $Амт| CAMPANIAN MAA. |PALEOCENE | EOCENE OLIGOCENE MIOCENE PLIO la = 1 FIGURE 7. First-occurrence data for genera in the Fagaceae based on fossil fruits and seeds. Conventions as in Figure 1, + indicates extinct genus. “Fagaceae,” possible fagaceous flowers and fruits under антен by Tiffney and Friis; (1978); Quercus L., Manchester (1981b); Castanea Mill., Sen ), see еса апа Crane (1983); Fagus L., Chaney (1927), Chandler (1957); Nothofagus L., Hill (1 ; Fagopsis Hollick, Manchester and Crane (1983); Trigonobalanus Forman, е Crepet and Daghlian (1980); Lithocarpus Bl. Axe 955. iley & H m Smiley and Huggins (1981); Castanopsis (D. Don) Spach., Mai (1964); Castanea ofagus Mill. van der "Burgh (1978 parently adapted for wind dispersal. The other members of the family, including the extinct mid- Tertiary Pseudofagus idahoensis Smiley & Hug- gins (Smiley & Huggins, 1981), are animal dis- persed. Betulaceae (Fig. 8). Of the six living genera, the fruits of A/nus Mill. and Betula L. are wind dispersed. Carpinus L. and Ostrya Scop. fruits may be wind dispersed [Ridley (1930) gave evi- dence that they travel respectable distances], or the bracts may be too small to permit effective dispersal. Thus, animal dispersal may also be important in these genera (Ridley, 1930; Martin et al., 1951). Corylus L. generally has a nut too large to be wind-borne by its bracts and is often animal dispersed (Ridley, 1930). The dispersal mode in Ostryopsis Decne. is not reported in the literature. Illustrations of the fruit (Li & Cheng, 1979) show a relatively large fruit in a winged involucre. In sum, dispersal mechanisms inter- grade from animal to wind in the family. The oldest fossil records involve Corylus L. (Brown, 1962; Koch, 1978) from North America and Greenland and the extinct genus Palaeocar- pinus Crane (Crane, 1981) from England and North America (Crane, 1984b), both appearing in the mid-Paleocene. In addition, Crane (1981) suggests that “Atriplex” borealis (Heer) Laurent of the Paleocene may also belong to the family. Alnus Mill. appears at the Paleocene/Eocene boundary (Takhtajan, 1982) and Betula L. in the Middle Eocene (Crane, 1984b). Carpinus L. may appear in the Late Paleocene (Chandler, 1961b), although Crane (pers. comm.) believes this rec- ord is of Palaeocarpinus, and that the first fossil fruit of Carpinus is of Late Eocene age (see Crane, 1981). With the exception of Corylus, all fossil members of the family are morphologically adapted to a greater or lesser degree for wind dispersal. Casuarinaceae. The seed is a small, one- seeded samara, well-adapted for wind dispersal. Christophel (1980) reported mature inflores- cences of Casuarina Adans. from the Eocene of 1986] TIFFNEY —HAMAMELIDAE FRUIT AND SEED DISPERSAL 407 1 Corylus | Alnus Betula | Т | | | Tubelat | í | Carpinus | | | | ! | Ostr | | | = | | | l'Atriplex" borealis І ' І | І | | Polaeocarpinust 1 ot 1 | | 1 Carpinicarpust | | —— = І + | Corylocarpinus | | j l | 2 Е L ЕТ M L E L E M] L O | ceno. |5 5 Ам“| САМРАМАМ | МАА. |PALEOCENE| | EOCENE OLIGOCENE MIOCENE PLIO. FE ] FIGURE 8. First-occurrence data for genera in the Betulaceae based on fossil fruits and seeds. Conventions as in Figure 1, omm.), first published report fro Chandler (1961b), P. Crane (pers. еер sugg C Straus (1969). Australia that are virtually identical to those of modern species. This suggests that the Eocene members were also wind dispersed. SUMMARY OF ABIOTICALLY-DISPERSED FAMILIES The Trochodendraceae, Cercidiphyllaceae, Platanaceae, Myrothamnaceae, Eucommiaceae, Barbeyaceae, Rhoipteleaceae, and Casuarina- ceae are dominated in the present day by forms with winged or very small fruits or seeds adapted for wind dispersal. The Hamamelidaceae have a few taxa that are wind dispersed, but the majority are ballistically dispersed. Where the above fam- ilies are known in the fossil record, they are abiot- ically dispersed. SUMMARY OF INTERMEDIATE FAMILIES Th eC 1 ARA i 4 VS that are often enveloped in | bracteoles. The fruits © ч E Г? could float or be wind dispersed. The often spiny Comptonia bracteoles might adhere to animals. The fruits of these groups could also be eaten and dispersed. The fruit of the Leitneriaceae could equally well attract animal dispersers or float. The fossil records of the Cannabaceae, Myrica- ceae, and Leitneriaceae indicate no change in morphology of the fruit in the recorded past. The Ulmaceae possess approximately 25% abiotically-dispersed, and 75% animal-dis- persed, species. The fossil record indicates that both dispersal modes appeared simultaneously, but that animal dispersal may have be fossil record of urticaceous fruits does not clearly indicate the nature of dispersal in the past. In the extant Juglandaceae, 63% of the species are biot- ically dispersed and 37% are abiotically dis- persed. Senonian fruits of juglandalean affinity 408 are probably abiotically dispersed. Whereas both wind- and animal-dispersed fruits are present in ie early Tertiary, there is a clear rise in domi- nance of biotically-dispersed species through the Tertiary (Fig. 9). Three genera of extant Betu- laceae are wind dispersed, two can be wind or animal dispersed, and Corylus is animal dis- persed. Abiotically- and biotically-dispersed bet- ulaceous diaspores appear simultaneously in the fossil record. SUMMARY OF BIOTICALLY-DISPERSED FAMILIES The Daphniphyllaceae, Didymelaceae, and Balanopaceae possess fleshy fruits suggestive of animal dispersal; they all lack fossil records. The Moraceae are almost entirely animal dispersed today. The fossil record is possibly equivocal but could be read to show no evidence of abiotic dispersal. The Cecropiaceae are entirely bioti- cally dispersed today and lack a fossil record. The Fagaceae are almost, if not entirely, bioti- cally dispersed in the present day. Possible mid- te Cretaceous fagaceous fruits are abioti- cally dispersed. The next genus to appear in the fossil record of the family (Fagopsis) is wind dis- persed. All others are animal dispersed DISCUSSION BIOTIC DISPERSAL AND PHYLOGENY At the outset, I outlined two assumptions; that the Hamamelidae reached a zenith in the Cre- taceous and are largely a relict group today, and that abiotic dispersal is primitive in angio- sperms. Given these premises, logic suggests that families dominated by biotic dispersal and in- cluded in the Hamamelidae may be mis-allied with the sub-class. Biotic dispersal is important in three orders of e glandales (Juglandaceae and Balanopaceae), and Fagales (Fagaceae and a small portion of the Bet- ulaceae). In addition, it dominates the Daph- niphyllaceae and Didymelaceae. With the excep- tion of the last two small “outliers,” all of the groups mentioned lie among the “advanced” Hamamelidae, and questions have been raised about the pro Il ofthese groups. The Urticales have been allied with the Mal- vales (Dilleniidae) rather than with the Hama- melidae (e.g., Thorne, 1973; Berg, 1977), a so- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 lution perhaps borne out з leaf architecture (Leo Hickey, pers. comm.). wever, separate evi- dence suggests жыл to the Hamamelidae (Cronquist, 1981: 186). The Juglandales have long been a bone of phylogenetic contention, with al- ternative ка allying them with the Ната- melidae or the Rosidae [near Anacardiaceae бараи) The dominance of biotic dispersal n modern representatives of the family would appear to support features of leaf architecture (e.g., Hickey & Wolfe, 1975) in arguing for their Rosid affinity. Again, however, there is much evidence for their association with the Hama- melidae (Cronquist, 1981: 204-207). The Fa- gales are not generally contested as Hamamelids, although Hickey and Wolfe (1975) did note that the leaf architectural affinities of the order (ex- cluding Balanopales and Betulales) were **uncer- tain." Of the smaller families in question, Hickey and Wolfe (1975) transferred the Didymelaceae to the Dilleniidae and Thorne (1976) referred it to the Euphorbiales (both paralleling its biotic dispersal), but Cronquist (1981) considered it st harmoniously placed in the Hamamelidae. The Daphniphyllaceae are another taxonomic football, alternating between placement in the Euphorbiaceae (Rosidae) and the Hamamelidae (see Cronquist, 1981), but without resolution. Here the biotic dispersal would join evidence for a euphorbian alliance In summary, dispersal evidence would appear to support arguments for exclusion of the Ju- glandales, Didymelaceae, Daphniphyllaceae, and several families of the Urticales from the Ham- amelidae and raises questions about the associ- ation of the Fagales with the sub-class. However, acceptance of these conclusions based on modern ispersal modes assumes that the families in question have not altered their primary mode of dispersal over time. The diversification of im- portant groups of modern dispersal agents in the early Tertiary indicates that plant lineages dating from the Late Cretaceous or Tertiary (e.g., Ju- glandaceae, Fagaceae, Urticaceae) may have passed through a period of intense selective pres- sure involving changing dispersal mechanisms (Tiffney, 1984). EVIDENCE FOR ALTERATION OF DISPERSAL MODE WITHIN A FAMILY Two, possibly three, examples in the fossil rec- ord and one from modern evidence suggest that the dominant mode of dispersal changed over 1986] TIFFNEY —HAMAMELIDAE FRUIT AND SEED DISPERSAL 409 1 | | | 11 | | | | | | | | | l | [ 1 1 [ І ! | ZA і | | | | 22 | | ABIOTICALLY DISPERSED | | 11 ! о! i | I [ | I 1 1 1 |— BUM ——— 9 9 13 UNO 15 IBIOTICALLY DISPERSED | | ET E'M L Г М | Е L (M! PALEOCENE !EOCENEI OLIGOCENE MIOCENE PLIAQ URE 9. Specific diversity of abiotically-dispersed and biotically- өл, aes Juglandaceae in the Tertia ary. Fic Data largely from Manchester (19 81a); additional data fr om sources cited in Figure 6. Diagonally-lined area in upper spindle demonstrates the proportionately large contribution of и to the diversity of abiotically- dispersed Juglandaceae in the later Tertiary. N umbers in each spindle indicate numbers of species present in each Epoch subdivision. Time convention as in Figure 1 time within families. The fossil examples will be considered first and in order of increasing clarity. Ulmaceae. The Ulmaceae i as divided iio the Ulmoi dez Y 5 abiotically Celtidoid (largel y biotically dispersed). Fruit- ing evidence of бош s is recorded from the middle Paleocene (Fig. 3). Zavada and Crepet (1981) described flowers of Celtidoideae from the Middle Eocene of southeastern North Amer- ica. In their discussion they noted that, in con- trast to other Eocene fossil flowers ofthe ““Amen- tifereae," which normally look modern at this time, these Celtidoid flowers were intermediate between insect-pollinated ancestors and the modern wind-pollinated flowers of the group. This conveys the impression that the Celtidoi- deae were still evolving in the mid-Eocene, and the hypothesis could be entertained that modern fruit and dispersal characters were also evolving in the group. However, such an interpretation ignores mosaic evolution and the likelihood that the evolutionary status of the flowers might have little to do with that of the fruits. Further, this = interpretation presumes that the ulmoid flowers of the time were “modern,” a conclusion for which no evidence exists. It is worth noting (Fig. 3) that, while two samaroid and two drupaceous genera are present in the Eocene, animal-dis- persed forms have come to dominate the family by greater than a 2:1 margin. I suspect the UI- maceae of having abiotic dispersal as the plesio- morphic state, but the actual evidence is very weak. Fagaceae. Fagaceae are currently almost en- tirely animal dispersed. The Tertiary fossil rec- ord involves seven extant and two extinct genera. f these, the extinct Fagopsis (Manchester & Crane, 1983) is both the earliest (Early Eocene through Late Oligocene) and is the only genus adapted to abiotic dispersal. The morphology of Fagopsis agrees with the cupules of other mem- bers of the family and argues for the derivation of structures for both biotic and abiotic dispersal from a common “cupule” morphology. Late Cretaceous flowers from Massachusetts (Tiffney & Friis, unpubl. data) are similar to flow- 410 ers of extant Fagaceae, particularly those of Lithocarpus. The associated fruits are tiny, very numerous, and lack a cupule. Their great num- bers and small size agree with Fey and Endress’s (1983: 451) prediction that the origin of the cu- pule involved the reduction of the “highly branched system of modified (compact, sterile) ultimate parts of the inflorescential cyme.” Such a reduction could also be linked to an increase in size of the remaining fruits, leading to the evolution of a single, large fruit. The tiny size and great а of these fruits argue for their abiotic dispersa The abiotic Sic of the putative Creta- ceous fagaceous fruits, together with the abiotic dispersal of the first clear fruit of the family to appear in the record, Fagopsis, suggests that the Fagaceae might be primitively abiotically dis- persed. The rapid appearance of biotically-dis- persed members of the family in the Tertiary could be seen as an adaptive response to the coeval radiations of mammals and birds. Juglandaceae. Manchester (1981a; see also Manchester & Dilcher, 1982 and Wing & Hickey, 1984) has provided a detailed review of the his- tory of the family with emphasis on its fruits. At the species level, abiotically-dispersed fruits dominated the early Tertiary record of the fam- ily; biotically-dispersed fruits diversified rapidly mid-Tertiary and dominate the family in the present day (Fig. 9). As found in the Fagaceae, this pattern is supported by Friis's (1983) data on three genera of fruits (Manningia, Antiquo- carya, and Caryanthus) allied with the Juglan- daceae in the Campanian. The mature fruits of all three are tiny and show no signs of fleshy parts leading one to assume abiotic dispersal. This evidence suggests that the Juglandaceae (or the family and its immediate ancestors) were abiotically dispersed in the Late Cretaceous and earliest Tertiary. With the appearance of biotic dispersal agents, the family experimented with mechanisms of both biotic and abiotic dispersal. Presumably selection acted for larger endosperm reserves and therefore seed size (Tiffney, 1984) in the early Tertiary. This led to the simultaneous evolution of large-nutted, biotically-dispersed, fruits (Juglans) and the evolution of larger abiot- ically-dispersed fruits borne on increasingly larg- er winged structures (e.g., Cyclocarya, Engle- hardtia). The forms with animal-dispersed fruits ultimately dominated the family. Of the wind- dispersed forms, only Pterocarya was successful in the later Tertiary, perhaps because of its small ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 fruit and wing size; in the only reference to its site ecology, Wang (1961: 91) remarked that terocarya stenoptera C. DC. occurs “in small patches of pure communities in forest openings" implying that it is an early to mid-successional tree. This ecology is in keeping with the pre- sumed early successional status (Tiffney, 1984) of the Late Cretaceous and Early Tertiary abiot- ically-dispersed members of the family oraceae, Cecropiaceae, Urticaceae. Turn- ing from the fossil record, there is evidence from the morphology of the Moraceae, Cecropiaceae, and to a degree, Urticaceae, that suggests these families underwent a transition from dominant abiotic to dominant biotic dispersal in their his- tory. The Cecropiaceae are entirely animal dis- persed today. In the extant Moraceae, 7896 of the genera and 8696 of the species are animal dis- persed. The Urticaceae are mixed, with 47.6% of the genera and 29.396 of the species biotically dispersed. The common fruit morphology in all three families is an achene, or if the fruit wall is fleshy, a drupe. However, drupes occur in only 1196 of the genera and 396 of the species in the Moraceae, only portions of one genus (Coussa- poa Aubl.) of 50 species in the Cecropiaceae (20.7% of total number of species in the family), and only 4.396 of the genera and 1.396 of the species in the Urticaceae. Animal dispersal based on adherent fruits (sticky, hairy, or hooked) ac- counts for no more than 696 of the genera and 0.396 of the species of Moraceae and 1396 of the genera and 12.896 of the species of Urticaceae. In all three families, the dominant structure as- sociated with attracting animal dispersal agents involves either fleshy, accrescent calyx parts or an inflated, fleshy, receptacle. Both structures form succulent and attractively-colored berry mimics, but without the participation ofthe fruit wall; the fruit remains an achene buried in the externally-derived flesh. Many reports attest to the success of these **pseudo-drupes" in attract- ing birds or mammals [e.g., Cudrania Trécul, Artocarpus Forst., Sloetia Teij. & Binn. (Streblus Lour., see Ridley, 1930)]. Such "fruits" made of fleshy floral parts account for 67% of the genera and 82.896 of the species of Moraceae, 87.596 of the genera and 79.396 of the species of Cecro- piaceae, and 30:396 of the genera and 15.2% of the species of Urticaceae. In some cases the floral parts aid in biotic dispersal by forming hairs or spines that stick to dispersal agents [e.g., Rous- selia Gaudich. and Soleirolia Gaudich. (Urtica- ceae)]. Abiotic dispersal in all three families in- 1986] volves both structures associated only with the achene and structures associated with the floral parts. In some genera [e.g., Metatrophis F. Brown (Moraceae), Hemistylus Benth. (Urticaceae)] the floral parts form wings, whereas in others [e.g., species of Streblus Lour., Dorstenia L. (Mora- ceae), Pilea Lindl., Procris Comm. ex Juss. (Ur- ticaceae)] the calyx exerts pressure on the mature achenes and forces the fruits out of their floral envelope with great force (Ridley, 1930). Ballistic dispersal occurs in roughly 296 of the genera and 1396 of the species of Moraceae, none of the Ce- cropiaceae, and 19.596 of the genera and perhaps as much as 5896 of the species of Urticaceae. The plesiomorphic status of the achene in this group is suggested but not proven by its wide distribution in all three families. It is also sug- gested by the fossil record, because the oldest fruiting remains of this group of families are achenes from the Late Cretaceous (Knobloch, 1971; Knobloch & Mai, 1984 If abiotic mechanisms of dispersal dominated in the Cretaceous, and if the Urticaceae extend back to the Late Cretaceous, then it is possible that the dry and ballistically-dispersed achenes of the present day reflect the ancestral dispersal mechanisms of the Urticaceae, and by inference, the Moraceae and Cecropiaceae. If so, this im- plies that these families met the selective pres- sures of the evolving mammals and birds of the early Tertiary in two ways. The first was the evo- lution (once, or repeatedly) of fleshy exocarps to form drupes. To judge from the present day, this solution was not widespread. The second re- sponse was to build upon the specialization of the perianth, presumably already evolved for abiotic ballistic dispersal. Again, this potentially could have evolved repeatedly in several lin- eages. The accrescent floral parts could become fleshy, mimicking a berry without a topological re-arrangement of mature flower and fruit, or the developmental switch from a hard exocarp to a fleshy one underlain by a hard meso- or endo- carp. This explanation is consistent with the im- portant role that flower parts play in abiotic dis- persal in the extant Urticaceae and to a lesser degree, Moraceae. It is also consistent with the perianth and receptacle-derived. pseudo- berries that seem to form an common to all three families. Tangentially, the early fossil record and the dominant abiotic dis- persal mode of the Urticaceae suggests that they retain the greatest amount of plesiomorphic fruiting characters among the Cecropiaceae, TIFFNEY —HAMAMELIDAE FRUIT AND SEED DISPERSAL 411 Moraceae, and Urticaceae. This could be related to their herbaceous growth habit. In either case, the ecological nature of the Urticaceae is consis- tent with the proposed abiotic dispersal ecology of the Cretaceous angiosperms (Tiffney, 1984). SUMMARY The fossil record of the Juglandaceae and its immediate predecessors strongly suggests a tran- sition from abiotic to biotic dispersal within the family. The fossil evidence for the Fagaceae is less strong, but also suggestive of such a transi- tion. Evidence for a similar transition in the Ul- maceae is inferential, but again suggestive. The dominance of perianth-derived pseudo-fruits in the Moraceae and Cecropiaceae and their pres- ence in the Urticaceae, the commonality of the achene morphology to all three families, and the mechanisms of abiotic achene dispersal in the Urticaceae, may be hypothesized to reflect an early Tertiary transition from abiotic to biotic dispersal in the Moraceae and Cecropiaceae, and the retention of more plesiomorphic characters in the Urticaceae. EVOLUTIONARY CONCLUSIONS RELATIONSHIP OF DIVERSITY AND DISPERSAL MECHANISMS The distribution of generic and specific di- versities in the Hamamelidae presents an in- teresting pattern (Fig. 10). Sixteen families have three genera or less, and 14 have only one genus. Only three of these families have more than ten species (Daphniphyllaceae—35 spp., Casuari- naceae— 50 spp., Myricaceae— 50 spp.). Four families have an intermediate diversity; the Ju- glandaceae with 9/60 (genera/species), the Ham- amelidaceae with 28/100 plus, the Betulaceae with 6/120, and the Ulmaceae with 18/150. Four families have high diversities: Cecropiaceae (8/ 275), Fagaceae (8/800), Urticaceae (46/1,255), and Moraceae Shad 1,31 3). d that in sect pollina tion is a primary cause of the present diversity of angiosperms. However, biotic dispersal of fruits and seeds offers a similar potential for animal- mediated diversification. The Hamamelidae provide a natural experiment to demonstrate the effect of biotic dispersal on diversity. With rare exceptions, the included families are dominated by anemophily (the Hamamelidaceae is primar- ily insect pollinated, and a few species in the 412 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 800 900 1000 поо 1200 1300 SPECIES PER FAMILY-HAMAMELIDAE FIGURE 10. Plot of number of species per family for families in the Hamamelidae sensu Cronquist (1981). Points for Cannabaceae and ее omitted in Horizontal axis, number of s n light of absence of clear-cut dominant mode of dispersal. s; Vertical axis, valueless; Open triangles, families dominated by abiotic dispersal; Closed circles, families oa by biotic dispersal; Open star, Urticaceae. Family identity indicated by the number by each symbol; the numbers corresponding to the order of families in Table 1 Moraceae, Urticaceae, and Cecropiaceae are en- tomophilous, but these are unusual; similarly, the Fagaceae includes a few species that are sec- ondarily entomophilous). In effect, the variable of insect pollination is controlled in the group. The only other variable that might confuse the proposed comparison is that of habit, however all the hamamelid families are dominantly (if not entirely) woody, with the exception of the Urticaceae, which are dominantly herbaceous. f the four families with substantially more than 150 species, only Urticaceae is not domi- nated by biotic dispersal. This anomaly may be explained largely by the herbaceous habit and y. Of the intermediate-sized families, Ul- Ri are dominated by biotic dispersal, Ju- glandaceae are dominated 2:1 by biotic dis- persal, and Betulaceae possess both dispersal modes in unclear proportions. The Hamameli- daceae are abiotically dispersed, but their diver- sity could be explained by the dominant ento- mophily of the family. The small-sized families (50 species or less) are dominated En to four) by abiotic dis l. A regression of ber per family against dispersal type (excluding data from the "generalized" Cannabaceae and Myricaceae) yields r = 0.225, below the 80% confidence limit. Elimination of the Urticaceae from the data set yields r = 0.423, which falls in the 90-95% confidence limit. Elimination of the Hamamelidaceae from consideration accen- tuates this effect. This pattern suggests the po- tential stimulating effect of biotic dispersal on the diversification of the angiosperms. The great diversity of the modern flora may be in large part a function of plant-animal interactions involving biotic diaspore dispersal, as we e more frequently cited ано of angiosperms with pollinators and herbivore DISPERSAL PLASTICITY Several examples demonstrate that dispersal mode changes within a family over time. Dis- persal modes appear stable within genera, but this is an artifact: one would not identify an iso- lated fossil fruit as a modern genus unless it had the morphology of that genus. The demonstra- tion of this plasticity is important on three counts. First, the appearance of animal-dispersed fruits within five to 15 million years after the radiation of mammals and birds in the early Tertiary is a measure of the reality of the concept of “соеуо- lution” in a loose sense (cf. Herrera, 1985) and its significance in angiosperm evolution. Second, the response of fruit and seed morphology to the appearance of new dispersal agents further i the importance of mosaicism in an- osperm evolution. For example, pollen data Е 1981) suggest an earlier time of first ap- pearance for many families than does that for fruits and seeds (Tiffney, unpubl. data). This sug- gests the separation of selective effects of polli- nators and fruit/seed dispersers. Third, the flex- ibility of dispersal mechanisms over time within a family emphasizes the need to seek important phylogenetic characters of fruits and seeds in the 1986] details of cell structure, cell layer sequence, etc., rather than simply in gross morphology. EVALUATION OF HYPOTHESES Two assumptions were made at the outset of this paper: (1) that the Hamamelidae originated and achieved their zenith in the Cretaceous, and (2) that Cretaceous angiosperms were largely abiotically dispersed and that biotic dispersal be- came important only in the early Tertiary. It is not possible to evaluate the first assumption from the fossil fruit and seed record, although fruiting material of some families (Platanaceae, Juglan- daceae, possibly Urticaceae, and Fagaceae) is known from the Cretaceous. The second as- sumption (discussed by Tiffney, 1984) is sup- ported by the apparent primitive status of abiotic dispersal demonstrated here for the Juglandaceae and possibly Fagaceae, and inferred for the UI- maceae and possibly Moraceae, Cecropiaceae, and Urticaceae. Given these assumptions, I predicted that (1) the majority of families in the Hamamelidae should have abiotic dispersal syndromes, and (2) that dispersal mode might provide an additional trait on which to assess the phylogenetic affinities of families whose alliance with the Hamamelidae has been questioned on other grounds. The majority of families associated with the Hamamelidae in the scheme of Cronquist (1981) are abiotically dispersed (Table 1; 13 abiotic, seven biotic, and four either with roughly equal proportions of biotic and abiotic, or with gen- eralized morphology and dispersal mechanism). However, evidence from the d record sug- he erhaps Fagaceae underwent a transition from a dnd) abiot- ic mode of dispersal to a biotic mode in the early Tertiary; more circumstantial evidence suggests that the same might be true of the Ulmaceae and the Urticaceae and associated Moraceae and Cecropiaceae. If the dominant mode of dispersal within a (аршу can change о over time, then the presens Spersa tan of the past. The primitive dispersal state, while inferable from neontological evidence, can be demonstrated only through the fossil record. us, only in the case of the Juglandaceae, and perhaps of the Fagaceae, can we state that the primitive mode of dispersal was abiotic and that the dispersal mode does not contest the affinities of the family with the Hamamelidae. A similar gests thatt TIFFNEY —HAMAMELIDAE FRUIT AND SEED DISPERSAL 413 transition from abiotic to biotic dispersal is sug- gested for the Ulmaceae by the fossil record, and for the Moraceae, Urticaceae, and Сесгоршобшеё е; based а$ on dispersal, the affinities of these families with the Hamamelidae cannot be evaluated. No fossil record exists for the Daphniphyllaceae, Didy- melaceae, or Balanopaceae, thus there is no way to evaluate whether the biotic dispersal mecha- nisms of these taxa are primitive or derived. LITERATURE CITED ANONYMOUS. 1980. Iconographia Cormophytorum Sinicorum, Volume 1. Academia Sinica, Beijing. AXELROD, D. I. 1966. The Eocene Copper Basin Flora of northeastern Nevada. Univ. Calif. Publ. Geol. Sci. 59: 1-125. BACKER, C. A. 1963. pos of feug (Spermatophytes only), Volumes 1-3. . P. Noordhoff, Gro- ningen, Netherlands. BERG, C. C. 1977. Urticales, their differentiation and systematic position. Pl. Syst. Evol., Suppl. 1: 349- 374. Bones, T. J. 1979. Atlas of fossil fruits and seeds m north central Oregon. Oregon Mus. Sci. Technol. Occas. Pap. Nat. Sci. 1: 1-23. Brown, R. W. 1939. Fossil leves: fruits, and seeds of Cercidiphyllum. J. Paleontol. 13: 48 9 New species and changes of name in some er M fossil floras. J. Wash. Acad. Sci. 30: 344- ' ie Alterations i in some fossil and living floras. J. Wash. Acad. Sci. 36: 344-355. —. 62. Paleocene flora of the Rocky Moun- tains n Great Plains. Prof. Pap. U.S. Geol. Surv. 75: 1-119. BURGH, сүн . VAN DER. 1978. The Pliocene flora of For- a ruits and seeds of EOM: Rev. Е естон Palynol. 26: 173-2 CHANDLER, M. E. J. 1925-1926. The us Eocene pora of Hordle, Hants. Palaeontogr. Soc. 78(363): Е Mu cey lake basin, cu Hist.), Geol. 3: 7 3. l 6la. ES e the Lower Headon beds of Hampshire and the Isle of Wight. Bull. Brit. Mus. (Nat. Hist.), Geol. 5: 91- 1961b. The Lower Tertiary Floras of South- ern England. I. Palaeocene Floras. London Clay Flora (Supplement). British Museum (Natural History), London. 62. The Lower Tertiary Floras of Southern England. II. Flora ofthe Pipe-Clay Series of Dorset p pu British Museum (Natural His- tory), London. . 1963. The Lower Tertiary Floras of Southern England. III. Flora of the Bournemouth Beds; the Boscombe, and the — чк British Ми- seum (Natural History), Lo . 1964. The Lower a Floras of Southern England. IV. A Summary and Survey of Findings The Oligocene flora of the Bovey Tra- ire. Bull. Brit. Mus. (Nat 414 in the Light of Recent Botanical ш . W. 1927. Geology and а of the Crooked River Basin with special reference to the rw Creek flora. Publ. Carnegie Inst. Wash. 346: 45-138. CHANG, ыза (editor). 1979. Flora Reipublicae Po- pularis Sinicae 35(2). Pittosporaceae, Hamameli- daceae, Eucommiaceae, Platanaceae. Academia , Beijing. CHRISTOPHEL, D. С. 1980. Occurrence of Casuarina megafossils in the Tertiary : South-eastern Aus- tralia. or J. Bot. 28: CRANE, P. 1981. © leaves and Jas from n И Upper Palaeocene. Bot. J. L Soc. oo 3-136. 1 А re-evaluation of Cercidiphyllum- like a “fossils from on British early Tertiary. Bot. J. Linn. Soc. 89: : | 984b. Early hal history of the Betulaceae: a preliminary report. Amer. J. Bot. 71(5,2): 109. [Abstract.] MANCHESTER. 1982. An extinct ju- glandaceous fruit from the Upper Palaeocene of southern England. Bot. J. Linn. Soc. 85: 89-101. R. ape 1985. Growth and repro- ductive biology of Joffrea speirsii gen. et sp. nov., к rcidiphyllum-like plant from the Late Paleo- 63: 340- of Alberta, Canada. Canad. J. Bot. —— A 1986. Morphology and develop- Cercidiphyllaceae. Ann. Missouri Bot. Gard. 73: 382-393. CREPET, W. L. & C. D. DAGHLIAN. 1980. Castaneoid inflorescences from the Middle Eocene of Ten- nessee and the diagnostic value of pollen (at the subfamily level) in the Fagaceae. Amer. J. Bot. 67: -757. Croat, T. B. 1978. Flora of Barro Secun Island. Stanford Univ. Press, Stanford, Californ CRONQUIST. A. 1981. An Integrated ncn of Clas- sification of Flowering Plants. Columbia Univ . R. FOSBERG (editors). 1981. hd Revised Handbook t ns Flora of Ce а Vol- Co., N Bia > ae W. Porrem JR. & W Ta QM EPET. 1976. Investigations of angiosperms from the Eocene of North America: juglandaceous winged 4. 1963. Tretichnye Flory Zapadnoi Sibiri. Izdat. Akad. Nauk S.S.S.R., Moscow and Lenin . 1969. Miotsenovaya Flora Mamontovoi Gory na eae Bot. Inst. V. L. Komarov. Izdat Nauka, DovLE, 3, Pu &L L. J. Hickey. 1976. Pollen and leaves from 1 Beck (editor), Origin and Early Evo- lution of Angiosperms. Columbia Univ. Press, New ELIAS, T. S. 1980. The Complete Trees of North merica. Field Guide and Natural History. Van Nostrand Reinhold Co., New York. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 ENGLER, A. & K. PRANTL. 1894. Die Natürlichen Pflanzenfamilien, Volume 3, 1. Verlag W. Engel- mann, Leipzig. & . 1930. Die Natürlichen Pflanzen- familien, 2nd edition, Volume 18a. Verlag W. En- elmann, Leipzig. EVERNDEN, J. F. & G. T. James. 1964. Potassium- iari dates and the Tertiary floras of North Amer- ica. Amer. : Sci. 262: 945-974. Fey, B. S. & P. К. Емокеѕѕ. 1983. Development and morphological interpretation of the cupule in Fa- gaceae. Flora pr 451-468. Friis, E. M. 1 r Cretaceous inira jen structures of Пе affinity containing № mapolles pollen. Rev. Palaeobot. Aime 39: 161- . Platanaceous inflorescences from the Late Cretaceous of Sweden and eastern North America. P. 14 in Abstracts of Contributed Papers, Intl. Org. Paleobot. Conf. Edmonton, Canada. GIVULESCU, К. 1980. Fructe $1 semințe cunoscute din Cainozoicul R. S. Románia. Muz. Brukenthal Stud. 4: 55 ——. 1984 „ A. V. Cox, P. G. LLEWELLYN, С. A. G. PICKTON, A. G. SMIT TH &R. WALTERS. 1982. A sis Time Scale. Cambridge Univ. Press, Cambridge HERRERA, C. M. 1985. Determinants of plant-animal f HrEvwoop, V. Н. (editor). 1978. Flowering Plants of the World. Mayflower Books, New HICKEY, L. J. & 1. A. DoYLE. 1977. Early ша t. Rev. (Lancaster) 43: 3-1 04. — £ J. A. Worrre. 1975 [1976]. The bases of angiosperm phylogeny: vegetative morphology. Ann. Missouri Bot. Gard. 62: 538-589. HILL, В. S. 1983. Nothofagus macrofossils from the Tertiary of Tasmania. Alcheringa 7: 169-183 нат Р, 1975, Оп ѕоте new species from the Mas- Hradek Me Nisou ene Basin, North Bohemia). Sborn. Nar. Mus. e 31: 109-122. HUTCHINSON, J. 1964. pis Genera of Flowering Plants (Angiospermae), Volume 1, Dicotyledones. Clar- endon Press, Oxford. Beal The Genera of Flowering Plants (An- giospermae), Volume 2, Dicotyledones. Claren- don "ren. Oxfor Flora of West Tropical e , Lon А edis M. 1960. siana, Volume 6, Part Juglandaceae. N. V. Dijkstra, Groningen, Neth- с s. JAHNICHEN, H., D. H. Mai & H. WALTHER. 1977. Blátter und Früchte von E ngelhardia Lesch. ex ВІ. с. Nov. Regni Veg. Beih. 88: 1980. Blatter und Früchte von Cercidiphyllum Siebold & Zuccarini im mit- tele а Tertiár. Schriftenreihe geol. Wiss. 16: 357- 1986] JUNG, W., H.-H. SCHLEICH & B. KASTLE. 1978. Eine neue, t tig pl iscl g ichert Fund dstelle für ir An- giospermen-Früchte und -Samen in der oberen Gosau Tirols. Mitt. Pru Staatssamml. Paláon- tol. Hist. Geol. 18: 131-142. KIRCHHEIMER, F. Braunkohlenzeit. W. Knapp Verlag, Halle (Saale). KNOBLOCH, E. 1971. Fossile Früchte und Samen aus der Flyschzone der máhrischen Karpaten. Sborn. Geol. Véd. Paleontol. 13: 7-4 . 1975. Früchte und Samen aus der Gosau- formation von Kóssen in Osterreich. Véstn. Us- tredního Ústavu Geol. 50: 83-92. 1977. Paláokarpologische Charakteristik der Fly schzone der máhrische oe . Sborn. Geol. Véd. Paleontol. 19: 79-1 & D. H. Mar. 1984. Neue pri — Früchten und Samen aus dem Cenoman bis Maa tricht (Kreide) von Mitteleuropa. Feddes i a ЖЕ? be. Beih. 95: 3-41 , K. J. NIKLAS, P. G. GENSEL & B. H. 1984. а са. апа patterns of evolution in early vascular plants. Ра- leobiology 10: 34-47 KocH, B. E. 1978. Some fossil angiospermous fruits from the Danian of West Greenland. Cour. Forsch.- Inst. Senckenberg 30: 47-53. KRasiLov, V. A. 76. Tsagayanskaya Flora Amur- skoi Oblasti. Izdat “Nauka,” Moscow Колдмс, K.-z. & A.-M. Lu. 1979. Flora Reipublicae Popularis Sinicae 21. Myricaceae, Juglandaceae. Academia Sinica, Beijing. р и of Vascular Plants. Macmillan Co., New Yor Li, P.-c. & S.-H. CHENG. 1979. Flora Reipublicae Popularis Sinicae 21. Betulaceae. Academia Si- Bei . kc le » Mastixioideen- Floren im Ter- ár der Oberlausitz. Paláontol. Abh., Abt. B, Pa- rye 2: 1-192. . Zwei ausgestorbene Gattungen im Ter- tiár Europas u nd ihre florengeschichtliche Be- deutung. Palaeontographica, Abt. B, Paláophytol. 123: 184-199. Der Formenkreis der Vietnam-Nuss [Carya poilanei (Chev.) LeRoy] in Europa. Feddes c. Nov. Regni Veg. Beih. 92: 339-385. H.-J. o 1982. Neue und interessante Arten aus dem Miozán von Salzhausen im Vo- gelsberg. Feddes Repert. Spec. Nov. Regni Veg. Beih. 93: 405-435. H. WALTHER. 1978. Die Floren der Ha- selbacher Serie im Weisselster-Becken (Bezirk Leipzig, DDR). Abh. Staatl. Mus. Mineral. Geol. Dresden 28: 1-200. 1985. Die obereozánen floren des Weisselster-Beckens und seiner Randgebeite. Abh. Staatl. Mus. Mineral. Geol. Dresden 33: 1-260. MANCHESTER, S. R. 1981a. Fossil History of the Ju- glandaceae. Ph.D. thesis. Indiana Univ., Bloom- ington. 1981b. Fossil plants of the Eocene Clarno Nut Beds. Oregon Geol. 43: 75-81. TIFFNEY — HAMAMELIDAE FRUIT AND SEED DISPERSAL 1957. Die Laubgewáchse der 415 1986. Vegetative and reproductive mor- hia i an те prih! ree (Platanaceae) from he E western North America. Bot. Gaz. cr k 147: (in press). P. R. CRANE. 1983. Attached leaves, inflo- rescences and fruits of Fagopsis, an extinct genus of fagaceous affinity from the Oligocene Florissant flora of Colorado, U.S.A. Amer. J. Bot. 70: 1147- 1164 L. DILCHER. 1982. Pterocaryoid fruits (Juglandaceae) in the Paleogene of North America and their evolutionary and biogeographic signifi- cance. Amer. J. Bot. 69: 275-286. MARTIN, A. C., H. S. ZM & A. L. NELSON. 1951. American Wildlife and Plants; A Guide to Wildlife abits. oya ae Boo , New York. MELCHIOR, H. 19 A. Engler’s Syllabus der Pflan- zenfamilien, 12th edition. Gebrüder Borntraeger, MILLER, N. G. 1971. The genera of the Urticaceae in the southeastern United States. J. Arnold Arbor : = MING, T.-L. . Flora Reipublicae Popularis Si- nicae er Daphniphyllaceae, Callitrichaceae, Empetraceae, Corariaceae, Anacardiaceae, Pen- rn Canada and rm United States. Univ. poen Press, Toro мов Ј. . Fossil сака а of extant an- giosperms. Bot. Rev. (Lancaster) 47: 1-142. NIKITIN, P. I. 1965. Akvitanskaya Semennaya Flora Lagernogo Sada ыы Izdat. Tomskogo Univ. Tomsk, U.S.S Owul, J. 1965. ds of Japan. Smithsonian Insti- tution, Meere si PALAMAREY, E. 1968. Karpologische Reste aus dem ozán мы Т aa oa Abt. B, Paldophytol 123: 200-212 Diasporen aus dern miozánen Kohle des cue Beckens (West-Bulgarien). Palaeonto- graphica, Abt. B, Paláophytol. 132: 153-164. . 1973. Fosilnata flora na Mi selva i сеш у x ene = Bot. Inst. (Sofia) 24: 75- PUL, E VAN D 1969. Principles of юс in er Pour 151 edition. Springer-Verlag, Berlin. PRAIN, D. (editor). 1917. Flora of Tropical West Af- , Volume 6, Section 2, Ulmaceae to Cycada- & yd "peg $ & С. В. BELL. 1968. tec da of the Carolinas E of the Dutch-Prussian border. Meded. Rijksopspo- ring Delfstoffen 6: 1-1 REID, E. M. & M. . CHANDLER. The London Clay Flora. British Museum ГОРЫ History), ES on 1930. The iA urs of Plants Throughout the World. L. Reeve & Co., Kent. SCHWARZWALDER, R., JR. & D. L DILCHER. 1981. Platanoid Лы, and infructescences from the pe Лен of Kansas. е Bot. Soc. Amer., 1977. Fossil fig-like objects from r Cretaceous sediments of the western inte- 416 rior of North America. Palaeontographica, Abt. B, ео. 161: 165-175. SMILEY, С. J. & L. M. HuGains. 1981. Pseudofagus idahoensis п. gen. et sp. (Fagaceae) from the Mio- E Clarkia Flora of Idaho. Amer. J. Bot. 68: 741- NUS E. Flora Malesiana, Volume 8, Part Ш Sit к шш Publs., Alphen 926. Trees and Shrubs of . Herb. 23(1-5): 1-1721. ——— 1928. Fl an the Panama Canal Zone. Contr. : 1-416. . 1946. Flora of Guate- mala. Fieldiana, Bot. 24(4): 1-493. = Flora of Guatemala. Fieldi- Bot. 24(3): 1-432. a G. L. 1984. Mosaic evolution, mosaic se- lection and angiosperm phylogeny. Bot. J. Linn. . 1969. Beitráge zur Kenntnis der Plio- zánflora von Willershausen (VID)—Die Angio- spermen-Friichte und -Samen. Argumenta Pa- = * 163-197. ZAFE 946. Flora Pliocenska z Kro$cienka тей — Akad. Umiejetn. Rozpr. Wydz. Mat.- 2s Dział B, Nauki Biol. 72 (1): 1- 162, (2): 1-2 4. Pli ж ЕЕ flora okolic Czorsztyna i jej stosunek do Plejstocenu. Prace Inst. Geol. 11: 1- 1974. Magnoliophyta l, Magno liaceae-Eu- TAK KHTAJAN, A. L. (editor). Fossilia U.R.S.S., Volume ningra (editor). 1982. Magnolio phyta Fossilia U.R.S.S., Volume 2, Ulmaceae-Betulaceae. Izdat. Nauka, Leningrad. THORNE, R. F. 1973. The "Amentiferae" or Hama- melidae as an artificial group: a summary state- ment. Brittonia 25: 395-405. . 1976. A ess avos classification of the an- giospermae. Pp. 35-106 i Steere & B. Wallace (editors) Evolutionary Biol- ogy, Volume 9. Plenum Press, New York. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 TirrNEv, B. H. 1977. Fossil oo fruits and seeds. J. Seed Technol. 2: 54- 1984 [1985]. Seed size, dispersal syndromes and the rise of the angiosperms. Ann. Missouri Bot. Gard. 71: 551-576. А . Asiatic ade ste ne n the Cainozoic flora of Eur Kongl. S Vetenskapsakad. Handl. 9: 3-87 Vink, W. 1957. Flora Malesiana, Volume 5, Part 3, Ha орна Noordhoff-Kolff N.V., Djakar- ta, Indo WALKER, E. "H. 1976. Flora of Okinawa and the Southern Ryukyu Islands. Smithsonian Institu- tion, Washington, D.C. WANG, C.-w. 1. The Forests of China. Publ. 5, Maria Moors Cabot Found., Harvard Univ., Cam- bridge, Massachusetts. WEYLAND, H. 1937. Beitráge zur Kenntnis der rhei- nischen Tertiárflora. II. Erste Ergánzungen und Berichtigungen zur flora der Blátterkohle und des Polierschiefers von Rott im Siebengebirge. Palae- ontographica, Abt. B, Paláophytol. 83: 67-122. WIGGINS, I. L. 1980. Flora of Baja California. Stan- d Univ. Press, Stanford, California. WiLLis, J. С. 1973. Willis’ Dictionary of the Flow- ering Plants and Ferns, 8th edition. Revised by H gobi e Univ. Press, Cambridge. WING, `$. L& Г. J. Hickey. 1984. The Platycarya perplex and the с of the Juglandaceae. mer. J. Bot. 71: 3. Fossil pes of Amentiferae. Brit- chronologic framework for Cenozoic megafossil floras of northwestern North erica and its relation to marine iei eed Special Pap. Geol. Soc. Amer. 184: 39-47. ZABLOCKI, J. 1930. Tertiáre Flora des Salzlagers von Wieliczka. (Zweiter teil). Acta Soc. Bot. Poloniae 7: 139-156. ZAVADA, M. S. & W. L. CREPET. 1981. Investigations of angiosperms from the Middle Eocene of North America: flowers of the Celtidoideae. Amer. J. Bot. 68: 924—933 PHYTOCHEMICAL ASPECTS OF PHYLOGENY IN HAMAMELIDAE! DAvID E. GIANNASI? ABSTRACT Chemical data for the Hamamelidae (sensu Cronquist) are numerous but scattered. Few large-scale comprehensive surveys of any particular group of compounds (micro- or macromolecular) exist for the Hamamelidae. This has limited the use of such data in drawing broad systematic conclusions beyond those based on extant morphological, anatomical, and palynological studies. Certainly, avail- maceae, Jugland snes sidus of angiospe at least from pres , have proven useful at the inter- and aceae, Urticaceae). However, the diverse and sometimes es (e.g., alkaloids, sesquiterpene lactones, polyacetylenes, glucosinolates) often rms are mostly lacking in the Hamameli biochemical conservatism (or alternatively, reduction) for the group dae. This implies, and an early and ene divergence from its more chemically diverse putative Magnoliid ances- tors. In an earlier review of the phytochemistry of the *Amentiferae," Mears (1973) catalogued the various classes of secondary metabolites for the group, including phenolics, sugars, various types of terpenoids (including iridoids), several alka- loids, and fatty acids. His major conclusions were that (1) insufficient ys of any class of secondary metabolites were available and thus (2) few correlations or putative relationships between taxa in the ““Amentiferae” could be drawn. More than ten years later (using computer- assisted and manual literature surveys), basically the same conclusions may be drawn despite a moderate increase in the number of new com- pounds discovered and an equally moderate in- crease in the number of families surveyed in de- tail for any single class of compounds (mostly phenolics). This is surprising for a group that has undergone considerable taxonomic Se and n- quist, 1981). Further, many of the Bons are isolated identifications (single species, a few compounds) and with few exceptions (Venka- taraman, 1972), little attempt has been made to summarize these scattered data. Most recent chemotaxonomic discussions of the Hamamel- idae have been placed within the broader context of angiosperm phylogeny in general (e.g., Ger- shenzon & Mabry, 1983; Gornall et al., 1979; Harborne, 1977; Harborne et al., 1975; Har- borne & Mabry, 1982; Young, 1981; Young & Seigler, 1981). o be sure, several families have been sur- veyed in detail, such as the Betulaceae (Wollen- weber, 1975) and Ulmaceae (Giannasi, 1978), as have several genera, for example, Fagus (Gian- nasi & Niklas, 1981) with, in some cases, em- phasis on different tissues such as wood chem- istry, for example, Moraceae (Venkataraman, 1972). All have been helpful at their respective taxonomic levels but of limited use above the family level. Macromolecular data for the Ham- amelidae, in the form of serological studies, are now available through the efforts of Fairbrothers and co-workers (Brunner & Fairbrothers, 1979; Petersen & Fairbrothers, 1979, 1983, 1985). However, many of the phytochemical correla- tions that do exist for the Hamamelidae still rest on earlier secondary metabolite surveys, pri- marily phenolics, and it is here that major em- ж continues (Egger & Reznik, 1961; Bate- Smith, 1962; Kubitzki & Reznik, 1966; Jay, 968 — This discussion is intended to: (1) provide a summarized update of the earlier review by Mears (1973) in terms of some of the new classes of compounds discovered in the Hamamelidae in the past ten years and (2) to highlight several types of chemical data that have recently been ! My thanks to Patricia Holmgren n (NY), Thomas a (BKL), and S. B. Jones and Nancy Coile (GA) for providing extensive herbarium and cultivated sam of a number of the taxa studied in this work. M thanks also to Joscelyn Hill for technical assistance on the sae studies. This work was supported by National 120515 Science Foundation grants DEB- and BSR-8120515. 2 Department of Botany, open of Georgia, Athens, Georgia 30602. ANN. Missouni Bor. GARD. 73: 417—437. 1986. E 1. Subclass Hamamelidae according to B Um ). ANNALS OF THE MISSOURI BOTANICAL GARDEN Subclass Hamamelidae Putative Relationships or Alternative Treatments? Trochodendrales ны ochodendraceae Hamamelidales Cercidiphyllaceae Eupteleaceae Daphniphyllales Daphniphyllaceae Didymelales Didymelaceae Eucommiales Eucommiaceae Urticales Barbeyaceae Ulmaceae Cannabaceae Magnoliidae Hamamelidaceae; Altingia, Liquidambar Trochodendrales Hamamelidaceae (Disan- thus Cercidiphyllaceae, Pla- tana agnoliidae (Schisandvacens) Euphorbiaceae Leitneriaceae, Euphorbi- ales, Thymelaceae Urticales, Hamamelida- ceae, Magnoliales Malvales alvales Fagaceae, Betulaceae Moraceae Moraceae, Urticaceae Moraceae Didymelales Rutales (Sapindales) Juglandales J Anacardiaceae (Juliana- uglandaceae ceae) Rhoipteleaceae Myricaceae, Fagaceae, Bet- ulaceae Myricales Juglandales, Fagales Myricaceae Fagales Balanopaceae Betulaceae Trigonobalanus, Fagaceae Casuarinales Casuarinaceae Betulaceae, Myricaceae 4 Miu in iil column — in alternative taxa in h the Hamamelidae (sensu Cronquist, 1981) have been d in by pris authors (see text at right for ref- es). applied to systematic problems in the Hama- su Cronquist (1981). The latter it innate pedagogical convenience as a frame- [Vor. 73 work against which other treatments may be compared. It certainly has provided a focal point for the spirited systematic discussion over the definition of the Hamamelidae. The historical development of the concept of several different treatments of the Hamamelidae (within larger angiosperm classifications) in con- temporary systematics (Conquist, 1968, 1981; Thorne, 1983; Dahlgren, 1980; Takhtajan, 1954, 1969, 1980). Some of these undergo regular re- visions (Dahlgren, 1977, 1980, 1983; Thorne, 1973, 1976, 1977, 1983). Others are more spe- cific reviews of the Hamamelidae alone (Abbe, 1974; Endress, 1977; Meeuse, 1975) or of specific orders along with other putative ada within the subclass (e.g., Berg, 1977). Merx- müller (1977) has succinctly commented on eae e the Hamamelidae contain the taxa shown in Ta- ble 1, the left-hand column of taxa representing Cronquist's treatment, the column to the right showing a selection of some other relationships suggested by other SN, Most authors concur that a basic “соге” of taxa including the Ham- amelidales and Fagales probably represent the true concept of Hamamelidae (and then perhaps conservatively only the type families). All the other orders (a number of which are monotypic or at least monogeneric) are moved with great frequency (and often with justifiable logic) to oth- er subclasses, orders, or families and back again. The phytochemist is often at a loss as to which and how many taxa to Masi to provide an adequate survey of w various taxonomists consider the Hamamelidae and related taxa. CHEMICAL REVIEW If any group has exploited the use of phenolic compounds, surely it is the Hamamelidae. Since many taxa in Hamamelidae are woody, or es- sentially so, this is perhaps not unexpected and Cronquist (1977) attributed this characteristic d of this subclass (Mears, 1973) as in, for example, the Moraceae (Fig. 1; cf. Venkataraman, 1972 owever, as Mears indicated, many of these compounds often are characteristic of only a few 1986] species or a genus or a family and thus of little taxonomic help beyond that taxonomic level, es- pecially in confirming or denying their position in the Hamamelidae. Alternatively, these com- pounds may be scattered throughout other dis- parate taxa in the angiosperms as a whole. The current known distribution of flavonoid bifla- vonyls is an example of such a class of com- pounds. In other cases, probably the majority, comprehensive surveys are lacking. Any heady phylogenetic relationship among plant taxa based on the presence or absence of a single or limited number of chemical characters often lasts only until the publication of the next phytochemical survey. In general, tannins, proanthocyanidins (e.g., prodelphinidin), ellagic acid, and especially my- ricetin (and other vicinyl-hydroxylated com- pounds), are common to the Hamamelidae (Fig. 1). They have been used most commonly to sep- arate the Hamamelidae (generally present) from the Magnoliidae (generally absent), although there are reports of their scattered occurrence in the Rosidae and Dilleniidae as well, with single iso- lated reports in the Asteridae and Liliopsida (al- though the Nymphaeales commonly have both ellagic acid and myricetin). A few species of the vones, but with flavones generally low in num- ber, at least by present survey data. O-methylation of flavonols and flavones in the Hamamelidae also seems to be low, with some exceptions (Bet- ulaceae, Wollenweber, 1975; Moraceae, Ven- kataraman, 1972), but certainly not to the extent seen in other subclasses (e.g., Rosidae, Dilleni- idae, Asteridae). Thus, the Hamamelidae possess a qualitatively conservative flavonoid comple- ment, again at least by present surveys— perhaps the most significant caveat for such conclusions. The occurrence of biflavonyls (Fig. 1) in Ca- suarina is unusual, as is their occurrence in a few other angiosperms. These flavonoid dimers are and some lower tracheophyptes (except ferns) and a moss or two. In angiosperms they occur in such disparate groups as Nandinaceae (Ra- ), Rhamnaceae, Euphorbiaceae, hymelaceae, Ochnaceae, Clusiaceae (Guttifer- ae), Anacardiaceae, Burseraceae, Caprifoliaceae, and most recently have been found in the mon- GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY 419 ELLAGIC ACID MULBERRIN (MORUS) MORIN (Horus) FIGURE 1. Examples of different types of phenolics in the Hamamelidae. ocots, the Amaryllidaceae (Liliaceae). Their scat- tered distribution offers little systematic value at this time and may represent isolated parallel syn- thetic capabilities or perhaps only a lack of com- prehensive surveys (Geiger & Quinn, 1975, 1982). More typical flavonoid monomers also occur commonly in Casuarina spp. consisting of a number of glycosidic variations based on a con- servative aglycone complement of myricetin, quercetin, and kaempferol (Saleh & El-Lakany, 1979). Quinones (Fig. 2) are found in several families in the Hamamelidae, especially the Juglanda- ceae, which produces the P. e dx juglone, and several allied compounds such a bis-juglone (Gupta et al., 1972; Pd 0 à Babu, 1978). Mixed phenolic-terpene (and ses- quiterpene) quinones such as the tic naph- thalenes of Ulmaceae heartwoods are known (Mears, 1973) along with some rare flavonol and flavanonol C-glycosides (Thomson, 1979; Hillis & Horne, 1966) and unusual C-methyl dihydro- chalcones in Myrica (Malterud et al., 1977; Uyar et al., 1978). ded acid patterns in nut oils of Car) Iso been employed with sys- tematic : success pidan et al., 1969) 420 OH о O 0н | | 3, 3'-BISJUGLONE JUGLANS REGIA OH 0 —- GLUCOSE LINOOL-9-0-GLUCOSIDE BETULA ALBA FiGURE 2. Quinone and monoterpene types in the Hamamelidae. Several types of terpenes are found in the Hamamelidae. Typical monoterpenes are found in Myrica (Myricaceae) and these have been РОЙ (Halim & Col 1973). Unusual monoterpene glycosides i E occur in Betula (Tschesche et al., 1977). Mono- terpene lactones or iridoids (Fig. 3) have Alo been found in the Hamamelidaceae (Liguidam- we i an (Daphniphyllum), Eu- ommiace Eucommia), and Didymelaceae Di a Undoubtedly more will be discov- ered as specific surveys continue (El-Naggar & Beal, 1980; Kaplan & Gottlieb, 1982; Gershen- zon & Mabry, 1983; Bianco et al., 1982). At this timo the presence of iridoids in a few Hama- noliidae, which currently sient to lack them. Sesquiterpene lactones appear to be absent from the Hamamelidae as well but present in the Mag- noliidae, providing yet another в character between the two subclas In terms of nitrogen saat nine, — roducts, glucosinolates appear to be absent from all Hamamelidae regardless of whose taxonomic ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 MS 0 ASPERULOSIDE DAPHN IPHYLLACEAE AcOCH» OGLUC H CO0CHz DAPHNIPHYLLOSIDE DAPHNIPHYLLACEAE OGLuc ACOCH, OISOMALTOSE т HOCH, 0н ULMOS IDE EUCOMMI ACEAE EUCOMMIOL EUCOMMI ACEAE снн RE 3. Examples of iridoids found in some Hos dae. scheme is used. Reports of cyanogenic glycosides (Fig. 7) rest mainly on color tests rather than extensive specific compound identifications. Such color tests have been reported for several families (Hamamelidaceae, Fagaceae, Juglandaceae). Complete identification is needed to clear this up especially in a biosynthetic sense (Hegnauer, 1973, 1977), but the few found are of the tyro- sine-derived types. Several types of alkaloids occur in the Ham- amelidae (Fig. 4). Most are characteristic of a genus or two and of limited use due to their re- stricted occurrence, inadequate survey, or scat- tered occurrence in seemingly unrelated (at least not closely related) taxa (Mears, 1973). These Ll tropine types (pseudopelletierine) in Fi- us (Moraceae) along with the tylophoric alka- lode which also occur in Tylophora (Asclepia- daceae) and Cryptocarya (Lauraceae). A series of diterpene alkaloids (e.g., daphniphylline, daph- nigraciline) have been reported from Daphni- phyllum gracile, mostly from bark, and as yet are 1986] GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY 421 HAMAMEL I DAE-ALKALOIDS PSEUDOPELLET IERINE (FICUS) TYLOPHORIN (Ficus) H CHa CHoNHo N 5-HYDROXYTRYPTAMINE (URTICA) O Pe H3C0 c—( N || 0 , 4-DIMETHOXY-W - (2' — ACETOPHENONE (BOEHMERIA) f STEROIDAL ALKALOIDS (DIDYMELES) FIGURE 4. Alkaloids in the Hamamelidae. taxonomically isolated in this occurrence (Grun- don, 1; Yamura et al., 1977, 1980). Steroid alkaloids have also been discovered in Didymeles (Ahond et al., 1980). The Hamamel- idae generally appear to lack the tyrosine/phe- nylalanine-derived (benzyl-) isoquinoline alka- loids of the Magnoliidae-Caryophyllidae-Lilliidae fo sn ening In this way, the Hamamelidae are e like Rosidae-Dilleniidae-Asteridae, in die n non-aromatic derived alkaloids (amino acids xd D cycle or terpenoids) begin to predom Ed. there is an unusual variety of sec- chemical studies of specific genera or families using single classes of compounds in detailed sur- veys have produced both interesting systematic results and grist for the chemist's mill, as indi- cated in the following discussion. CHEMOSYSTEMATIC STUDIES — MICROMOLECULAR Some of the more recent comprehensive stud- ies in the Hamamelidae have centered on the Ulmaceae. An early flavonoid study of Ulmus and several related genera (Bate-Smith & Rich- TABLE 2. Generic distribution of flavonoids in the Ulmaceae (Giannasi, 1978) Flavonoids Affinity’ Flavonols 1. Ampelocera C ашы С 3 arbeya (—) 4 ris MR (—) 5 Hemiptelea U 6 Holoptelea* U 7 Mirandaceltis (—) 8 Phyllostylon U 9 Planera U 10. Ulm U 11. Zelkova U 12. Gironniera: Galumpita* Glycoflavones 12a Gironniera: Gironniera C 13. Celtis C 14. Chaetachme C 15. Lozanella C 16. Parasponia C 17. Plagioceltis (—) 18. Pteroceltis 19. Trema а According to Grudzinskaya (1965); C = Celtoid, U = Ulmoid, (—) = not considered by Grudzinskaya. > Data from Bate-Smith and Richens (1973). * Placed in Aphananthe by some authors. ens, 1973) showed that flavonoid evolution in the genus probably proceeded by reduction in flavonoid types and content (mostly flavonols). Bate-Smith and Richens also noted that several other related genera differed in their possession of flavone compounds but did not pursue it fur- ther. Subsequent studies by Giannasi and Niklas (1977) suggested that a flavonoid dichotomy ex- isted between U/mus (flavonols) and Celtis (gly- separate families (Ulmaceae and Celtidaceae) rather than the more common treatment of two subfamilies (cf. Giannasi, 1978, for discussion and references). Later, SEM pollen analyses by Zavada supported Grudzinskaya’s treatment (Zavada, 1983). As Zavada indicated, these data, along with fossil evidence (Zavada & Crepet, 1981), suggest that the two subfamilies have had и v5 and perhaps earlier. 422 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE 3. Leaf flavonoid? distribution in genera of the Juglandaceae (Giannasi & Niklas, unpubl. data). Fla- Glyco- Flavonols vones flavones Flavanonols Phenolics Taxon’ M Q K A A DQ DK EA GA Platycarya (1) + + + + + Pterocarya (6) + + + + + Alfaroa (3) + + + + + Juglans A. (7) + + 4 + + + + + B. (6) + + + + + + Carya 2 (6) + + + + + + B. (6) + + + ? + + Oreomunnea A. (1) + + + + + + . + + + + Engelhardia (3) + + + 4 + + a Abbreviations: M = myricetin, Q = quercetin, K = kaempferol, A = apigenin, DQ = dihydroquercetin, DK = dihydrokaempferol, EA = ellagic acid, GA = gallic acid derivatives, + = present, ? = not completely confirmed. ^ Number in parentheses indicates number of species examined in each genus. The Juglandaceae are another recently studied ing the taxonomic affinities of a Miocene fossil group. Cronquist (1981) placed them in their own leaf compression as that of a Juglans and its order along with the Rhoipteleaceae. Thorne putative relationship to North American taxa. (1983), however, placed the family in his super- The presence of this rather conservative leaf order, Rutiflorae, suborder Juglandineae, not far flavonoid complement in the Juglandaceae cer- from the Anacardiaceae (suborder Rutinae); tainly allows the family to lie comfortably within Dahlgren (1980) places them in his Rosiflorae the Hamamelidae. A suggested relationship of along with most of the Hamamelidae (sensu the Juglandaceae with or close to the Anacar- Cronquist, 1981). diaceae, based primarily on the presence of el- In a recent flavonoid study ofthe Juglandaceae — lagic acid and myricetin in both taxa, does not (Giannasi & Niklas, unpubl. data) it was found seem strong at this point, especially, when com- that mostly common flavonol glycosides, includ- pared with the large number of unusual flavonoid ing those of myricetin, quercetin, and kaempferol types found in the Anacardiaceae (including along with two flavanols were produced in the Julianiaceae) by Young (1976, 1979). The Ana- leaves of the Juglandaceae (Table 3). Flavones cardiaceae, for example, possess anthochlor and glycoflavones apparently are absent, or if pigments, methylated flavonols, and 5- an present occur in trace amounts that are difficult — 7-deoxyflavonoids not found in the Juglanda- to recover. The several genera examined may be ceae. Also, the lack of 5-methoxy flavonoids in separated into two major groups based on the the Anacardiaceae suggests that the Juglandaceae presence or absence of myricetin glycosides, as (which do possess them) are more compatibly shown in Table 3. All ofthe genera that produce retained in the Hamamelidae (at this time). The myricetin occur in the New World. If the pres- presence of biflavonyls in the Anacardiaceae ab- ence of myricetin is considered a primitive char- solutely sets this family apart from the Juglan- acter, then one-character chemotaxonomy would аасеае in which they are unknown. suggest that the family may have originated in A comprehensive survey of leaf bud flavonoid temperate North America. Indeed, the only ex- exudates has been carried out on the Betulaceae, ception to this is the Asian genus Engelhardia, including the genera Betula, Alnus, and Ostrya which does produce myricetin. However, thisap- by Wollenweber (1975). All three genera could parently “Asian” taxon producing myricetin was be distinguished on the basis of their flavonoids, widely represented in North America during and considerable interspecific flavonoid differ- Eocene times (Dilcher et al., 1976; Crepet et al., | ences were observed within each genus. What adi its current remaining Asian “endemism” was most interesting was the very large number ing a secondarily derived or simply fortuitous of O-methylated flavonols that occurred in these relictual distribution. The myricetin marker genera, as well as a few methylated flavones, a compounds also proved to be useful in confirm- flavonoid character of advancement (including 1986] 6-hydroxylation) observed only in some Rosidae and Asteridae, but not in other Fagales based on published surveys. Strangely enough, myricetin glycosides are not reported in the bud scales but commonly occur in the mature leaves along with several of the flavone and flavonol types cited for bud scales (Giannasi, unpubl. data). The fam- ily, therefore, seems to s some advanced (or specialized) biosynthetic слао (О- Bain ylation, 6-substitution) wit the Hamameli- dae, although more uite myricetin pud sides do appear in the leaves of some species of Betula. Beyond these few most bé vanus are based on n older broad кечын of ydrolysis in angiosperms (Bate-Smith x 1962: Lebreton, 1965; Jay, 1968) i in wbidh the flavonol aglycones are reported. A few (Kubitzki & Reznik, 1966; Gurni & Kubitzki, 1981; Egger & Reznik, 1961) do identify other flavonoids and their glycosides. Flavones or other compounds generally are not cited even though they are pres- ent (see below), which can give rise to spurious interpretations as we shall see. agly one CHEMOSYSTEMATIC STUDIES— MACROMOLECULAR Available macromolecular data on taxa in the Hamamelidae emanate from the efforts of Fair- brothers and colleagues (Brunner & Fairbrothers, 1979; Petersen & Fairbrothers, 1979, 1983) and have been obtained at several taxonomic levels. In a serological study of the Corylaceae (Brun- ner & Fairbrothers, 1979), serological affinities of representative taxa from Alnus, Betula, Car- pinus, Corylus, and Ostrya were examined. Us- ing four serological techniques the genera could be divided into three major groups: (1) A/nus, (2) Betula, and (3) Carpinus, Corylus, and Os- trya. Betula proved to be the most serologically isolated taxon of the five but showed closest af- finities with A/nus. Alnus, though distinct, was most similar to Corylus of Group 3. Overall sim- ilarities between all five genera suggest that they be retained within a single family (as tribes cor- responding to the serological groupings) rather than elevating Group 3 to familial status, that is, Corylaceae. In a second study (Peterson & Fairbrothers, 1979), an attempt was made to determine if the Juglandaceae, Myricaceae, and Fagaceae were closest to a Hamamelid origin (Cronquist, 1981; Takhtajan, 1969; Hutchinson, 1959) or if the Juglandaceae and Myricaceae are of a Rutalean GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY 423 origin near the Anacardiaceae (Thorne, 1976) with only the Fagaceae and Hamamelidaceae re- tained within the Hamamelidae, or one of sev- eral other possibilities mentioned by the other workers. Serology indicates that the Fagaceae and Myricaceae are closely related and show close similarity with the Juglandaceae as suggested by Cronquist (1981) and Takhtajan (1980). Little similarity between these three families and the Anacardiaceae was observed, thus failing to sup- port such a relationship. As mentioned earlier, our own flavonoid surveys of the Juglandaceae also fail to support any strong relationship with the Anacardiaceae. In a third study, an attempt was made to deal with one of the peripheral taxa in the Hama- melidae (sensu Cronquist, 1981), the Leitneri- ales, a monotypic order (Peterson & Fairbroth- ers, 1983) placed close to the Hamamelidales and near the Fagales-Myricales-Juglandales by Cron- quist. In fact, serology suggests =. the strongest affinity of the Leitneriales lies w and Picrasma of the аг ae thus it is of Rutalean origin rather than of Hamamelid ori- n. With the limited macromolecular data avail- able, support is given in various examples for a road concept of the Betulaceae, a solid rela- tionship of Hamamelidales-Fagales-Myricales within, or as, the Пе (or at least as а natural taxon f whose treat- ment is followed) and the removal ofa peripheral group, Leitneriales, to the vicinity of the Sima- roubaceae (Petersen & Fairbrothers, 1985). The results are encouraging and we can only hope that further studies will be conducted. HAMAMELIDAE — CURRENT SURVEYS I also undertook a limited survey of phenolics and flavonoids in the Hamamelidae both to check the results of earlier studies, especially the mon- umental work by Bate-Smith (1962) as well as others (Lebreton, 1965; Jay, 1968) and to add as for a few additional taxa where possible. that employed was that of Giannasi ud . It was observed in the earlier studies that usu- ally only the presence of flavonol aglycones was reported although from my own studies it was often obvious that other flavonoid classes were also present. My own studies further indicated that these other compounds were glycoflavones and flavones, in addition to the flavonols. Thus, the earlier studies, which emphasized only fla- 424 HAMAMELIDAE (sENsU CRONQUIST, 1981) URTICALES JUGLANDALE S MYRICALES * EUCOMMIALES * LEITNERIALES ? DIDYMELALES CASUARINALES * DAPHNIPHYLLALES HAMAMEL I DALES * TROCHODENDRALES | FIGURE 5. Putati H lid (sensu Cronquist, Tw Asterisk Xndicstes lack of el- lagic acid and myricetin compounds. vonols, inadvertently left out other flavonoids of considerable potential taxonomic importance, especially when comparing different subclasses, as we shall see below. In my own survey I was unable to confirm the presence of some aglycones cited in earlier studies of the same taxa. In some of these taxa I found additional aglycones not previously noted. In these cases such differences are most likely attributable to natural infraspe- cific variation exhibited by some taxa. The pres- ence of flavonol aglycones (myricetin, quercetin, kaempferol) often depended on the number of posite flavonoid score from several or more col- lections to characterize a taxon. For example, in the extensive literature survey by Gornall et al. (1979) the Casuarinaceae were said to lack my- ricetin (Bate-Smith, 1962), but a contemporary detailed survey (Saleh & El-Lakany, 1979) as well as my own survey clearly document the presence of myricetin and the relative small quantities of biflavonyls produced in the leaves of Casuarina species. Therefore, all of these general surveys, including my own, that included taxa not sam- pled previously, must be considered provisional. Nevertheless, some correlations may suggest several phyletic trends in the evolution of the Hamamelidae and among its related subclasses. If we consider Cronquist's phylogenetic treat- ment o we find that d data (Table 6) at the ordinal level suggests a backbone" group consisting of the Hamamel- idales-Fagales-Juglandales-Myricales-Casuari- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 TABLE 4. Glycoflavone distribution in some Ham- amelidae Family Genera Glycoflavones Cecropiaceae Cecropia, Pourouma Hamamelidaceae Sinowilsonia Moraceae Helicostylis, Ficus Urticaceae Urtica(?) Glycoflavones/Flavonols Hamamelidaceae Hamamelis Leitneriaceae Leitneria(?) oraceae Helicostylus, Cudrania Glycoflavones/Simple Flavones Moraceae Ficus, Dorstenia, Brousso- nettia Glycoflavones/Flavones/Flavonols Cannabaceae Humulus nales-Urticales. Similarly, many of the **periph- boon debated, lack both ellagic acid and myric- etin, a fact confirmed by earlier and present stud- ies. Care must be taken in using these generali- zations, however, since not all species within a genus (e.g., Myrica) possess these characters al- though most do (Table 5), nor do all families within the “core” orders characterized by cai constituents diee s them (Table 6; e.g., cales, Hamamelidales). Indeed, in dealing ate genera ae may contain several hundred taxa, existing studies are certainly provisional. Despite these caveats, several correlations and resultant hypotheses for the taxonomic grist mill are warranted based on current evidence. For example, the presence of myricetin and ellagic acid is considered a primitive chemical character 7). This also may represent (1) a separate subclass, but closely core 1 ог subclasses (see Table 1). Tiffney (1986) also in- dicated that there is little or no overall phyletic correlation in fruit dispersal mechanisms in the 1986] GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY TABLE 5. Compound distribution in taxa of the Hamamelidae sampled for phenolics and flavonoids.^ Fla Phenolics Flavonols Flavones Glycoflavones nonols Taxon GA EA M О K A L T D A L C DQ DK Trochodendraceae Trochodendron araliensis + cb Tetracentraceae Tetracentron sinense + + + Cercidiphyllaceae Cercidiphyllum japonicum + + + + Eupteleaceae Euptelea pleiospermum + + Е. polyandra + + Platanaceae Platanus на ще. p P. occidentali + + + Hamamelidaceae Corylopsis spicata + + + + C. pauciflora + + + + C. sinensis + * + + 2 vars. о ern + + + + + 4 + + pine sinensis + + p + + + + + + F. garden + + ? + + + ИЕ vernalis + + + + + H. о (+) + + + + + + ? H. macrophylla + ? + + + 2 Liquidambar styraciflua + + + + (+) L. + + + (+) a chinense - + + + (+) Sycopis sinensis (+) + + + + Parrotia persica + + Sinowilsonia йепгуй + + + + Myrothamnaceae Myrothamnus flabellifolium + + + + + ? ? Daphniphyllaceae Daphniphyllum teigmensis $ .9 + + D. glaucescens ? 7? + + р. calycinum ? 9 + + Didymelaceae Didymeles spp. (unknown?) Eucommiaceae Eucommia ulmoides + + Barbeyaceae” + + + Ulmaceae” F + E + + + + + Cannabaceae* Humulus americana + + + H. japonicus T P Е + + 426 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE 5. Continued. Flava- Phenolics Flavonols Flavones Glycoflavones nonols Taxon GA EA M О K A L T D A L C DQ DK Moraceae Broussonettia papyrifera + + + + ? udrania tricuspidata + + + + Dorstenia foetida + + + Fatoua spe + Ficus aur + + F. benjamina + F. brevifolia + + Е. caprifolia + F. caprica + F. citrifolia + + 4 F. gemin + + + + + F. laevigat + + F. llewellynii + + F. macrophylla + + F. nitidifolia + + F. pumila + F. webbiana + Helicostylis elegans + + + H. scabra + + Morus alba + + + M. rubra + + + Cecropiaceae Cecropia peltata + + Pourouma phaeotricha + + P. palmata + + Urticaceae Boehmeria cylindrica + + Laportea canadensis + + + + Urtica dioica os 7 (2 vars Leitneriaceae Leitneria floridana + + + + + ? Juglandaceae“ + + + + + ? + Rhoipteleaceae Rhoiptelea chiliantha + + + + Myricaceae Myrica asplenifolia (^ Comp- tonia perigrina) + + + + M. cerifera + + + 7 М. gale + + + + M. heterophylla + + + М. тоаога + + + + M. rubra + + + + M. serrata + + + + + Balanopaceae Balanops (unknown ?) 1986] GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY 427 TABLE 5. Continued. Flava- Phenolics Flavonols Flavones Glycoflavones nonols Taxon GA EA M Q K A L T D A L C DQ DK Fagaceae Quercus* * + + + + + + Castanea* + + + Fagus‘ ds. E Betulaceae! Betula + + + + + + (+) Casuarinaceae^ Casuarina equisetifolia + + + + C. glauca + + + + + C. cunninghamia + + + + + a Abbreviations: GA = m ag EA = ellagic acids, M = myricetin, Q = quercetin, K = kaempferol, A = lin metin, C = chrysoeniol, DQ = or occasionally present, ? = not completely co apigenin, L = luteoli ад = tri = diosm kaempferol, + = prese ace amounts dihydroquercetin, DK = dihydro- nfirmed. ь See Giannasi (1978) n detailed distributions and Table 2. Also see Bate-Smith and Richens (1973). 1979). * See also Clark € Bohm 4 See Table 3 (Giannasi & Niklas, unpubl. data). * See Niklas and Ganndsi (1978). f See Giannasi and Niklas (1981). в See also Wollenweber о Present study also includes some methylated and/or 6-substituted flavones and flavonols as per Wollenwe ^ See also Saleh and El- PR (1979). Hamamelidae and similar suggestions may be gleaned from discussion of pollination mecha- nisms (Whitehead, 1969). Indeed, protein serol- ogy (Petersen & Fairbrothers, 1983, 1985) sug- gests that the affinities of the Leitneriales, for example, lie near or within the Simaroubaceae (Rosidae, sensu Cronquist, 1981). Any one of the alternatives is possible for each of these pe- ripheral orders, especially since most of these orders are monotypic or at least monogeneric. Often there are fewer intermediates that might more clearly suggest more direct interordinal re- lationships. In addressing the flavonol bias of some earlier кене it may be observed from Table 5 that he glycoflavones represent a second major class a flavonoids occurring in the Hamamelidae. As vonols and/or flavone O-glycosides, characterize a number of species and genera in various fam- ilies. This is especially striking in the Daphni- phyllaceae in which glycoflavones occur exclu- sively; a character state considered advanced or derived over the presence of flavonols alone or the intermediate state of flavonols/glycoflavones (e.g., Leitneriaceae). The same trends (i.e., fla- vonols — glycoflavones) may also be observed at various taxonomic levels within orders (Urti- cales), families (Moraceae), and genera (Hama- melis, Ficus) and thus may represent the major chemical trend of advancement within the sub- class. This contrasts with earlier literature, which state that flavonols characterize the Hamameli- dae as a whole, implying more of a conservative flavonoid capability than really exists. Excep- tions do exist, as in the Betulaceae, for example. In A/nus, Ostrya, and especially Betula, flavo- noids from bud scale excretions contain a large number of variously methylated and 6-substi- tuted derivatives of the flavonols quercetin and kaempferol and to a much lesser degree flavones (apigenin) and flavanones (naringenin). Many of these compounds also occur in the leaves of Bet- ula species (Giannasi, unpubl. data) along with the more archaic myricetin glycosides, which ap- parently do not occur in the bud excretions. Thus, the Betulaceae have retained primitive charac- ters (flavonols) along with specialized characters (6-substitution, 6-methylation, flavones). These highly specialized flavonoids apparently do not commonly occur in the related members of the Fagaceae that have been surveyed (Niklas & Giannasi, 1978; Giannasi & Niklas, 1981; Gian- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 428 + + (+) (+) E + + +++++ + + ++++ + +++++ (+) (+) (+) + ++ + + +++++ + (+) + (¿ umouyun) + ITIDPDMÁJNA so[pouAJQ oeaoea[aidioq y эвээтрие[8п[ $әүерие[8п f эеээвцэииэ7] SILLI 629911] 92291019125 эеээвАэатея sa[e9n1/] 9eooerururoonq so[eruruioong desde [SWAPIG зэтеэшАр эвээвпАЧатчае 5эгеплАчатачае әрәәришещолАр{ әрәзр]Ацатртәләг) so[epljawewepy эезэвлриэроцоол]. эвэовцизепэ Y зэтелриэрочоол | NS ч "> Q vw => A RF " 8&8 g^ ПАП СИУ INV HO Od VO УЗ 500119121514] prouoAe[ pue orouaud uoxe[ (1861 ‘Isibuos> JO 1841 s«oj[oj Ашоцохе1) aeprjoureureH oy) ut sprouoAep pue sorououd jo uonnquisrp Kreurung '9 7149Y L 1986] GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY 429 nasi, unpubl. data). Therefore, at this time, these eis compounds in the Betulaceae seem to represent WSA 3 d я a unique event in the Hamamelidae. сев Also notable is the occurrence of a large пит- Ad qom 5. ©, І ber of flavones (and а few xanthones) that are Jg 2 ox a substituted at various positions by isoprene (5C) d I p units rather than sugars, methoxy or sulfate units dd 3 <4 in the root bark of Moraceae (e.g., Nomura et 58 8 al., 1976, 1977, 1978а, 19785; Konno et al., 1977; СИП + SEST Deshpande et al., 1973). Similarly substituted = | © 3 flavones (or flavanone or flavanonol analogues) IWT + = EE also are found in the Rosidae (Fabaceae, Ruta- © g y Q ceae) and Asteridae (Asteraceae) and thus are not TWV + чаи T unique to the Hamamelidae but are unique with- E 3 Ма їп а single family of the Hamamelidae. That most INV di 5 5 E > of these prenylated flavones in the Hamamelidae 2, = E 5- have thus far been isolated only from root bark 5 1 as Е B tissue of the Moraceae further emphasizes the Е = e E РОЗН of inter-tissue a ndn {= 1 + —-9.5.—- ust be considered in chemica E FERE studies (Gornall 9 al., 1979). Indeed, leaf fla- - У + E ы 39 vonoids, in this case glycoflavones, seem to be e| чо " YU Sx “normally” substituted with C-glycosyl sugars. 2 Өрү The occurrence of these prenylated flavones in E | 494 4 Na b = the Moraceae may represent simply an isolated E 2 Sez specialization in the Hamamelidae, as is sug- 2 lama ++ м a Bei gested by the isolated occurrence of biflavonyls Q e Е Е E in the Casuarinaceae. The latter compounds, too, £ чї Н = 3 JI. are unique to the Casuarinaceae in the Hama- £ | © 35 melidae but do occur in other subclasses (Rosi- ч B O ES > dae, Dilleniidae). The possibility that the Mora- ELIO. ceae do not belong in the Hamamelidae is also Я dea +1881 = > possible (see р. 433 for additional discussion). >23 z me, Their scattered occurrence within the angio- © ++ E: ga 3 if sperms makes them of questionable taxonomic D Е <> value at this time. ә m + E _ Е р " HAMAMELIDAE — GENERAL CONSIDERATIONS HƏ TT БЕ E & E The Hamamelidae, in terms of their phenolics, P 3 El - F 9 seem to represent a primitive group in the general eid E vp d 7 "y > E: presence of proanthocyanidins, ellagic acid, my- vo 8 £ 1 20 i ricetin compounds, and a general conservatism Z вые“ E. in other flavonoids. Variation is based on gly- va Ese + $ Egi cosylation patterns of a few simple flavonol and d = E Е > S flavone types with a moderate substitution, in 3 23 < 8 some cases, by the evolutionarily intermediate E с b “ B 5 glycoflavones. Proanthocyanidins, ellagic acid, 8 9 3 pes PE M and myricetin flavonoids are considered primi- | S 9 2. 383/25 EA tive chemical characters and are often found to S S E 3 © ЗЕ 5 & БЕД be characteristic of woody plants (Bate-Smith, E 3 Е S E 5 3 9 E 383 1962; Harborne, 1977). Thus, the Hamamelidae E AS 2 91991 8 > are distinct from the Magnoliidae and Liliopsida, m O x 2c which generally lack one or more of these com- pounds. Yet these same Hamamelid characters 430 are found to some degree among the Rosidae and Dilleniidae, inet ting a more than casual rela- tionship. predominance of these chemical characters in pd Hamamelidae, however, act more as a mark of exclusion of other subclasses because of the preponderance of their occurrence in the Hamamelidae, rather than as any absolute qualitative distinction (or requirement of *prim- itiveness"). Since non-chemical characters are n used similarly, such correlations among chemical characters are probably just as legiti- mate. For example, recent ultrastructural studies of phloem sieve-tube plastids show the uniform presence of the S-type in the Hamamelidae (with the exception of U/mus species), but the same type is also found in some Magnoliidae and Rosi- dae (Behnke, 1973, 1977). e problem of production of rather rare and unusual compounds in the Hamamelidae (es- pecially the Moraceae), which are of restricted taxonomic use, has already been mentioned. Dif- ficulties arise in the use of chemical data only in their simple distributional form without dad con- sideration of the classes of nvolved. For example, if the Hamamelidae are os ized by primitive phenolics that are either absent or only moderately represented in the other sub- classes, then the Hamamelidae must be a more primitive group (at least with respect to certain secondary compounds) and the others advanced, generally showing a decrease or loss in synthesis of these compounds (cf. Kubitzki £ Gottlieb, 1984a, 1984b). Also, the older surveys portray ama as being primitive in their flavonoid chemistry due to the overwhelming reporting of flavonols. Yet recent studies on the Winteraceae (Williams & Harvey, 1982), Idiospermaceae (Sterner & Young, 1980), and the Eupomatiaceae (Young, 1983) show not only the presence of flavones in these taxa but also a number of methylated flavones, both ad- vanced characters. If, in fact, flavones and meth- ylated flavones are common in the Magnoliidae, this, along with the absence of the primitive my- ricetin and ellagic acid (or nearly so), actually suggests a less primitive taxonomic position for the Magnoliidae, and that the more primitive Hamamelidae are not a derivative of the Mag noliidae (sensu Cronquist). Closest chemical similarities of the Hamamelidae lie with some osidae, Dilleniidae, and a few Asteridae that ave similar compounds, but again, in decreas- ing amounts (evolution by loss) indicating a more direct relationship with these taxa, or at least ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 parallel biosynthetic capability (Kubitzki & Gottlieb, 1984a, 1984b). To put it simply, one cannot delineate a group (Hamamelidae) as hav- ing primitive chemistry and then have it evolve from a group that is possibly more advanced (Magnoliidae) in its chemistry. Alternatively, the possibility still exists that early in their evolution the elidae and the Magnoliidae may have been more similar in concentrating on synthesis of phenolics. Subsequently, the Magnoliidae and Hamamelidae may have diverged, with the latter retaining emphasis on more primitive phenolics, and the former emphasizing alkaloid synthesis and advanced types of flavonoid substitutions. his, however, is a larger hypothetical “if” (and not orovable) than the former alternative, which like all current studies at least deals with factual, contemporary, comparative data. Even in the al- ternative case, common divergence, rather than derivation, is the logical conclusion to be drawn. ndeed, Kubitzki and Gottlieb (1984a) suggest that the neolignans of the Lauraceae and reduced virolane flavonoid types of the Myristicaceae may represent remnants of an earlier protoangio- sperm emphasis on shikimic acid (phenolic) syn- thesis in these members of the Magnoliidae, where benzylisoquinoline alkaloids are generally ab- sent. Other compounds (Table 7) such as glucosi- nolates, sesquiterpene lactones, and polyacety- lenes are either limited or erratic in distribution among the angiosperms, providing limited gen- eral clarification in angiosperm systematics. However, if we examine biosynthetic and dis- tributional aspects of the nitrogen-containing compounds among the angiosperm subclasses, several interesting comments can be m Much has been made of the benzylisoquino- line alkaloids as being characteristic of the Mag- noliidae. These compounds (Fig. 6) are derived from elated iso- quinoline and similar types are found in many of the monocots, and apparently in the Cary- ophyllidae as well. Indeed, if one considers the tyrosine-derived betalains simply as colored al- kaloids (Mabry, 1977), then these three taxo- nomic subclasses are closely related in their biosynthetic origin for these compounds. The Hamamelidae lack these tyrosine-derived alka- loids (at least by present surveys) and thus appear to be less than a direct offshoot of the Magno- liidae. Instead, the Hamamelidae produce most- ti 1 saa + Р РР СИ o — M 45. 4) emanating from the citric acid cycle (TCA) or 1986] Summary of secondary chemistry of angiosperms. TABLE 7. Compounds? AK GF CGN OM FO EA Taxon Liliidae O (+) (+) Magnoliidae (+) Ac/k Ac/k Ac/k GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY 431 Ac/k Ac/k + Rosidae +) Q.K (M) (+) Asteridae present, — a Abbreviations: + alkaloids, O = aromatic-tyrosine or phenylalanine precursor, Ac/k = non-aromatic amino acid precursor from polyacetylenes. iridoids, S = sesquiterpene lactones, P acetate or Krebs (TCA) cycle, I BENZYLISOQUINOLINE ALKALOILS PAPAVERINE PAPAVER LIRININE LIRIODENDRON _ MORPHINE PAPAVER FiGURE 6. Alkaloid types in the Magnoliidae. carbohydrate precursors. This is much more like the Rosidae, Dilleniidae, and Asteridae, in which tyrosine-derived alkaloids are absent or begin to decrease in representation in favor of TCA-de- rived alkaloid precursors, or more importantly in some other groups, toward exploitation of the terpene-steroid-derived pseudoalkaloids. Also, of course, not all Magnoliidae produce isoquin- oline alkaloids either. Consideration of cyanogenic glycosides, how- ever, does suggest a relationship between Ham- amelidae and Magnoliidae. Those aromatic cyanogens derived from tyrosine (Fig. 7) are found in the Hamamelidae, Magnoliidae, and Liliidae (Liliopsida) as well as in some Rosidae and As- teridae (Saupe, 1981). Oddly enough, those cyan- ogens derived from phenylalanine occur in Mas апар the Rosidae and Asteridae. The gymnosperms cine, isoleucine) begin to predominate in the Rosidae, Dilleniidae, and Asteridae, and thus may represent the more advanced forms restricted to advanced taxonomic groups, a trend not unlike that observed in the evolution of alkaloids from aromatic to non-aromatic precursors. At least in the cyanogens a more direct relationship between Hamamelidae and Magnoliidae is suggested, al- though it would be interesting to see if the Ham- CYANOGENIC GLYCOSIDES PRECURSOR EXAMPLE TYROSINE DHURRIN x 0-GLuc C=N PRUNASIN PHENYLALANINE CH; N N 0-GLuc VALINE CHz 2 LINAMARIN =N 0-GLuc I SOLEUC INE C2H5 LOTAUSTRALIN C=N IGURE 7. Major classes of cyanogenic glycosides in а and their biogenetic precursors amelidae contain more than one type of aromatic cyanogen and perhaps both aromatic and non- subclasses (and non-angiosperm seed plants), de- pending, of course, on the placement of certain taxa within one of the current angiosperm clas- sifications. Thus, some caution should be exer- cised in emphasizing the systematic significance of these compounds. Finally, in terms of iridoids, these monoter- pene lactone glycosides occur in all of Cron- quist’s subclasses except Magnoliidae, Cary- ophyllidae, and Liliidae (Liliopsida). In this case the Hamamelidae are again isolated from the Magnoliidae with suggested similarities closer to the Rosidae. Looking at Table 7 again, and considering the limitations in conclusions to be drawn from available phytochemical data, I would make the ollowing statements: (1) the Hamamelidae ap- pear to be a primitive subclass of plants at least as old as the Magnoliidae if not older and perhaps ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 MAGNOLIOPSIDA ASTERIDAE ROSIDAE HAMAMEL I DAE DILLENIIDAE CARYOPHYLLIDAE MAGNOL I I DAE NYMPHAEALES LILIOPSIDA "PROTOANG IOSPERMS" FIGURE 8. Phylogenetic relationship of the Ham amelidae to other subclasses (sensu Cronquist, 1981) based on micromolecular data not as directly related to the Magnoliidae as sug- Basted, (2) mE presence. ots a number of com- ty (affinity) with the Rosidae/Asteridae, at least at a primal level (this unique and hig in some Hamamelidae reinforce the notion of an early divergence of the Hamamelidae from the other subclasses and (4) the current concept of angiosperm monophylesis in the simple sense of a single botanical **Noah's Ark” may require a slightly larger boat or a small but closely inte- grated fleet, that is, a broader concept of the an- giosperm ancestral pool rather than limitation to the Magnoliidae alone (cf. Dilcher, 1979; Retal- lack & Dilcher, 1981). The data would suggest an arrangement like that shown in Figure 8, a conclusion similar, at least in part, to the cladistic analysis of non- chemical data for these subclasses by Nixon (un- publ. data), and to conclusions drawn in earlier studies (e.g., Hegnauer, 1977). More recent pub- lications have also come to similar conclusions when attempting to put micromolecular data within biosynthetic and distributional frame- works, recent papers by Kubitzki and Gottlieb (1984a, 1984b), being the most еи апа provocative discussions of the proble A considerable obstacle to the i ee of +. +1 t between phytochem- istry and m Dilcher (1979) suggests that the reduced, ophilous, unisexual flower types on o like SEM. иан that are found in some possible angiosperm-like Cretaceous fossils represent an alternative ancestral type for modern Hamamelids. However, although the pollen in these fossils may have been wind borne they are monosulcate. Thus they are similar to the earliest presumably angiosperm monosulcate pollen types as found in the Magnoliidae. The presumably more advanced triaperturate pollen of the Hamamelids does not occur as early in the fossil record. Based on pollen distributions then, Hamamelidae still are currently considered to be a derived group (from the Magnoliidae). The problem of opposition in two essentially single morphological character approaches (as well as micromolecules versus pollen) seems insoluble at this point. However, just as phytochemical conclusions may change with each survey, so too each new palynological find may alter current concepts. The chemotaxonomic debate is by no means finished either, as evidenced by the recent se- rological review of the angiosperms by Jensen and Greven (1984). These authors indicate that their serological results support a conservative m Interestingly, these serological data also suggest that the Betulaceae are a discordant taxon within the Fagaceae (as do flavonoids), showing a great- er similarity to the Magnoliidae. Certainly the rather complex flavonoid chemistry of the Bet- ulaceae (Table 6) does set the family apart within (and perhaps the Urticaceae; cf. Table 6), whose flavonoid chemistry is quite unusual within the Urticales and the Hamamelidae generally. Se- rological work by Petersen and Fairbrothers (1985), in fact, suggests that the Moraceae, Can- nabinaceae, and perhaps Urticaceae as well, do not fit in the Hamamelidae, but are better placed near or in the Malviflorae (sensu Dahlgren et al., Y Mn раж the isoprenyl flavonoids in the Moraceae are found elsewhere only in taxa of the Rosidac/Dilemid (sensu Cronquist) lines. Jensen and Greven (1984) discussed interre- ити among other subclasses as well, and indicated that an expanded survey is desirable (only three taxa tested from the Hamamelidae). Further, they recognized the possibility of changes GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY 433 in their interpretation with the inclusion of more taxa in the test sample. Some considerable Кел ation in precipitation reactions was observ among species from the same genus (Passiflora) with the Magnolia seed protein employed. Sim- ilar exceptional reactions were also observed in some families (Solanaceae). These exceptions and the use of a single seed storage protein still argue for caution in the interpretation of these data. It is clear from both micro- and macromolecu- lar studies that the Hamamelidae (sensu Cron- quist) probably still do not represent a totally natural and homogeneous taxon, and both mi- cromolecular and macromolecular data have much to contribute towards the taxonomic re- finement of the concept of the Hamamelidae. LITERATURE CITED ABBE, E. C. 1974. Flowers and inflorescences of the Amentiferae. Bot. Rev. (Lancaster) 40: 159-261. AHOND, А., . DEBRAY, F. Picot, C. POUPAT, V SANCHEZ & P. POTIER. 1980. Alcaloidiques ste- roidiques di Didymeles cf. madagascariensis. Pl. Med. 39: 204. BATE-SMITH, E. G. 1962. The phenolic constituents of plants and their taxonomic significance. Bot. J. Linn. Soc. 58: 95-173. К. H. RicHENs. 1973. Flavonoid chemistry and taxonomy in U/mus. Biochem. Syst. Ecol. 1: 141-146. BEHNKE, H.-D. 1973. Sieve-tube plastids of Hama- melidae. я 22: 205- 1977. nsmission ele ctron microscopy and flowering plants. Pl. Syst. Evol. systematics of Suppl. 1: 155-178 BERG, C. C. 1977. Urticales, their differentiation and systematic position. Pl. Syst. Evol. Suppl. 1: 349- 374. BIANCO, A., C. C. BONINI, C. IAVARO NE & C. TROGOLO. 19 c BRUNNER, F. BROTHERS. ical investigation of de Corylaceae. Bull. Torrey Bot. Club 106: 97-103. CLARK, M. N. & B. A. BoHM. 1979. Flavonoid vari- ation in Cannabis L. Bot. J. Linn. Soc. 79: 249- 257 СкЕРЕТ, W. L., О. L. DILCHER & F. W. POTTER. 1975. investigations of angiosperms from the Eocene of North : а catkin with juglandaceous affin- ities. yea T “Bot. 62: 813-823. MEI A. 1968. The Evolution and Classifica- n of Flowering Plants. Houghton-Mifflin, New York. . 1977. On the taxonomic significance of sec- ondary metabolites in angiosperms. Pl. Syst. Evol. Suppl. 1: 179-189. 1981. An Integrated System of Classification of Flowering Plants. Columbia Univ. Press, New York. 434 DAHLGREN, R. 1977. A commentary on a diagram- matic presentation of angiosperms in relation to the distribution of character states. Pl. Syst. Evol. Suppl. 1: 253-283. 1980. A revised system of classification of the angiosperms. Bot. J. Linn. Soc. 80: 91-124. 1983. General aspects of angiosperm evo- lution and macrosystematics. Nordic J. Bot. 3: 119-149. 1981. ENDAHL-JENSEN & B. NIELSON. D. S. Seigler (Editore) Plivtochemistry and Angio- sperm Phylogeny. Же New York. DESHPANDE, у ,A. V А КАО, M. VARDAN & K. VENKATARAMAN. 1973. Wood and bark phe- nolics of Maclura pomifera: four new xanthones. Industr. J. Chem. 11: 518-524. DILcHER, D. L. 1979. Early angiosperm reproduc- tion: an introductory report. Rev. Paleobot. Pal- ynol. 27: 291- ——, F. W. POTTER, JR. & W. L. CREPET. о rth America: juglandaceous winged fruits. hace J. Bot. 63: 532 EGGER, K. & H. NIK. 1961. Die flavonolglykoside er Hamamelidaceen. Planta 57: 239-249. EL-NAGGAR, L. J. & J. L. BEAL. 1980. Iridoids: a review. J. Nat. Prod. 43: 649- ENDRESS, P. 1977. Evolutionary trends in the Ham amelidales-Fagalese group. Pl. Syst. Evol. Suppl. 1: 321-347 GEIGER, H. & C. QUINN. 1975. Biflavonoids. Pp. 692- 742 in J. B. Harborne, T. J. Mab H. Mabry (editors), The Flavonoids. Chapman and Hall, London. 1 1982. Biflavonoids. Pp. 505—534 in J. B. Harborne & T. J. Mabry (editors), The Flavonoids: Advances in Research. Chapman and Hall, London GERSHENZON, J. & T. J. Masry. 1983. Secondary metabolites and higher classification of angio- sperms. ers Bot. 3: 5-34. GIANNASI, D. E. 1978. Generic relationships in the Ulmaceae uscd on flavonoid chemistry. Taxon 27: 331-334. . NIKLAS. 1977. Flavonoid and oth chemical constituents of fossil miocene Celtis кар Ulmus (Succor Creek Flora). Science 197: 765- 67. & . 1981. Comparative paleobiochem- istry of some Po and extant Fagaceae. Amer. J. Bot. 68: 7 GORNALL, R. J., E A. BOHM & R. DAHLGREN. 1979. The distribution of eod in the angiosperms. 32: ot. Not. MAE SKAYA, I. A. 1965. The Ulmaceae and rea- ns for distinguishing the Cu as a sep- =. family Celtidaceae tus Bot. Zurn. (Moscow & Leningrad) 52: 1723-17 GRUNDON, M. F. (Senior А 1977. The Al- kaloids (A Specialist Periodical Report), Maru 7. The Royal Society of Chemistry, Londo (Senior Reporter). 1981. The Alkaloids (A ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Specialist Periodical Report), Volume 11. The Royal Society of Chemistry, London GUPTA, S. R., B. RAVINDRANATH & T. R. SESHADRI. 1972. Polyphenols of Juglans nigra. Phytochem- istry 11: rae ў Совм, А. А. & К BITZKI. 1981. Flavonoid chem- istry and DH of the Dilleniaceae. Bio- chem. Syst. Ecol. 9: 109-114. HARBORNE, J. B. 1977. Flavonoids and the evolution noids. Advances in Research. Chapman and Hall, London. & H. Masry (editors). мн The Flavonoids. Chapman and Hall, Lon HALM, А. Е. & В. P. Сошлмѕ. 1973. Ferd oil analysis of the Myricaceae of the eastern United States. suey neue 12: 1077-1083. HEGNAUER, 1973. Die cyanogenen verbindungen der ае und Magnoliatae- Magnoliidae: zur уап ogenese. Biochem. Syst. Ecol. 1: 191-197. . Cyanogenic compounds as systematic markers i in tracheophyta. Pl. Syst. Evol. Suppl. 1: -200. HiLtis, W. E. .H. S. Horn. 1966. The structure of Keyakinin. Austral. J. Chem. 19: 705-708. HUTCHINSON, J. 59 e Families of Flowering lants е І, Dicotyledons. Oxford Univ. Press, ord. Jav, M. 1968. Distribution des flavonoides chez lez Hammelidacees et familles affines. Taxon 17: 136- 147 JENSEN, U. & G. GREVEN. 1984. Serological aspects and phylogenetic E ofthe Magnoliidae. Taxon 33: 563-577 KAPLAN, М. А. C. & O. В. GOTTLIEB. 1982. Iridoids as systematic markers in dicotyledons. Biochem Syst. Ecol. 10: 329-347. Коммо, C., У. OSHIMA & H. HIKINO. 1977. Moru- sinol, isoprenoid flavone from Morus root barks. Pl. Med. 32: 118-124. Ковітхкі, К. & О. R. GOTTLIEB. 1984a. Micromo- lecular patterns and the evolution and major clas- sification of angiosperms. Taxon 33: 375-391. & 9 “ee icon pen of n. Act t. Neerl. angiosperm origin an 33: 457-4 & H. REZN NIK. 1966. Flavonoid-muster der Polycarpicae als VPE P EE merkmal. Beitr Biol. Pflanzen 42: 445-470. LEBRETON, P. 1965. а de chimiotaxomie bo- tanique. 2. Cas de flavonoides chez des Urticales; conclusions generales. Bull. Soc. Bot. France 111: 80-93. е T.J. 1977[1978]. The аы ea Bot. Gard. 64: 210-220. Мы К. E., T. ANTHONSEN & С. B. LORENTZEN. 1977. Two new C-methylated flavonoids from Myrica gale. Paice 16: 1805-1809. 73. Chemical constituents and sys- 75. Floral evolution in the Ham ameli dae. Ш. Hamamelidales and виа 1986] groups including Urticales, and final conclusions. Acta Bot. Neerl. 24: 181-191. MERXMÜLLER, H. Summary lecture are ы zm Evolution). РІ. Syst. Evol. Suppl. 1: 397- $, К. J. & D. E. GIANNASI. 1978. Angiosperm paleobiochemistry of the Succor Creek Flora (Miocene) Oregon, USA. Amer. J. Bot. 65: 943- NOMURA, T., T. Fukai & M. KATAYANAGI. 1977. Ku- wan , B, Ca and oxydihydromorusin, four new flavones from the root bark of the d mul- berry tree (Morus alba L.). Chem. Pharm. Bull. 25: 529-532. 1978a. Studies on the con- stituents of the cultivated mulberry tree. III. psp lation of four new flavones, kuwanon A, B, oxydihydromorusin from the root bark of mre alba L. Chem. Pharm. Bull. 26: 1453-1458. ——— , — ——, S. YAMADA & M. КАТАҮАМАСІ. 1976. Phenolic constituent of the cultivated mulberry tree (Morus alba L.). Chem. Pharm. Bull. 24: 2898- 2900. ———, 1978b. Studies on the constituents of the cultivated mulberry tree. Ee new prenylflavones from the root bark of на, Chem. Pharm. Bull. 26: 1394-1402. ие , M. & M. H. Basu. 1978. A new bisjuglone Som Juglans regia root bark. Phyto- chemistry 17: 2042-2043. PETERSEN, F. P. & D. E. FAIRBROTHERS. rological -— y: selected Amentiferous з xa. Syst. Bot & e A E appraisal of Amphipterygium and Leitneria —4wo amentif- erous taxa of Rutiflorae (Rosidae). Syst. Bot. 8: 8. & ————. 1985. A serotaxonomic appraisal of the “Amentiferae.” Bull. Torrey Bot. Club 112: —52. RETALLACK, С. & D. L. DiLCHER. 1981. Arguments fora glossopterid ancestry of angiosperms. Paleo- biology 7: SALEH, N. A.M M. & M. H. EL-LAKANY. 1979. A quan- titative variation in the flavonoids and phenolics of some Casuarina species. Biochem. Syst. Ecol. 7: 13-15. SAUPE, S. G. 1981. Cyanogenic compounds and an- giosperm phylogeny. Pp. 80-116 in D. A. Young & và S. Seigler (editors), elle era and An- rm Phylogeny. Praeger, New York. 1973. Development of ы yim т ous concept. Brittonia 25: 316-333 STERNER, В. W. & D. A. YOUNG. 1980. Flavonoid chemistry and phylogenetic relationships of the Idiospermaceae. Syst. Bot. 5: 4 : STONE, D. E., С. A. ADROVNY & В. Н. FLAKE. 1969. New World Juglandaceae. II. Hickory nut oils, phenetic similarities, and a implica- tions in the genus Carya. Amer. J. Bot. 56: 928- 9 35. TAKHTAJAN, Á. 54. The Origin of Angiospermous Plants, 2nd edition, Soviet Sciences Press, Mos- cow. а Russ 969. Flowering Plants, Origin and Disper- GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY 435 sal. Oliver and Boyd, Edinburgh. [English trans- lation from the Russian by C. Jeffrey. . 1980. Outline of the classification of flow- ering plants (Magnoliophyta). Bot. Rev. (Lancas- ter) 46: 225-359. Thomson, R. H. 1979. Recent advances in the chem- istry and biochemistry of quinone pigments. Pp. 287-312 in T. Swain, J. B. Harborne & C. F. van Sumere (editors), Recent Advances in Phytochem- e12 istry, Volume 12, Bioc mum of Plant Phenolics. Plenum Press, N Уу THORNE, К. F. 73. The г ЗЕТЕ ог Ната- melidae as an artificial group: a summary state- ment. сери 25: 395—405. phylogenetic og ofthe an- giospe В ы 35-106 in M. K. Hecht, Steere & B. Wallace (editors), о, Biol- ogy, Volume 9. Appleton-Century-Crofts, New or 1977. Some realignments in the Angiosper- mae. Pl. Syst. Evol. Suppl. 1: 299-319. —. 1983. Proposed new realignments in the an- giosperms. Nordic J. Bot. 3: 85-117 TirFNEY, B. H. 1986. Fruit and seed dispersal and the evolution of the Hamamelidae. Ann. Missouri Bot. Gard. 73: 394-416. TSCHESCHE, R., Е. CIPER & E. BREITMAIER. 1977. Monoterpen-glucoside aus den blattern von Bet- ula alba und den Fruchten von Eur ja- ponica. Chem. Ber. 110: 3111-311 Uyar, T., К. E. MALTERUD & T. ee 1978 Two new Podes Sg ree from Myrica gale. Phytochemistry 17: 2011 VENKATARAMAN, K. 1972. е phenolics in the chemotaxonomy of the Moraceae. Phytochemis- try 11: 1571-1586. WHITEHEAD, D. В. 1969. Wind pollination in the angiosperms: evolutionary anne nvironmental considerations. Evolution 23 : WILLIAMS, C. A. & W. J. HARVEY. 1982. Leaf fla- vonoid patterns in the Winteraceae. Phytochem- т 21: 329-337 ау vonoidmuster in knos- J. A. LAMBERTON, H. IR Daphniphyllum alkaloids with a new pg het- erocyctic skeleton. Chem. Soc. Japan 50: 1836- 1840. ——— WA, K. ENDO & Y. HIRATA. fen Three new т alkaloids with a ó-unsaturated e 4 group from Daphniphiilm pore 393-396. 76. onoid chemistry and the phylogenetic ida А of the Julianiaceae. Syst. Bot. 1: 149-162. 79. Heartwood flavonoids and the infra- generic бег о us (Anacardiaceae). Amer. J. Bot. 66: 502-51 ————. 198 1. The usefulness of flavonoids i in angio- ł . 205- 232 in D. A. Young & D. S. Seigler (editors), Phy- tochemistry and Angiosperm Phylogeny. Praeger, New York. 436 1983. Leaf flavonoids of Eupomatiaceae. Biochem. Syst. Ecol. 11: 209-210. & D. S. SEIGLER (EDITORS). 1981. Phytochem- istry and Angiosperm Phylogeny. Praeger, New York ork. ZAVADA, M. 1983. Pollen Morphology of Ulmaceae. Grana 22: 23-30. & W. L. CnEPET. 1981. о of an- giosperms from the Middle e of Nort America: flowers of the А ae J. Bot. 68: 924— APPENDIX I Voucher specimens used in chemical studies. Trochodendron aralioides Sieb. & Zucc. UGA Botanical Garden, cultivated Tetracentron sinense Oliv. Fang 2725, 6705 (NY) Cercidiphyllum japonicum Sieb. & Z Murata et al. 37168, Wood & Do 3929 (GA) Euptelea pleiospermum Hoo & Thomas Brooklyn Botanic ai cultivated E. polyandra Sieb. & Zuc Boufford 22245, V wen 44430 (GA) Platanus "mibi Willd. e cultivated, Shugrue 55 (GA) P. al UGA члены cultivated C ae spicata Sieb. & Zucc. ooklyn Botanic Garden a sinensis Hemsl Brooklyn Botanic d cultivated C. pauciflora Sieb. & Brooklyn Botanic Garden cultivated Distylium lepidotum Naka Murata et al. 320 HOR D. racemosum Sieb. & Zucc. Brooklyn Botanic Garden, cultivated Fortunearia sinensis Rehd. & E. H. Wils. nic Garden, cultivated Fothergilla major Lodd Radford 34675, Stewart 1554, Wilbur 7012 (GA) F. gardenii Murr Duncan 5115 (GA) Hamamelis vernalis Sarg. Chase 9928 (GA) . virginiana L. Faircloth 4235 (GA) H. macrophylla Pursh. Ewan 21059 (GA) Liquidambar styraciflua L. GA Botanical Garden L. formosana Hance UGA Campus, cultivated Loropetalum chinense (R. Br.) Oliv. eyer 16441, Wigginton s.n. 24-XI-51, Coile 2139 (GA) ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 APPENDIX I. Continued. S к sinensis D. Oliver. n Botanic Garden, cultivated Pontus persica C. A. Mey. Brooklyn Botanic oe cultivated podere henryii Hemsl. B n Botanic бае, нав М роты flabellifolia W Brass 16132, Cronquist e 2 Winter 11608 (NY) «b teijmanens rooklyn Botanic ia cultivated D. rt Blume yn Botanic Garden, Tanaka € Shimada 17827 D. Е аиа Benth. Brooklyn Botanic oa Levine 1624 Eucommia ulmoides O Brooklyn Botanic ee cultivated; Gillis 14345 GA Humulus americanus Nutt. Chase 12151 He A) H. japonicus Windler & ene РУЙ (СА) Broussonetia papyrifera L'Her larke Co., cultivated Cudrania tricuspidata (Carr.) Bur. ex Lavallee ason s.n. XI-81, cultivated Dorstenia foetida Schweinf. (= D. obovata Hochst.) U Greenhouses, cultivate Fatoua villosa (Thunb.) N Godfrey 7235 é Thieret 10227 (GA) Ficus aurea Nut — d cm F. ben Proust Я (СА) Е. brevifolia N Scull s.n. 27- 1 y (GA) F. caprifolia D Russel 2054 m Е. carica L Nob x 13 (GA) F. d Mil mbach 97 ^ (GA) Es gemina Ruiz ex Miq. in Mart. Rimachi 2790 (GA) F. del ro Vahl. Smith 966 (GA) F. llewellyni Standl. Stimson e PUE (GA) F. nitidifolia B prins “Bodenheim 15156 (GA) F. perfora a 373s (GA) F. pumila C anord et al. 959 (GA) F. webbiana Miq Са 9178 (СА) 1986] GIANNASI—HAMAMELIDAE PHYTOCHEMISTRY APPENDIX I. Continued. Helicostylis elegans (Macbr.) C. C. Berg. (GA) H. scabra (Macbr.) C. C. Berg. Rimachi 2898 (GA) Morus alba L. Faircloth 5308, Redfearn 3712, Clarke Co., culti- vated (GA) Duncan 5131, Clarke Co., cultivated Cecropia peltata L Duke 12500 (GA) Pourouma palmata P. & E. Rimachi 2724 (GA) P. phaeotricha Mildbr. Rimachi 2725 (GA) а cylindrica (L.) Sw s & Duncan 19488, Hardin 14286 (GA) Laporte — (L)W 0-VII-58, pave p McVaugh 857 (GA) Urtica ak Pursh Thieret 32699 (GA) U. dioica L. Clokey 8322, Swendsen 487 (GA) Leitneria floridana Chapm. Demaree 45267E, McDaniel 903; E. L. Rich- ards 9749 (STAR) т chiliantha Diels & Handel-Mazzetti 40 om Ching Myrica o = Comptonia perigrina) Ahles 75334, aga p» Hunt MA 180 (GA) M. cerifera L Faircloth 3444, Lane 142 (GA) Miller E4315, е 89166 (СА) М. oo dd ná js 14573, 20724 (GA) д Вагїг. Falreloth 732, Godfrey & Harrison s.n. 9-III-57 (GA) M. rubra Wigginton s.n. 24-XI-51, s.n. 30-XII-52 (GA) . serrata Lam. Russel 2093 (GA Casuarina equisetifolia Forst. 9-IX-40, Ward & Crosby s.n. 9-VIII- GA) C. glauca Sieb. ex Spreng. Baum & Wilson 161 (GA) C. cunninghamiana Miq. Duncan 30462 (GA) PHYLOGENY OF THE HAMAMELIDAE: TAXONOMIC INDEX COMPILED BY Davip L. DILCHER AND DANIEL MACKLIN! Ailanthus, 400, 423 Alfaroa, 367, 368, 422 Alnus, 277, 278, 279, 362 367, 368, 406, 407, 422, 423, 427 Altingia, 325-328, NEU 339-345, 351, 356, 367, 368, 378, 380, 399, 4 Altingiaceae, 325, 326, P Amentiferae, 227, 229, 325, 417, 418 Ampelocera, 367, 368, 370, 380, 421 на 348, 369, 370, 371, 408, 418, 419, 422, Antiquocarya, 404, 405, 410 Aphananthe, 367, 368, 380, 401, 421 Aquilapollenites, 279 0 —— 420 Astera 29 рента 409 421, 423, 429, 430, 431, 432 Atriplex, 406, 407 creta 300 Balanocastanon, 238, 250 Balanopaceae, 260, 276, 348, 357, 368, 370, 396, 398, 405, 408, 413, 418, 426, 429 Balanopales, 408 Balanops, 357, 362, 366, 367, 368, 369, 371, 372, 373, 8, 42 Banksites, 401 PO 355, 361, 367, 368, 371, 372, 373, 374, 378, Barbeyaceae, 260, 348, 355, 366, 368, 370, 371, 396, 398, , 407, 418, 425, 428 a 402, 403 Betula, 277, 278, 279, 362, 367, 368, 373, 378, 406, 407, 420, 422, 423, 426, 427 Betulaceae, 228, 238, 249, 251, 252, 261, 276, 277, 279, 348, 362, 366, 368, 369, 370, 371, 373, 396, 398, 406, 408, 411, 412, 417, 418, 419, 422, 423, 03, 4 403, pe 421, 426, 437 Broussonettia, "in 403, 424, 426, 436 e, 37 Buxaceae, 371 Callaeocarpus, 241 Cannabaceae, 348, 355, 368, 369, 370, 396, 398, 402, 404, 407, 412, 418, 424, 425, 428 Cannabidaceae, 260, 261 Cannabinaceae, 433 Cannabis, 355, 362, 367, 368, 374, 378, 402 Caprifoliaceae, 419 Carpinicarpus, 407 Carpinus, 362, 406, 407, 423 Carpinuspollenites, 277 Carpites, s Carpolithus Carya, 278, a 367, 368, 369, 404, 405, 419, 422 Caryanthus, 404, Caryophyllidae, 421, 430, 432 Casholdia, 398, 404, 405 Castanea, 229-233, 238-245, 248, 250, 251, 252, 263, 264, 272, 277, 279, 284, 285, 288, 290, 292, 293, 295, 366, 367, 368, 405, 406, 426 Castaneae, 229, 230 Castaneoideae, 230, 231, 234, 239, 240, 241, 245, 250, 251, 252, 262, 263, 268, 272, 279, 359 Castaninae, 230 Castanopsis, 230, 231, 232, 239-246, 248-252, 262, 268, 272, 284, 285, 288, 293, 294, 295, 367, 368, 06 4 Castanoxylon, 239 Casuarina, 362, 366, 367, 368, 378, 406, 419, 424, gts 260, 277, 279, 348, 362, 368-372, 396, 406, 407, 411, 418, 424, 4 29 Casuarinales, 276, 362, 366, 367, 398, 418, 424, 429 Cathayambar, 328, 329, 343 Cecropia, 424, 426, 437 Cecropiaceae, 260, 348, 357, 394, 396, 398, 403, 408, 410—413, 418, 424, 426, 428 Celtidaceae, Celtidoideae, 355, 366, 369, 370, 371, 401, 402, 409 Celtis, 362, 367, 368, 373, 380, 401, 402, 421 Cercidiphyllaceae, 260, 279, 319, 348, 349, 367, 370, 373, 382, 393, 396, 398, 399, 407, 418, 425, 428 Cercidiphyllales, 349, 382 Cercidiphyllites, 277 Cercidiphyllum, 297-300, 303-309, 312, 313, 315-321, , 367, 368, 378, 380, 382, 383, 385-393, 399, 425. 436 Cerris, 232, 247, 261, 262 Chaetachme, 367, 368, 380, 421 01 Chrysolepsis, 230, 231, 232, 240, 241, 242, 246, 250, , 252, 261, 264, 268, 272, 367, 368 owe 342, 353 us, 407 Соор, T 305, 353, 356, 367, 368, 378, 380, 400, 4 Corylus, 2 362, 367, 368, 406, 407, 408, 423 ' Department of Biology, Indiana University, Bloomington, Indiana 47405. ANN. MISSOURI Bor. GARD. 73: 438-441. 1986. 1986] Pridie put Cryptocary Cudrania, ioi в 424, 426, 436 Cupuliferae, gu 0 Cupuliferites, 239 Cyclobalanopsideae, 230 Cyclobalanopsis, 230, 232, 243, 246, 247, 250, 262 Cyclobalanus, 230 Cyclocarya, 398, 404, 405, 410 Daphniphyllaceae, 260, 348, 353, 368, 370, 371, 396, 398, 8, 411, 413, 418, 420, 425, 427, 428 Daphniphyllales, 227, 353, 366, 367, 398, 418, 424, 428 M ig ай 358, 359, 367, 368, 378, 380, 420, m un EM 78 ebeya- Dewalquea, 248, 252 Didymelaceae, 260, 277, 279, 348, 353, 372, 373, 396, 398, 408, 413, 418, 420, 425, 428 Didymelales, 276, 353, 398, 400, 418, 428 Didymeles, 420, 421, 425 To 369, 408, 419, 421, 429, 430, 431 Diospyro. Disanthoidea, 325, 342, 344, 345, 353 Disant 342, 353, 356, 367, 368, 378, 380, 382, 400,4 T Distincta poni Ee 400, 425 m 357, 362, 367, ps ver 378, 402, 411, 424, 426, 436 Dryophyllum, 233, 239, 242, 245, 252, 278, 279 Embothrites, 401 Engelhardtia, 357, 367, 368, 404, 405, 410, 422 Engelhardtieae, Erythrobalanus, 230, 232, 247, 261, 262, 264 Eucastanon, 238, Eucommia, 353, 360, 366, 367, 368, 372, 378, 380, 398, 400, 401, 420, 425, 436 Eucommiaceae, 260, 279, А 372, 373, 396, 398, 400, 407, 418, 420, 425, 428 Eucom miales, 353, 398, 418, 424, 4 upomatia 0 Euptelea, 297-300, 303-309, 312, 313, 315, 316, 318, 319, 320, 321, 349, 351, 367, 368, 378, 380, 425, Eupteleaceae, 348, 349, 370, 373, 396, 398, 399, 418, 425, 428 Eupteleales, 349 Eupteliaceae, 260, 319 Euquercus, 246, 247 ат 342, 343, 344, 351, 353, 356, 367, 368, 378, 380, 399 Exbucklandioideae, 325, 342, 344, 345, 351 Fabaceae, 429 Fagaceae, 228-232, 234—241, 245, 246, 248-253, 261, 262, 263, 276-280, 284, 285, 295, 296, 348, 359, 366, 368-373, 394, 396, 398, 405, 406, 408-413, 418, 420, 423, 426, 429, 433 DILCHER & MACKLIN — TAXONOMIC INDEX 348, 353, 367, 368, 370, 439 Fagales, 276, 357, 366, 367, 408, 417, 418, 423, 424, 429 E 230 Fagineae, 230 Еаройдсае, 229, 230, 233, 250, 251, 262, 263, 268, 362 Fagophyllum, 278, 27 Fagopsis, 229, 236, 251, 252, 278, 280, 398, 405, 406, , 410 Fagoxylon, 2 278 Fagus, 229—234, 237, 238, 239, 245, 250, 251, 252, 261, 263, 277, 278, 280, 284, 362, 366, 367, 368, 374, 405, 406, 417, Fatoua, 426, 436 Ficus, 402, 403, rr Aon 424, 426, 427, 436 Fortunearia, 400, 4 Fothergilla, 353, WO vd 368, 378, 380, 400, 425, 436 Fraxinifolae, 248 Fraxinus, 400 Galumpita, 421 Ginko, ie Girardinia, 403 DID 367, 368, 380, 401, 402, 421 Guttiferae Gy mnostoma, 362, 366, 367, 368, 378 Haloragacidites, 277 mamelidaceae, 260, 277, 279, 297, 303, 305, 306, 308, 315, 318, 319, 325, 326, 344, 345, 348, 351, 356, 367, 368, 369, 370, 373, 393, 395, 396, 398, 399, 400, 407, 411, 412, 418, 420, 423, 424, 425, 42 8 Hamamelidae, 225, 226, 227, 276, 277, 297, 303, 309, 315, 321, 322, 348, 366, 369-373, 375, 382, 383, 394, 395, 396, 398, 399, 408, 411, 412, 413, 417- 421, 423, 424, 425, 427-433 Hamamelidales, 276, 297, 303, 318, 321, 322, 349, 366, 367, 371, 382, 398, 417, 418, 423, 424, 428 Hamamelidiae, 278 , 278 Hamamelidoideae, 326, 342, 344, 353, 399 Hamamelidopsidae, 23 е 303, 353, 356, 367, 368, 378, 380, 399, 400, 424, 425, 427, 436 Helicostylis, yu 426, 436 Helicostylus, Hemiptelea, e 368, 421 Hemist dips Hicoreae, Holoptelea, pM 368, 421 Humularia, 402 mulus, 402, 424, 425, 436 UE 238 diospermaceae, 430 Poe um, 313 Jenkinsella, 319 Joffrea, 319, 382, 383, 385, 387, 389, 391, 392, 393, 399 Juglandaceae, 260, 279, 348, 357, 368-371, 373, 394, 440 395, 396, л 404, 408-411, 413, 417-420, 422, 423, 426, 4 ein 291 357, 366, 367, 398, 408, 418, 423, 428 Р» ed 405 mum 422 uglans, 367, 368, ~ 374, 398, 404, 405, 410, 422 НЕ 418, 4 Klikovispermum, 400 Laportea, 403, wer 426, 437 430 419 Leitneria, 227, 353, 360, 367, 368, 371, 372, 374, 378, 404, 424, 426, 437 Leitneriaceae, 261, 348, 353, 368, 371, 373, 396, 398, 404, 407, 418, 424, 426, 427, 428 Leitneriales, 227, 353, 366, 367, 398, 418, 423, 424, , 428 Lepidobalanus, 232, 247, 261, 262 Liliaceae, 419 Liliopsida, 419, 421, 429, 431, 432 Lilliidae, 421, 431, 432 Limaipollenites, 375 Limlia, 244, 251 Liquidambar, 277, 325-329, 331, 333-337, 339-345, 367, 368, 378, 380, 399, 400, 418, 420, 425, 436 Liquidambaroideae, 325, 327, 328, 341, 342, 344, 345, 348, = __ on 368, 373 Lithocarpe DM A 231, 232, 239, 240, 242-246, 248- 252, 261—264, "s 272, 284—289, 291—295, 367, 368, 405, 406, 4 Loropetalum, 353, EUR 367, 368, 378, 380, 425, 436 Lozanella, 367, 368, 380, 421 Macrobalanus, 230, 232, 247, 262 Magnolia, 433 Magnoliaceae, 297, 299 Magnoliales, 306, 318, 321, 322, 382, 418 Magnolidae, 321 Magnoliid, 433 Magnoliidae, 227, 297, 303, 309, 315, 321, 322, 394, 418, 419, 420, 421, 422, 429—433 Mesobalanus, 232, 247, 261 Metatrophis, 411 Microcarpolithes, 403, 404 Mirandaceltis, 367, 368, 380, 421 Momipites, 277 MOM 260, 261, 348, 357, 366, 369, 373, 394, 396, 398, 402, 403, 404, 408, 410, 411, 412, 413, 417, 418, 419, а 424, 426, 427, 428, 429, 430, 433 о rus, 402, ^ 419, 426, 437 ue 277, 279, 357, 362, 367, 368, 369, 379, 404, 76, 279, 348, 357, 368, 370, 371, 373, 396, 398, 404, 407, 411, 412, 418, 420, 423, 426, 428 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Myricales, 276, 357, 366, 367, 398, 418, 423, 424, 428 Myristicaceae, 430 Myrothamnaceae, 260, 318, 348, 349, 366, 367, 368, 370, 372, 396, 398, 400, 407, 418, 425, 428 Myrothamnus, 351, 354, 366, 367, 368, 371, 379, 380, 25, 436 Mytilaria, 342, 344, 351, 353 Mytilarioideae, 325, 342 Nandinaceae, 419 Nordenskioldia, 399 Nothofagaceae, 228, 231, 250, 251, 252, 279 Nothofagidites, 235 Nothofagoxylon, 237, 278 Nothofagus, 228-238, 249-252, 261, 263, 264, 268, 272, 276-281, 348, 362, 366-371, 405, 406 Nymphaeales, 419, 432 K^ cda 382, 383, 385, 387, 389, 391, 392, 393, Ochnaceae, 419 DORIA. 367, 368, 422 Ostrya, 362, 406, 407, 422, 423, 427 Ostryopsis, 362, 406 Ovicarpum, 402, 403 Paleocarpinus, 406, 407 eo 4 onia, 367, 368, 369, 380, 421 Parrotia, 305, 353, 400, 425, 436 Phyllites, 237, 245 ое 367, 368, 421 Picra 42 Pilea, 405 лее 411 Pubs о Me 371, 379 о 115, 42 ега, 355, 2 368, 401, 421 ‚аы 260, 278, 279, 318, 345, 348, 349, 367, 368, 370, 373, 396, 398, 399, 407, 413, 418, 425, 4 Platanus, 227, 345, 349, 351, 352, 367, 368, 379, 380, 425, Platycarya, 351, ч 368, 369, 404, 405,422 оасеае, 371 Polyptera, 398, 404, 405 Populus, 362, 367, 368, 369, 370, 379 Pourouma, 424, 426, 437 388 Procris [ir diet 232, 247, 248, 262 nus, 400 Pseudofagineae, 229, 231, 251 Pseudofagus, 229, 231, 251, 406 de oec 357, 367, 368, 369, 374, 404, 405, 409, 410 © 367, 368, 380, 401, 402, 421 1986] Querceae, 230 Quercineae, 229, 230 Quercoideae, 229, 230, 231, 234, 245, 250, 251, 252, 262, 263, 268, 272, 362 Quercophyllum, 248 Quercus, 229-232, 239, 240, 242-252, 261-264, 268, 272, 279, 280, 284, 285, 288, 290, 291, 293, 295, 367, 368, 374, 398, 405, 406, 426 Ranunculaceae, 309 Ranunculidae, 419 Rhamnaceae, 419 Rhodoleia, 326, 342, 343, 344, 351, 356, 367, 368, 0 Rhodoleioideae, 325, 342, 344, 345 Rhoiptelea, 357, 367, 368, 369, 374, 426, 437 Rhoipteleaceae, 348, 357, 366, 368—371, e 398, 404, 8 Rhoipteliaceae, 260 Ros ida ae, 419, 421, 423, eh 429—432 osidae-Hamamelidae, 2 21 Ip 369, 371, ri ш 2 Rutiflorae, 371, 422 Rutinae, 42 Salicaceae, 227, 348, 369, 370 Salix, 367, 368, 369, 370 Sapindales, 227, 408, 418 ou T 418 Schizocolp Semiliguidambar 326, 327, 336, 337, 341-345 Simaroubacea: Sinowilsonia, 278, 393, 424, 425, 436 410 pectae, 24 Sycopsis, "305, 353, 425, 436 Symingto 4 SIminstonioidese, 325 DILCHER & MACKLIN — TAXONOMIC INDEX 441 Tetracentraceae, 260, 349, 370, 396, 398, 399, 418, Tetracentron, 278, 297-301, 303-309, 312-315, 318- 321, , 349, 367, 368, 370, 372, 380, 389, 399, 425, e Thymeleceae, 418, 4 Trema, 367, 368, rg 380, 401, 421 пие 277 gonobalanoideae, 229, 231, 251 ое 229-232, 239, 242, 244, 245, 246, 250, 251, 252, 262, 268, 272, 284, 285, 367, 368, 398, 405, 406, 418 Trimeniaceae, 300 Trochodendraceae, 260, 297, 299, 319, 348, 349, 367, 370, 396, 398, 399, 407, 418, 425, 428 Trochodendrales, 297-300, 303, 305, 306, 318-322, 349, 366, 367, 370, 372, 373, 382, 398, 399, 418, Trochodendreae, 297, 319 Trochodendrocarpus, 382, 383, 387, 388, 389, 391, 92,3 Trochodendron, 297-310, 313-316, 318-321, 349, 368, 399, 425, 436 Tubela, 407 Tylophora, 420 Ulmaceae, 260, 261, 276, 279, 348, 355, 366, 368, 369, 371, 373, 394, 396, 398, 400, 401, 407, 408, 409, 411, 413, 417, 418, 419, 421, 425, 428 Ulmoideae, 355, 369, 370, 401, 402, 409 S, 277, 278, 362, 367, "un 401, 421, 430 unica 403, 421, 424, 426, 4 Urticaceae, 260, 261, 348, ©. 366, 373, 394, 396, 398, 403, 404, 407, 408, 410—413, 417, 418, 424, 426, 428, 43 Urticales, 227, 355, 366, 367, 370, 371, 398, 408, 418, 424, , 428, 433 Urticicarpum, 403 Urticoidea. 403, 404 Virentes, 263 Winteraceae, 297, 309 Zelkova, 367, 368, 401, 421 PHYLOGENETIC STUDIES USING PROTEIN SEQUENCES WITHIN THE ORDER MYRTALES! Р. С. MARTIN? AND J. М. Dowr? ABSTRACT N-terminal sequences (40 amino acids) of ribulose bisphosphate carboxylase small subunit are phylogenetic trees embracing all groups were up but both showed relationships to LUNES Within Euphorbiales and Malvales, grouping of ieri conformed well with expectation A symposium on “The Order Myrtales” was held at the XIII International Botanical Congress in 1981 and, when introducing the published proceedings, Raven (1984) summed up the con- sensus opinion that this order is a clearly defined group. This suggests that the families that com- prise the Myrtales may be favorable subjects for a study in which a comparison is to be made between phylogenies derived from morphologi- cal, anatomical, and micromolecular character- istics and from an independent approach such as the computerized analyses of macromolecular sequence data. The most comprehensive attempt to build an angiosperm phylogeny from sequence data is that of Martin et al. (1985), which combined into a single analysis sequences of four proteins and one RNA for up to 11 families. Phylogenetic trees derived from a single macromolecule did not agree with each other and consistency was main- tained only when a fourth macromolecule was added to the combined sequences of three pro- teins. The explanation advanced for this result was that, for four of the macromolecules (the exception was the small subunit of ribulose bis- phosphate carboxylase, henceforth RBC-SSU), a family was often (17 out of 33 cases) represented by only a single sequence. Martin et al. (1985) suggested that errors were less likely if a family node was derived from more than one member ofa family. The present investigation was carried out to test this idea that replication is effective, a measure of the correctness ofa phylogeny being available in Myrtalean taxonomy. However, in our laboratory it is practicable to sequence only C-SSU and there are no sequences of other macromolecules available for the Myrtales. Therefore the test is limited to what is achievable with only one sequence. It can be claimed that, with more sorts of sequence data, greater pre- cision would follow. MATERIALS AND METHODS The 27 species that have been investigated are listed by family in Table 1. Methods are set out in detail in Martin and Jennings (1983) and the following is only a very brief account. One hundred grams of fresh leaves were treated using the “pungent-leaf method." After maceration, species of Onagraceae were usually very muci- laginous so that more than the usual amount of extracting buffer was necessary to prevent solidi- fication before gel-chromatography. After puri- fication of the “‘fraction 1" protein, S-carboxy- methylation, and separation of the subunits, five milligrams of small subunit were sequenced on the Beckman Automatic Sequencer 890C. No attempt was made to progress beyond the first 40 amino acids. Amino acids were identified us- ! We acknowledge with oir eae from the Missouri Botanical Garden and a grant from the Australian Sc to Pet n fi search Grants eme; grant recently BSR82- 148, have ro support Botanic Gardens. Peter Raven has given rom the United States National Science Foundati ? Department of Botany, University of pedet "Box 498, G.P.O., Adelaide. a y o 5001. ANN. MISSOURI Bor. GARD. 73: 442-448. 1986. 1986] BLE Species studied. ABG and MBG indicate that the leaves were шее from plants growing in t. Lou the hat is, Milsouri (MBG), respectively. AD indicates that leaves were collected elsewhere and a voucher specimen has en lodged in herbarium AD. For two species col- lected in the field there were only enough leaves for processing, not for voucher specimens Myrtaceae Eucalyptus microcarpa (Maiden) Maiden AD Acmena smithii Poit. ABG Melastomataceae Melasto . à Tibouchina semidecandra Cogn. ABG Combretaceae Quisqualis indica L MBG ombretum decandrum Roxb. ABG Lythraceae Lythrum sa ABG Woodfordia aie: (L.) Kurz ABG ao Daphne odora Thunb. ABG Pimelea prid ‘Hook. Onagraceae Ludwigia peploides (Kunth) Raven AD Lopezia semeiandra (Plitmann) Raven & Breedlove MBG Circaea cordata Royle MBG Fuchsia hybrida Voss AD Epilobium ciliatum Raf. AD Epilobium canum (Greene) Raven ABG Hauya elegans MBG Clarkia unguiculata Lindl. AD Clarkia rubicunda (Lindl.) Lewis & Lewis AD Gaura lindheimeri Engelm. & Gray ABG Oenothera stricta Ledeb AD Buxaceae sensu lato Simmondsia chinensis (Link) Schneid. ABG Buxus sempervirens L. ABG Euphorbiaceae к кои Muell. Arg. ABG nus commu ABG Tiliacea pi occidentalis L ABG Sparmannia africana L. f. ABG ing high y liquid cl hy and thin- layer chromatogeanhy: Sequence » data were ana- lyzed by computer methods that have been de- scribed by Martin et al. (1985). Briefly the amino acid sequences were converted to inferred nu- cleotide sequences from which differences be- tween species were derived. The program esti- MARTIN & DOWD—MYRTALES PHYLOGENY 443 arn C. unguiculata C. rubicunda Oenothera Gaura —" y auya Epilobium E. glabellum Fuchsia Circaeal Lopezia Ludwigia SCALE ÉAE — proONAGRACEAE | | nucleotide difference ROOT FIGURE 1. Phylogenetic tree for 11 species of Ona- graceae. For specific names see Table 1. The two equal- _ that the branches rrying Gaura and Hauya were ee Note a in these trees angles are meaningles mates the lengths of all possible Steiner trees and the minimal phylogenetic tree is deemed the most probable. Consensus trees are derived from the shortest trees and this procedure is especially im- portant when more than one tree of minimal length occurs. Thus the method is tially one of maximum parsimony. It should be noted that the number of possible trees increases exponen- tially with the number of taxa and, with most computers, usually only 11 taxa can be analyzed simultaneously. For taxa that group together constantly no matter what others are present, sequences of their common node can be derived and used in other analyses, so making the tree- building process more practical. RESULTS General. N-terminal sequences (40 amino acids) are listed in Table 2. More than 100 other 444 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE 2. N-terminal sequences of RBC-SSU. For other specific names and also names of families see Table 1. A, alanine; D, aspartic acid; E, glutamic acid; F, phenylalanine; G, glycine; I, isoleucine; K, lysine; L, leucine; M, methionine; N, e P, proline; Q, glutamine; R, arginine; S, serine; T, threonine; V, valine; W, tryptophan; Y, tyrosi — ә — © — — — N — w — A — CA е — - — oo Eucalyptus Acmena sia Epilobium ciliatum Epilobium canum Hauya Clarkia unguiculata Clarkia rubicunda Gaura Oenothera Simmondsia B Acalypha Ricinus Grewia Sparmannia S $ & СЕ R6ROOOOZZZZZZZZZZmROOOORFOOOO|w» <<<<<<<< << <<< SSS SKS SJ] Z£££££2£222222£2£2£222££££z£££2£z£z££z£|. AAA UU UG SAA Uo << 7 ше ыы еее < үе AZAMDAANANAAANDAAANANAAAZOAANAOZZOAO!o@ m= CAPOVNOCOOCOOACOCOCAZOEARCEAARAARAA| © RRRRRRRRERRERRRRRURRRRRRRERRERK Кекеч ккк ккк т КК Ж єз "Tj "rj "rj" Жез Жыз 7j "n ы, "n Un n onm om onm om omm omn om фо; om om Mes Mes Mes m trj m m m m m m m m m rm m m m m m m m m m m m m m m m m AA AAA AAA AAA AAA Е Е НЕ Е ЕЕ Е ЕЕ eee al най чый о о ыр чый e чы | el ll чеш бый баш е кй с се Б еъ с со А ка кы, AMMA о о ор о о о UA YA UM MUNI UM UN UU UM UM UN UM UM UU UU LALA RR IRR RR RR RR RR RR RR RR MR e чый чый ай чый ый лый ll тай ча ай ы ll >> эш Л ll a ее = species of angiosperms have now been analyzed, from a wide range of families, and so it is possible to say that the range of variation exhibited in Table 2 is quite normal in all respects except one. The occurrence of phenylalanine and asparagine at positions one and two in most species of Ona- graceae is unknown elsewher Onagraceae. In Raven’ as) classification of Onagraceae, Zauschneria is combined with Epilobium and Godetia is combined with Clark- ia. The results support this taxonomy; prelimi- nary analyses on these sequences showed that Epilobium canum (formerly Zauschneria cali- fornica) grouped closely and consistently with Epilobium ciliatum. Similarly Clarkia rubicunda (formerly Godetia amoena) grouped consistently with Clarkia unguiculata. Considering these re- sults, and the taxonomic position, we derived the sequences at generic nodes for Epilobium and Clarkia and used these in subsequent analyses. This reduced the number of taxa from 11 to nine and for these there were 57 trees of minimal length. This situation arose because there were two adjacent internodes of very short and equal length so that many different combinations were possible without change of overall tree length. For similar reasons there were two consensus trees. We have rationalized the situation and in Figure | have presented a single tree with a circle of uncertainty at the relevant node and showing the difference between the two consensus trees. According to Raven (1979) there are seven tribes; Onagreae (represented here by Clarkia, Oeno- thera, and Gaura), Epilobieae, and Hauyeae arise from this circle of uncertainty. The remaining four taxa in our analysis represent four mono- generic tribes and the analysis shows them as separate. The point at which other families join into this tree (see below) is taken to be the root of the Onagracean tree and this makes the first divergence between Ludwigia and the rest. This is in part because Ludwigia is the only species 1986] TABLE 2. Continued. MARTIN & DOWD—MYRTALES PHYLOGENY 445 — © N e N — N N N uy N A N л N © N з N oo N o чө © ы — we N ө > „ө > ы CA uy е > ч w со w o > > — Я meammmmmmuogmuommugmmeem-mmomumgogmtm № AAA E ыы а ы ла ыы ы ы ы ые ше эш ал pp epp ® oe Me Mos Mis Mc Aol ioe Mee Mos BA: Mos Bios Mos Mos Mee cle ce ee ES CO о<чочУкччр чя коооручичикииити DDR FAD э б с с обок ккк опсо О О pp ы ыйа айы е p ай ай ы] а pi айы зы сай pu pe ps py pt ps pt ps Г >>> ASS "FAP A гр р RAPP PRARAKRKRARAKRKRRARAKRAKRAKRAAAKRA mmmommtmtmmmmmummmmmmmummmmootmtm я о Ugmmommmgoggmmmmmogomoomoo»»uog DIT Seer Seer EErEE ECE eee re err ree eee eS ot чый ae te ee ee ee ee ee eae ee ie eee DOC COOAARADAAANYDYARAOOPYRZOR ARZZZYZZZZALZZZZARALZZARAZLZYLAAR RAAQAZAAAZAAAAAAZARKXKRAAKRANAAA SESE SES SSS LEE LE LE LLE LEE LE LE LE LEE LEE E ааа cmH << "Y DDD DDD DDD "а-а a Наа - а, that we have sampled from the family that has a conventional N-terminus, the other ten species having the highly distinctive Phe-Asn, which alone accounts for three nucleotide differences. Four Myrtalean families and Thymelae- aceae. In this study four families which, as dis- cussed above, are agreed to be members of the order Myrtales, were chosen and, from = two species, each from a different genus, we lected. Dahlgren and Thorne (1984) АИРИ that, though closely related, Thymelaeaceae was not a Myrtalean family, so it was chosen as an “out-group” (to define the base of the tree) and two species were studied. When these ten species were analyzed simultaneously, there were two consensus trees out of 17 minimal trees (Fig. 2). In one tree the species pair perfectly into the five families, whereas in the other two species of Ly- thraceae lie side by side. Sequences at the five familial nodes shown in Figure 2 were derived and used in an analysis with Ludwigia, Lopezia, and Circaea and other external angiosperm nodes (see below) to give the familial node for Ona- graceae shown in Figure 1. The grouping of the six familial nodes is shown, with other infor- mation, in Figure 3; five ofthe families arise from a single junction but Onagraceae diverge sepa- rately though close to the five. The surprising result in this tree is that Thymelaeaceae appear to be a very definite member of the same group as most of the Myrtalean families and not, as had been expected, an “out-group.” Euphorbiales. Іп one ofthe currently respect- ed angiosperm taxonomies, Cronquist (1981) did indeed place Thymelaeaceae in Myrtales, al- though he did so because the family fits in there more easily than anywhere else, not because he considered it a clear-cut Myrtalean family (Cron- quist, 1984). Dahlgren and Thorne (1984) stated that a number of taxonomists believe that Thy- melaeaceae approach most closely Euphorbiales and they themselves considered that Thymelae- 446 - Ti hina “жы... mane Melastoma i M \ Woodfordia | P d x y РА Е. А / „ый. EQUALLY c dedos IN / ana. Woodfordia \ Acmena = | nucleotide difference ve of the familial nodes used in deriving Figure 3 were obtained from this tree. aceae should be placed near Euphorbiaceae, and that these two families are related to Malvales. It therefore seemed appropriate to investigate Euphorbiales and Malvales To represent Euphorbiales, two species were chosen from Euphorbiaceae and two from Bux- aceae, although it was understood that one of the latter, Simmondsia, is often placed in a family of its own. The four species were first analyzed in conjunction with the Myrtalean familial nodes о vs all four diverged from a com- node, confirming the reality of the order нти eer the two Euphorbiaceae ap- propriately grouped. However, the two putative Buxaceae did not have a common familial node suggesting that the separation of Buxaceae and Simmondsiaceae might indeed be valid. In the Malvales, three species of Malvaceae had already been studied (Martin & Dowd, 1984) so, in order to broaden the representation, two species of Tiliaceae were sequenced. These wer analyzed with the Malvaceae species and from the minimal tree (Fig. 4) both familial nodes were derived. The analysis of the familial nodes for the five oO for two other “external” families, Fabaceae and Brassicaceae (Martin et al., 1983) (Fig. 3). Thy- melaeaceae do not move from their position in the Myrtales and Onagraceae still diverge slightly separate from most Myrtales. The major surprise ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 EUPHORBIALES SCALE — 1 nucleotide difference Buxus Simmondsia proTHYMELAEACE AE proMELASTOMATACEAE proMYRTACEAE proCOMBRETACEAE proLYTHRACEAE proONAGRACEAE proTILIACEAE m-— MALVALES FIGURE 3. Unrooted minimal phylogenetic tree for ten familial nodes and four species from Euphorbiales. The familial nodes were derived from the analyses rep- resented in Figures 1, 2, and 4. proFABACEAE proBRASSICACEAE is that Euphorbiales diverge from the Myrtalean tree. We have carried out similar analyses sub- stituting other families for the two “external” ones without changing this result. Thus, except for the junction with the Euphorbiales and Thy- melaeaceae, four of the Myrtalean families (Ly- thraceae, Combretaceae, Melastomataceae, and Myrtaceae) form a natural group as would be expected for an order like Myrtales. DISCUSSION As an exercise for testing the ability of objec- tively-generated phylogenies derived from se- quence data to reflect the conclusions of taxon- omy, this must be regarded as at least a partial success. Most notable is the correct grouping of species into families as displayed in Figures 2, 3, and 4. In the more extensive study of Onagra- ceae, two successes can be recorded; the correct grouping of two species each into the genera Clarkia and Epilobium; and the identification of the first divergence, which confirms several other lines of strong evidence that Ludwigia is the “‘sis- ter group” ofall other Onagraceae (Raven, 1979). The separation of the taxa into the other six tribes (Raven, 1979) is more uncertain and three of them (Onagreae, Hauyeae, and Epilobieae) de- part from the tree so closely that they are effec- En оши according to Raven (1979) there is a possible relationship between Hauyeae and Onagreae. The other three tribes (Lopezieae, Circaeeae, and Fuchsieae), which are 1986] Grewia Sphaeralcea Althaea 7 proMALVACEAE proTILIACEAE Sparmannia SCALE | nucleotide difference FIGURE 4. Unrooted minimal phylogenetic tree for species of Malvaceae and Tiliaceae. The familial nodes for these two families, used in Figure 3, were derived from this tree. all monogeneric, are not confused but Fuchsia is thought to be the most generalized and might have been expected to diverge earliest. These are rather stringent tests of our methods because, when relationships are close and differences are small, chance is more likely to operate to give a misleading result; for this reason, Martin et al. (1983) made a strong point of deriving familial nodes from which to generate inter-familial phy- logenies, the primary purpose of this research. When considering the comparison of our phy- logenetic trees with those derived for the Myr- tales from taxonomy, the best source of com- parison appears to be four trees derived from Johnson and Briggs (1984). These authors do not appear to have a clear preference for one or more of these so all are shown in Figure 5; only the families that overlap with this study are repre- sented, viz. Onagraceae, Lythraceae, Е = bre- taceae, Melastomataceae, and Myrtaceae. There appears to be enough variation | among these four trees to accommodate most of the dif- ferences in our tree. The main discrepancy is that in some of their trees Myrtaceae diverges near the base; the main similarity is that Onagraceae and Lythraceae usually diverge early. However, it must be said, without in any way implying criticism, that the main objective of the whole exercise has been partly frustrated by the lack of a clearly defined phylogeny to act as a model against which we can test our methods. The inclusion of Thymelaeaceae as an “‘out- group" for the Myrtales, as suggested by the pa- per of Dahlgren and Thorne (1984), did not serve that purpose because in our study the family grouped with Myrtales. This was indeed the opinion of Cronquist (1984), so we were able to perform a test, in the best scientific tradition, to distinguish between the hypothesis of Cronquist and the alternative one of Dahlgren and Thorne MARTIN & DOWD—MYRTALES PHYLOGENY CMB MRT MLS CMB LYT ОМА MRT MLS CMBLYTONA = МАТ MLS CMB LYT ОМА Y Y v FiGURE 5. A comparison of phylogenies 1 Johnson and Briggs (1984) (a, b, d, and e) with equiv- a and b are Tom from their figures 6 and 7. MRT, Myrtace Melastomataceae; CMB, Combretaceae; LYT, Lythra- ceae; ONA, Onagraceae. (1984), that of a relationship of Thymelaeaceae to Euphorbiales and Malvales. The results of our test supported Cronquist’s hypothesis in that Thymelaeaceae did group with Myrtales, but this result should be considered in the light of the fact that there are sequence data for less than 15% of dicotyledonous families; given tests against a much wider range of variation, the fam- ily may well group elsewhere. We believe that the fact that Thymelaeaceae and Euphorbiales do not group with Malvales is stronger evidence against the hypothesis of Dahlgren and Thorne (1984) In another respect the test has been partly suc- cessful because the families of Myrtales do form a natural group separate from others, always ex- cepting Euphorbiales and, possibly, Thymelae- aceae. It is true that Onagraceae seem to diverge separately from the rest but additional data could conceivably change this. It must be acknowl- edged that Hutchinson (1973) placed Onagraceae in an order of their own but this seems to be quite contrary to the great majority of evidence produced in the symposium at the XIII Inter- national Botanical Congress (Raven, 1984). 448 LITERATURE CITED CRONQUIST, A. 1981. An Integrated System of Clas- sification of Flowering Plants. Columbia Univ. Press, New Yor Ў 1984 [1985]. A commentary on the definition of the order Myrtales. Ann. Missouri Bot. Gard. 7 2. 1: DAHLGREN, К. M. T. & К. Е. THORNE. 1984 [1985]. The order Myrtales: circumscription, variation, and relationships. Ann. Missouri Bot. Gard. 71: 633- 699 HUTCHINSON, J. 1973. The Families of Flowering Plants, 3rd edition. Clarendon Press, London. JOHNSON, L B.G. BRIGGS. 1984 [1985]. esi tales and Myrtaceae ysis. Ann Missouri Bot. Gard. 71: 700-756. MARTIN, P. G. & J. M. ipn ie The study of plant phylogeny using a cid sequences of ribulose-1,5- bote Cu Mee III. Ad- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 dition of Malvaceae and Ranunculaceae to the phylogenetic tree. Austral. J. Bot. 32: 283-290. — —— & A. С. JE ENNINGS. 1983. The study of plant PHUYIUBLLY 1,5-bisphosphate carboxylase. I. Biochemical methods and the patterns of variability. Austral. J. Bot. 31: 395-409. , О. BOULTER & D. PENNY. 1985. Angiosperm phylogeny studied using sequences of five mac- romolecules. = axon —400. J S. J. L. Stone. 1983. The study of plant paruit using amino acid se- quences of ribulose-1,5-bisphosphate carboxylase. II. The analysis of small subunit data to form phy- logenetic trees. Austral. J. Bot. 31: 411-419. Raven, P. H. 1979. A survey of reproductive biology in Onagraceae. New Zealand J. Bot. 17: 575-593 1984 [1985]. The order Myrtales: a sympo- sium. Ann. Missouri Bot. Gard. 71: 631-632. PALEOGENE PHYTOGEOGRAPHY AND CLIMATOLOGY OF SOUTH AMERICA! EDGARDO J. ROMERO? ABSTRACT Paleog M ichti and Mi fl d. They are located on continental margins, where al sedimentary basins developed. The affinities of the fossil genera to living ones allow the recognition of three paleofloras: Neotropical, Mixed, and Antarctic. These co uld be the forerunners, respectively, of the presently more humid dominions in Neotropical Eon (Caribbean, Amazonic, nd and Guayano of the Antarctic Region. The S American 0), of the drier ones Lig ia and Andino-Patagonico climate, as indicated by the fossil floras, showed a trend Eocen АА: AAI to higher and Upper Eocene. The purpose of the present work is to review the Paleogene taphofloras of South America and to analyze their paleophytogeographical and pa- leoclimatical significance. This problem has interested botanists and pa- leontologists in the past. Berry (1921, 1938, 1940) published several reviews as a direct result of his work in the continent. More recently, Menéndez (1964, 1969, 1971) detailed the taphofloras and the areas occupied by the different types of vege- tation, and Volkheimer (1971) gathered evidence about the paleoclimatology of Argentina. Later, Archangelsky and Romero (1974) analyzed the environmental conditions based on the pollen record of the southern South American Paleo- gene, and Aragon and Romero (1984), Romero (1978), Romero and Arguijo (1981b, 1982), and Romero and Dibbern (1984) analyzed some taphofloras of the same region. Therefore, much new geological, botanical, and paleontological evidence has been gathered, warranting a new review. The present paper is arranged in two parts. The first one reviews every published outcrop with impressions known to the author from the Paleogene of South America, complemented with information about fruit casts, petrified wood, and palynology. Data about some Maastrichtian and iocene floras have been added to show more lications. The second part discusses the phyto- geographical areas that could have existed during that time, and the climate under which they have developed. gh the Middle There is not an agreement on the terminology to be used in biogeographical discussions, es- pecially among paleobotanists. I shall use a few of them, in an attempt to reach more accuracy and clarity, but without discussing their ante- cedents or convenience, which surpasses the lim- its of the present paper. They are: Flora. The list of the plants living together in a given area at a given moment. They may be fossil or living. tation. Plant cover in a given area at a given moment, fossil or living. Their association, ecology, and life forms, rather than their taxo- nomic identity, are considered. Taphoflora. The list of the fossil plants pre- served in a given outcrop. They represent the original flora, as affected by dispersal, burial in that place, diagenesis, and differential alteration. They are the smallest real evidence to be used in paleophytogeography. Paleoflora. A fossil flora that may be char- acterized by its systematic list, and so differen- tiated from other fossil floras. Each paleoflora may comprise several taphofloras. From the characteristic systematic list a characteristic plant association may be hypothesized, and so vege- tational features suggested. A REVIEW OF THE PALEOGENE TAPHOFLORAS OF SOUTH AMERICA Figure 1 shows the positions of the known Pa- margins of several marine basins that existed ! I thank S. Archangelsky and A. Bertels for reading parts of the manuscript and making useful suggestions. A. Bertels and the Asoc ? Departamento de Ciencias Biológicas, Aires, Intendente Guiraldes 2620, Investigaciones Cientificas (CONIC ANN. MISSOURI Bor. GARD. 73: 449—461. 1986. ciación ee Argentina kindly permitted me to use published figures. Facultad de Ciencias Exactas y Naturales, Universidad de Buenos 1428 iri Aires, Repüblica Argentina. Member of Consejo Nacional de ET). ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 450 2 ^ l go» 70° o 60° 1 50° 140 о " Ч 2 o 10* “$ с . i | iq ч | i ae. 1 a? los Ma ГА ^ ГА ‘very Р ў A EMT NW MM I ¡e N N 7 i ; ` РА > ‘> \ SS D 0° E ЄТ : „=“ 0° > = yaw a 9 aa e ` > ~ DN 2 я " = í » uini i Al (у и” mil a f ( 12 V ` ` 134 10° 7.25 21 q ) pu [ uc did 4 0] \ tee J a \ s= ( >i ү ( ! Г) | a EN 7 ) А t P y és mE Я “=” TRE [ / O MAASTRICHTIAN А / / ¿ r - - 4 [3 4 Nat Na „=“ 20 A PALEOCENE Y ` t ~ O iw. EOCENE d dd E. ( Р = © м. EOCENE ! M e: A Upp. EOCENE / И t ГА B OLIGOCENE ! PS ү Г ^ > 30° ` ^ у, Y " ‘ ( + 1 N 2 ГА \ a 14 Li o (А 18 p 40 Y үе “г 190 “ee 10204 2A, yo i 1986] during the Upper Cretaceous and Lower Tertiary (Fig. 2). These basins are related to similar ones of the Upper Cretaceous and the Oligocene- Mio- continent. Most of the fossil floras have been buried in deltas, coastal swamps, or lagoons. MAASTRICHTIAN Included here are the outcrops of the Gauduas originally described as respond to the Maastrichtian according to van der Hammen (1954). These outcrops are located in Cipacon (Fig. 1.6) and Zipaquina (Fig. 1.7) (Department of Cundinamarca), El Infierno (Fig. 1.5) (Department of Boyaca), and Falls of Te- quendama (Fig. 1.8) near Bogota. Fruit casts and Selling, 1945); among the latter, Geonomites zi- paquinensis, Ficus andrewsi, Coussapoa camar- goi, C. ampla, Nectandra lanceolata, and Theo- broma fossilium (Berry, 1929a; Huertas, 1960). Humiria was also found in Belén, Perú (Lower Eocene) (Berry, 1927, see below). The palyno- logical record shows that northern South Amér- ica was a province rich in American elements, where palms dominated (van der Hammen, 1954) In Brazil, Pa/mocarpon luisi is quoted in the Paraiba Group (Fig. 1.12) in Paraiba state (Mau- ry, 1936). In southernmost South America, the tapho- flora of Cerro Guido (Fig. 1.23) has approxi- mately 30 species of leaf imprints, most of which have been attributed to Lower Cretaceous species of North America. They have been studied by Kurtz (1899), with a few additions by Hunicken (1971), and Menéndez (1966, 1972a, 1972b). though of Coniacian age (earlier than the Maas- ROMERO—PALEOGENE PHYTOGEOGRAPHY AND CLIMATOLOGY 451 trichtian) (Romero, 1978), are also worth men- tioning because they present some species that migrated to Australasia during the Paleogene. Among them, some have or have had a cos- a distribution (Hymenophyllum, Po- New Zealand (Blechnum, Laurelia) during the Tertiary Fossil leaves of Nothofagus have not yet been found in the South American Maastrichtian, but a few pollen grains have been found, already differentiated into three types (fusca, brassi, and menziesii) in sites some 1,500 km apart (Arch- angelsky & Romero, 1974; Romero, 1973). The site of the plants that produced these pollen grains has not yet been discovered. PALEOCENE Paleocene taphofloras are not known in north- ern South America, but the palynological evi- dence (Muller, 1970, 1981) already indicates the presence of Bombacaceae (Bombax type). In northeast Brazil (Fig. 1.13), fruits of Nypa per- nanmbucensis and Celtis santosi and imprints of Psidium have been found (Beurlen & Sommer, 1954; Dolianiti, 1955). In southern South America the oldest Tertiary deposit is that of Funes (Fig. 1.20) of Danian age (Berry, 1937a; Romero, 1978), "ipo together with that of Sur del Río Deseado (Fig. 1.21) (Ar- guijo & Romero, 1981; Hunicken, з Spe- gazzini, 1924). In these, 11 and three species have been discovered respectively, of which only one is found in other deposits. All the species seem to belong to subtropical genera of the Neotrop- ical Region. The information on these taphofloras may be complemented with the results of palynological (Archangelsky, 1973, 1976a, 1976b) and paleo- xylological studies (Petriella, 1972), which cul- minated with a paleoenvironmental reconstruc- tion (Petriella & Archangelsky, 1975). This =— FiGURE 1. Paleogene taphofloras of South America.— Venezue la.—3. Maracaibo, акы —4. Dpto. Bolivar, Colombia. — mbia. — 7. Zipequina, Cundinamarca, Colombia. — 8. Tenquendama, one — — 10. s. de Talara, Peru.— 1 l. Belén . Mata Amarilla, Argentina.—23. Co Gui 1. Santa Barbara, Trujillo Venezuela. — 2. NE 5. El Infierno, Boyacá, Colombia.— ambúco, Bras sil. — , Peru. —12.P araiba, Brasil. — 13. Pern Argentina. — 25. Rio Guillermo, Argentina. — 26. Loreto, Chile. — 27. Rio Leona, Argentina. — 28. Ia 25 de Mayo, Argentina. — 29. la Seymour, Argentin 452 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 MAASTRICHTIAN PALEOCENE Falcón Geosinclinal de Bolív . L Foz de Amazonas ` Geosinclinal : Peruviano APA Barreirinhas 7 Recife/ J, Pessoa / Espiritu ' Santo EOCENE OLIGOCENE Oriental Sergipe/ J” Alagoas y Sur Bahia 7... Colorado FIGURE 2. Marine basins in South America. From Bertels (1979), slightly modified. 1986] reconstruction showed in the region of Golfo San Jorge (latitude 46°S) a vegetation section similar to the one found today in southern Brazil, some 2,000 km and 20° further north. It includes man- grove communities, swamp forest, tropical rain forest, mountain rain forest, = forest, and то forest or savanna. The climate ust have been subtropical hd This coin- ies with the findings of crocodiles, which in- dicate that the 10? July isotherm passed through the region at that time, whereas at present it is some 1,500 km further north (Volkheimer, 1971). Among the genera that constituted these com- munities, several were of Australasian origin such as Dacrycarpus, Dicranopteris, and Gunnera, and Tricolpites fillii, similar to Beauprea, which ex- isted in aland during the Cretaceous. Other elements appear simultaneously in New Zealand and South America, such as Anacolosa and Nypa, although they were not as frequent in South America as in Australasia. Nypa must have had a wide distribution because it has also been found in Pernambuco (Brazil) in Paleocene de- posits. Other genera appear in these sediments for the first time, reaching New Zealand during the Eocene and Oligocene (Retidiporites cama- choi, las affinities with Banksia, Weinman- nioxylon, etc.). The taphoflora of Laguna del Hunco (Fig. 1.18), in northwest Patagonia, is at least 59 mil- lion years old, as established by isotope dating (Archangelsky, 1974). It has some 25 described species (Berry, 1925), to which Petersen (1946) added a list of over 70 determinations. It has 69% of entire margined leaves. This percentage corresponds to the Paratropical Rain Forest ac- cording to Wolfe (1971, table 1) and to the Trop- ical Seasonally Dry climatic zone according to Dilcher (1973a, fig. 4). This taphoflora includes several genera of the present Neotropical Region (Annona, Cochlospermum) and others of the present Antarctic Region (Lomatia, Peumus), al- though it does not contain Nothofagus. Further more, endemics to the present Australian Region (Casuarina and Akania, Frenguelli, 1943; Ro- mero & Hickey, 1976) and subtropical genera of xeric type (Schinopsis) have also been found in this taphoflora. Ata latitude of some 22° and 2,000 km further south is the taphoflora of Seymour Island (Fig. 1.29) (Isla C. Marambio in Argentine maps; Du- sen, 1907). Its age is probably equivalent to that of Laguna del Hunco, since it seems to belong to the Paleocene Cross Valley Formation, ac- ROMERO—PALEOGENE PHYTOGEOGRAPHY AND CLIMATOLOGY 453 cording to Elliot and Trautman (1982). Dusen described 25 species and 64 forms determined to genera, especially of Pteridophyta, and he was the first author to point out the existence of a mixture of "temperate species" with “species similar to those of southern Brazil" or “subtrop- ical" (Dusen, 1907). This taphoflora is presently found at a latitude of 64?31'S. It has 4296 ofentire leaves, which corresponds to a Subtropical For- est (Wolfe, 1971, table 1) or to Warm Temperate, probably moist, climatic zone (Dilcher, 1973a, fig. 4) LOWER EOCENE In northern South America the Lower Eocene is represented in Peru by the deposits of Sur de Talara (Fig. 1.10) and Belén (Fig. 1.11). The for- mer was described (Berry, 1929b) as belonging to “Restin Formation" but in fact it corresponds to the Chacras Formation (Hoffstetter, 1956; Bertels, 1979). Berry described some of the fruits he studied as Attalea olsoni, Iriartites restinensis, and Carpolithus jathrophaformis. The second deposit (Belén) lies in the Parinas Sandstone of Lower Eocene age (Hoffstetter, 1956). From these Berry, (1927, 1937c) described 11 species, be- longing to genera such as Palmacarpon, Hu- muria, Sapindoides, Cupanoides, and Anacar- dium. The climate was “humid, lowland tropical," according to Berry (1927). The deposits of Lota and Coronel (Fig. 1.14) in Chile are at a latitude of 37°S. Their age is Lower Eocene, although some authors consider it to be Paleocene (Palma Heldt, 1980). Engelhart (1891, 1905) and Berry (1922) described over 100 species, all of them assigned to Neotropical genera. Berry expressed the opinion that “its elements are represented by modern species of tropical South America east of the Andes, and in the main of Denim dwelling in the Amazon basin, espe- cially toward the Peruvian part of the basin, and extending southward into eastern Bolivia ... I would estimate the Chilean fossil flora as indi- cating a change from the present climatic con- ditions corresponding to from 10? to 12? of lat- itude at sea level." About the climate he said that “there was never frost" and it was “very much warmer and with more sunshine than at the pres- ent time." The percentage of species with entire margined leaves (70%) would indicate that it was a Paratropical Forest, in the sense of Wolfe (1971) or a Tropical, Seasonally Dry climatic zone in the sense of Dilcher (1973a). 454 In northwest Patagonia, the Lower Eocene is represented by two important deposits: Rio Pi- chileufú (Fig. 1.16) and Río Chenqueniyeu (Fig. 1.17). Three species described from the outcrop of Alto Rio Nirihuau (Fiori, 1939) may be con- sidered with the last one. All are thought to be- long to the Ventana Formation (Aragón & Ro- mero, 1984). They inciude some 140 species and 40 species respectively (Berry, 1938; Fiori, 1940) with a mixture of Neotropical and Antarctic ele- ments. The first of these deposits is essentially similar to that of Laguna del Hunco, described above, and its percentage of entire-margined leaves is 6996. Berry (1938) pointed out the pres- ence of genera that are a) tropical or subtropical, b) temperate, c) of semiarid environments (Cu- pania, Schinopsis, and Schinus), d) “from the temperate rain forest region of southern Chile," and e) 14 genera of possible lianas. The condi- tions were of humid and warm climate, judging from the flora, but the mixtures of elements did not allow Berry to classify them with certainty. The taphoflora of Rio Chenqueniyeu is also mixed, and very similar to the previous one, but it does include Nothofagus. Somewhat further south, in the Basin of Golfo San Jorge, lie the deposits of Canadon Hondo (Fig. 1.17), of which Berry (1932a) studied sev- eral small taphofloras. Their age is considered to be Eocene according to Andreis (1978) and they contain Neotropical and Subantarctic elements (see Romero, 1978, for a discussion of unpub- lished material). Finally, some 2,000 km to the south, leaf im- prints also are found (Orlando, 1963) in the ta- phoflora of Isla 25 de Mayo (King George) (Fig. 1.28) of the Antarctic Peninsula, at a latitude of 64°S. It contains Antarctic elements such as Lau- relia and Nothofagus, but together with Neo- tropical ones such as Schinopsis and Nectandra. It may be especially compared with the tapho- flora of Rio Pichileufü. The total number of de- scribed species is only ten, but seven are entire margined. Although the number of species is not sufficient for a statistical analysis, it seems evi- dent that the association was not less warm than subtropical. The age, given originally as Mio- cene, is considered to be Lower Eocene (Arguijo & Romero, 1981; Romero, 1978). MIDDLE EOCENE In the Department of Bolivar, Colombia (Fig. 1.4), Berry (1924b, 19292) described several fos- sil fruits that he attributed to the Middle Eocene. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 These are Sapindoides peruviana, Carpolithus bolivariensis, and Celtis bolivariensis. In Vene- zuela near Escuque (Fig. 1.2), Berry (1920) dis- covered Entada boweni, of the Escuque For- mation (Middle Eocene, Hoffstetter, 1937). In Ecuador, the area of Ancón (Fig. 1.9), Peninsula de Santa Elena, yielded fruits of angiosperms Berry, 1929c, 1932b). According to the collector (Sheppard, 1937) the fossils came from the So- corro Sandstone (Middle Eocene, Hoffstetter, 1936). The material includes Astrocarym shep- pardi, Ventanea sheppardi, Palmocarpon bravoi, Annona peruviana, and Sapindoides peruvianus. The last three are also found in the deposit of Belén, and the last is also found in the Depart- ment of Bolívar, Colombia. The genera belong to a Neotropical lineage, and the environment that they indicate is a lowland rain forest. In Patagonia, the fossil floras of the Middle Eocene are represented in the Rio Turbio For- mation (Fig. 1.24). This important formation contains several levels with imprints, which were studied by Berry (1937b), Frenguelli (1940), and Hunicken (1955, 1967). The levels “s” and "t," which were studied specially by the = author, contain a mixture of Neotropical and Antarctic genera, including Nothofagus. The pollen anal- ysis indicated that in several levels the pollen grains of Nothofagus may be dominant (Archan- oo, percentage of entire-leaved s would indicate a ala ¡Rain Forest (Wolfe, 1971) or a Warm Temperature, moist, climatic zone (Dilcher, 1973a). UPPER EOCENE The Upper Eocene is represented in Venezue- la, near Santa Barbara (Fig. 1.1), Zulia State (Ber- ry, 1936), and in the basin of Maracaibo (Fig. 1.3) (Berry, 1939). Berry described leaf imprints of nine species already known in, or linked to, the floras of the Coastal Plain of North America. They belong to the genera Apocynophyllum, Bur- serites, Cedrela, Chrysophyllum, Ficus, Inga, Terminalia, and COREG, suggesting men ex- 6 Law 1; q d war INCI of the equatorial zone.” In northwest Patagonia, approximately at the latitude at which paratropical rain forest devel- oped during the Lower Eocene, several outcrops with leaf imprints were found in the Ñirihuau Formation (Fig. 1.15) (Aragón & Romero, 1984; e 928; Feruglio, 1941; Fiori, 1931, 1939; Romero & Anm 1981a). Altogether these out- 1986] crops have yielded some 45 species of Antarctic genera. They are strongly dominated by genera th creases to 27% representing a Mixed Mesophytic Forest according to Wolfe (1971) and Cool Tem- perature, moist, climate zone according to Dilch- er (1973a). OLIGOCENE No confirmed records of Oligocene plants are known to me from northern South America. Some deposits attributed to that age are actually older, such as that of Pariñas Sandstone in Belén, Peru, or that of Falls of Tequendama, Colombia, mentioned above. In the southern extreme of the continent there are plant imprints in several taphofloras of the Frenguelli, 1940; Gilkinet, 1955). The first of these lies unconformable on the Río Turbio Formation mentioned above, and its age was established as Oligocene-Miocene by Hunicken (1955). The Río Leona Formation is considered to be correlated with the lower section of the Río Guillermo Formation (Russo et al., 1980). The Loreto Formation is considered to be Upper Eocene or Oligocene (Fasola, 1969), slightly more modern than the Río Turbio For- mation. Arguijo and Romero (1981) found that the fossils contained in these three formations are very similar (Jaccard index ca. 50). Conse- quently, it seems logical to consider the tapho- floras of the three formations together, and at- tribute them for the moment to the earlier Oligocene, or perhaps to the later Eocene. The leaf imprints represent 27 species of Fa- gaceae and associated plants of the Antarctic Re- gion, although Dusen (1899) pointed out that the greatest affinities of the flora are with the asso- ciations presently limited to the northern ex- treme of the Subantarctic Province, some 1,000 km north of the fossiliferous outcrops. The per- centage of entire margined leaves is similar to that of the Nirihuau taphoflora, of the Upper Eocene, but the size of the leaves seems to be substantially smaller, generally nanophylls, or at the most microphylls. MIOCENE Outcrops of Miocene age with plants are abun- dant in South America and some of them are ROMERO—PALEOGENE PHYTOGEOGRAPHY AND CLIMATOLOGY 455 classical, such as those of Loja and Cuenca Basins (Ecuador), but their consideration lies outside the scope of the present work. It is noteworthy that they are absent from Pat- agonia and the southern part of Chile. Thus they reflect a change in the deposition regimes as was pointed out by Yrigoyen (1969), and Pascual and Odreman Rivas (1971), based on different kinds of evidence. Furthermore, many of them that indicate humid, indi tropical forest appear in present day highlands or desert areas, thus making it iade. to establish, together with oth- ers of Pliocene and Pleistocene ages, the ише at which the And tarted to rise, p that lead to the fe ti f deserts and highland environments (Menéndez, 1971; Axelrod, 1979). DISCUSSION PALEOPHYTOGEOGRAPH Y The living vegetation of South America be- longs to two botanical Regions: Neotropical and Antarctic (Cabrera & Willink, 1973). The Neo- tropical Region covers most of the continent un- der an array of very variable climatic conditions. Wet forests comprise the Caribbean, Amazo- nian, and Guayanan Dominions; drier and poor- er forests correspond to open areas without trees and comprise the Chaqueño and Andino-Pata- gonian Dominions. The Antarctic Region con- tains a wet to rather dry forest, in the only Sub- antarctic Dominion. Figure 3 shows the taphofloras discussed in the previous section placed according to their age and approximately at their present latitudes. I used present latitudes because the plotted out- crops belong to different ages. Furthermore, the South American continent has not migrated very much since the Cretaceous. According to Valen- cio et al. (1971), in Cretaceous times it was about 10? more to the north, and a slow, uniform, non- rotating displacement has occurred since then. Three main sectors may be observed in Figure 3. The first one at the upper left contains all the taphofloras with genera of the Neotropical Phy- togeographic Region or their ancestors, exclu- sively. These taphofloras constitute the Neotrop- ical Paleoflora. The second one, at the upper right, contains the taphofloras with genera char- acteristic of the Antarctic Phytogeographic Re- gion, or their ancestors. This is the Antarctic Pa- leoflora. Between them, the third sector contains the floras that seem to present a mixture of species of both lineages. This is the Mixed Paleoflora. 456 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 N 10° | +2 + 1 +, +3 + 5-8 0°. *9 +10 : +1 10° | ыы 413 NEOTROPICAL 20°44 PALEOFLOR|A S? $ go 30°, wo” MIXES ^7 Northern p Pd d P d 40°- PALEOFLO а №? 1% 50° Southern LEOFLORA *22 2 + 23 (60%) +25 (di T 26 (125%) 27 Ей +28 (70% +29 (42% Maastrichtian Paleocene Lw. Eocene M. Eocene Upp. Eocene Oligocene FIGURE 3. Age of Paleogene taphofloras of South America. See numbers in caption of Figure 1. Solid lines separate the paleofloras. Dashed lines separate probable provinces within them. Percentages of entire-margined leaves and the presence of Nothofagus (N) are denoted. The neotropical taphofloras cover almost all South America during the entire period consid- ered in this paper. The antarctic ones covered only the southern area from the Upper Eocene onward. The mixed ones covered almost all of Patagonia to the Antarctic between the Upper Paleocene and the Middle Eocene. According to Dusen (1899), mixed floras were explained by an altitudinal zoning of the species, where those belonging to the Antarctic Region occupied the uplands, and those of the Neotrop- u Ln оО which I called Mixed Paleo- po" Dilcher (1973a) pointed out that in the past, both Asa Gray and R. W. Brown added genera that they considered temperate to the floras of Kentucky and Tennessee. But rather than mix these temperate forms in to so-called subtropical coastal flora, they postulated their source area in the Appalachian uplands. At pres- ent, several authors have ceased to consider those floras subtropical, and Dilcher is of the opinion that they were **probably temperate forms pres- ent in the lowland community which we would not recognize as elements of a lowland flora to- day." Also, Kemp (1978) found that, in Australian Eocene outcrops known palynologically, there is an *anamalous mixture of tropical or subtropical rainforest types with those of cool temperate rainforest communities . . . (which) . . . suggests that the early Tertiary vegetation cannot be rep- resented by a single modern forest type. It may be that the climate regime which supported this forest has no modern analog. I suggested earlier (Romero, 1978) that the Mixed Paleoflora would be characterized by a mixture of genera from other units, whose species would have had, however, ecological require- ments different from the living species of the same genera, and by genera of its own, linked with the present Chaqueno Dominion. I espe- 1986] ROMERO—PALEOGENE PHYTOGEOGRAPHY AND CLIMATOLOGY Lower Tertiary Upper Paleoflora: Cretaceous Antarctic Mixed Species 22 26 25 Мео- 16 18 24 28 19 17 tropical * Schinopsis sp. Schinopsis patagonica balansiformis moronghifolia sp. 1 sp. cf. Schinopsis Schinus molliformis Anacardites pichileufuensis rittoni patagonicus Astronium argentinum Roophyllum serratum nordenskjoldi E 4. Evolution of Anacardiaceae in southern FIGUR Romero (1978). See numbers in caption of Figure 1. cially emphasized the evolution of the Anacar- iaceae family in Patagonia, which, with only one species in the Coniacian (Upper Cretaceous), radiated during the Lower Tertiary, having five genera and 14 species in taphofloras of the Mixed Paleoflora (Fig. 4). Among those genera are Schinopsis, Schinus, and Astronium, which are conspicuous today in the Chaco, Caatinga, and Monte Provinces (Cabrera & Willink, 1973). In the Neotropical Paleoflora a few taphofloras present elements now endemic to the Australian Region as discussed above. Upon this basis, it is possible to separate a northern, purely American Province, from a southern one, subject to mi- gration (Fig. 3). In the Mixed Paleoflora, the ab- sence of Nothofagus from taphofloras in Figure 3.16, 3.18, allows the separation of two prov- inces. Another separation is tempting, isolating taphofloras in Figures 3.24, 3.29, on the basis of their low percentage of entire-leaved species. However this is a physiognomic characteristic, and phytogeographic limits are being established in this paper on a floristic basis. I therefore tentatively suggest that the Neo- iie Paleoflora developed during the Lower iary and was the forerunner of the more hu- 2 Dominions of the Neotropical Region (Ca- South América from Cretaceous to Paleogene, from ribbean, Amazonian, and Guayanan); that the Mixed Paleoflora, from the Upper Paleocene on- ward, gave origin to the drier Dominions (Cha- queno and Andino-Patagonian), and that the Antarctic Paleoflora, since the Upper Eocene, was the antecedent in arctic Dominion. The real significance of further differences within the paleofloras is not yet ap- parent. CLIMATIC CHANGES Figure 3 shows the taphofloras discussed above and the corresponding paleofloras. The presence of Nothofagus was also denoted in several of them. Furthermore, the percentage of entire-leaved species was added in other cases. The merits of this last feature has been discussed by Dolph (1979), Dilcher (1973a), and Wolfe (1971, 1979) but ever since the observations of Bailey and Sinnot (1916), the high percentage has been con- sidered a good indication of warm environment. The areas covered by the different climatic types circumscribed on the basis of percentages of entire-margined leaves in South America do not coincide with the phytogeographical limits 458 of the Neotropical, Mixed, and Antarctic Paleo- floras, as here defined. As an example, warm climates indicated by high percentages are pres- ent in both Provinces of the Neotropical Paleo- floraand both Provinces of the Mixed Paleoflora. This is in accordance with the climatic variations within present Phytogeographic Regions (the Neotropical Region extends from the Amazo- nian wet warm forest to the Patagonian dry cold steppe) today and reflects the criteria used in this paper, whereas the Phytogeographic Regions were defined, as the recent ones are, on a taxonomic basis, and the climatic types were defined on a physiognomic basis. The physiognomic analysis is relevant in this paper only for those taphofloras of Patagonia, because the rest, from northern South America, contain too few species for statistical analysis (Wolfe, 1975). The percentages used are mostly those given by Volkheimer (1971). As shown in Figure 3, for the Maastrichtian, the taphoflora of Cerro Guido (Fig. 3.23) in southwest Patagonia, has 60% of species with entire-margined leaves. For the Paleocene and early Eocene there are records in northwest Patagonia (taphofloras in Fig. 3.14, 3.16, 3.17) of 70% with entire-mar- gined leaves. In the Antarctic peninsula there is an increase from 43% (Paleocene, Fig. 3.29) to about 70% (early Eocene, Fig. 1.28). Finally, dur- ing the late Paleocene, eastern Patagonia had a subtropical humid climate, with mangroves, crocodiles, etc., as explained above. Therefore, it may be accepted that since Maastrichtian to lower Eocene times a progressive temperature increase occurred. For the Middle Eocene there is only a record in the Rio Turbio taphoflora (Fig. 3.24), south- western Patagonia, ue a decrease up to 4996 of leaves. For the upper Eocene the Ñirikuau mon (Fig. 3.15) in northwest Patagonia show a further decrease to 1796. And during the Oligocene, again in southwest Pata- gonia, the same percentage remains, with a dom- inance of small-sized leaves. Therefore, a dete- rioration of climate since Middle Eocene to Oligocene may be supposed. Roughly speaking, the sequence of floras in Patagonia would imply variations from the equivalents of a Paratropical Rain Forest (Lower Eocene) to a Subtropical Forest (Middle Eocene), then to the Mixed Mesophytic Forest (Upper Eocene), and finally to a Mixed Northern Hard- wood Forest (Oligocene) (sensu Wolfe, 1971). -maroined ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 In other parts of the world, a progressive tem- perature increase was detected during the Paleo- cene and Lower Eocene (Wolfe, 1978), although “cool intervals occurred during the late Paleo- cene, the late early to early Middle Eocene and the early Late Eocene" (Wolfe, 1978). The most important increase, during the late Eocene, is correlated with the Bertonian, although its **du- ration and exact timing is unknown" (Wolfe, 1975: 50). dun brief shift was called the Eocene Terminal Eve Finally, a major shift during the Oligocene was recorded in Europe, western Siberia, New Zea- land, and western North America (Wolfe, 1971). It was more marked during the middle Oligocene and produced a spectacular transformation in the high latitude floras, which changed from having high percentages of large, evergreen broad leaves with entire margins to high percentages of den- tate, small, deciduous leaves. The data from South America are not abun- dant enough to show minute changes during the late Paleocene to the early late Eocene. However, a coincidence with the general trends to higher temperatures in other continents may be detect- ed during the Paleocene and Lower Eocene, and then a deterioration through the Middle and Up- per Eocene. The subsequent major shift during the Oligocene, however, is not yet apparent in this continent. LITERATURE CITED ANDREIS, R. 1978. Geología y Sedimentologia de Canadón Hondo Pcia. de Chubut (Obra del Cen- 1 . 1984. Geología, paleoam- bientes y paleobotánica de yacimientos Terciarios del occidente de Río Negro, Neuquén y Chubut, Repüblica Argentina. Actas IX Congr. Geol. Ar- gent. 4: 475-507. ARCHANGELSKY, S. 1973. _ Palinología del drehen de Chubut. I. D eghi- niana 10: 339-399. 1974. Sobre la edad de la tafoflora de Laguna del Hunco, Provincia de Chubut. Ameghiniana 11: 413-417 76a. Palinologia del Paleoceno del Chubut. II. Diagramas Polínicos. Ameghiniana 13: 43-55. . 197 ali . I. Introducción у matrices de similitud. Ameghiniana 13: 169-1 joe Ў Los iste mas antiguos s de Nothofagus (Fagaceae) de Patagonia eo y Chile). Bol. Soc. Bot. México 33: 13- ARGUIJO, M. H. & E. J. Romero. 1981. Análisis bies- tratigráfico de formaciones portadoras de tafo- 1986] floras terciarias. Actas VII Congr. Geol. Argent. 4: 691-717. AXELROD, D. I. he Desert bi eg its age and origin. Pp. 2 in J. R. Goo Or- thington E Arid Plant о ‘Inter r- national Center Arid & Semi Arid Land Studies, Lubbock, Texas. BaiLEv, I. W. & W. E. Sinnot. 1916. The climatic distribution of certain types of Angiosperm leaves Amer. J. Bot. 3: 24-39. . 1920. Log i fossil plants from Ven- ezuela. Proc. U.S. Natl. Mus. 59: 553-579. 1921. Tertiary formations of western South America. Pp. aon in Pan Pacific Science Con- ference, Washingt 1922. The зм of the Concepción-Arauco Coal — "i Chile. Johns Hopkins Univ. Stud. Geo ph 73-1 24 | са fruits from the eastern Andes of Көч к Bull. Torrey Bot. Club 51: 61—67. 24b. A fossil Celtis from Colombia. Tor- reya 24: 44-46. —. 1925. А Miocene flora P Patagonia. Johns Hopkins Univ. Stud. Geol. 6: -252. 1927. Petrified fruits and purs viue Oligo- cene of Perú. Pan-Amer. Geol. 47: 121- Tertiary fossil plants from is Argen- tine Republic. Proc. U.S. Natl. Mus. 73: 1-27. 929a. Tertiary fossil plants from Colombia, South America. Proc. U.S. Natl. Mus. 75: 1-12. 1929b. y es plants from Restin formation of Perú. Pan-Amer. Geol. 51: 4. 1929c. Fo ssil fruits i in the is sandstone of. Ecuador. J. Paleontol. 3: 298-301. 1932a. ыш plants from Chubut Territory Patagonian Expedition. Amer. 1-10. collected by Scarri Mus. Novit. 1932b. A new palm from the poo Eocene of Ecuador. J. Wash. Acad. Sci. 22: 327-329. . 1936. Tertiary fossil plants from Venezuela. II. Proc. U.S. Natl. Mus. 83: 335-360. —. 1937a. A Paleocene flora from urs gaa Johns Hopkins Univ. Stud. Geol. 12: 22-50. 1937b. Eocene "o from Rio т. in the Territory of San , Patagonia. Johns Hopkins Univ. Stud. Geol. "12: 91-98. 1937c. The Pariñas sandstone of Perú. Johns Hopkins Univ. Stud. Geol. 12: 99-106. . 1938. Tertiary flora from the Rio deii Argentina. Special Pap. Geol. Soc. Amer. 149. 1939. Eocene Plants from a well core in Ven- ezuela. Johns Hopkins Univ. Stud. Geol. 13: 157- 162. . 1940. Mesozoic and Cenozoic plants of South rica ntillas. Proc. BERTELS, A. 1979. niferos del Cretacic о 16: 273- 256. BEURLEN, К. & Е. W. SoMMER. 1954. Restos vegetais 055615 = tectónica de Bacio Calcarea de Itaborai, Estado do Río js jar Bol. Div. Geol. Mineral. Minist. Agric. 16-20. CABRERA, A. L Paleobiogeografía de los forami- de América del Sur. P WILLINK. 1973. Biogeografia ROMERO— PALEOGENE PHYTOGEOGRAPHY AND CLIMATOLOGY 459 de América Latina. Monografia 13, Ser. Biologia. Organización de Estados Americanos, Washing- ton. Риснек, О. 1973a. A paleoclimatic interpretation of the Eocene Floras of southeastern North Amer- Graham ши Vegetation M6 DE rn Latin Am . El- e Eocene Flora o southeastern North America. Расон 20: 7- 18. [1971 DoriANrrI, E. 55. Frutos de Nipa no Paleoceno de Pe rna anbuco. Brasil. Bol. Div. Geol. Mineral. Mi- —36. Variation in leaf margin with respect to climate in Costa Rica. Bull. Torrey Bot. Club 106: 104-109. Dusen, P. 1899. Uber die tertiare Flora der Magel- lanslan der I. Wiss. Ergebn. Schwed. Exped. Ma- кши 1895—1897, 1: 87—108. . Uber die tertiare Flora des Seym с Wiss. Ergebn. Nordsk. Schwed. Südpolar- 1-27. №) е E v т Q m © 1. е, iarpflazen уоп Chile. TE Naturf. Ges. 16: 629-692. ——— 19 merkinger zu chilenischen Tertiarp- en. Abh. Naturf. Ges. Isis 1905: 69-72. ErLLioT, D.H. & T TRAUTMAN. 1982. Lower ter- arctic Geoscience. Univ. Wisconsin Press, Madi son. FasoLa, A. 1969. Estudio palinológico de la Fm Lo- a 2 (Terciario Е гоуіпсіа de Magallanes, e. Ameghiniana —19. o E. А E preliminar sobre la hoja geológica “San Carlos de Bariloche” (Patagonia). Bol. Inform. Petrol. 18: 27-64. Fiort. A. 1931. Fillite terziare della Patagonia. I. Fil- lite della riva meridionalle del Lago Nahuel Hua- pí. Giorn. Geol. 4: 101-116. . 1939. Fillite terziare della Patagonia. II. Fil- lite y Río Nirihuau. Giorn. Geol. 13: 940. Fillite terziare della Patagonia. ш. Fil- lite ad Río Cheuquefiiyen. Giorn. Geol. 14: 94— 143. FRENGUELLI, J. 1940. Nuevos elementos я del N de Patagonia Austral. Notas Mus La Plata, Secc. Paleontol. 6: 173-202. 1943. Restos de Casuarina en el Mioceno en El “Mirador, Patagonia Central. Revista Mus. La Plata, Secc. Paleontol. 8: 349-354. GILKINET, A. Quelques arces fossiles des Terres Magallaniques. Exped. Antarc. Belge, Re- sult. Voyage S.Y. Belgica, 1897- 1898. Rapp. Sci. HOFFSTETTER, R. 1936. p gy cub ал In- ernatio que Latine. Fasc. 37. Lexique: Stratigraphique International, Volumen V, > e Latine. Fasc. 5c, Vene zuela. CNRS, P . 1956. ne Stratigraphique hg asi Volumen ы, Amerique Latine. Fasc. 5b, Per NRS, Par HUERTAS, G. 1960. De la flora fosil de la Sabana. 460 Bol. Geol, Fac. Petrol., Univ. Industr. Santander 7 : 53- Нома КЕМ, M. 1955. Depósitos neocretácicos y Ter- os del Extremo S.S.W. de Santa Cruz (Cuenca Corboatfers de Rio Turbio). vs Inst. Nac. Invest. Ci. Nat., Ci. Geol. 4 Flora Terciaria e " Estratos de Río Turbio, Santa Cruz (Niveles Vd ide del arroyo Santa Flavia). Revista Fac. Ci. E , Fis. Nat. 1968. Sobre los tipos de Carolites patago- a y Ameghinoites desiderata Speg. Ameghi- niana 5: 447—456. 1971. Paleophytologia Kurtziana III— Atlas de la flora fósil de Cerro Guido (Cretácico Superior), Ultima Esperanza, Chile (especimenes examinados por F. Kurtz). Ameghiniana 8: 231- Kemp, E. M. . Tertiary climatic evolution and vegetational history in the Southeast Indian Ocean ares dera Palaeoclimatol. Palaeoecol. 24: 8. KURTZ, E 1899. Contribuciones a la Paleophytologia Argentina II. Sobre la existencia de una Dakota- a en la Patagonia Austro-occidental (Cerro Guido, Gobernación de Santa Cruz) Revista Mus. La Plata, Secc. Paleontol. 10: 43-60. dieras C. J. . O Cretaceo de Sergipe. Monogr. v. Geol. Mineral. Minist. Agric., Brasil 11: 44. MenénDEz, C. A. 1964, Paleobotanical evidences in gard to the origin of the flora of Argentina. yi on. Res. Rep. Geol. Lab. Univ. Arizona 5: 23. 1966. La presencia de Thyrsopteris en el Cre- tácico Superior de Cerro Guido, Chile. Ameghi- ma 4: 299-301. . 1969. Die fossilen Floren Sudamerikas. In а. und Okologie іп Sudamerika. Monogr. Biol. 2: 161-200. 971. Floras Loue de la Argentina. Ameghiniana 8: 357- 1972a. ска Kurtziana III-8. La e Cerro Guido, Chile 2; 1972Ь. Paleophytologia Kurtziana Ш-9. DW del Cretácico Superior de Cerro Guido, Chile (3-7). Ameghiniana 9: 289-297. MULLER, J. 1970. Palynological gs on d . Biol. ; 63. La flora fósil en las ТРЕЕ € s de la Península Ardley, Isla 25 de Mayo, Islas Shetland del Sur. Contribución 79. Instituto 198 Contribución al conoci- а palinológico de los mantos carbónicos del Terciario de Arauco-Concepción, Chile. Actas II E. rgent. Paleontol. Bioestratigr., Latinoamer. Paleont., II: 175- e PascuaL, К. & O. E. ODREMAN Rivas. 197 lución de las comunidades de los a del Terciario argentino. Los aspectos paleozoogeo- I Congr. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 graficos y paleoclimaticos relacionados. Ameghi- niana 8: 372-412. m C. A. 1946. Estudios geológicos de la E ón del Río Chubut medio. Dir. Gen. Min. Geo Bol. 59: 1-137. PETRIELLA, B. T. P. . Estudio de maderas petri- ficadas del Terciario inferior del área Central de Chubut, (Co Pororo.: Revista Mus. La Plata, Secc. Paleontol. n.s. S. ARC 197 5. Vegetación y am- bientes en el he eno vr Chubut. Actas I Congr. Argent. dede dee iden 2: 257-270. RAVEN, P. & D. Axe 14. Angiosperm bio- geography and p tal movements. Ann. Missouri e Gard. 61: ROMERO, E. J. 1 Polen fosil de Nothofagus (Noth- ofagidites) a Cretácico y Paleoceno de Patagonia. vista Mus. La Plata, Secc. Paleontol. n.s. 7: 291- 03. 1977. Polen de Gimnospermas y og sein de la Formacióon Rio Turbio (Eoceno) Sa 'TUZ Argentina. Fecic Editorial, Buenos MA 978. Paleoecología y Paleofitogeografía de las tafofloras del Cenofitico de Argentina y áreas vecinas. Ameghiniana 15: 209-227. & RGUUO. 1981a. Adición de la tafo- flora del Yacimiento “Bariloche” (Eoceno) Pcia. de Río Negro, Repüblica Argentina. Anais II Congr. Latinoamer. Paleontol. 2: 489—495 Ib. Nota sobre algunos prob- y pea de Austrosudamérica. Bol. Asoc. Latinoa- r. Paleobot. Palinol. 8: 21-33. Me 982. Análisis оа de las tafofloras del Cretacico Supe y Cretacico (editors), Cuencas Sedimentarias del Jurasico y Cretacico de América del Sur 2: 393- 406. M. C. DiBBERN. 1984. Floras Cenozoicas. Pp. 373-382. In V. Ramos (editor), Geología y Recursos Naturales de la Provincia de Río Negro. Buenos Aires. & J. Hickey. 1976. A fossil leaf of Akaniaceae from Paleocene beds in Argentina. Bull. Torrey Bot. Club 103: 126-131. Russo, A., M. A. FLORES & H. Di BENEDETTO. 1980. Patagonia Austral Extraandina. Geología Regio- nal Argentina. Acad. Nac. Ci. Córdoba II: 1431- 14 SELLING, O. H. 1945. Fossil remains of the genus Humi Ecuador. or. Muby, 9 7. VALENCIO, D. A., piu) 1 ne y evolución del Cont nente Gondwana. Sobre la base de paleo. magnéticos de la propagación de los fondos pa los Océanos. Revista Asoc. Geol. Argent. 26: 5-23. VAN DER HAMMEN, T. 1954. El desarrollo de la flora colombiana en los períodos geológicos. F: Maes- trichtiano hasta Terciario más inferior. (Una in- 1986] ара palinológica de la formación de Gua- das y ivalentes). Bol. Geo 1. Inst. Geol. Мас. Colombia . 2: 49-106 VOLKHEIMER, W. 1971. Aspectos paleoclimáticos del Terciario Argentino. Revista nt. Ci. Nat. Bernardino Rivadavia, Secc. Paleontol. 1: WOLFE, J. A. 1971. Tertiary climatic fluctuations and methods of analysis of Tertiary Floras o- geogr. Palaeoclimatol. Palaeoecol. 9: 27- 1975. Some aspects of plant а of the northern hemisphere during the late Cretaceous and Tertiary. Ann. Missouri Bot. Gard. 62: 264— 1978. A paleobotanical interpretation of ROMERO- PALEOGENE PHYTOGEOGRAPHY AND CLIMATOLOGY 461 Tertiary climates e the Northern Hemisphere. Amer. Sci. 66: 694-703. Temperature parameters of humid to mesic forest of eastern Asia and relation to forest of other regions of the Northern Hemisphere and Australasia. Profess. Pap. U.S. Geol. Surv. 1106: 1-37. ee M. R. 1969. _ Problemas estratigráficos del co (editors), Cuencas Sedimentarias del Jurásico y Cretácico de América del Sur. 1: 9-44. NEW TAXA OF CALADIUM, CHLOROSPATHA, AND XANTHOSOMA (ARACEAE: COLOCASIOIDEAE) FROM SOUTHERN CENTRAL AMERICA AND NORTHWESTERN COLOMBIA! MICHAEL H. GRAYUM? ABSTRACT Six new infrageneric taxa of Araceae, subfamily oe are described. eee ham- osta Ri meliana is described from a restricted area of Pana . СРО and Panama, and Caladium lindenii (André) dee id var. prudens These represent the first reported indigenous occurrences of the genera C. тинин outside of South America. In addition, Chl/orospatha gentryi a na ssp. croatia from Panam and C. croat D. described from northwestern Colombia. Chlorospatha croatiana and C. Rain are the aide and fifth 1 А species of their genus known to have compound leaves; a key is provided to all five. brief review of generic distinctions within the tribe Caladieae precedes the oe of the second known peltate- leaved species of Xanthosoma, X. caladioides, from eastern Panam The tribe Caladieae (sensu Madison, 1981), belonging to the subfamily Colocasioideae of the overwhelmingly tropical family Araceae, con- sists of six genera and 70-75 species, all confined to the New World. Three of the genera, Aphyl- larum, Jasarum, and Scaphispatha, are mono- this paper. The remaining three genera, Caladi- um, Chlorospatha, and Xanthosoma, comprise the bulk of the species and range more widely, though only Xanthosoma has been heretofore re- ported to extend beyond South America (Croat, 1979; Madison, 1981). The Caladieae appear to be especially diverse in the Andean regions of northern South Amer- ica, and recent work on newly available collec- tions from that area (Madison, 1981) has led to the clarification of generic concepts in the group. These refined concepts have, in turn, facilitated the placement of new taxa collected in South America and other regions, including those de- scribed below. CALADIUM Caladium lindenii (André) Madison is well known in cultivation as an ornamental plant with whitish or silvery leaf venation. This species is more commonly, but improperly, known as Xan- thosoma lindenii (André) Engl., the transfer to Caladium having been made only recently (Mad- ison, 1981). It was first described, as Phyllotae- nium lindenii André, in 1872, from plants supposedly collected in Colombia; although per- sisting in cultivation, the species was not found again in the wild until 1939, when a form with plain green leaves was collected in Chocó De- partment, Colombia (Ki//ip 35140, COL). Sub- sequent collections have extended the known natural range of this species—and consequently of the genus Ca/ladium —into central Panama. All collections of the plain-leaved form have, until recently, either been misidentified or left undetermined. he “wild type," plain green-leaved form of Caladium lindenii is described below as a new variety. ' I am grateful to Thomas B. Croat for assistance in my work at the Missouri Botanical Garden and for . Gentry for . Hammel supplied fresh pollen samples from the field, in addition to herbarium specimens, and Frances Mazanec critically reviewed and rewrote the Latin diagnoses for all taxa except Xanthosoma caladioides; to both of these individuals I am deeply appreciative. The investigation of pollen morphology in Ca/adium lindenii, Chlorospatha croatiana, and Chlorospatha hammeliana was facil- itated by National Science Foundation grants to James W. Walker as well as an Albert DeLisle Scholarship иа of Massachusetts) to the author. ? Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. ANN. MISSOURI Bor. GARD. 73: 462-474. 1986. 1986] Caladium lindenii (André) Madison var. syl- vestre Grayum, var. nov. TYPE: Panama. Panama: El Llano—Carti Rd., 9 mi. from turnoff of PanAmerican Hwy., ca. 300 m, Hammel et al. 11341 (holotype, MO- 2982267). Var. sylvestre typico immaculatis viridis laminis dis- tinguenda. Lamina simplex, sagittata, inferiore pagina n iii eg uh sions petiolus 40-95 cm lon- , ferrugineo-furfuraceus; lobus anticus 25-35 cm prolis 15-25 cm latus; nid posticus 9-18 cm longus. Inflorescentia solitaria; pedunculus 6-17 cm longus; spatha 8-12 cm longa; tubus cm longus; spathae lamina 5-8 cm longa; spadix pars mascula 4—4.5 cm longa. Terrestrial herb, ca. 75 cm tall. Stem ca. 2-2.5 cm diam. (Croat 52023, 56094), largely or en- tirely subterranean. Petiole purplish at the base, “narrowly flattened” (Croat € Cagallo 52150) to "narrowly and obtusely sulcate" (Croat 56094) adaxially, with scant milky sap or sap not evident at all, 40-95 cm long, covered with sparse, tan- gled, reddish brown scurf. Lamina sagittate, not peltate, the anterior lobe 25-35 cm long, 15-25 cm broad, the posterior lobes 9-18 cm long (mea- sured along main vein). Leaves uniformly green on both surfaces (midrib and proximal parts of major veins white in Croat 52023), matte and dark green above, much paler below and mi- nutely covered with reddish brown scurf; major veins sunken above, the smaller veins conspic- uous beneath, darker than the lamina. Inflores- cence solitary (or perhaps occasionally paired), peduncle 6-17 cm long. Spathe 8-12 cm long; tube green to purplish, 4—5 cm long; blade white, 5-8 cm long. Spadix ca. 7.5-9.5 cm long; female part of spadix ca. 1.5-2.5 cm long; sterile part 1.5-2.5 cm long; male part 4—4.5 cm long, white. Pollen in monads, inaperturate, subspheroidal, perfectly psilate, 34-43 um (mean 38 um) diam., starch-bearing, trinucleate. Distribution. Caladium lindenii var. syl- vestre is currently known from the departments of Antioquia and Chocó, Colombia, and the Province of Panamá, Panama. It has been col- lected mostly in deep, shaded primary forest, on slopes near creeks, from near sea level to about 600 m; the label on Denslow 2672 reads “сот- mon in second growth." The population of Caladium lindenii recently discovered along the El Llano-Carti Rd. in cen- tral Panama represents the first known indige- nous occurrence of this genus outside the South GRAYUM— ARACEAE 463 American continent (see Croat, 1979; Madison, 1981). The seven flowering collections were made in February, March, April, August, and October. he vulgar name “picha de gato” is recorded on the label of Folsom et al. 5785. Caladium lindenii var. sylvestre differs from the “typical” var. lindenii in its plain green, un- variegated leaves. According to the description of C. lindenii provided by Madison (1981), var. also appear to differ in Danus cially with regard to leaf measurements, ов length, peduncle length, spathe length, and the length of the male portion of the spadix. In all of these categories, the ranges of measurements for the two varieties overlap scarcely if at all. That the two forms are nonetheless conspecific is indicated by their distinctive sagittate leaf shape and tangled, reddish brown, scurfy indumentum, as well as by the tendency toward whitening of the major veins seen in one collection of var. sylvestre. In any case, the purported size differ- ence is spurious, since many specimens of var. lindenii (e.g., Hutchison 8456, MO) are every bit as large in all respects as var. sy/vestre. Madison’s measurements must have been taken from a poorly grown specimen, or a particularly small cultivar. The rediscovery of Caladium lindenii in the wild as a plain green-leaved form suggests the possibility that other rare aroid species currently known only from variegated specimens (e.g., Zomicarpella maculata N. E. Br. ) may also prove o hav уер li in the wild state. Horticulturists have long favored oddities such as variegated foliage, and it is not unreasonable to suggest that collectors in search of ornamental subjects might have bypassed whole populations of plain green-leaved individ- uals, pausing only to dig up the occasional at- tractive, yet anomalous specimen—which, when shipped back to Europe, may well have come to “typify” certain species. This is clearly what must have happened in the case of Caladium lindenii and is also suspected in other aroid species such as the familiar “pothos” of horticulture, which is unknown in the wild state. Known variously as Scindapsus aureus Engl., Rhaphidophora au- rea (Engl.) Birdsey and, most recently, Epiprem- num aureum (Engl.) Bunting, it is now believed to represent nothing more than a variegated form of the widespread Asian species Е. pinnatum (L.) Engl. (Nicolson, 1978). Lond 464 Specimens examined. COLOMBIA. ANTIOQUIA: municipio of Anori, Soejarto et al. 3197 (MO); Rio Anori valley, Shepherd 533, 647 (WIS), Denslow 2672 (WIS); between San Luis and Puerto Triunfo, Croat 52023 (MO). cHocó: between Quibdó and Istmina, Croat € Cagallo 52150 (МО); near Río Iró, S of Ist- mina, Croat 57398 (МО); region of Río Baudó, Fuchs 22269 (COL); Serranía de Baudó, between Las Animas and Pato, Croat 56094 (M PANAMA. PANAMÁ: El Llano-Cartí Rd., Folsom et al. 5785 (MO), Hammel et al. 11341 (MO), Hammel & Kress 13392 (DUKE) CHLOROSPATHA The genus Chlorospatha, regarded for nearly a century as monotypic, owes its present status as a moderately sizable genus to Madison's (1981) realization that C. kolbii Engl, the original species, differed from several species then con- stituting the genus Caladiopsis only in its pos- session of compound leaves. This being a char- acter of no significance at the generic level among New World Colocasioideae, the two genera were combined under the older name. As indicated previously, the genus Chloro- spatha has never before been reported from out- side the South American continent. Plants be- longing to this taxon have, in fact, frequently been collected from Panama over the past cade; the specimens, however, have generally been misfiled with the more familiar genus Xan- thosoma. The two genera may be easily distin- guished on the basis of vegetative (Xanthosoma is frequently cormose; CA/orospatha never is) and floral characters (Chlorospatha has more slender and elongate inflorescences on more delicate pe- duncles, with the female portion of the spadix usually at least partly adnate to the spathe). For further details, the reader is referred to the im- portant paper of Madison (1981). Chlorospatha and Xanthosoma are also paly- nologically distinct; indeed, the first indication that the two mainly Panamanian species de- scribed below were members of the genus Chlo- rospatha was provided by an examination of fresh pollen, without recourse to sporophytic material. The two genera are unique in Araceae in shed- ding pollen in permanent tetrads; as Madison (1981) has indicated, the individual grains (as well as the tetrads) of Chlorospatha average con- siderably smaller than those of Xanthosoma— more than 5096 smaller, in fact (Grayum, 1984). In addition, pollen of Chlorospatha lacks starch and is basically binucleate, whereas pollen of Xanthosoma is always starchy and frequently ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 trinucleate (see Grayum, 1985, 1986). In all of these respects, the two species described here are typical members of the genus Chlorospatha. Chlorospatha croatiana Grayum, sp. nov. TYPE: Panama. Coclé: N slope and summit of Cer- ro Pilón, 900-1,173 m, Croat 22932 (ho- lotype, MO; isotypes B, K, PMA, Planta үс caulis erectus, subterraneus, saltem ad 22 cm 5-9- и glabra; petiolus (35-)50-70(-83) cm longus, purpureo-maculatus; lamina segmentum me- dium subaequilaterale, (17-21-3143) cm longum, (6-)9-15(-20) cm latum, anguste ad late ellipticum; segmenta lateralia inaequilateralia. Inflorescentia ad usque 5-6; pedunculus 13-33(-48) cm longus, pur- Sali spatha sub anthesi angusta, 7-9 cm longa; tu- us (3 -)4-6 cm longus, ad 7-9 cm longus i in statu fruc- ee aliquanto cucullata; spadix 3.9-6.1 cm longa; pars feminea 1.2- 2.4 cm longa, in dimidio inferiore spatha adnatae; pars sterilis 0.5—1.2 cm longa; pars mascula 2-2.7 cm longa; fructus albus; semina 15-25, alba, striata, ovoidea, 1- 1.5 mm longa Terrestrial herb 0.5-2 m tall, the sap milky (e.g., Churchill 3804). Stems basally decumbent to erect, 5-22 cm or “sometimes са. | m r 22932) long, 0.6-2.5 cm wide (dried), largely subterranean, clothed with weathered brownish and whitish cataphyll fibers. Leaves solitary or few, the petioles (35-)50-70(-83) cm long, terete, green and mottled with purple, at least toward the base, or else solid purple or purple-brown, vaginate in the lower one-fourth to one-half; lamina of adult leaf pedately 5—9-foliolate, gla- brous, the central lobe more or less symmetrical, (17-)21-31(-43) cm long, narrowly to broadly elliptical, (6-)9-15(-20) cm broad, usually with 6-7 primary lateral veins per side. Lateral lobes progressively smaller than the central lobe, asymmetrical, the laminar tissue decurrent onto the upper side of the costae at base. Outermost lobes (second pair) acute to Boone cordate or hastate at the base on the p or side, only rarely (Croat 49172) dg basally into an additional (third) pair of lobes. Cataphylls lanceolate, 9-13(-20) cm long, drying reddish brown, persistent and becoming fibrous. Inflo- rescences up to 5-6, the peduncles 13-33(-48) cm long, purple. Spathe 7-9 cm long at anthesis, narrow, (4.5-)6-9 mm broad, slightly constricted centrally; spathe tube (3—)4—6 cm long at anthe- sis, green, purplish within, to 7-9 cm long in fruit; blade white to cream, curved forward and some- what hooded, abscising soon after anthesis. Spa- dix 3.9-6.1 cm long, stipitate basally for 2.5-6 1986] (-10) mm, the female portion 1.2-2.4 cm long, adnate to the spathe in the basal half (or appar- ently not at all in Folsom 5870); pistils more or less conical; sterile region 0.5—1.2 cm long, pink (Dressler 4884), staminodia all connate, the syn- androdia drying pale with orange chromoplasts, subprismatic, flat-topped; male portion of the spadix 2—2.7 cm long, broadest at the base or toward the middle. Spadix (male portion?) var- iously reported as white or cream (Hammel 3528, Dressler 4884, Knapp & Dressler 347 1) or purple (Hammel 2571). Fruits (in Madison 775, Sytsma et al. 4244, Churchill 3804) white, ca. 3-5 mm diam., 15-25- t Seeds ovoid, white, lon- gitudinally striate, 1-1.5 mm long, minutely brown- о Pollen (pictured in Grayum, 1984) in planar tetrads, inaperturate, coarsely foveolate or reticulate, the individual grains 24— 31 um (mean 27 um) diam., starchless, binucle- ate. Distribution. Chlorospatha croatiana ranges from the Atlantic slope of central Costa Rica, southward throughout едни and into north- western Colombia. The new species is named for Thomas B. Croat of the Missouri Botanical Garden, the fore- most authority on Central American Araceae, one of whose ample collections has been selected as the holotype. Chlorospatha croatiana can be divided at the outset into two subtaxa. Since these appear to be allopatric, they are here accorded the rank of subspecies. Chlorospatha croatiana Grayum ssp. croatiana. Figures 1, 2 The *typical" subspecies is distinguished by its consistently 5-foliolate (very rarely 7-folio- late) leaves and Central American distribution. Distribution. The first collection of ssp. croa- tiana was made in Limón Province, on the lantic slope of Costa Rica, in 1924 Tope 37356). This remains the only known collection of the species from Costa Rica, and also repre- sents the most northerly station for the grons as a whole. This st llection might for Xanthosoma pentaphyllum (Vell.) Engl. (= X. hoffmannii Schott; see Bunting, 1979), which oc- curs on the Pacific slope of Costa Rica; the lo- cality is suspicious, however, and the inclusion of the apical 2-3 cm of the clearly non-cormose caudex bearing the whitish and reddish brown GRAYUM- ARACEAE 465 cataphyll fibers characteristic of C. croatiana leaves little doubt as to the identity of the spec- imen. This subspecies is most widely distributed in Panama. It is now known from every province except Herrera, Los Santos, and Veraguas— and will certainly turn up in the latter eventually. C. croatiana ssp. croatiana has been collected on wet slopes and along creeks, mostly in cloud for- est, on either side of the continental divide, from 300 to 1,400 m (but mostly from 700 to 1,000 m). The Darién and Puerto Obaldía (San Blas) stations are so near the Colombian border that the са will undoubtedly be found there sooner or lat Flowering po of ssp. croatiana have been made in all months except November, Jan- uary, May, and September. The three known fruiting collections were made on 2 May and 22 and 26 October. Chlorospatha croatiana ssp. croatiana is dis- tinct from all other members of the genus except C. gentryi (described later in this paper) in its quinquefoliolate leaves. Only three other species of CAhlorospatha — C. kolbii, C. mirabilis (M. T. Mast.) Madison and, the recently described, C. corrugata Bogner & Madison (Bogner, 1985)— are known to possess compound or essentially compound leaves. Both C. kolbii and C. mirabi- lis were briefly in cultivation during the late nine- teenth and early twentieth centuries, and are be- lieved to have originated from northern South America (see Madison, 1981). For the better part of the past hundred years, however, these species have been known collectively from just three her- barium specimens and a few illustrations (Mad- ison, 1981). Though the possibility was consid- ered that the Panamanian entity might represent one of these forgotten species, it does not agree well with the description of either (see Engler & Krause, 1920; Madison, 1981). C hlorospatha kolbii differs from ssp. croatiana о with the female portion fully adnate to the spathe and the lower sterile flowers distinct. Chlorospatha croatiana ssp. croatiana is more similar in overall dimensions to C. mirabilis. Two previously unidentified and uncited collections of the latter species were discovered and studied 466 ing the course of the present se qud Kip 35091 (COL) and Croat 56133 (COL), both m the Chocó region of Colombia. Living ma- ew of the latter collection is presently in cul- tivation at the Missouri Botanical Garden. Yet another recent Colombian collection is repre- sented by living material at the Munich Botanical Garden and has been pictured by Bogner (1985). Chlorospatha mirabilis differs strikingly from C. croatiana in its trifoliolate, usually cream-spot- ted leaves and ei leaflets, frequently broadly confluent at the bas The Colombian Chiesa corrugata 1s equally distinctive in its trifoliolate, deeply ru- gose laminae (see Bogner, 1985). The available material of C. croatiana ssp. croatiana is somewhat variable, especially with respect to leaflet size and shape; but although occasional specimens appear distinctive in cer- tain aspects (most notably Hammel & Kress 13401, from 300 m), a considerable range of this sort of variation is evident even within single populations (see especially Luteyn & Kennedy 1804). Thus I have little doubt that the collec- tions listed hereunder all pertain to a single tax- on. Vegetatively, C. croatiana ssp. croatiana might be confused with some species of Xanthosoma with pedately compound leaves. The only one of these occurring in Panama is the widespread X. helleborifolium (Jacq.) Schott, which, unlike ssp. croatiana, is cormose and generally has nar- rower and more numerous (11-17) leaflets. Many species of Syngonium have pedately compound leaves, however all are scandent. Specimens examined. | COSTA RICA. LIMON: vicinity of Guápiles, Standley 37356 (US). PANAMA. BOCAS DEL TORO: Fortuna-Chiriquí Grande (DUKE), Croat 49172 (MO), Knapp & Dressler 3471 (MO); N of El Valle de Antón, Luteyn 3155 (DUKE); Cerro Pilón, Croat 22932 (MO). co rro Cam pana, Bartlett & Lasser 16938 (MICH, MO), Luteyn & Kennedy 1804 (DUKE), Madison 775 (HUH), Croat 22815 (MO); Río Pequeni, 10-15 min. upstream from hydrographic station by motor, Dressler 4884 (MO); ano-Cartí a (DUKE); Churchill 3804 (MO). SAN BLAS: forest SE of ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Puerto Obaldia, Croat 16818 (SCZ); headwaters of Río angandi, Hammel & de Nevers 13593 (MO). Chlorospatha croatiana ssp. enneaphylla Gray- um, ssp. nov. TYPE: Colombia. Boyacá: El Humbo, 130 mi. N of Bogotá, Lawrance 794 (holotype, K). A Chlorospatha croatiana ssp. croatiana lamina 7- 9-pedatisecta differt. A few Colombian collections exist of an entity that appears to be conspecific with Chlorospatha croatiana. Though agreeing with the Central merican collections of ssp. croatiana in floral morphology and most vegetative characters, it differs consistently in having more leaflets. All three Colombian specimens with mature leaves are 7-9-foliolate, whereas only a single Central American specimen of the 23 examined exhib- ited even as many as seven leaflets. It seems appropriate, then, that the Colombian entity should receive taxonomic recognition. Distribution. Chlorospatha croatiana ssp. enneaphylla is known only northwestern olombia, from the departments of Antioquia and Chocó and the western panhandle of Boyacá. Recorded elevations range from 350 to 600 m. The only dated flowering collection (Shepherd 899) is from 5 August. This taxon may be distinguished from vege- tatively similar species of Xanthosoma inhabit- ing the region by its caulescent, rather than cor- mose, habit and its slender and delicate peduncles and inflorescences. The two collections from Chocó are of sterile and presumably juvenile material; they are placed here speculatively, since they are atypical in leaf size and morphology Specimens examined. COLOMBIA. ANTIOQUIA: Rio Anorí M a ba en BOYACÁ: El Humbo Lawrance (K). осо: between Medellin and Quibdó, tte 49311. 55930 (MO). Chlorospatha hammeliana Grayum & Croat, sp. nov. TYPE: Panama. Coclé: just N of sawmill above El Copé, ca. 1,000 m, Hammel & Kress 13465 (holotype, MO). Figures 3-6. Planta terrestris; caulis saltem ad 8 cm longus; pe- tiolus 40-65 c cm latus; lobus posticus (9-)14-18 cm longus. Inflorescentiae 2-3; pedunculus 25-40 cm longus; spa- 1986] GRAYUM —ARACEAE 467 FIGURES 1-4.— 1, spatha hammeliana Grayum & Croat, sp 2. Chlorospatha croatiana Grayum, sp. nov., . nov., Croat 44589 (th ssp. croatiana, Croat 22815.—3, 4. Chloro- is is a greenhouse-grown a the leaves are disproportionately small as compared with wild-collected material). Photos by Thomas B. Cro tha angusta, 9-12 cm longa; tubus intus purpureus; spathae lamina alba vel subviridis; spadix 5.4—7.5 cm о быз yes alba, aliquanto clavata, 1.6-2 cm longa. Terrestrial herb; stem appearing erect, 1-1.5 cm thick when dried, to at least 8 cm tall. Petioles 40-65 cm long, terete, somewhat spongy (i.e., easily compressible), glabrous, vaginate only in the basal one-sixth to one-fourth. Lamina of flowering individuals glabrous, simple, ovate, prominently cordate to sagittate basally, the an- terior lobe 23-35 cm long, 16-27 cm broad, the posterior lobes (9-)14-18 cm long (measured along main vein). Lamina drying membrana- ceous and strongly bicolored (much paler below, with dark primary and secondary veins). Pos- terior costae denudate for 0.5-1.5 cm. Inflores- cences 2-3, the peduncle 25-40 cm long. Spathe 9-12 cm long, narrow, constricted at summit of tube. Spathe tube greenish externally, purplish within; spathe blade white or greenish white, ab- scising after anthesis. Spadix 5.4-7.5 cm long; female portion 2.4—3.5 cm long, adnate to spadix 468 only in the basal one-fourth to one-third; female иен pale, more or less conical, lois 2mm Jong, as the base oftl the ovary; stigmas yellowish; sterile part of spadix 1.4-2.2 cm long; synandrodia co- lumnar or irregularly prismatic, 0.7-2.2 mm diam., more widely spaced than the fertile male flowers, whitish with reddish chromoplasts, the basal ones sometimes 4—6-lobed or divided. Fer- tile male region ca. 1.6-2 cm long, white, some- what clavate; synandria prismatic, 4—5-thecate, truncate at apex, closely spaced. Pollen in planar tetrads, inaperturate, minutely rugulate or ver- ruculate, the individual grains 23-25 um (mean 24 um) diam., starchless, binucleate. Distribution. Chlorospatha hammeliana is known only from two isolated areas astride the Continental Divide in west-central Panama. At the type locality, near El Copé in Coclé Province, it occurs on steep slopes in very wet cloud forest on the Atlantic slope at 900-1,000 m. Two other sterile collections are from the area of Cerro Tute, near Santa Fe, Veraguas Province, at 700 m on the Atlantic slope and 1,050-1,150 m on the Pacific The single wild flowering collection (the ho- lotype) was made on 25 August. (Croat 44589 also includes fertile material, however the orig- inal wild collection was sterile; a live voucher flowered later in the greenhouse at the Missouri Botanical Garden — see Figs. 3—6 —and was added to the collection. Chlorospatha hammeliana must be consider- ably rarer than C. croatiana, or perhaps it occurs in a more inaccessible habitat. Both species grow in the vicinity of the sawmill above El Copé, a popular collecting locality; but although C. croa- tiana seems to have been collected by —_— every botanist who has visited = site, C. ham meliana has been found but tw The new species is dedicated ш S E. Ham- mel, now of the Missouri Botanical Garden, ar- dent and perspicacious collector and monogra- pher of the flora of southern Central America, who first brought this plant to our attention. Chlorospatha hammeliana is perhaps most similar to the Ecuadorian C. besseae Madison (1981; validated in Selbyana 7: 353. 1984, and again by Bogner, 1985), which likewise has sim- ple, cordate leaves. That species, however, has shorter petioles and peduncles, smaller laminae, a smaller inflorescence, and orange male flowers. Furthermore, the female portion of the spadix ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 in C. besseae is said to be completely adnate to the spathe Specimens examined. AMA. COCLÉ: N of saw- mill above ong trail to top of Cerro Tute, Croat 48895 (МО), along Santa Fe-Ca- lovébora rd., Croat 48993 (MO). During the course of the present investigation, yet another undescribed species of Chlorospatha was discovered among the collections of the Mis- souri Botanical Garden. This particularly dis- tinctive species, from the Cordillera Occidental of northern Colombia, is described below. Though the male portion of the spadix (hence, the pollen) is unknown, the specimens conform in all other important respects to the most recent circumscription of Chlorospatha (Madison, 1981), to which the new species is therefore as- signed without hesitation. Chlorospatha gentryi Grayum, sp. nov. TYPE: Colombia. Antioquia: trail from Encarna- ción to Parate Nacional de los Orchideas, W slope of Cordillera Occidental, 1,600- 1,800 m, Gentry & Renteria A. 24585 (ho- lotype, MO-2715461; isotypes, COL, HUA). Figure 7. Planta terrestris; caulis erectus saltem ad 10 cm lon- lateralis, ellipticum, 6.5-9.5 cm lon segmenta lateralia valde inaequilateralia. Inflorescen- tiae 2-3; pedunculus 6.5-9 cm longus, furfuraceus praesertim versus apicem; tubus spathae 2-2.5 cm lon- gus, non nisi dorso extrinsecus dense furfuraceus, vi- ridis, tenuibis longitudinalibus ei Era lamina alba; spadix pars feminea ca. m longa, sessilis, in dimidio inferiore spathae he spadix pars mascula ignota. Terrestrial herb; stem erect, to at least 10 cm tall, 6-9 mm thick when dried. Petioles 12-20 cm long, slender, vaginate in basal one-fourth to one-third, clasping the stem, scurfy-pubescent with more or less flattened, branched, multicel- lular hairs in the apical one-half to one-fourth. Lamina pedately 5-foliolate, the margins cris- pate-undulate; lamina glabrous above, paler be- low and crispy-puberulent along the veins of all orders; costae naked or obscurely alate, with an indumentum like that of the upper petioles. Cen- tral lobe of lamina symmetrical or nearly so, el- liptical, 6.5-9.5 cm long, 2.5-4 cm broad; lateral 1986] GRAYUM —ARACEAE 469 d FIGURES 5-7.— 5, 6. Chlorospatha hammeliana Grayum & Croat, sp. nov., Croat 44589. Photos by Thomas B. Croat. —7. Chlorospatha gentryi Grayum, sp. nov., Gentry & Rentería 24585 (Type). Photo by Alwyn H. Gentry 470 lobes strongly asymmetrical, progressively smaller; all lobes acute at the base. Inflorescences 2-3, the peduncles 6.5-9 cm long, more or less densely crispy- -puberulent especially in the apical part, t ding in a dense patch onto the dorsal surface of the spathe tube; spathe tube 2-2.5 cm long at anthesis, ca. 5 mm diam., greenish, finely longitudinally many-nerved; blade whitish, abscising soon after anthesis. Fe- male part of spadix ca. 1.9 cm long, sessile (1.e., not stipitate), adnate to the spathe in the basal half. Male part of spadix unknown. Distribution. To date, Chlorospatha gentryi is known only from the holotype, collected at 1,600-1,800 m on the western slope of the Cor- dillera Occidental in Antioquia Department, Co- lombia. Chlorospatha gentryi, in its small, pedately- compound leaves with crispate-undulate mar- gins, immediately recalls the elusive C. kolbii. It differs from that species, however, not only in its 5- rather than 7—9-pedatisect leaves, but also in its even shorter petioles and peduncles, non- stipitate spadix and crispy-puberulent indumen- tum (Engler & Krause, 1920, described the lam- ina of C. kolbii as “subholosericea,” but only on the upper surface). e new species is named for the collector of the holotype specimen, Alwyn H. Gentry of the Missouri Botanical Garden, one of the most ac- tive and authoritative field botanists working in the neotropics and a specialist on the flora of Pacific South America. Specimens examined. COLOMBIA. ANTIOQUIA: trail from Encarnación to Parque Nacional de las Orchi- deas, W slope of Cordillera Occidental, Gentry & Ren- tería A. 24585 (MO, COL, HUA). Inasmuch as the present paper significantly augments the number of compound-leaved Chlorospatha species known, a key to this assem- lage is provided below: KEY TO THE SPECIES OF CHLOROSPATHA H COMPOUND LEAVES la. Lamina trifoliolate, the leaflets narrowly to broadly confluent basally 2a. Lamina deeply rugose, concolorous; lat- eral lobes nearly equaling median lobe; petiole less than ни cm lon C.c rugata Bogner & Madison . Lamina relatively ned y spot ted with yellow or cream; median lobe nearly DUE as long as lateral lobes; petiole more than 50 cm lon MES C. mirabilis (M. T. Mast.) Madison N c ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 16. Lamina 5-9- denis ш leaflets narrowly confluent ee e base or not at а]... 3a 15 cm long and 5 cm broad; lea plane; petioles more than 33 cm lon 4a. Lamina 5-(rarely 7- )pedatisect; southern Central America .. Chlorospatha croatiana Grayum SSp. cro oatiana 4b. Lamina 7-9-pedatisect; Colom Chlorospatha ыша enneaphylla Grayum . Lamina small, the о lobe less than 15 cm long and 5 cm broad; leaf margins crispate-undulate; petioles less than 33 m long 5a. Lamina 5-pedatisect; petioles less than 25cm long, peduncles less than o с spathe tube crispy-puberulent; spa- dix sessile a Chlorospatha gentryi Grayum . Lamina 7-9-pedatisect; petioles u с crispy-puberulent; spadix m C CE pared Engl. With the addition of Chlorospatha corrugata Bogner & Madison and the three species just described to Madison’s (1981) previous total of 11, the number of species now attributable to this suddenly burgeoning genus stands at 15. This represents nearly a fourfold increase in less than a decade (cf. Madison, 1978), occasioned by the installation of a clearly articulated generic cir- cumscription coincident with the exploration of rich new habitats. Judging from the present rate of discovery and the number of intriguing sterile specimens filed away in herbaria, it seems likely that Chlorospatha, which lay dormant for nearly a hundred years, will eventually come to com- prise as many as 20-25 species. The Chocó re- gion of Colombia, which appears to be a strong- hold of the genus, is one of the most botanically diverse and least-known areas remaining on the planet (Gentry, 1982), and can be s to yield many interesting new species olbii, which, somewhat perplexingly, continues to elude rediscovery. XANTHOSOMA Over the past ten years, generic concepts in the tribe Caladieae have been arduously tested, mainly due to a major influx of specimens, com- prising many new species, from the Andean re- 1986] gions of Colombia, Ecuador, and Peru. Chloro- spatha, not even recognized in its present circumscription a decade ago, has ironically emerged as perhaps the most distinctive of the three polytypic genera. It is well characterized by its caulescent habit, slender inflorescences with the female portion of the spadix usually fused with the spathe, widely spaced female flowers with thin, deliquescent, mantle-like styles dotted with red chromoplasts, oddly shaped synandro- dia and small, starchless pollen borne in tetrads (Madison, 1981; Grayum, 1985) The definitions of Caladium and Xanthosoma have, on the other hand, become more nebulous over the past ten years. In 1976, Michael Mad- ison, who has made a specialty of this group and significantly clarified its taxonomy, published a list of six mainly vegetative and habital criteria that, he believed, could be used to distinguish Caladium from Xanthosoma. In all of these at- tributes— plant size (less than 0.5 m tall), habitat (not weedy), stem morphology (a globose tuber), peduncle length (relatively long), inflorescence number (solitary), and petiole attachment (pel- tate)—the species described below would fall more or less clearly into Caladium. In actual fact, however, several species of Xanthosoma were already known that violated one or more of these rules, and, in subsequent years, numerous ad- ditional aberrant species were described. The distinctions between the two genera finally be- came so blurred that, by 1981, Madison was forced to admit that “the single inflexible char- acter separating them is the shedding of pollen in tetrads in Xanthosoma and in solitary grains in Caladium." With the discovery of the following species, unique in violating all six ofthe above guidelines, the beleaguered generic concepts of Xanthosoma and Caladium have been subjected to their most formidable challenge yet—from which they emerge, I believe, unscathed, vindicated, and even reinforced. Xanthosoma caladioides Grayum, sp. nov. TYPE: ma. Comarca de San Blas, coast NE of Puerto Obaldia towards Colombian border, 8°40'N, 77°25'W, sea level, Knapp & Mallet 4687 (holotype, MO-2982678). Figures 8, 9. Planta terrestris vel epilithica, cormifera; petiolus longus, 9-25 cm latus; lobus posticus 3.9-12.5 cm lon- gus, rotundatus; distantia insertione petioli sinui la- GRAYUM— ARACEAE 471 minae 2.7-5.5 cm. Inflorescentia solitaria; pedunculus (10.5-)12.5-33 cm longus; spatha sub anthesi 6. - 11 purpureus; spathae lamina (3.3-)4.3-6 a 5.3-9.8 cm longa, sessilis; pars feminea 1.6-2.9 m longa, spathae vix adnatae; styli lati, discoidei, pa- uli spadix pars sterili 1.4-2.7 cm longa, inferne di- la 2 .6 cm lon- ga, exe m diametro, alba. Semina ca. 40—50, brunnea, tenuiter dud rinata, ovoidea, O. st mm longa. Pol- linis granula quaternatim aggrega Terrestrial or epipetric, acaulescent, cormose herb, ca. 25-50 cm tall. Corm subglobose, 1.5— 2.5 cm diam. when dried. Leaves and “stems” (presumably petioles) а о (Knapp & Mallet 4687); petioles 17-58 cm long, vaginate in basal one-eighth to one-third, peltately attached. Lam- ina of adult leaf glabrous, drying membrana- ceous, glaucous below, apparently plain green above (Allen 4629 appears to have darker mark- ings, but the label is uninformative), simple and ovate, cordately lobed at the base, asymmetri- cally acute apically, the margins entire; anterior lobe (measured from insertion of petiole) 11-25 cm long, 9-25 cm broad, the posterior lobes (measured along main vein from insertion of pet- iole) 3.9-12.5 cm long, rounded; distance from petiole insertion to base of sinus 2.7—5.5 cm, the sinus more or less acute. Inflorescences 1 or 2 per plant, solitary in the sheaths, the peduncles (10.5-)12.5-33 cm long. Spathe 6.5-11 cm long at anthesis, constricted centrally; spathe tube 2.5-5 cm long at anthesis, ca. 1. road when pressed, green externally, purs within, to ca. 6 cm long in fruit; blade (3.3-)4.3-6.3 cm long, white, abscising after anthesis. Spadix 5.3- 9.8 cm long, slightly shorter than the spathe, ses- sile, the female portion 1.6-2.9 cm long, adnate to the spathe for 5 mm or less; pistils broadly conical, surmounted by a broad, spreading, dis- coid style, the ovary apparently with 3 partitions; sterile region of spadix 1.4—3.5 cm long, the basal third thicker than the female region, with the flowers bulbously enlarged, the apical two-thirds stipitate, narrower than the male region, the flowers vertically elongated and flattened; male portion of the spadix 2.7-4.6 cm long, widest in middle, ca. 4-7 mm broad when pressed, white; synandria flat-topped, roughly polygonal, with the margins sinuate, 0.5-2.2 mm in greatest diam., the anthers ca. 1.5 mm long and some- what curved, dehiscing by apical pores. Fruits (in Stern et al. 444) са. 3—3.3 mm diam. and whitish, when dried; seeds ca. 0.8-1 mm long, ovoid, brown, with 12-14 narrow, longitudinal keels; at least 45 counted in one intact fruit (Bristan 107 4). 472 Distribution. Xanthosoma caladioides was first collected by Paul Allen in 1947 from Darién rovince, Panama. It is still known from only eight collections, all from the Province of Darién the Comarca of San Blas, in easternmost Panama. The three San Blas collections are all from near or at sea level, from near the Colom- bian border west to the vicinity of Playón Chico. Two of the Darién collections are from below 100 m in the valley of the Río Tuira, but Allen 4629 was collected at 500—700 ft. on the inland slope of the Atlantic coastal range, and Bristan 1074 appears to have come from the same gen- eral area. Hammel 7314, on the other hand, is from about 500 ft. in the remote Serranía del Sapo, fronting the Pacific coast. Thus, though little collected in this poorly explored end of Pan- ama, X. caladioides is clearly widespread there. It certainly ranges into adjacent Colombia as well, at least along the Urabá coast, and perhaps into the Río Atrato basin, but no definite South American collections have been seen. A recent sterile collection (Hammel 14162, MO) of an unknown peltate-leaved aroid from primary for- est on limestone at about 600 m in southwestern Costa Rica may also belong here. Xanthosoma caladioides apparently may be either terrestrial or epipetric. The holotype, from San Blas, was collected in “tropical dry forest” with “strong sea winds”; the other two San Blas collections were from *'coastal rocks . . . near the sea" (D'Arcy 13692) and in sandy soil, in deep shade in secondary growth (Stier 34). Two of the Darién collections were from along streambanks (“very common on steep, shaded banks,” A//en 4629; **on slopes along stream," Hammel 7314), whereas Stern et al. 444 was “growing on a ver- tical stone wall." All of the recognizable collections of Xantho- soma caladioides are fertile, and all were made between 18 April and 30 June—the later collec- tions representing mostly fruiting material. This period marks the onset of the rainy season in most of Panama (Foster, 1985). It is conceivable that X. caladioides passes much of the dry season in a leafless condition, as do several other species of the genus (pers. observ.). No information is available on pollination of Xanthosoma caladioides. 1 have noted that in- florescences at the anther-dehiscent stage (e.g., Allen 4629, Stier 34) tend to be lacking the en- larged basal sterile flowers, suggesting that these may serve as food-rewards for the pollinators (presumably dynastine scarab beetles). ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Although one would not have anticipated the existence of any ethnobotanical information for such an obscure species, the label of Stier 34 records the following: “kuidar [presumably a common name, in some unspecified indigenous language] — root used against fever." The new species resembles the familiar, cul- tivated species Caladium bicolor (Ait.) Vent. so strikingly — at е оп herbarium sheets— that it ined repeatedly by several leading aroid е а including Madison himself. Vegetatively, the two species are vir- tually indistinguishable from dried material— though the posterior laminar lobes of X. cala- dioides are generally somewhat smaller and less divergent, and presumably the leaves are less frequently and/or less extensively variegated, if at all. (Regrettably, no plant of this species has ever been collected or, to my knowledge, even beheld alive by any aroid specialist, and none of the specimens is accompanied by fully adequate notes.) The resemblance also extends to the ex- ternal aspect of the inflorescence. Careful study of the inflorescence morphology, however, clearly reveals not only that the new species is not Caladium bicolor, but that it be- longs unequivocally in the genus Xanthosoma. The first indication of this was provided by an examination of pollen under a compound light- microscope: pollen from all four polleniferous collections (Stier 34, Allen 4629, Stern et al. 444, Hammel 7314) was observed exclusively in tet- rads. Two other technical, floral characters serve to distinguish Caladium and Xanthosoma in the vast majority of cases, and in both of these, the new species clearly agrees with the latter genus. First, broad, discoid, spreading styles, charac- teristic of Xanthosoma but lacking in Caladium, intact fruit of X. caladioides (from Bristan 1074)—considerably above the maximum of 20 tolerated for Caladium. The multiovulate nature of Xanthosoma ovaries, incidentally, is probably closely correlated with the retention of pollen in permanent tetrads (Madison, 1981). Two additional floral characters, though tax- onomically insignificant at the generic level, may be of practical aid in distinguishing specimens of Caladium bicolor an anthosoma cala- dioides: the synandria of the former species are deeply notched marginally, whereas those of the latter are merely sinuate; and, the sterile region in Caladium bicolor is not so sharply differen- 1986] Ficures 8-10. GRAYUM—ARACEAE —8, 9. Xanthosoma caladioides Grayum, sp. —8. Hammel 7314.—9. Knapp & Mallet 4687 (Type), inflorescence.— 10. Caladium bicolor (Ait.) Vent., ps 640 (MO), Venezuela, inflorescence. tiated into a narrowly stipitate apical portion and a bulbous basal portion, though there is a slight tendency in this regard (Figs. 9, 10). Xanthosoma caladioides is not the first species of Xanthosoma known to have peltate leaves. The precedent was set a decade ago with the description of X. peltatum Bunting (1975) from Venezuela, about which the author later re- marked: “el encuentro de esta especie andina (Bunting, concept thus altered, the new species is neatly accommodated in Xanthosoma. It does not, however, appear closely related to X. peltatum— a much larger and coarser, arborescent plant of higher elevations—and one must suppose inde- pendent derivations of peltate leaves in these two species. The general aspect and inflorescence structure of Xanthosoma caladioides appear to ally it more closely with X. mexicanum Liebm. and related species, intractable violators of Mad- ison's six criteria, from which it may be readily distinguished by its glabrous foliage, in addition to its peltate leaves. Specimens examined. PANAMA. DARIÉN: Río Tu- Pucuro, Duke 1314 1 oa Rio San Antonio, up- stream G hi mel 7314 (M О) sa SAN BLAS: Playón Chico, Molia, cds "M (MO); coastal rocks be- tween Puerto Obaldía and Puerto Armila, D’Arcy 13692 (MO); coast NE of Puerto Obaldía, Knapp & Mallet 4687 (MO) LITERATURE CITED ANDRÉ, E. 1872. Phyllotaenium lindeni. Ill. Hort. 19: 3-5. BOGNER, J. 1985. A new Chlorospatha species from Colombia. Aroideana 8: 48-54. BuNriNG, G. S. 75. Nuevas especies para la revi- sión de las Araceas venezolanas. Acta Bot. Venez. 11: 263-335. 1979. Sinopsis de las Araceae de Venezuela. 1 3- 132 inA. ‘Engler (editor, Das at 71: (IV c W. Engelmann, Leipzig. FOSTER, 3 985. Plant seasonality in the forests of else p. 255-262 in W. G. D'Arcy & M. D. Correa A. йу өе | Тһе Botany and National His- Pig rindi gree patterns as evidence for a Choco refuge. Pp. 112-136 in G. T.P ), Biological D th Tropics. Columbia Univ. Press, New Yor GRAYUM, M. H. 1984. Palynology and Phylogeny of the Araceae. Unpubl. Ph.D. Thesis. Univ. of Mas- sachusetts, Amherst. 85. Evolutionary and ecological signifi- cance of starch storage in D of the Araceae. Amer. J. Bot. 72: 1565-1577 474 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 . 1986. Phylogenetic implications of pollen nu- clear number in the Araceae. Plant Syst. Evol. 151: ME MADISON, 976. Luctatio aroideis I. Caladium ind Xanihoon ma. Phytologia 35: 103-107. A synopsis of Caladiopsis. Contr. Gray Herb. 2s 95-98. 1981. Notes on Caladium (Araceae) and its allies. Selbyana 5: 342-377 NICOLSON, D. 1978. Araceae. Pp. 345-348 in A. C. Smith, A precursor to a new flora of Fiji. Allertonia 1: 331-414. NOTE ADDED IN PROOF: Chlorospatha croatiana ssp. croatiana has recently been recollected in Costa Rica: HEREDIA: forest between Río Peje Aoi Río Sar- dinalito, Atlantic slope of Volcán Barva, 10?17.5'N, 84°04.5'W, 700-750 m, Grayum 6657 (MO). NEW TAXA IN OENOTHERA (ONAGRACEAE)! WARREN L. WAGNER? ABSTRACT new aia three new sections, and one new subsection of Oenothera are described and a new An combination scribed. T O. flava is made here so that the names — erum, veh Bru from north-central Texas to the southern part of the pan on O. flava subsp. taraxacoides (sect. Lavauxia) is made for large-flowered, will be available. Oenothera coryi (sect. nhandle, is com modas outcrosing populations of O. flava from three iy a areas: 1) Mogollon Plateau, Arizona rn New to Catron Mexico; ., New Mexico; 2) Sacramento M Madre ac idental from Chihuahua to а га As part of a genus-wide ге- and ierr: evaluation of the infrageneric classification, three new sections, sect. Ravenia, sect. Eremia, and sect. Contortae, are proposed for spe ognized, and su sect. Lavauxia A number of concurrent studies of Oenothera are in progress, including studies of flavonoids, seed anatomy and morphology, and pollen mor- phology. dn new names are published here in order to e them available prior to the publication a ail revisions of the sections in which they occur. With the publication of this paper and an upcoming paper on the systematics of subsect. Raimannia and a related new sub- section (Dietrich et al., submitted) all taxa in the genus will have been published that have been recognized in a series of detailed systematic stud- ies initiated in the mid-1960s by Peter Raven. Detailed discussion ofthe new taxa and one com- bination presented here will be made in the re- spective sectional revisions. OENOTHERA SECT. MEGAPTERIUM This section consists of four species. The most o is Oenothera macrocarpa Nutt., a morphic species of the Great Plains that is ikd into four subspecies (Wagner, 1983). The other three species were previously included > actually consists of three allopatric species: О. brachycarpa, a diploid (n = 7) from western Tex- as to southeastern Arizona and northern Mexico; O. howardii (A. Nels.) W. L. Wagner, consisting of tetraploid, hexaploid, and octoploid plants (п = 14, 21, 28) from eastern Colorado, Utah, and eastern Nevada (Wagner, 1983); and O. coryi, cies formerly included in subg. Pachylophus, which is no longer rec- ubsect. Australis is described for the two white-flowered South American species of described here, a tetraploid occurring from north- central Texas to the southern Texas Panhandle. Oenothera coryi W. L. Wagner, sp. nov. TYPE: U. TEXAS: Taylor Co., eroded red clay and, Camp Barkeley, 1 July 1943, W. L. Tolstead 7537 (MO-1266818, holotype; BH 2 sheets, GH, MICH, NEB 2 sheets, NY 3 sheets, SMU, TEX, isotypes). ial pidio As var. jg din sensu Munz = ma- except the type), A . J. Bot : 368. 1930, non n (as to tipi ae Folia linearia ad angusto lanceolata margine integro vel in dimidio inferiore remote pinnatilobato; flores vix odori, petalis flavis mox dilute aurantis; capsula abrupte ad rostrum apicale producta, valva quaque ala marginali 4—6 mm lata praedita Acaulescent or caulescent and caespitose pe- rennial herbs from a stout woody taproot with a usually branched caudex sometimes producing clusters of rosettes 10—60 cm across. Leaves lin- ear to narrowly lanceolate, 5-16 cm long, (2-)3- 5(-7) mm wide, densely strigillose, apex long- attenuate, acute to rounded, margin entire to the lower half remotely pinnately-lobed. Flowers 1- 3, rarely more, opening near sunset, weakly scented. Buds with free sepal tips 0.7-1.2 mm ong. Floral tube (5.5—)7.5-10(-12.5) cm long. Petals yellow, fading orange, drying lavender to purple, broadly obovate, 3.5-4.3 cm long, 3.7- 4.2 cm wide, sometimes with a terminal tooth ca. 2 mm long. Capsule ovoid in outline, 2.5-3 cm long, usually abruptly constricted to an apical ! This Daum is based upon research supported by a grant to Peter H. Raven from the National Science Foundatio poesi AN of Botany, Bernice P. Bishop Museum, P.O. Box 19000-A, Honolulu, Hawai'i 96817. ANN. MISSOURI Bor. GARD. 73: 475—480. 1986. 476 beak, each valve with a marginal wing 4-6 mm wide. Seeds in 2 distinct rows per locule, often reduced to one row near the apex, or occasionally only one row throughout, obovoid to subcuboid, angled, dark brown, 2.5-4 mm long, the surface corky and coarsely rugose to furrowed especially on the abaxial surface toward the proximal end, the distal end with a thick ridge on the abaxial and lateral sides of the seed; the testa very thick especially at the distal end, this area with a con- spicuous internal cavity adjacent to the embryo. Self-incompatible. Mitotic chromosome num- ber, 2n = 42 Distribution. Open grassland or о areas such as roadcuts from ох, Throckmorton, Nolan, Taylor m ОШ counties in north-central Texas and Crosby and Garza counties in the Texas Panhandle, 350—920 m. Flowering in April and May. Oenothera coryi is named in honor of V. L. Cory (1880-1964), who recognized this plant as 1 it. This new brachycarpa var. typica (1930). The type of O. brachycarpa A. Gray [Pl. Wright. 1: 70. 1852. TYPE: Between present-day Val Verde and El Paso counties, Texas, July to November 1849, e Wright s.n. (GH, lectotype, a MO, TEX US, isolectotype; Munz, Amer. J. Bot. 17 : 368. 1930).] is not from а Texas as Munz thought, but, based on Charles Wright’s itinerary for 1849 (McKelvey, 1955), it appears to have been collected in southwestern Texas between Val Verde and El Paso counties. Thus the plants from west-central Texas need a name. Appar- ently Munz, who did not know Wright’s route at the time, placed the type specimen with others from west-central Texas because of its narrow leaves, although he commented (1930, p. 368) that the capsule wings of the type of O. brachy- carpa were narrower than all other collections that he placed in his var. typica. Narrow wings are typical of O. brachycarpa. 1 am describing this entity endemic to west-central Texas as Oe- nothera coryi. Not only is its range completely separate from that of O. howardii to the north and west in Colorado, Utah, and Nevada, and that of O. brachycarpa to the south, in western Texas to southeastern Arizona and northern Mexico, but it is also morphologically distinct. sistently oblanceolate, elliptic, or sometimes lan- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 ceolate leaves, strongly scented flowers with a sweet odor, free sepal tips 1-3(-4) mm long in bud, plants usually at least sparsely hirsute, bril- liant yellow and usually larger petals (3-)4-7.3 cm long that fade deep red and dry deep reddish purple or reddish brown, and its capsules that gradually taper to a sterile beak. Oenothera brachycarpa differs from O. coryi in its grayish green leaves usually with a terminal lobe, longer floral tube (9-)12-21(-22) cm long, longer free sepal tips 1-7 mm long, broadly rhombic-ob- ovate petals, and presence of hirsute as well as strigillose pubescence. OENOTHERA SECTS. RAVENIA, EREMIA, CONTORTAE AND PACHYLOPHUS In the most recent evaluation of Oenothera subg. Pachylophus (Raven, 1970) eight species were included, O. muelleri Munz, O. tubifera Sér., O. xylocarpa Cov., O. primiveris A. Gray, caespitosa Nutt., O. cavernae Munz, O. bran- degeei (Munz) Raven, and O. macrosceles A. Gray. In that paper these species were tentatively divided into four groups: O. muelleri and O. tu- bifera with white flowers and stout, obtusely an- gled, nontuberculate capsules; O. UM and риетиует миа yCIIOW cu angled, кулан capsules; О. н лий О. cavernae, апа О. brandegeei with white flow- ers and stout, tuberculate capsules; and O. mac- rosceles with yellow flowers and slender, quad- rangular, nontuberculate capsules. Raven (1970) suggested that these eight species were related and could be considered as constituting one somewhat heterogeneous subgenus or alterna- tively the four subgroups could be recognized as distinct sections. Recent study of crossing relationships, vege- tative, capsule, and seed morphology (Stubbe & Raven, 1979; Dietrich et al., 1985) suggests that O. macrosceles is best grouped with O. maysil- lesii Munz, O. stubbei Dietrich, Raven & W. L. Wagner, and O. organensis Munz in sect. Oe- nothera subsect. Emersonia (Munz) Dietrich, Raven & W. L. Wagner. The other species placed in subg. Pachylophus fall into four crossing groups whose members cannot be successfully hybrid- ized with species of a different group (Wagner, 970): 1 cutely d O. cavernae (Raven, 1970; Stockhouse, 1973; Wagner et al., 1985). Morphology, especially of 1986] capsules and seeds, and a cladistic analysis sup- port the four crossing groups (Wagner et al., 1985; Wagner, unpubl.). Anatomical evidence from a genus-wide study of seeds (Tobe et al., in press) showed that each of the crossing groups has dis- tinctive anatomical features, although O. xylo- carpa is very similar to O. maysillesii and O. primiveris. The O. caespitosa group and the O. muelleri group have conspicuous, unique, de- rived features. In view of the narrow sectional concept established by Lewis and Lewis (1955) for Clarkia and subsequently followed in all sys- tematic studies of the family, these four groups of Oenothera should be recognized as sections. Three of the four sections are described here as new, and sect. Pachylophus is now delimited to include O. caespitosa and four related species (Wagner et al., 1985). Subgenus Pachylophus is no longer recognized. With these adjustments f Oenothera is divided into 14 sections, each o which is composed of very closely related species that share numerous morphological features, an- atomical features, and, based on crossing studies, js дан Lied and usually similar plastomes. At present no subgenera d, but wor in progress (Wagner & Raven, ‚ unpubl. ) sum- marizing all available data for the genus suggests that it may be possible to subdivide it into two subgenera. x © I. Oenothera L. sect. Ravenia W. L. Wagner, sect. nov. TYPE: Oenothera muelleri Munz. тг subg. Pachylophus sensu Munz, А тег. J. Bot. 18: 728. 1931, pro parte; N. Amer. Fl. II. 5: 98. 1965, pro parte. O h bg. Rai ] Munz, N. Amer. Fl. II. 5: 104. 1965, pro parte. Plantae perennes hirsutae strigillosaeque vel sub- glabrae; tubus floralis recurvus; petala alba; capsula oblongo- lanceolata angulis acutis ad rotundatis; sem- ina respectu morphologia saepe irregularia 3-7 mm M supervicie adaxali costis longitudinalibus prae- Fleshy-leaved perennial herbs; stems several, arising from the rosette, decumbent to ascending. Pubescence of two types: hirsute, the hairs usu- ally with a purple pustulate base and strigillose, the leaves sometimes glabrous. Buds curved downward by the recurved floral tube. Petals white. Capsules oblong-lanceoloid, somewhat curved, the angles acute to rounded, sessile. Seeds in 1 or 2 rows per locule, basically obovoid to oblong or oblanceoloid, often somewhat irreg- ular, 3-7 mm long, the abaxial surface with lon- WAGNER—OENOTHERA 477 gitudinal ribs; the testa very thick above the raphe and at the distal end, the thickened area with a cavity not visible externally or rarely appearing as a distal pore and/or a raphial groove. Self- incompatible (O. muelleri) or self-compatible (O. tubifera), outcrossing or modally Merian p respectively. Basic chromosome number, x = 7 The two species of this section, Oenothera muelleri and O. tubifera, have disjunct and pre- sumably a ranges in Mexico from 2,300 to 2, m adrean woodland vegetation. Oenothera muelleri occurs in Nuevo Leon, Ta- maulipas, and Coahuila, and O. tubifera occurs further south and west in the states of Hidalgo, Puebla, Guerrero, and Durango. This section ap- pears to represent a lineage that diverged early in the evolution of the genus from ancestors sim- ilar to sect. Oenothera subsect. Emersonia (Ra- ven, 1970; Stubbe & Raven, 1979; Wagner et al., 1985) and it is the sister group to sects. Eremia, Contortae, and Pachylophus (Raven, 1970; Wag- ner et al., 1985). The sectional name is intended to honor Peter H. Raven, who initiated this series of revisionary works on Oenothera over 25 years ago and who first placed the species of sect. Ra- venia together as close relatives in 1970. II. Oenothera L. sect. Eremia W. L. Wagner, sect. nov. TYPE: Oenothera primiveris A. Gray. dais iius ч Pachylophus sensu Munz, Amer. : Bot. 18: 728. E pro parte; N. Amer. FI. II. 98. 1965, pro part Plantae annuae hirsutae ec res et glanduloso- puberulae a flava; capsula fere lanceola ta ad hin га angulis pow semina obo- voidea ad oblanceolata, 3-3.5 mm longa crasse rugosa, superficie apu sulco conspicuo area crassa cum for- ma U circumcin Winter annual herbs, acaulescent or caules- cent; stems when present usually simple, but oc- casionally with secondary branches arising from ‚1 hairs usually with a reddish purple pustulate biis: strigillose; and glandular puberulent. Buds curved downward by the recurved floral tube. Petals deep yellow. Capsule lanceoloid to ovoid, falcate or curved to nearly straight, the angle acute, sessile. Seeds in 2 rows per locule, obovoid to oblan- ceoloid, 3-3.5 mm long, coarsely rugose on the distal half of the abaxial side, the raphial face 478 with a conspicuous groove surrounded by a U-shaped, thickened area terminating at a pore near the distal end, the entire surface papillose, epressed; the testa very thic above: the raphe and at the distal end, the thick- ened area with a cavity that appears as a pore near the distal end and a groove along the raphial face. Self-compatible, rarely self-incompatible, outcrossing to autogamous. Basic chromosome 7. r, Oenothera primiveris, the only species of this section, is scattered or occasionally locally com- mon in the Mojave, Sonoran, and Chihuahuan deserts, 30—1,600 m, in sandy soils in low desert to mountain foothills. Based on a ie anal- ysis, especially utilizing seed morphology, sect. Eremia appears to be most closely a to sect. Pachylophus (Wagner et al., 1985), a rela- tionship supported by the recent analysis of seed coat anatomy (Tobe et al., in press). The sectional name is the Greek word eremia, meaning desert, in reference to the restriction of this section to the North American deserts. III. Oenothera L. sect. Contortae W. L. Wagner, sect. nov. TYPE: Oenothera xylocarpa Cov. — deii Pachylophus sensu Munz, Amer. J. 8: 728. 1931, pro parte; N. Amer. Fl. II. 5: з 1965, pro parte. Кн А tha sutae; tubus floralis erectus; petala flava; capsula fere spicue rugosa; semina grosse rugosa superficie adaxali cristis duabus parvis longitudinalibus praedita. Acaulescent perennial herbs. Pubescence of two types: short-hirsute, the hairs erect to curved and somewhat appressed; occasionally also sparsely hirsute especially on floral parts. Buds erect. Pet- als bright yellow. Capsule lanceoloid, flexible, falcate, tapering gradually to a long slender sterile apex, the angles acute, contorted and twisted, the surface conspicuously wrinkled, sessile. Seeds in nearly the length of the seed. Self-compatible, outcrossing. Basic chromosome number, x = 7 The only species in sect. Contortae, Oenothera xylocarpa, is locally t from 2,250 to 3,050 m in open meadows, flat areas, or on slopes in loose granitic gravel, sand, or pumice. It grows ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 in Pinus jeffreyi Balf. forest with Artemisia tri- dentata Nutt. or in Pinus contorta Dougl. ex Loud. subsp. murrayana (Balf.) Critchf. to Abies mag- nifica Andr. Murray forest and is known from three disjunct areas in California and adjacent Nevada: 1) Mount Rose, Washoe Co., Nevada, 2) southern Sierra Nevada, southwestern Mono Nevada, west-central Inyo and eastern Tulare counties, California, area bounded by Horseshoe and Big Whitney Meadows to the east and north, and Casa Vieja and Volcano Meadows to the south and west. The sectional name is derived from the Latin word contortus, twisted, in ref- erence to the unique contorted capsules of O. xylocarpa. Oenothera xylocarpa is here placed in a mono- typic section because, like O. primiveris, most of its specialized features are not shared with any e papillose surface, it definitely appears to be part of the lineage that includes sects. Ravenia, Ere- mia, and Pachylophus. Hybrids cannot be formed in crosses with any of these species, although abortive seeds are sometimes formed in crosses between O. xylocarpa and O. primiveris or O. maysillesii (sect. Oenothera subsect. oe A recent study of seed anatomy (Tobe et al., press) showed that the seeds of O. xylocarpa and O. primiveris differ from those of O. maysillesii primarily only in having thinner endotesta and irregularly swollen exotestal cells (papillose seed surface), but that O. primiveris differs further from the other two species in its thicker, 1 or 2 cell- iverged from these species relatively late in the evolution of the group and subsequently devel- oped a number of unique derived features (Wag- ner et al., 1985). OENOTHERA SECT. LAVAUXIA Oenothera sect. Lavauxia is a distinctive sec- tion of five species. They are characterized by narrowly ovoid, ellipsoid to rhombic-obovoid capsules with oblong wings that extend essen- tially throughout the capsule body or triangular wings confined to the upper one-quarter to 1986] two-thirds of the capsule. The seeds are asym- metrically cuneiform and the surface is minutely beaded. The collective range of the three North American species, O. flava (A. Nels.) Garrett, O. triloba Nutt., and O. acutissima W. L. Wagner, extends no further south than Guanajuato and Hidalgo, Mexico. The other two species, O. cen- tauriifolia (Spach) Steud. and O. acaulis Cav., are disjunct in Argentina, Uruguay, and Chile. The two groups differ in features such as petal color and proportion of the capsule that is winged. The North American species have yellow petals and wings extending two-thirds or throughout the capsule length, whereas the South American species have white petals and wings extending only one-quarter to one-half the capsule length. The following subsection is proposed for the two South American species. Oenothera L. sect. PANNE (Spach) Endl. ne sect. Australis? W. L. Wagner & Dietri subsect. nov. TYPE: Oenothera а ру (Spach) Steud. Petala alba; plantae brevi-villosae; capsulae alae abrupte ad medium truncatae; intra medium carentes. Petals white; plants short-villous, the hairs transparent to translucent, rarely creamy white, .1-0.4(-0.5) mm long, mixed with minute erect transparent glandular hairs, 0.1-0.2 mm long, and sparsely hirtellous, the hairs transparent to translucent, 0.7-1.7 mm long, these sometimes with a pale reddish purple pustulate base, only rarely strigillose with creamy white hairs 0.1—0.3 mm long; capsule wings abruptly truncated \- V; of the way from the capsule apex, and absent or nearly so from the lower half of the capsule. In addition to the new Subsection: of sect. La- vauxia, а ded for one of the North American taxa it in this section. Oe- nothera flava occurs from Saskatchewan and Al- berta, Canada, to western parts of the Great Plains, nearly throughout the Rocky Mountains to eastern Oregon and California, and south to the states of Guanajuato and Hidalgo, Mexico. Throughout most of its range it is modally au- togamous, growing in seasonally moist sites or along streams; however, in montane areas of Ar- 3 Treatment of this new subsection in collaboration with W. Dietrich, Botanisches Institut der Universitat, Universitátsstrasse 1, D-4000, Düsseldorf 1, Germany WAGNER—OENOTHERA 479 izona, New Mexico, and the Sierra Madre Oc- cidental, Mexico, a large-flowered, modally out- crossing form replaces the autogamous plant. Previously these outcrossing plants had been considered a distinct species, O. taraxacoides. Careful study of these plants throughout their nge (Wagner, unpubl.) showed that they inter- grade extensively with O. flava, especially in Ar- izona near Flagstaff and throughout northern Mexico. Because of this intergradation and be- cause it differs primarily only in flower size and breeding system, it is here treated as a subspecies of O. flava KEY TO THE SUBSPECIES OF OENOTHERA FLAVA la. Stigma elevated above or outside the ring of ers; petals usually obcordate, (2.5-)3- 4.5(-5) cm long; sepals often flecked with red- dish purple splotches and with free sepal ~~ 7-)2.5-10(-12) mm long; seeds ove 2.5 mm long subsp an ds lb. M surrounded by ni. ен? usually obovate with a terminal tooth, (0.7-)1-2.6 СЭВ), cm long; sepals generally lacking red- dish purple splotches and with free sepal tips 1-2(-5) mm long; seeds 1.8-2.2(-2.6) mm long subsp. flava Oenothera flava (A. Nels.) Garrett subsp. tar- axacoides (Woot. & Standl.) W. L. Wagner, comb. et stat. nov. Lavauxia taraxacoides Woot. & Standl., Contr. U.S. Natl. Herb. 16: 155. 1913. TYPE: U.S.A. NEW MEXICO: Otero Co., James Canyon, Sacramento Mountains, 2,800 m, 6 July 1899, E. O. Wooton s.n. (US-563856, holotype, photo MO; POM, RM, isotypes). Oenothera tar- axacoides (Woot. & Standl.) Munz, Amer. J. Bot. 17: 362. 1930. Oenothera flava subsp. taraxacoides is colon- ial, rarely abundant in rocky clay- or sandy-loam soils of montane meadows to gravelly or sandy sites along seasonal or permanent watercourses in ponderosa pine, spruce-fir, or pine-oak forests, from three disjunct areas: 1) the Mogollon Pla- teau in Arizona to Catron Co., New Mexico; 2) Sacramento Mountains and Sierra Blanca, Lin- coln and Otero counties, New Mexico; and 3) the Sierra Madre Occidental from northern Chi- huahua south to Durango, Mexico LITERATURE CITED DIETRICH, W., P. Н. RAVEN & W. L. WAGNER. 1985. Revision of Oenothera sect. Oenothera subsect. Emersonia (Onagraceae). Syst. Bot. 10: 29-48. 480 Lewis, H. & M. E. Lewis. 1955. The genus Clarkia. Univ. Calif. Publ. Bot. 20: 241-392. LE . Botanical Exploration of the Trans-Mississippi West, 1790-1850. Arnold Ar- boretum, Jamaica Plains, Massachusetts. UNZ, P t in Onagraceae V. The North can species of the subgenera Lavaux- me ia and pred of the genus Oenothera. Amer. J. Bot. 17: 358-370. n P. H. 1970. aah ie Gin e from Baja ifornia, Mexico and a review of subgenus Dace а Сеа. Madroño 20: 350-354. STOCKHOUSE, R. E. 1973. Biosystematic Studies of Oenothera L. Subgenus Pachylophus. Ph.D. dis- sertation. Colorado State Univ., Fort Collins. STUBBE, W. & P. H. RAvEN. 1979. A genetic contri- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 bution to the taxonomy of Oenothera sect. Oe- nother d subsects. Euoenothera, Emer- 5 imannia and Munzia). Pl. Syst. Evol. 133. za 59 Tose, H., W. L. WAGNER & H.-C. CHIN. 1986. A systematic and evolutionary study of Oenothera (Onagraceae): seed coat anatomy. Bot. Gaz (Crawfordsville) (in press). WAGNER, W. L. New species and combinations in the genus Oenothera (Onagraceae). Ann. Mis- souri Bot. Gard. 70: 194-196. OCKHOUSE & W. M. KLEIN. 1985. The systematics and evolution of the Oenothera caes- pitosa species complex O ан Syst. Bot., Missouri Bot. Gard. 12: STUDIES IN THE ARALIACEAE OF NICARAGUA, AND A NEW WIDESPREAD SPECIES OF OREOPANAX' M. J. CANNON AND J. F. M. CANNON? ABSTRACT w information on the аи а from the preparation of an account of the family for the Flora ¢ de Nicaragua is reviewed i Oreopanax nicaraguensis is a ake es During the preparation of an account of the Araliaceae for the forthcoming Flora de Nica- ragua we have encountered a number of prob- lems, but as the concise format of that work pre- vents their wider discussion, some explanatory notes are presented here together with the de- scription of a new species. These decisions were taken in the context of a broader study of the family in Central America, and this will appear in due course in Flora Mesoamericana. OREOPANAX DECNE & PLANCH. The considerable difficulties in separating the entire-leaved species of Oreopanax have been noted by various authors, notably A. C. Smith (1936) and L. O. Williams (1966). In Nicaragua, the entire-leaved specimens are easily divided into two groups, one of which is referable to O. capitatus (Jacq.) Decne & Planch., which, in our opinion, must also include O. liebmannii Mar- chal. The other appears to be a new species, here described as O. nicaraguensis. Oreopanax lieb- mannii has generally been separated from O. capitatus on the basis of its narrow leaves and smaller staminate and hermaphrodite capitula, but it has proved quite impossible to maintain this distinction in our area, there being a com- A. C. Smith (1936) has drawn attention to the difficulty of separating O. liebmannii in the southern part of its range (Costa Rica and Pan- ama), and throws doubts upon its specific status. We are in complete agreement with this view and include O. /iebmannii within the synonymy of O. capitatus. Marchal cited specimens of O. liebmannii from wider context of studies for the Flora Mesoamericana, and Copenhagen in his original description, together with others from Paris and Leningrad. There are two sheets of Liebmann no. 14 at Copenhagen, one of which has two separate branches, one bearing staminate heads, the other hermaphro- dite heads, with the locality recorded as Donguia (a probable misspelling for Donaguia in Mexico). We designate this as the lectotype. The other sheet from Donaguia bears a staminate branch, while packets hold part of a hermaphrodite in- florescence and several leaves. A third sheet from Copenhagen bears a specimen of a staminate in- florescence labelled no. 11, and three separate leaves, one of which is labelled no. 12. One leaf closely resembles those of Liebmann no. 11 from Donguia and probably has been incorrectly mounted on this sheet. The flowering branch (no. 11) on this sheet does not match the Donguia specimens very closely, whereas the other two leaves may be from another species. The locality for no. 11 on this sheet is quoted as *?” and that for no. 12is Alpatlahua. The specimen from Paris, Hahn from ‘Forét de Perote', is a staminate branch that does not resemble the Donguia spec- imens very closely; we have not seen the sheet from Leningrad. All the specimens of Oreopanax we have seen from Nicaragua with palmately lobed leaves are referable to O. geminatus Marchal. Comparison of type material of O. geminatus with O. lach- nocephalus Standley and with other specimens from Honduras shows many similarities. These species have been separated mainly by the pres- ence or absence of peduncles, but examination of the type material of the former (Oersted no. 7 from Segovia, Nicaragua), which is preserved at Copenhagen, disclosed the presence of a few pe- dunculate capitula. We therefore propose the in- clusion of O. lachnocephalus within O. gemi- ! We are grateful to colleagues in the following institutions for the opportunity to examine material from their collections: A, C, F, FCME, GH, K, M ? Department of Botany, British Museum (Natural History), Cromwell Road, London SW7 5BD, England. ANN. MISSOURI Bor. GARD. 73: 481-485. 1986. 482 natus, thus confirming the tentative suggestion made by L. O. Williams (1966). A new species of Oreopanax is described here from Nicaragua. It has also been found to the south of our area in Costa Rica and Panama, but, as yet, has not been collected in Guatemala or Honduras. The Oreopanax species of Nicaragua can be separated by the following key: la. Leaves digitate, inflorescence racemose .......... O. xalapensis (Kunth) Decne & Planch. lb. Leaves simple, entire or lobed, inflorescence ропе ог compound umbellate. 2a. Leav n lobed, styles DE herma dite flow O. geminatus Marchal 2b. Leaves 2 styles of map di te flowers 5-1 3a. Peduncles glabrous, leaf bases ob- tuse to cordate, disc end A E ee a cup .... O. nicaraguensis M. J. Cannon 1. F. M. Cannon 3b. Peduncles often puberulent, leaf bases pine to a disc neither fleshy nor formin s O. c oie np у Decne & Planch. Бйр еа nicaraguensis M. J. Cannon & J. F. in non, Hs nov. TYPE: NICARAGUA. DEPT. JIN : Cam a Aranjuez, Santa iube) aoe 24"N, 85°54) 57"W, alt. ca. 400 m. Arbol 40 m de alto frutos morados, Nov. 1983. Hermaphrodite inflorescence, Vega & Quezada 197 (holotype, BM; isotypes, HNMN, MO). Figures 1, 2. eopanax capitato (Jacq. ) Decne & Planch. affinis, simplicia, lam -16 x , suborbicularis vel late ovatis vel rhombo-ovatis, apice acuminatis ac mine inusque г to el E basi nervis, KON supra saepe nitidis; petiolus 4-15 cm longus, stria , brac teis dessin 2-3(-10) cm longis hes por vel nut: lis, ramis s pedunculisque Striatis, ig vel icd SSIS, ш, һе cuspidatis; capitulum ad 5 mm diametro, limbo а lobato vel undulato; petali 1.5 mm longi, u is 4—6 cm m longis, crassis, Кыш. M усы Үн ad 3mm ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 longis acutio. Fructus 6-10 x 4—6 mm, ovoideus vel subglobosus, carnosus, 2-5(-8) in quoque capitulo, bracteis subfructibus 5-8 mm longis, suborbicularis, mar- inis suberosis; discus carnosus urceolatus; styli (6—)7(-9), 2-3 mm longi, in cupula inserti, liberi vel basi crassi et кан apice arcuato reflexi. Semina albumine rumina Trees to 40 m tall, or rarely vines, glabrous throughout or the inflorescence only pubescent, the hairs stellate, sessile, long-branched. Leaves entire, blade 7-16 x 7-14 cm, suborbicular to broadly ovate or rhomboid ovate, tip with a short more or less rounded or acute acumen, base rounded to cordate, conspicuously (3-)5-7(-9)- nerved from the base, coreaceous, upper surface shiny; petioles 4—15 cm long, striate, somewhat swollen at the base. Staminate inflorescence more or less umbellate or paniculate, bracts subtending branches 2-3(-10) mm, ate or absent, bescent, peduncles m long, stout, bracts subtending peduncles 2-4 mm long, very broadly ovate, shortly cuspidate, heads to 5 mm diam., subglobose, bracts sub-orbicular; calyx obconic, limb broadly lobed or undulate; petals 1.5 mm long, deltoid; filaments 1.5 mm long, anthers 1.5 mm long, oblong; styles 1. Fruiting inflorescence compact, usually more or less um- bellate, very rarely paniculate, branches 4-6 cm ong, peduncles 5-20 mm long, stout, bracts sub- tending peduncles to 3 mm, apex acute. Fruits 6-10 x 4-6 mm, ovoid to subglobose, fleshy, 2- 5(-8) per head, bracts subtending fruits 5-8 mm long, suborbicular, at first ciliate, emarginate and or swollen and touching at the base, the apices arcuate-reflexed. Endosperm ruminate. The new species differs from the common and widespread O. capitatus (Jacq.) Decne & Planch. by its very broad leaves, the leaf venation, the styles within a cup-shaped hollow at the apex of the fruit, and its compact, large-fruited inflores- cence. It differs from O. obtusifolius L. Williams the shiny upper leaf surface, the acuminate leaf apices, the styles within the cup-shaped hol- low and the larger fruited, more compact inflo- rescence. While O. obtusifolius seems to be found to the north of the region in Mexico, Belize, Gua- temala and Honduras, O. nicaraguensis appears to have a more southerly distribution in Nica- ragua, Costa Rica, and Panama. 1986] CANNON & CANNON—NICARAGUAN ARALIACEAE 483 FIGURE 1. Oreopanax nicaraguensis, holotype, showing fruits, Vega & Quezada 197 (BM). 484 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 mearaguen: Farar нит» ЗЕМ E FROAN: ad MA No. 20278 19 May 1981 ARALIACEAE Depar tà to de ESTEL Lh ca "yon if ana Laguna ratore "s; approximately 13015'N, 86016" W, elev 1100-1300 ergreen forest and Рэг а - ong small stream stream bank, tr $ m aii, flowers green=white, "hoja de pied a" 1 2 _ _ - Cohected by " м ә а ai Ame |" Hw with J. Henrich FIGURE 2. Oreopanax nicaraguensis, paratype, showing flowers, Stevens & Henrich 20278 (BM). 1986] Paratypes: NICARAGUA. ESTELI: ca. 7.3 km beyond Rio Esteli on road between Highway | and Laguna de Miraflores, ca. 13?15'N, 86°16’W, alt. 1,100-1,300 m. 19 May 1981. Male inflorescence. Stevens & Henrich 20278 (BM, HNMN, MO); El Tisey, 1,500 m, Laguna 292 (BM, HNMN). JINOTEGA: Sanatorio Aranjuez, 1,300-1,350 m, Sandino & Martinez 4328 (BM, HNMN, MO). MATAGALPA: Cerro El Picacho, 1, 420- 1,520 m, Stevens 22123 (BM, MO). MADRÍZ: Cerro Ar- enal, 1,500 m, Atwood & Neill 274 (HNMN, MO). ZELAYA: Comarca Waslala 520-560 m, Moreno 17261 (BM, H COSTA RICA. LIMON: Chiripo National Park, 2,800 Cordillera de Tilaran, 1,500-1,560 m, Dryer 1360 (F). ANAMA. CHIRIQUÍ: Boquete, 2, 900-2, 950 m, Croat 34885 (BM, MO). SCHEFFLERA J. R. & G. A. FORST. The northern limit of this genus in Central America occurs in Nicaragua, with the sole ex- ception of S. morototoni (Aubl.) Maguire, a ermark & Frodin, which, until recently, was i cluded in the genus Didymopanax. However, nd genus is indistinguishable from Schefflera other | by stigma number, which is itself variable, en within one plant, as has been clearly estab- lished by Frodin (pers. comm.) in a wide ranging survey of Schefflera and related groups. Schef- flera nicaraguense (Standley) A. C. Smith was described from Nicaragua and has now been found also in Costa Rica. A further, apparently undescribed, species of Schefflera has been col- lected in the Department of Rio San Juan in southern Nicaragua: Rio Indio, 5.5 hours up riv- er from San Juan del Norte (25 Hp) 11?07'N, 83*50—52'W, alt. less than 20 m, Finca las De- licias de Alfonso Crespo Aragon, 8 Sept. 1982. Riviere 292 (MO). This ous is epiphytic, with about 10 leaf- lets, 22-23 6.5 cm, oblong-lanceolate, gla- brous, tips acuminate (acumen 1 cm), bases acute CANNON & CANNON—NICARAGUAN ARALIACEAE 485 to rounded, lateral veins 10-12, petiolules 9-11 cm long, terete. Petiole and ligule not seen. In- florescence at least 6-, probably many, branched, ferruginous pubescent, bracts subtending branches 1-2 mm long, truncate, branches ca. 30 cm long, striate; peduncles slender, 6-8 mm long, ebracteate just after anthesis; pedicels very slen- der, 3-4 mm, (4-)5 per umbel; calyx teeth cus- pidate; calyptra very pointed and cuspidate at the apex, densely ferruginous pubescent. Stylar cone of very young fruit 1-2 mm long, styles 5, becoming free above for about half of their length. It may be related to the very variable species Schefflera systyla (Donn. Smith) Viguier from Costa Rica and Panama, from which it seems to differ by its many-branched inflorescence, its very slender pedicels and peduncles, and its densely from South башкыны 2 the leaflets differ greatly in shape and ve Although by de а of this difficult and complex genus, this taxon seems relatively dis- tinct, we do not propose to give it formal rec- ognition at this time, as the single known spec- imen is both immature and seriously incomplete, and the purposes of Science will be better served by waiting until further material is available for consideration, leading perhaps to the designation cussion of the Schefflera problems in the light of his very extensive long-term monographic stud- ies on the genus. LITERATURE CITED MancHar, E. 1879. Révision des es cap améri- caines. Bull. Acad. Belg. Ser 7. Notes on jens Araliaceae. Araliaceae. In P. C. Standle WILLIAMS, L. 6. . О. Williams, Flora of Guatemala. Fieldiana 24: 13-14 NOTES CHROMOSOME NUMBERS OF NEW CALEDONIAN PLANTS Chromosome numbers of 26 collections that represent first reports for 23 species embracing 17 families of New Caledonian flowering plants are presented and discussed. Chromosome num- rs are documented for the first time in the genera Adenodaphne (n = 12), Agatea (n = 8), Dubouzetia (n = ca. 90), Kibariopsis (n = 19), Montrouziera (n = ca. 29), Oncotheca (n = 25), Phelline (n = 17), and Sphenostemon (n = ca. 26). Two of these represent the first reports for Oncothecaceae and Sphenostemonaceae. Num- bers not previously reported are established in Belliolum (n = ca. 43), Garcinia (n = 33), Gre- villea (n = ca. 22), and Zygogynum (n = ca. 43). Although the flora of New Caledonia is one of great potential value in the interpretation of phy- togeographic and phylogenetic relationships, few species are cytologically known. In an effort to contribute to a better understanding of these re- lationships, the Missouri Botanical Garden in- cluded as part of its 5-year-long field program on the island the collection of flower buds for cy- tological study. The purpose of this paper is to present and discuss the 26 chromosome counts that these collections have yielded so far. MATERIALS AND METHODS Floral buds for study were preserved and stored in modified Carnoy’s fixative (6 chloroform: 3 absolute ethanol: 1 glacial acetic acid; v:v:v). Meiotic, and in two cases, mitotic chromosome determinations were made from anther material macerated in acetocarmine. Hoyer’s solution (Beeks, 1955) was added to the preparations in order to increase quality and durability. The in- dexes of chromosome numbers that were con- sulted for interpretive and comparative data in- clude Fedorov (1974), Goldblatt (1981, 1984), and Moore (1973, 1974, 1977). The sequence of families in the list follows Cronquist (1981). All collection numbers (Table 1) are G. Mc- Pherson’s. Voucher specimens are deposited at MO RESULTS The results are presented in Table 1. Where there was some question about the cytological Anat а 13 A 4 11 1 JL ANN. MISSOURI Bor. GARD. 73: 486—489. 1986. high numbers, inadequate fixation, or stickiness, the number reported is listed as approximate (prefaced with ca.). These are the reports in great- est need of additional confirmation. There were no clear indications of meiotic abnormalities in any of the material studied. DISCUSSION Winteraceae. The report here of л = ca. 43 (Table 1) for Belliolum crassifolium is the second for this species and differs from the earlier record of n — ca. 86 (Ehrendorfer et al., 1968). Our determination of n = ca. 43 for Zygogynum po- miferum subsp. balansae is also a lower ploidy level than previously known in this New Cale- donian endemic genus. The only other report available is n — 86 for Z. baillonii V. Tiegh. (Ehrendorfer et al., 1968). Monimiaceae. Hedycarya parvifolia has n = 19 (Table 1), a number that has also been re- ported in two other species of the genus (Ehren- dorfer et al., 1968). Hair and Beuzenberg (1959) have recorded n = 57 in a third species of He- dycarya. The report here of n — 19 for Kibariopsis caledonica is the first for this endemic New Cale- donian genus. This number has now been estab- lished in at least five genera of Monimiaceae. Lauraceae. „The count here of n = 12 for Ad- enodaphne tly represents the first report for this genus. “The same number char- acterizes many genera of the family. Chloranthaceae. The report here of n = ca. 14 for Ascarina rubricaulis agrees with an earlier one of 2n = 28 (Ehrendorfer et al., 1968). This number is also known in Chloranthus (Sugiura, 1931). Fagaceae. The count of n = 13 for Notho- Јавиѕ discoidea (Table 1) is the same as that found in all seven species of the genus already inves- tigated (Armstrong & Wylie, 1965; Ono, 1977). Other genera in Fagaceae reportedly have n = 11, 12, and 24 Oncothecaceae. The report of n = 25 for On- cotheca balansae (Table 1) is the first for this endemic New Caledonian family. At diakinesis ring pairs of chromosomes ranged from about 1.5-2 um in diameter while rod pairs ranged from about 2-3.3 um in length. The inclusion of On- 1986] NOTES TABLE 1. Chromosome numbers of New Caledonian plants. 487 Taxon N 2n Collection data Winteraceae Belliolum crassifolium (Baillon) van Tieghem ca. 43 Thy Valley, 1852 Zygogynum pomiferum Baillon subsp. balansae (van Tieghem) Vink? ca. 43 Mt. Dzumac, 3880 Monimiaceae Hedycarya parvifolia Perk. & Schltr 19 Mt. Do, 200 Kibariopsis caledonica ED. Jérémie? 19 Mandjélia, 2546 Lauraceae Adenodaphne uniflora (Guillaumin) Kostermans? 12 Yaté Dam, 3886 Chloranthaceae Ascarina rubricaulis Solms ca. 14 Plaine des Lacs, 46/2 Fagaceae Nothofagus discoidea (Baum.-Bodenh.) van Steenis* 13 Thy Valley, 2487 Oncothecaceae* Oncotheca balansae Baillon^ 25 Thy Valley, 3/30, 3301 Clusiaceae Garcinia puat (Montrouzier) Guillaumin* Mt. Mé Ori, 2013 Montrouziera sphaeroidea Pancher ex Planch. & Triana" ca. 29 Plaine des Lacs, 3014 Elaeocarpaceae Elaeocarpus speciosus Brongn. & Gris? 30 Mt. Mé Maoya, 2950 Dubouzetia elegans Brongn. & Gris^ ca. 90 Mt. Mé Oni, 3029 Violaceae Agatea sp.^ 8 Riviére Bleue Valley, 3670 Symplocaceae Symplocos baptica Brongn. & Gris* 11 Mt. Do, 3810 Pittosporaceae Pittosporum cf. dzumacense Guillaumin* 12 Mt. Do, 3813 Proteaceae Beauprea neglecta R. Virot® 11 Houailou, 3261 Grevillea meisneri Montrouzier* ca. 22 Népoui Valley, 7896 Olacaceae Olax hypoleuca Baillon? 12 24 Goro, 3631 Aquifoliaceae Phelline sp.” 17 Mt. Mé Ori, 3041 Phelline sp.” 17 Col de Mouirange, 3599 Sphenostemonaceae* Sphenostemon comptonii Baker f.^ ca. 26 Mt. Mé On, 3047 Malpighiaceae Acridocarpus austrocaledonicus Baillon* 9 Tiébaghi, 3306 Apocynaceae Alstonia plumosa Labill.* 11 Thy Valley, 7877 A. vieillardii van Heurck & Muell. Arg.? 11 ntagne des Sources, 1640 Rauvolfia semperflorens (Muell. Arg.) Schlechter* 11 Mt. Mou, 7871 a Denotes species not recorded in chromosome indexes. ^ Denotes genera not recorded in chromosome indexes. * Denotes families not recorded in chromosome indexes. 488 cothecaceae in Theales by Cronquist (1981) is compatible with cytological data because n = 25 is also known in Ternstroemia (Theaceae) (Mor- inaga & Fukushima, 1931). Clusiaceae. The count of n = 33 for Garcinia puat (Table 1) is unique in the genus. However, n — 22, 24, ca. 27, ca. 28, ca. 29, 36, 38, 40, and 48 have been reported for Garcinia (cf. Fedorov, 1974; Goldblatt, 1984). The count of n = ca. 29 for Montrouziera sphaeroidea (Table 1) appar- ently represents the first for this endemic New Caledonian genus. In addition to the numbers listed above for Garcinia, n= 28 and ca. 28 have been reported for Allanblackia (Mangenot & Mangenot, 1957) and Pentadesma (Mangenot & Mangenot, 1962), respectively. Elaeocarpaceae. The count of 2n = 30 for Elaeocarpus speciosus (Table 1) agrees with re- ports for at least three other species of the genus (Mehra, 1976; Rattenbury, 1957). Other num- bers reported for E/aeocarpus are n= 12 and n = 14 (Arora, 1961; Ono, 1975). Our report of n = ca. 90 for Dubouzetia elegans is apparently the first for this genus. This number may represent dodecaploidy based on n = 15. Violaceae. The report here for Agatea of n = 8 is the first for the genus. It agrees with reports for Hybanthus, Ionidium, and a couple of species of Viola (Moore, 1973, 1974; Goldblatt, 1981, 1984). Two species of the Hawaiian endemic ge- nus Isodendrion also have n = 8 (Carr, 1985). Symplocaceae. Thecount of n= 11 for Sym- plocos baptica (Table 1) agrees with the number established in several other species in the genus. Counts of л = 12 (Nevling, 1969) and 2л = ca. 90 (Nooteboom, 1975) have also been reported for Symplocos. Pittosporaceae. The first report here of n = 12 for Pittosporum cf. dzumacense agrees with nearly every report for the family. Only Citrio- batus differs by having n — 18 (Fedorov, 1974). Proteaceae. The count of n = 11 for Beau- prea neglecta (Table 1) accords with the report for B. paniculata Brongn. & Gris. (Johnson & Briggs, 1963), the only other count for the genus. Counts of n = 11 are common in Proteaceae, but n — 22 has not been reported previously. Thus, the count of п = ca. 22 given here for Grevillea meisneri is noteworthy, but needs verification. In any case, it contrasts markedly with n = 10, the only other number known in the approxi- mately 27 species of Grevillea that have been investigated (cf. Fedorov, 1974). ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Olacaceae. The count of n = 12 for Olax hypoleuca (Table 1) agrees with the only other report available for the genus. Olax nana Wall. and Schoepfia fragrans Wall. appear to be the only other species in the family reported to have n = 12 (Mehra, 1976). Aquifoliaceae. The report here of n = 17 for two collections of Phelline represent the first for this genus. The same number has also been re- ported in a few species of Цех, including J. ma- tanoana and I. mertensii (Ono & Masuda, 1981). Sphenostemonaceae. Our count of = ca. 26 for Sphenostemon comptonii is apparently the first for this family. At diakinesis ring pairs of chromosomes ranged from about 2.3-3.4 um in diameter while rod pairs were about 3 um long. Historically, the family has been included in Tri- meniaceae, Clusiaceae, Aquifoliaceae, or Thea- ceae. Cronquist (1981) included Sphenostemon in the Aquifoliaceae. However, strictly in terms of chromosome numbers, the greatest accord ap- pears to be with Theaceae (cf. Ternstroemia ja- ponica, n — 25; Morinaga & Fukushima, 1931) or Clusiaceae, but obviously the data are equiv- ocal. Malpighiaceae. Acridocarpus austrocaledon- icus is here reported to have n = 9. This number has also been reported for the two other species in the genus that are known cytologically (Man- enot & Mangenot, 1962). Tristellateia and Stig- maphyllon appear to be the only other genera of Malpighiaceae reported to have species with this number (Fedorov, 1974). Apocynaceae. First reports of n = 11 are giv- en here for Alstonia plumosa and A. vieillardii. This number and n = 22 and 44 have also been found in other species of the genus (Fedorov, 1974). A first report here of n = 11 for Rauvolfia semperflorens agrees with the number in several other species in Rauvolfia, in which n = 22, 33, 44, and 88 are also known (Fedorov, 1974; Gold- blatt, 1981). The authors acknowledge aid provided by the Laboratoire de Botanique, ORSTOM, Nouméa. Support was provided in part by National Sci- ence Foundation Grant DEB-8022179 to G. McPherson. We are grateful to Peter H. Raven for initiating this cooperative effort and for his assistance in coordinating the project. We thank Elisabeth Rabakonandrianina for help with some of the chromosome preparations. 1986] LITERATURE CITED ARMSTRONG, J. M. & A. P. WYLIE. 1965. A new basic chromosome number in the family Fagaceae. Na- ture 205: 1340-1341. ARORA, C. M. 1961. New chromosome report. II. Bull. Bot. Surv. India 3: 37. ВЕЕК$, В. M. . Improvements in the squash technique for plant chromosomes. Aliso 3: 131- 134 CARR, G. D. 1985. Additional chromosome numbers of Hawaiian flowering plants. Pacific Sci. 39: 302- 306. CRONQUIST, A. 1981. An Integrated System of Clas- sification of Flowering Plants. Columbia Univ. ress, New York. EHRENDORFER, F., F. NDL, E. HABELER & W. SAU- ER. 1968. г me numbers and evolution in ig itive Angiosperms. Taxon 17: 337-353 or). nu T nograp Senate Bu: from the Missouri Botanical Garden, rd e 5. 984. Indes to Plant Chromosome Numbers . Monographs in Systematic Botany from the Missouri Botanical Garden, Volume 8. J. BE Contributions to a chromosome atlas of the New Zealand flora. 2. New Zealand J. Sci. (Wellington) = 148—156. JOHNSON, L. A. & B ; s Evolution in the Proteaceae. Austral. J. Bot. : 21-61. MANGENOT, S. & G. MANGENOT. 195 N chromosomiques nou ico- tylédones et Манка окан d Afrique occiden- tale. Bull. Jard. Bot. Etat 27: 639-654. & Enquéte sur les nom n chromosomiques dans une collection s tropicales. Rev. Cytol. Biol. Vég. 25: 411-4 mbres NOTES 489 МЕНВА, P. М. 1976. Cytology of Himalayan Hard- woods. Sree Saraswaty Press, Calcutta. Moore, В. J. (editor). 1973. Index to plant chro- mosome numbers 1967-1971. Regnum Veg. 90: 1-530. ————. 1974. Index to plant chromosome numbers for 1972. Regnum Veg. 91: 1-108. 1977. Index to plant Parc numbers for 1973/74. ip gum Veg. 96: 1-257. MORINAGA, T. & E. FUKUSHIMA. . Chromosome numbers in cultivated plants. n Bot. Mag. (To- 9. 1969. Ecology of ап elfin forest in Puerto Rico. 3. Chromosome numbers of some flowering plants. J. Arnold Arbor. 50: 99-103. NOOTEBOOM, H. P. . Revision of the Symplo- caceae of the Old World Excluding New Caledo- nia. са Pers, Leiden. [Leiden Botanical Series No. Ono, M. * 1975. " Chromosome numbers of some en- demic species of the Bonin Islands I. Bot. Mag. ict 88: Pim 328. th neis rican | species of Norhofagus = Bot. Mag. (Tokyo) 90: 313-316 У. Masupa. 1981. Chromosome numbers of some endemic species of the Bonin Islands II. gasawara Res. 4: 1-24. мы J. A 1957. Chromosome numbers in New Zealand Angiosperms. Trans. & Proc. Roy. New Zealand 84: 936-938. SUGIURA, T. 1931. A list of chromosome numbers angiospermous plants. Bot. Mag. (Tokyo) 45: —355. — Gerald D. Carr, Department of Botany, Uni- versity of Hawaii, 3190 Maile Way, Honolulu, Hawaii 96822; and Gordon McPherson, Mis- souri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. A NEW COMBINATION IN LUDWIGIA SECT. MICROCARPIUM (ONAGRACEAE) The following new combination is proposed in advance of ongoing monographic studies of North American Ludwigia in order to make the name available for numerous floristic projects and for a biosystematic study of Ludwigia sect. Microcarpium (Peng, 1987). It will be discussed and justified in my subsequent publication, and is offered at this time with minimal synonymy. Ludwigia glandulosa Walter subsp. brachycarpa (Torrey & A. Gray) Peng, comb. nov. L. cylindrica Elliott 8. brachycarpa Torrey & A. Fl. N. Amer. 1: 524. 1840 non Jus- . torreyi Munz, Bull. Torrey Bot. Club 71: 164. 1944, illeg. subst. LECTOTYPE: U.S.A. Texas: Aus- tin Co., San Felipe, “third collection," 1833- 1834, T. Drummond 84 [GH; isolectotypes, GOET, K (2 sheets), W; here designated]. Munz (1944) renamed this taxon because he thought that Torrey and Gray (1840) based it on Jussiaea brachycarpa Lam. (= L. glandulosa Walter subsp. g/andulosa), which, however, is doubtful. In any case, they were free to use the same epithet for a variety, so that “brachycarpa” has priority at the varietal level. Here I am taking it up for the short-fruited, western subspecies of Ludwigia glandulosa. Munz (1944) chose a lec- totype for this entity (Hall 219, from Hempstead, Waller Co., Texas) that was not collected until more than 30 years after Torrey and Gray wrote; his choice must therefore be disregarded. Of the two specimens they cited, only the lectotype I have chosen meets the criteria of this entity as we now understand it, having short mature fruits. In this respect it agrees more fully with Torrey and Gray's diagnosis. Chapman's collection (no. 38) from middle Florida, without definite local- ity (GH, NY), is an immature specimen of subsp. glandulosa, which Torrey mis 2 considered to be their short-fruited vari Ludwigia glandulosa Mn С РЕНА dif- fers from the subsp. g/andulosa in its smaller stature, narrower leaves, and smaller flowers and capsules. In addition, the seed surface pattern of these two entities—a feature that is often diag- ANN. MISSOURI Bor. GARD. 73: 490. 1986. nostic in Ludwigia sect. Microcarpium —is sharply distinct. In L. glandulosa subsp. glandulosa, the cells are longitudinally elongate, whereas in subsp. brachycarpa, they are transversely elongate. Ludwigia glandulosa subsp. brachycarpa oc- curs along the Gulf Coast from extreme south- western Louisiana to Nueces County, Texas, and north ward through eastern Texas to southcentral Oklahoma. Subspecies g/andulosa, in contrast, has a much broader distributional range: it oc- curs throughout the Atlantic and Gulf Coastal Plains and the Mississippi Embayment, west- ward to eastern Texas and southeastern Okla- homa. Ludwigia glandulosa subsp. brachycarpa thus has a range that lies along the western edge of that of the subsp. g/andulosa in northern Texas and Oklahoma. It grows in the same areas as subsp. glandulosa throughout much of its range, but extends farther south and west. The distinc- tiveness of these two subspecies is probably maintained by self-pollination. I am indebted to Drs. Peter H. Raven and Peter C. Hoch for valuable discussions and crit- icisms during the preparation of this paper. This research is supported in part by the National Science Foundation through grants to Dr. Peter H. Raven. I am also grateful to the Institute of Botany, Academia Sinica, Taipei, Taiwan, Re- public of China for granting me a leave of ab- sence to work on this study. LITERATURE CITED Munz, P. A. 1944. Studies in Onagraceae — XIII. The American species of Ludwigia. Bull. Torrey Bot. Club 71: 152-165 PENG, C. I. sect. о (Опаргасе Bot. Gard. 74: (in press). TORREY, J. An GRAY. 1838-1840. A Flora of North merica: Containing Abridged Descriptions of all 1987. A biosystematic study of Ludwigia ae). Ann. Missouri to the Natural Em Volume 1. Wiley & Put- nam, New Y — Ching-I Peng, Institute of Botany, Academia Sinica, Nankang, Taipei, Taiwan 11529, Re- public of China. DESCRIPTION OF A NEW SECTION AND SUBSECTION IN CLARKIA (ONAGRACEAE) In their ph of Clarkia, Lewis and Lewis (1955) described two new sections in the genus, sects. Peripetasma and Fibula. In addition, Sa described four subsections under C. sect. Per petasma: subsects. Peripetasma, swell Micranthae, and Prognatae. The type of both C. sect. Peripetasma and C. subsect. Peripetasma is Godetia bottae Spach [= C. bottae (Spach) Lewis & Lewis]. The type of C. sect. Fibula is Godetia deflexa Jepson [= C. deflexa (Jepson) Lewis & Lewis]. Raven and Parnell (1977) re-examined the type of Godetia bottae and concluded that it 1s con- specific with the type of G. deflexa. Because G. bottae has priority over G. deflexa (1835 versus 1907), the correct name for the taxon Lewis and Lewis (1955) treated as Clarkia deflexa is Clark- ia bottae. In addition, the taxon treated as C. bottae by Lewis and Lewis was provided with a new name, Clarkia lewisii Raven & Parnell, since no other specific epithet was available. Further nomenclatural adjustments must re- sult from the work of Raven and Parnell (1977). Article 10.1 of the International Code of Botan- ical Nomenclature (ICBN) provides that “the type of a name of any subdivision of a genus is the type of a name of a species . . . for purposes of designation or citation ofa type, the species name alone suffices ....” Thus, Raven and Parnell’s investigation demonstrated that the type of Clarkia sect. Peripetasma, which is also the type of C. subsect. Peripetasma, and the type of Clark- ia sect. Fibula are conspecific. Since *'the appli- cation of names of taxa of the rank of family or below is determined by means of nomenclatural types. " (ICBN, Art. 7.1), the description ac- companying the publication of C. sect. Peripe- tasma and C. subsect. Peripetasma is simply in error insofar as it does not apply to the taxon including the type of C. bottae; C. sect. Fibula and C. sect. Peripetasma are taxonomic syn- onyms; and the section and | including C. lewisii are without a na Article 57.2 (ICBN) es that if two taxa bearing names of equal priority are united, as we propose to unite Clarkia sect. Fibula and C. sect. Peripetasma, one of them must be chosen, and the choice must be followed by subsequent work- ers. In the interests of nomenclatural stability we ANN. MISSOURI Bor. GARD. 73: 491-494. 1986. choose C. sect. Fibula as the name for the section including the types of C. deflexa and C. bottae. In addition, we describe a new section and sub- section to accommodate those taxa that Lewis and Lewis (1955) included in C. sect. Peripetas- ma and C. subsect. Peripetasma. A synopsis of the taxa involved is provided below Clarkia Pursh sect. Fibula Lewis & Lewis, Univ. Calif. Publ. Bot. 20: 333. 1955. TYPE: Gode- tia deflexa Jepson, Univ. Calif. Publ. Bot. 2: 332. 1907. [= C. deflexa (Jepson) Lewis & Lewis, = C. bottae (Spach) Lewis & Lewis] Clarkia Pursh sect. "eek arity Lewis & Lewis, Uni Calif. Publ. . 20: 313. 1955. Clarkia Pursh subsect. Peripetasma Lew wis & Lewis, Univ. Calif. 15. 1955. TYPE: Godetia bottae n. Mus. Hist. Nat. 4: 393, 1835. [= C. Foil (Spach) Lewis & Lewis] Erect herbs; st lal d gl ; rachis of the inflorescence erect; buds deflexed, becom- ing erect as the flowers open; floral tube obcon- ical, 2-3 mm long, the ring of hairs at the apex; sepals remaining united and deflexed to one side at anthesis; petals fan-shaped, 1-3 cm long, flecked, the claw very short or obscure; stamens 8, in two series, the inner usually with cream- colored pollen, the outer at first divergent, the pollen blue; anthers obtuse, shorter than the fil- aments; filaments slender; immature capsule te- rete or 4-grooved 1. Clarkia bottae (Spach) Lewis & Lewis, Ma- no 12: 33. 1953. Godetia bottae Spach, Nouv. Ann. Mus. Hist. Nat. 4: 393. 1835. Oenothera godetia Steudel, Nomencl. Bot., 2nd edition, 2: 206. 1841, nom. illeg. TYPE: U.S.A. California: s.d., Paolo Emilio Botta s.n. (P; fragment, DS). Godetia deflexa Jepson, Univ. Calif. Publ. Bot. 2: 332. 1 Godetia bottae var. deflexa (Jepson) Hitch- cock, Bot. Gaz. (Crawfordsville) 89: 355. 1930. Clarkia deflexa (Jepson) Lewis & Lewis, Madrono 3. 1953. TYPE: U.S.A. California: sandy plains of Los Angeles, 1854, Lobb s.n. (K). 2. Clarkia jolonensis Parnell, Madrono 20: 322. 1970. TYPE: U.S.A. California: Monterey Co., 9 mi. NW of Bradley along Jolon road, 3 492 June 1963, Thorne & Everett 32186 (holo- type, LA; isotype, CAS). Clarkia Pursh sect. Sympherica Holsinger & Lewis, sect. nov. TYPE: Clarkia lewisii Raven & Parnell, Ann. Missouri Bot. Gard. 64: 642. 1977 [1978]. Herbae erectae; caulibus pubescentibus raro sub- glabris; deg tium axe in apice recurvato; calycis tubo obconico, 0.5-5 mm longo, annulo pilorum ad apicem pp in medio ornato; calycis limbo sub anthesi connato et declinato; petalis 5-35 mm longis, oblanceolatis vel obovatis vel flabelliformis interdum bilobis, plerumque ferme purpureopunctulatis, ungui- culis brevis, staminibus 8, circilis duo dissimilibus; antheris obtusis quam filamenta brevioribus, filamen- tis gracilibus; ovario 4- vel 8-canaliculato vel 8-costato. Erect herbs; stems puberulent, at least above, with short upwardly curled hairs, rarely glabrate; rachis of the inflorescence recurved in bud, the buds pendulous; floral tube obconical, 0.5-5 mm long, the ring of hairs at the middle or more frequently at or near the apex; sepals remaining united and deflexed to one side at anthesis; petals 5-35 mm long, oblanceolate to obovate or fan- shaped, sometimes bilobed, usually flecked with reddish purple, the claw short or obscure; sta- mens 8, in two series, the inner usually with cream-colored pollen, the outer with blue pollen; anthers obtuse, shorter than the filaments; fila- ments slender; immature capsule 4- or 8-grooved or 8-ribbed The name for this section is taken from the Greek, ovupépov (useful) and the adjectival suffix -Kov (fitness or ability), referring to the usefulness this group has had in several evolutionary stud- ies. The results of a comparison of Clarkia lin- gulata Lewis & Lewis and Clarkia biloba (Dur- and) Nelson & Macbride using the techniques of enzyme electrophoresis (Gottlieb, 1974) were consistent with the hypothesis that C. /ingulata is a recent derivative of C. biloba (Lewis & Rob- erts, 1956; Lewis, 1962). All ofthe diploid mem- bers of this section, with the exception of C. ros- trata Davis, have two loci coding for subunits of the cytosolic isozyme of phosphoglucoisomerase (PGI) instead of only one locus, as is character- istic of most diploid plants (Gottlieb, 1982). This is also the only section in C/arkia in which there are differences between species in the number of loci coding for cytosolic PGI (Gottlieb & Wee- den, 1979). Four of the eight diploid species in this section have only one locus coding for sub- units of the cytosolic isozyme of 6-phosphoglu- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 conate dehydrogenase (6PGD), while the re- mainder of the diploid species in the genus have two loci coding for cytosolic 6PGD. In addition, two of the diploid species in this section have only one locus coding for subunits of the plastid 6PGD, while other species in the genus have two loci coding for plastid 6PGD (Odrzykoski & Gottlieb, 1984). Finally, a recent analysis based on restriction mapping of chloroplast DNA in- dicates that Heterogaura heterandra (Torrey) Coville and Clarkia dudleyana (Abrams) Mac- bride are more closely related to one another than either is to any other extant species (Sytsma & Gottlieb, 1986). Clarkia Pursh subsect. Lautiflorae Lewis & Lew- is, Univ. Calif. Publ 20: 319. 1955. TYPE: Oenothera biloba Durand, J. Acad. Nat. Sci. Philadelphia, Ser. 2, 3: 87. 1855. [= Clarkia biloba (Durand) Nelson & Mac- bride] Petals 10-30 mm long, lavender to pink, not differentiated into bands or zones of different color; mature stigma usually held above the sta- mens, the stamens maturing first; immature cap- sule conspicuously 8-ribbed or grooved. 3. Clarkia biloba (Durand) Nelson & Macbride, Bot. Gaz. (Crawfordsville) 65: 60. 1918. Oe- nothera biloba Durand, J. Acad. Nat. Sci. Philadelphia, Ser. 2, 3: 87. 1855. Godetia biloba (Durand) Watson, Bot. Calif. 1: 231. 1876. Oenothera prismatica var. biloba Lé- veille, Monogr. Onothera 264. 1908. LECTOTYPE: grown from seed at Louisville, Kentucky, U.S.A., 1855, Short s.n. (P). 3a. Clarkia biloba subsp. australis Lewis & Lewis, Univ. Calif. Publ. Bot. 20: 322. 1955. TYPE: U.S.A. California: Mariposa Co., Hwy 140, Merced River, 1 mi. W of junction with the South Fork of the Merced River, 18 June 1949, Lewis & Lewis 628 (LA). 3b. Clarkia biloba subsp. biloba 3c. resins T subsp. brandegeae (Jepson) Lew s, Univ. Calif. Publ. Bot. 20: 323. 1955. bis dudleyana f. brandegeae Jepson, Univ. Calif. Publ. Bot. 2: 334. 1907. Godetia dudleyana var. brandegeae (Jepson) Jepson, Man. Fl. Pl. Calif. 675. 1925. TYPE: U.S.A. California: Eldorado Co., Simpson Ranch, Sweetwater Creek, 29 May 1907, Brandegee s.n. (UC). 1986] 4. Clarkia dudleyana (Abrams) Macbride, Contr. Gray Herb. 56: 54. 1918. о Abrams, Fl. Los Angeles 267. 1904. TYPE: U.S.A. California: Los Angeles Co., i» Santa Anita Canyon at 2,500 ft., s.d., Abrams 2625 (DS). Godetia bottae var. usitata Jepson, Univ. Calif. Publ. Bot. 2: 332. 1907. TYPE: U.S.A. California: San Bernardino Co., s.d., Parish 3672 (holotype, UC; isotype, GH). 5. Clarkia lingulata Lewis & Lewis, Madrono 12: 35. 1953. TYPE: U.S.A. California: Mar- iposa Co., Merced River, 0.2 mi. W of bridge over South Fork ofthe Merced River, 8 June 1947, Lewis & Lewis 334 (LA). 6. Clarkia modesta Jepson, Man. Fl. Pl. Calif. 73. 1925. Godetia epilobioides var. modes- ta (Jepson) Jepson, Fl. Calif. 2: 585. 1936. Phaeostoma modesta (Jepson) Heller, Leafl. W. Bot. 2: 221. 1940. TYPE: U.S.A. Califor- nia: Fresno Co., Waltham Creek, San Carlos (Diablo) Range, s.d., Jepson 2690 (JEPS). Clarkia Pursh subsect. Micranthae Lewis & Lewis, Univ. Calif. Publ. Bot. 20: 330. 1955. TYPE: Oenothera epilobioides Nuttall ex Tor- rey & Gray, Fl. N. Amer. 1: 511. 1840. [= Clarkia ean oe ex Torrey & Gray) Nelson & Macbride] Petals white, as much as 10 mm long, not flecked; mature stigma not held above the sta- mens; stigma and stamens maturing more or less simultaneously; anthers adhering to the stigma and depositing pollen directly upon it; immature capsule subterete, not conspicuously ribbed. 7. Clarkia epilobioides (Nuttall ex Torrey & Gray) Nelson & Macbride, Bot. Gaz. (Craw- fordsville) 65: 60. 1918. Oenothera epilo- bioides Nuttall ex Torrey & Gray, Fl. N. Amer. 1: 511. 1840. ари арн ерїїо- bioides (Nuttall ex Torrey & Gray) Walpers, Repert. Bot. Syst. 2: 78. 1843. Godetia epi- lobioides (Nuttall ex Torrey & Gray) Wat- son, Bot. Calif. 1: 231. 1876. TYPE: U.S.A. California: San Diego, s.d., Nuttall s.n. (NY). Clarkia Pursh subsect. Prognatae Lewis & Lewis, v. Calif. Publ. Bot. 20: 332. 1955. TYPE: C pem similis Lewis & Ernst, Madrono 12: 89. 1953 NOTES 493 Petals pale pink to nearly white, flecked with purple in the lower half, not differentiated into distinct zones of color, 6-12 mm long; anthers usually free, maturing with the stigma; stigma not held above the stamens. 8. Clarkia similis Lewis & Ernst, Madrono 12: 89. 1953. TYPE: U.S.A. California: San Die- go Co., 7.6 mi. W of Ramona, 22 Apr. 1951, Lewis, Lewis, Ernst & Mathias 773 (LA). Clarkia Pursh subsect. Sympherica Holsinger & Lewis, subsect. nov. TYPE: Clarkia lewisii Raven & Parnell, Ann. Missouri Bot. Gard 64: 642. 1977 [1978]. Petalis 10-35 mm longis, e 15 mm lon- gioribus, duobus vel tribus coloris zonae; staminibus plerumque quam stylo ascia proterandrus; ova- rio plerumque 4-canaliculato Petals 10-35 mm long, usually more than 15 mm long, differentiated into two or three zones of color; mature stigma held above the stamens, the stamens maturing first; immature capsule usually 4-grooved. 9. Clarkia cylindrica (Jepson) Lewis & Lewis, Madrono 12: 33. 1953. Godetia bottae var. cylindrica Jepson, Univ. Calif. Publ. Bot. 2: 332. 1907. Godetia cylindrica (Jepson) Hitchcock, Bot. Gaz. (Crawfordsville) 89: 352. 1930. TYPE: U.S.A. California: Fresno Co., Waltham Creek, near Alcade, s.d., Jep- son 2656A (JEPS). 9a. Clarkia cylindrica subsp. clavicarpa Davis, Brittonia 22: 283. 1970. Type: U.S.A. Cal- ifornia: Tulare Co., Exeter to Springville road, 1.5 mi. E of the County Fire Station, 6 June 1947, Lewis & Lewis 300 [holotype, LA (the specimen on the far right); isotypes, LA, RSA]. >» 9 = . Clarkia cylindrica subsp. cylindrica = > . Clarkia lewisii Raven & Parnell, Ann. Mis- souri Bot. Gard 64: 642. 1977 [1978]. TYPE: U.S.A. California: Monterey Co., Point Lo- bos, along the trail to China Cove from the end of the road, 26 June 1947, Lewis & Lew- is 498 (LA). p LJ . Clarkia rostrata Davis, Brittonia 22: 281. 1970. TYPE: U.S.A. California: Mariposa Co., Hell Hollow, Merced River Canyon, 3.3 mi. 494 N of Bear Valley on California Hwy. 49, 17 May 1968, Lewis, Bloom & James 1424 (LA). Dan Nicolson provided valuable nomencla- tural advice, and Joh homas commented on an earlier version of this paper. This research was supported in part by NIH grant GM 28106. LITERATURE CITED ue L. D. 1974. Genetic confirmation of the n of Clarkia lingulata. Evolution 28: 244- 1982. Conservation and duplication of iso- zymes in plants. Science 216: 373-380. & М. Е. WeeDEN. 1979. Gene duplication and phylogeny in C/arkia. Evolution 33: 1024-1039. Lewis, H. 1962. Catastrophic selection as a factor in speciation. Evolution 16: 257-271. ANNALS OF THE MISSOURI BOTANICAL GARDEN nia 94305; Bio [Vor. 73 & M. E. Lewis. 1955. The genus Clarkia. Univ. Calif. Publ. Bot. 20: 241-392. . R. RoBERTS. 1956. The origin of Clarkia li ngulata. Evolution 10: 126-138 ODRZYKOSKI, &L cations of gen p hydrogenase in Clarkia (Onagracea Raiden ad а Syst. Bot. 9: 479-489. Raven, P. Н. & D. PARNELL. 1977 [19 78]. Reinter- ten of the type of Godetia bottae Spach. Ann issouri Bot. Gard. 64: 642-643. о L. О. GOTTLIEB. 1986. Chloroplast NA evidence for the origin of the genus Hetero- ura from a species of Clarkia (Onagraceae). (Submitted) —Kent E. Holsinger, Department of Biological Sciences, Stanford University, Stanford, Califor- and Harlan Lewis, Department of ology, ao of California, Los Angeles, California 90024. HOLSTIANTHUS, A NEW GENUS OF RUBIACEAE FROM THE GUAYANA HIGHLAND Up to the present time 11 genera of Rubiaceae are known to be endemic to the flora ofthe Guay- ana Highland. They include the following: Cephalodendron, Maguireocharis, and Neblina- thamnus (Steyermark, 1964, 1972) from Cerro de la Neblina; Aphanocarpus Steyermark (1965) from sandstone table mountains and adjacent plateaus of Estado Bolívar; Coryphothamnus Steyermark (1965) from Auyan-tepui; Duidania Standley (1931) from the о mountains of Duida, Huacl on- drococcus Steyermark (1907) [now known as Coccochondra Rauschert (1982)] from the Ser- гапіа Part; Maguireothamnus Steyermark (1964) from various table mountains of the Venezuelan Guayana; Merumea Steyermark (1972) from Cerro Sipapo of the Venezuelan Guayana and the Merume Mountains of Guyana; Pagameop- sis Steyermark (1965) from various tepuis of the Venezuelan Guayana and adjacent northern Bra- zil; and Chalepophyllum Hook. f. (1873a, 1873b) from southeastern Venezuelan Guayana and ad- jacent Guyana. Although such genera as Glea- sonia Standley (1931), Platycarpum Humb. & Bonpl. (1809), Sipaneopsis Steyermark (1967), and Dendrosipanea Ducke (1935) show the great- er part of their diversity within the Guayana Highland, they may also occur in lowland sa- vannas, as in the case of Sipaneopsis and Den- drosipanea, or in Amazonian Colombia and Bra- d Marahuaca; CA Holstianthus Steyermark, gen. nov. TYPE: H. arbigularis Steyermark Inflorescentia 1—3-flora plerumque axillaris pedun- culata basi bibracteata. Calycis lobi a inae- Corollae extus salm a rubrae, Altus» pendicem subulatam abrupte desinentes. Genus monotypicum The genus is named for Bruce Holst, who col- lected the type material and served as botanical assistant on the expedition to Cerro Marahuaca. ANN. Missouni Bor. GARD. 73: 495—497. 1986. Holstianthus barbigularis Steyermark, sp. nov. TYPE: Venezuela. Territorio Federal Ama- zonas: Depto. Atabapo, forested steep sand- stone SE-facing slopes and bluffs, above branch of Caro Negro, Cerro Marahuaca, S-central portion, downstream from *'Sima Camp," 3?43'N, 65°31'W, 1,220-1,350 m, 23-24 Feb. 1985, Steyermark & Holst 130637 (holotype, MO; isotype, VEN). PARATYPE: same locality and date, Steyer- mark & Holst 130664 (MO, VEN). x 0.6-2 metralis; foliis lanceolato- SR vel о -ellipticis apice acutis vel acum natis basi subacutis vel obtusis 4.5-11.5 cm lon i 5-3.5 cm latis, subtus hirtellis; inflorescentia o axillari, floribus pedicellatis, pedicellis 4-12 mm longis dense cano-tomentosis; bracteis sub pedicellis dibus acutis vel acuminatis 8-18 mm longis 4—6 mm latis; € sub pedunculo majoribus; calycis lobis 5 imbricatis, 3 exterioribus majoribus late lanceolatis vel tirs ree ceolatis acuminatus 2-2.5 cm longis 8-9 mm latis extus strigillosis intus glabris, 2 interioribus lanceolatis 2- cm longis 8-9 mm latis; corolla infundibuliformi же ст е extus glabra, tubo 2.3 cm longo 3.5-4 m lato intus glabro, limbo intus orificio dense bar- e aliter glabro, lobis ovato-lanceolatis acuminatis 10 mm longis 5 mm latis; staminibus 5 superne in- sertis, antheris 4.5-5 mm иси е endocarpio te- nui 0.2 n mm crasso osseo me io 0.5 mm instructo oblongo- шешш. 75-1.5 m longis. Shrub 0.6-2 m tall; young stems densely hir- tellous. Stipular sheath subdeltoid, 1-3 mm long, 2-4 mm wide, densely strigillose-hirtellous with- out, abruptly terminating in a subulate, strigil- lose appendage 3-7 mm long, on sterile leafy shoots longer. Leaves а lanceolate- oblong or oblong-elliptic, acute to acuminate at apex, subacute to obtuse at base, 4.5-11.5 cm long, 1.5-3.5 cm wide, glabrous above except strigose along midrib and on some lateral nerves; petioles 1-4 mm long or the uppermost leaves subsessile, densely strigillose. Inflorescence ter- minal or axillary, 1-3-flowered, 2-bracteate, pe- dunculate; peduncle divaricate, 7-14 mm long, the upper ones shorter, densely canescent. Bracts subtending each pedicel ovate, acute to acumi- nate, 8-18 mm long, 4— m wide, glabrous above, sparsely hirtellous below, minutely cil- iolate on margins; bracts subtending peduncle larger, 18-22 mm long, 8-10 mm wide. Calyx 496 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURE 1. Holstianthus barbigularis Steyermark.—A. Habit.—B. Stamen, abaxial and lateral views. С. Stipule.—D. Interior of upper portion of corolla limb and lobes.—E. Calyx lobes and disk, seen from above.— F. Calyx and hypanthium. —G. Vertical section through ovary.—H. Transverse section through ovary.—I. Seeds. 1986] lobes 5, imbricate, persistent, somewhat un- equal, 3 outer ones broadly lanceolate to ovate- lanceolate, acuminate, 2-2.5 cm long, 8-9 mm wide, finely longitudinally 5-nerved, moderately substrigillose without, glabrous within, 2 inner lobes lanceolate, 2-2.5 cm long, 8-9 mm wide. Hypanthium subcampanulate, 3-6 mm long, 5- 7 mm wide, densely gray-hirtellous without. Co- rolla salmon-red, infundibuliform, 6.3 cm long, glabrous without, tube 2.3 cm long, 3.5-4 mm wide, expanded above into a limb 3 cm long, 10 mm wide at the summit, densely barbate within at the throat with elongate yellow hairs 8-10 mm long exserted beyond the orifice, lobes equal, ovate-lanceolate, acuminate, 10 mm long, 5 mm wide at the base. part of the limb; anthers linear, dorsifixed, rounded at both ends, attaining summit of orifice but not exserted, 4.5-5 mm long, glabrous; fil- aments 3-4 mm long, glabrous, inserted 7-8 mm below orifice. Disk fleshy, annular, 5 mm across. Style filiform, 7-7.5 mm long, glabrous; stigma lobes 2, erect; ovary 2-celled; ovules numerous, arranged on all sides of a rounded thickened pla- centa, pas gd oblong; fruiting hypanthium 7- , 7-8 mm wide. dip E osseous, en thè endocarp firm, , 0.2 mm thick; mesocarp 0.5 mm thick; s adis yellow- brown, somewhat angled and dorsiventrally sub- compressed, subelliptic-oblong, 0.75-1.5 mm long. The genus Holstianthus can be properly ac- commodated in the tribe Rondeletieae by virtue of its dry fruit, numerous, exalate or exappen- diculate seeds, and contorted corolla lobes. Among most other genera of the tribe having contorted corolla lobes, it is at once separated by its indehiscent fruit. From Sipaneopsis, a member of the tribe with indehiscent fruit, Hol- stianthus differs in the contorted corolla lobes, ous, subcompressed, subangulate seeds, densely barbate orifice of the corolla with elon- gated yellow hairs, subunequal, prominent calyx NOTES 497 lobes, bibracteate pedicels, large, reddish infun- dibuliform corolla, and mainly axillary inflores- cences The new genus herewith described was col- lected on a recent expedition to a previously unexplored sector of Cerro Marahuaca in Ter- ritorio Federal Amazonas. The trip was made possible through helicopter support provided by the Terramar Foundation of Caracas, Venezuela. Grateful appreciation is hereby given to Arman- o and Fabian Michelangeli of the Foundation for having furnished the logistics on this expe- dition. LITERATURE CITED Ducke, А. 1935. Plantes nouvelles ou peu connues de la region amazonienne (VIII serie). Arg. Inst. Biol. Veg. 2: 69. Hooker, W. J. 1873a. Rubiaceae. T enm & Hooker, Genera Plantarum 1873b. Icones Plantarum. ш Lon- don. (Pl. 1148.] HUMBOLDT, 1809. Plantae Ae- . & A. BONPLAND. TD 2: 81. t. 104. 1982. Nomina nova generica et com- nati iones novae spermatophytorum et pterido- p! ytorum O 31: 554-563. STANDLEY, vs . Rubiaceae of Venezuela. Publ. Fie us. ré Hist., Bot. Ser. 7: 372. на J. 1964. Rubiaceae. In B. Maguire & Collaborators. The Botany of the Guayana High- land. Mem. New York Bot. Gard. 10(5): 193, 220, . 1965. Rubiaceae. In B. Maguire & Collab- orators. The Botany of the Guayana Highland. Mem. New York Bot. Gard. 12(3): 263, 264, 267. . 1967. Rubiaceae. Jn B. Maguire & Collab- orators. The Botany of the Guayana Highland. em. New York Bot. Gard. 17(7): 230, 284. 1972. Rubiaceae. Jn B. Maguire & Collab- orators. The Botany of the Guayana Highland. Mem. New York Bot. Gard. 23: 220, 228, 230, 232, 403 —Julian A. Steyermark, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. A NEW SPECIES OF SOLANUM (SOLANACEAE) FROM MADAGASCAR Solanum mahoriensis D’Arcy & Rakotozafy, sp. nov. TYPE: Cultivated, Missouri Botanical Garden, D'Arcy 15487a. Seed from Mada- gascar. Antsiranana es forest of Mahory, ca. 6 km S of Marataolana (ca. 12 km S of Anivorano Nord), 12?48'S, 49°14’E (holotype, MO; isotype, TAN, du- plicates to be distributed). Frutex ad | m altus, sparse pubescens pilis stellatis, spinis rectis nemio armatis. Folia lobata, armata, gla- brata. Flores purpurei, staminibus angustis. Bacca vi- ridis odorata, calyce accrescenti tecta. Robust herb or shrub | m tall; branching from near the base, branches slender, arching, wand- like, to 1.5 m long, pubescent with occasional long-stalked often porrect, pauciradiate hairs that lack midpoints, copiously armed with acicular, sometimes flattened, straight, retrorse spines to 12 mm long, these often with a terminal stellate hair or with one or more hairs near the base. Leaves obovate, to 25 cm long, 16 cm wide, basally dimidiate, 1-3-pinnately lobed, with ca. 4 major lobes on each side, lobed three-fourths way to the midrib, on aged leaves lobed three- fourths way to the midrib, the sinuses rounded to slightly angular, the lobes apically obtuse or acute; major veins drying elevated and darker beneath, the costa and major lateral veins bear- ing numerous spines as those on the stem, the lamina glabrate above but with sparse short- stalked multangulate hairs beneath, the margin and veins with copious stellate hairs; petiole slen- der, 10-35 mm long, stellate pubescent and spar- ingly armed. Inflorescences short racemes in- serted near top of internodes, the peduncle slender, 8 mm long, stellate-pubescent and with numerous small straight spines, the pedicels 8— 10, ca. 7 mm long, resembling the peduncle but with smaller spines. Flowers heterostylous, calyx cupular, basally colored with copious short red- dish brown, dark-based spines and reddish brown stellate hairs, whitish distally with stellate hairs, the tube cupular, 7-8 mm long (3-4 mm long in short-styled flowers), whitish mauve, the lobes subulate, 10 mm long (4 mm long in short-styled flowers), the apical one-half or one-third green; corolla mauve, darker at the center and drying with fainter tips, the lobes darker with the costa elevated and darker, the sinuses rounded to ANN. Missouni Bor. GARD. 73: 498-500. 1986. slightly angulate, the lobes ca. 10 mm long (8 mm long in short-styled flowers), glabrous with- in, stellate-pubescent outside except on the si- nuses where puberulent with reduced simple tri- chomes; stamens 5, equal, the anthers yellow, subsessile, narrow, 8 mm long (4.5 mm in short- styled flowers), the terminal pores minute, the filaments purplish as the corolla, 0.5 mm long, glabrous; ovary white, ovoid, 2 mm long, the apical half glandular with minute, gland-tipped, simple hairs, the styles white, ca. 18 mm long, overtopping with stamens and apically recurved (those of the short-styled flowers rudimentary with stigmas not differentiated), with a few great- ly reduced simple, gland-tipped hairs and occa- sional branched hairs on the basal portion, stig- ma subglobose, dark green. Fruit usually one per inflorescence, pendant, a perfumed globose ber- —)3 cm across, 2-locular, loosely enveloped by the accrescent bladder-like calyx, the calyx 4 cm across, 4 cm long, proximally sulcate at the major veins, whitish, copiously armed with acic- ular spines; pedicel 15 mm long, 2 mm thick; seeds numerous, stramineous, compressed, 4 mm ong. Solanum mahoriensis is a member of Sola- num subg. Leptostemonum (Dun.) Bitt., which includes spiny species with slender anthers. Its probable sectional placement is unclear The heterostyly in this species resembles that well known in many other species of subg. Lep- tostemonum: the first flowers have long styles and bear fruit, the later short-styled flowers have greatly reduced ovaries and do not bear fruit. Anthers of both types of flower are polleniferous. The spiny lobed leaves and inflated fruit-en- veloping calyx of Solanum mahoriensis are strongly reminiscent of Solanum sisymbriifolium Lam. of South America, but that species has a white flower, a smaller, scarlet fruit, and a less spreading growth habit. The fruit of this species remains green, and it has a strong perfume-like scent, an unusual condition for berries in the Solanaceae. However, the cut surface of the juicy erry has a tomato-like очор Тһе fruiting calyx is truly echinoid g more than a spiny sea urchin—and when old, the dark red- dish brown spines contrast with the nearly white calyx walls. To judge from a reduction in pu- 1986] NOTES 499 FIGURE 1. Solanum mahoriensis D'Arcy & Rakotozafy.—A. Habit. — B. Fruit. [After D'Arcy 16045 (MO).] bescence in the apical portion of the pedicel, the of Marataolana which is са. 12 km S of Anivo- elongation of the pedicel in fruit may be confined rano Nord. This locality (12?48'S, 49?14'E) is to this apical region. tropical, in the extreme north of the island, far Solanum mahoriensis was found on level from the temperate climates of southern Mad- ground under а 60% canopy in the “Foret de араѕсаг where the unusual endemic solanums of Mahory," a wooded area ca. 6 km S ofthe village sect. Croatii D'Arcy occur. 500 Specimens examined. MADAGASCAR. ANTSI- RANANA: vestige forest tiére, sur basaltes, prés de Ma- rovato au N de l'Ankarana, 9/11/66, бо dud 24561- 5 Е); forét de Ma F (MSF, P, TE hory, 6 km S of Ma- rotaolana, ca. 12 km S of Anivorano Nord, D'Arcy 15487 (MO, P, TAN); collines et plateaux calcaires de l'Ankarana, forét tropophile, 300 m, Humbert 18936 (MSF, P). Supported by a grant from the National Geo- graphic Society. Thanks are due also to Steve ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Kingsley, Antsiranana, who helped locate the species in the — William G. D'Arcy, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63110; and Armand Rakotozafy, National de Recherches de Tsimbazaza, B.P. 4096, Antananarivo, Mad- agascar. A NEW PASPALUM (POACEAE) FROM MESOAMERICA The enormous pantropical and warm temper- ate genus Paspalum includes at least 400 species (Chase, 1929). Paspalum is recognized by the inflorescence of one or more unilateral racemes, the panicoid spikelets being plano-convex, often lacking lower glumes, and occurring solitary or in pairs on the lower side of the raceme rachis, with their upper glumes and fertile lemmas turned toward the center line of the rachis. The species described occurs on Cerro Uyuca, Honduras, an ancient volcanic peak near the town of El Za- morano. Because of its proximity to the — Herbarium of the Escuela Agricola Panameri cana, Uyuca has been thoroughly botanized and is the type locality for a number of new species. Paspalum uyucensis Pohl, sp. nov. TYPE: Hon- duras, 26 July 1951, Freytag s.n. Herbarium of G. B. Van Schaack 3274 (holotype, MO- 3212543). p perenne caespitosum rhizomata curta, no- nodi soa. barbati; ligula truncata membranosa 0.5-0.8 m longa; ‘aah aminae planae, papilloso-pilosae, 8-15 cm x 8-11 m sx dn dpi sine о Inflores- centia termin 6-9 cm alta; emi , ascen- dentes, laxi; rachis ец ciliata, tenuis. Spiculae mm longae, obo 7 o <~ ES S © supernum (fertile) 1.5 m longum, stramineum,; palea striatula. Caespitose perennial from short knotty rhi- zomes; basal scales papillose-pilose; culms 40— 65 cm tall, erect, the bases swollen; internodes glabrous; nodes retrorsely bearded. Lower sheaths papillose-pilose, keeled; upper sheaths glabrous, the uppermost elongated, bladeless. Ligule a uncate membrane, 0.5-0.8 mm long; blades loosely papillose-pilose on one or both surfaces, 8-15 cm by 8-11 mm, flat, the midrib prominent beneath. Inflorescence terminal, 6—9 cm long, the peduncle slender, arcuate, exserted 5-12 cm; ra- cemes 4—6, ascending or spreading, slender, 3-7 cm long; rachis triquetrous, ca. 0.3 mm wide, bearing scattered elongated hairs; pedicels usu- ally somewhat united, the terminal one to 1 mm ae Spikelets paired, brownish purple, 1.4-1.5 m long, obovate 1.5-1.7:1; lower glume ab- sent; upper glume 3-nerved, sparsely appressed- pubescent, slightly shorter than the spikelet; low- er lemma sterile, 3-nerved, glabrous; upper lem- ma fertile, 1.5 mm long, stramineous, elliptical 1.4: 1; palea minutely striate; caryopsis elliptical 1.4: 1, brownish. Other тюр ехатї E ue HONDURAS. Cerro e Uyuca, Freytag s. rb. Van Schaack 3273 (MO- 2311478); Cerro a cun 147 A (MO- 2926547). This species is presently known only from up- per pine and cloud forests on Cerro Uyuca and Cerro Monserrat in central Honduras at eleva- tions of 1,400-2,000 m. Paspalum uyucensis belongs to the informal group Caespitosa of Chase (1929). It is closest to P. umbratile Chase, from which it differs in its bearded nodes, more numerous racemes, small- er, broader spikelets, and 3-nerved glume and lower lemma. Journal Paper Number J-12090 of the Iowa Agriculture and Home Economics Experiment Station, Ames, Iowa; Project 1833. LITERATURE CITED СНАЗЕ, A. 1929. The North American d of Pas- palum. Contr. U.S. Natl. Herb. 28: 10. — Richard W. Pohl, Professor of Botany and Cu- rator of the Herbarium, Department of Botany, Iowa State University, Ames, Iowa 50011. Volume 73, No. 1, pp. 1-224 of the ANNALS OF THE MISSOURI BOTANICAL GARDEN, was published on 28 May 1986. ANN. MISSOURI Bor. GARD. 73: 501. 1986. › 4 INFORMATION The ANNALS publishes original manuscripts in systematic botany and related fields. Authors ked to follow tl ggesti below in order to expedite editing and publication. Manuscripts not prepared properly may be returned for re- vision prior to review. If an author feels that his manuscript presents special problems, he should write the editor concerning the best way to han- dle these before submitting the manuscript. Page costs are $60.00 per page; charges may be re- duced under special circumstances, and the ed- itor will help authors to seek additional funding if necessary. Corrections on manuscripts in the proof stage will cost $2.50 per line and are not subject to negotiation. Manuscripts must be typewritten on one side of good quality (not erasable) 8%” x 11" paper. The manuscript should have at least one inch margin all around and be double-spaced through- out, including the abstract, notes, legends, ta- bles, lists of specimens, and literature cited. Fig- ure legends and tables should be typed separately and placed at the end of the text. Legends for each group of figures should be typed as one bles. Include author’s name and page number in upper right-hand corner of each page. Submis- A of the original and two carbon or xerograph- 1€ Copies of the manuscript is required, and the ut should retain a copy of the final, typed FOR AUTHORS style found in other ANNALS issues, e.g., in Lor- ence, 1985 (72: 1-166). Keys should be dichot- omous, indented, with numbered couplets. Cite representative specimens as follows: Mexico. - axaca: Sierra San Pedro Nolesco, Talea, Jur- gensen 865 (BM, G, K, US). Use herbarium ac- ronyms as listed in Index Herbariorum (Regnum Veg. 106). Use authority abbreviations as listed in Draft Index of Author Abbreviations compiled. at the Herbarium, Royal Botanic Gardens, Kew — (1980). ee ои Abbreviations should be checked for consis- = tency and to make sure they are unambiguous. = Periods are used after all abbreviations except — measures, compass directions, and herbarium designations. Send copies of illustrations; origi- nals will be requested after the manuscript is accepted. All illustrative materials should be mounted on stiff cardboard. A maximum size of 68 mm) or full page width (5'2" or ca. 140 mm). | Figures are numbered consecutively with Arabic. numbers, in the order they are referred to in the text. Photographs should be sharp, glossy, black- and-white prints. Abut edges of photographs in composite plates. Do not mix line copy and con- tinuous-tone illustrations on one plate. Maps should include a metric scale and reference to latitude and longitude; where appropriate, size and scale should be included in photographs or drawings, not in figure legends. All original il- lustrations will be returned after publication. The bibliography should be compiled hai care. The author s 4L + + г 1 Acknowle A Е é or + used in the manuscript. } nr nar npany each paper oth- ert m У : han “Notes. The abstract should succinctly ummariz e D Ba tay Use one pa XOn, author, g AA 188 N. A sions of e the purpose, findings, and conclu- o! the paper and should be completely com- - Б due issue _ Style, entry 101108 phy is rel ec dr 5 and each reference to a paper in the text is entered in the bibliography. Citations of periodical lit- erature should appear as follows: author’s last For other aspects of style, consult a recent ‚ of the ANNALS; The Chicago Manual of 12th or 13th edition, University of Chi- res: . Chicago; or write to the editor ripts s f the Missouri Botanical Garden, . Louis, Missouri 63166. — Contents continued from front cover Giannasi e la creel ie iden С ^ Phylogeny of the Hamamelidae: Taxonomic Index eher & реше ОН intr в ea ke NE ERG MC MON UI unus о Palbcatile pilares and Climatology of South America a IROMO ыык ки Se сс с New Taxa of Caladium, Chlorospatha, and Xanthosoma (Arsenal Colo dn d from Southern Central America and Northwestern Colom bia chael H. Grayum e ecce RR ; New Taxa in рга (Опаргасеае) Warren L. Wagner —- Studies in the Araliaceae of Nicaragua, and a New boa Зреск Отеорапах M. ah Cannon & J. F. M. Cannon .. х МОТЕЅ Chromosome зне of New Caledonian Plants Gerald D. ordon 1 McPherson. AM а CN ы | A New Combination in ee Sect. Microcarpium (Ona Chine] Feng: o de ne hn tl np rer Description of a New Section and Subsection in Clarkia (Onagraceae) Kent E. Holsinger & Harlan Lewis erem Holstianthus, a New сезш. of Rubiaceae from the Guayana Julian A New Species of Salam Койке) Bom Madagascar D'Arcy & Armand. Rakotozafy ... a A T er (Posse) from Mesoamerica ANNALS SOURI BOTANICAL GARDEN puME 73 1986 NUMBER 3 Sketches of cacti by G. Engelmann . CONTENTS T y m in North American Botany, A Symposium Commemorating George ngelmann: The Thirty-First Annual Systematics Symposium far ' Shall R. Crosby —— i Instructions for the Collection РЕ Preservation rof Botanical Spec eet- mens George Engelmann кык T S р асре 3 Changing Botany in North America: 1835-1860. The Role oE George Engelmann Elizabeth A. Shaw ...— 508 Augustus Fendler (1813-1883), ролю Plant | Collector: иче orrespondence with George panem Michael T. Stieber & Carla Lange .... 920 The Ever-Changing PA of Cactus Systematics Arthur C. Gib- - son, Kevin Es Spencer, Renu Bajaj, & Jerry Ez McLaughlin . = 534 accas and Yucca Moths—A Historical Commentary Herbert G. в г. 556 Cenozoic 385 of. Goat Western American Pines ‘Daniel | 1, elrod . 565 A | Historical Sketch of the min of the Мой ‘and | Col lorado Deserts of the American Southwest Robert Е. Thorne ................... 642 VOLUME73 AUTUMN 1 icon NUMBER ANNALS OF THE MISSOURI BOTANICAL GARDEN The ANNALS, published quarterly, contains papers, primarilyin — systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be ac- cepted. Authors should write the Editor for information concerning - arrangements for publishing in the ANNALS. Instructions to Authors. are printed on the inside back cover of the first issue of this volume. " amne EE n ——— —G—ÉÓÓMN PES a Se —_-_ EDITORIAL COMMITTEE Nancy Morin, Editor Missouri Botanical Garden MARSHALL R. Cro Missouri Botanical cando RRIT DAVIDSE praises Botanical Garden JouN D. DwYER 4 Missouri Botanical Garden & Saint Louis University . PETER GOLDBLATT = Missouri Botanical Garden For subscription information contact Department 11, . Box 299, St. Louis, MO 631 166. Subscription price is $70; per volume U. $ day Cda and Mexic $80 all other co | untries оой ае ыц $35 per v olume. Fou Price эрт to ааа without notice LG ANNALS MISSOURI E BOTANICAL GARDEN VOLUME 73 1986 NUMBER 3 TOPICS IN NORTH AMERICAN BOTANY, A SYMPOSIUM COMMEMORATING GEORGE ENGELMANN The thirty-first annual Systematics Sympo- sium of the Missouri Botanical Garden was held on 19 and 20 October 1984, the year that marked the centennial of the death of George Engelmann. Since George Engelmann was so important in the establishment of the basis of a scientific pro- gram at the Missouri Botanical Garden, and since he made so many early contributions to the knowledge of the flora of North America, we decided to commemorate him in the thirty-first annual Systematics Symposium. The papers pre- sented at the Symposium are devoted to various aspects of botanical topics that interested En- gelmann: Yucca and its pollination by moths; the pines; the family Cactaceae; aspects of the phy- togeography and vegetation of the American West; and the interactions of Engelmann with his Harvard contemporary, Asa Gray. Certainly a one-day symposium could not hope to do full justice to a consideration of the contributions ANN. MISSOURI BOT. GARD. 73: 503. 1986. that Engelmann made to North American botany and the progress in the same areas in the century following his death, but the papers contained in this Symposium, when considered in connection with Engelmann’s own publications and publi- cations based on botanical specimens that he caused to be collected in the West, give an in- dication of the progress that has been made and the state of our knowledge in these areas. Dr. Engelmann appeared at the closing session of the Symposium and presented a lecture, il- luminated with slides, of his life and work. This Symposium was supported in part by grant BSR-8311392, Gerrit Davidse, Principal In- vestigator, from the National Science Founda- tion to the Missouri Botanical Garden. — Marshall R. Crosby, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166. INSTRUCTIONS FOR THE COLLECTION AND PRESERVATION OF BOTANICAL SPECIMENS! GEORGE ENGELMANN? In gathering plants you will do well to pay attention to all the plants you come across, whether showy or unsightly. Do not neglect the latter on account of their appearance. Collect if possit fthe same plant, partly to show different states of the same species, and partly to be able to distribute them among different botanists. Do not be deterred from gathering what ap- pears to be the same species at different places and seasons. It may prove not to be the very same species, but only an allied species; or even if identical it is interesting for the study of geo- graphical botany to have the same species from distant localities. On the whole collect only such plants as you find in flower or fruit; but trees and shrubs ought to form an exception, as also smaller plants, if they afford some particular interest, either by their medicinal or other properties, great pre- ponderance in certain districts, etc The most important part of the plant is the flower and fruit. Get if possible such specimens as present both states, flower and fruit, or both on different specimens. You will find plants which have fertile and sterile flowers distinct, they oc- cur either on the same plant, as in the oaks, hick- ories, etc., or in different plants, as in the willows, cottonwoods and others. In both cases it is im- portant to collect specimens which show each of them. Many plants develope the leaves after the flow- ers, as the oaks, redbuds and many others. In these the flowers must be preserved, and later in the season, the leaves with the fruits; but great care must be taken to get them from the same species. If the specimen you obtain is not too large, gather it entire, with the root or at least with part of it, so as to show the nature of that organ. Try to have the lower as well as upper leaves com- plete on the specimen, especially if they should differ from each other. In case the specimen is too large for a sheet of paper, say more than 17 inches high, it may still be preserved entire, by bending or rather break- ing (without entirely severing the parts) the stem in an acute angle. If necessary, this may be re- peated and branches or leaves may also be treat- ed in the same manner. This is better than cutting it in different segments, as these might become separated and much confusion ensue from this cause. Of still larger plants, shrubs, trees, it is possible to take only a part, a branch, etc.; but if there should be different leaves on the plant, it will be necessary to cut off such leaves with a small piece of stem attached, and preserve them with the other specimens. Make the specimen large enough to present a fair sample of the plant, its manner of growth, ramification etc. It will be well to put your specimens in paper as soon as gathered; their parts are then fresh and still and are easily spread out in a neat way; but if they become flaccid they present much diffi- culty, and the dried specimens will appear un- sightly. Large specimens with thick stems or roots (bulbs especially) or even very clumsy flowers (as ! These instructions exist as a manuscript in the George Engelmann Papers at the Library, — са Сагаеп. They are e printed here verbatim and unedited for spelling and punctua tion. The m cript is scrivener’s hand, but there are a few corrections and ин ei Engelmann, and he Е ап ión of = of p how to bend a specimen that 1s "too large fora shee et eh KL 11 Josia h Gregg wrote to Engelmann: “I could ane no o stiff ИМ € portfolios" de Febru nfin that your piod En in every regard, have been of i system of num collector could take to as ip directed" (14 April 1849). (See. Diary E Letters of Tosiah 1, 1944, University of Oklahoma Press, Norman.) The Instructions provide a glimpse of him. For a 1847); “I repeat ite use to me” (13 August 1847); “I tae followed the Gregg, M. G. F heart. The legends to the illustrations were written on Dr. Engelmann’s behalf by M. R. Crosby, who remains close to him through his correspondence and other papers. ? Address reprint requests to: George Engelmann, M.D., % M. R. Crosby, Missouri Botanical Garden, P.O. Box 299, St. Louis, MO 63166. ANN. MISSOURI Bor. GARD. 73: 504-507. 1986. 1986] ENGELMANN —PRESERVATION OF BOTANICAL SPECIMENS 505 PEO - ч—— M BOTANICAL Garden 3 SEE MA eH ‘ners “i b 1 Мы = [ob ИМ у үа рб Anearly St. Louis a After кон my medical degree from Wiirzburg and spending a year studying i me grati 24, 1832 and immedia С. proceeded to the Shiloh Valley of St. Clair County, Illinois. bows next two years were ^ real though I made an interesting trip to the Arkansas Territory, losing my money, gun, and health along the . Returning to Illinois, I decided to recross the River to St. Louis. Settling there in November 1835, I established my panes h 3rc = Bee т aene, ved to remain long enough in this frontier town of 8,300 Anglo-Americans to sav ney to return to my native Germany. But, my practice flourished and bus plants of is "We st were pip more A than those of Europe. In 1847, against the advice of my s, I moved to the far western part of town, to the southwest corner of 5th and Elm Streets, where Busc i Memori IS di m now stands. Here my medical practice pire. to expand, and here I carried out many of my botanical р The sketch As here is among my Papers, now preserved in the Library of the Missouri Botanical Garden, and shows my herbarium, ar adjoined the medical office. As later workers have found, I u reused paper (see Аа . Missouri Bot. Gard. 61: 907. 1974), and this sketch is on the back of later detailed drawings of Jsoétes. It is labelled To ai Elm & 5th" at the top; the жале аин погі pn The up-side-down scale along the bottom is in English feet. The outline of the room indicate windows on the north side an one each on the east and south sides of the room. Doors are located авна асгоѕѕ fom e ach other in the northeast and southwest corners. The dark oblong near the west wall is the free-standing, coal-burning stove, connected to a a by a short, dark pipe. There were two wor pine indicated id diagonal marks and circles in each corner as legs, one on the north wall, shelf above, and one near the center of the room. A third table, “Chemisch rebels Tisch," was placed perpendicular Ки pud Bii wall, next to the work table with the shelf. The eight nearly square, cross-hatched structures along t, south, and west walls are herbarium cases, as is the block of six, free standing, diagonally marked sucre "between the center table and the south wall. These cases’ capacity was more or less equivalent to the steel ones in common use in the late twentieth century, so the room held about 14,000 specimens, when filled to pest a common problem even 150 years ago! There was a book shelf in the northeast corner, next to the door. In 1869 I moved farther west yet, to Locust and Garrison Streets, and A course the herbarium went with me. I willed my herbarium to my only child, George Julius Engelmann, also a physician, but not a botanist. At that time the collection numbered about 98,000 specimens, inc luding epis nonvascular cryptogams. George J. donated the Herbarium to the Missouri Botanical Garden before 1890. 506 large thistle heads) often require to be split lengthways, so as to make them less bulky and injurious to the other plants in the herbarium. Large fruits may be also split, or they may be preserved separately, wrapt in paper In putting up the specimens, Bread them out in such a manner that all the different parts are seen, and the flowers, or some of the flowers, are laid open. If, however, time should be wanting, the plant may be laid in the paper just as it is. The object of pressing plants, being to keep all the parts spread out smooth, and free from shrinking and wrinkling, but not to crush the more delicate organs, the flowers especially: the pressure should be moderate, say from 25 to 40 or at most 50 pounds in weight, so as to compress not destroy the organs, that they may afterward be examined. In traveling, two boards tightly strapped together will be quite sufficient to press plants. At home any weight of 30 or 40 pounds will do the same service. After the specimen has been put in paper and pressed a while, it becomes necessary to change the layers of paper as soon as they have become damp from the moisture absorbed from the plant and to substitute dry ones for them. This ought to be repeated daily till the specimen is com- pletely dried. he most convenient method is to put the specimen in finer paper, say printing paper, then a layer of 2, 4 or 6 sheets of coarser bibulous paper, then a finer sheet with a plant and so forth. In changing the plants dry layers are substituted for the moist ones without removing the speci- men from the finer sheet immediately touching it, which would be a tedious job, often injuring the specimen. The damp layers are then hung up or spread out and dried. The dried specimens are or sheets of paper, as Many in one sheet a as may be put there without injuring each other. When you have got a suffi- ciently sized bundle together, pack it either in a box of convenient size or in a fresh skin of some animal (hair inside) which will harden and shrink and form an easily handled and safe package. A specimen is of much less value if not prop- erly labeled. Therefore as soon as collected or when put up, attach a piece of paper to it, (the most simple method is to stick the stem through a hole in the paper,) on which you note at least date and locality; but if possible also every thing you can ascertain about the plant and which does not appear in the dried specimen itself: colour of flower, taste, smell, time of opening and clos- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 ing of flowers, size of the plant, height, е of stem (in trees); nature of the soil (swamp, sa rocks, open places, shade, etc.), whether бош or rare. In parasitic plants it should be stated on which plant they grow. Besides this it will be well to number your plants as you collect them. This number will stand for a name and can always be referred to, especially if you keep a journal or some other memoranda of your collections, or in correspondence with other botanists, to whom specimens may be communicated. You wul further materially oo our | fthe veget ou could collect any parts of plants or pain of plants, which may be valuable or curious, such as me- dicinal roots, barks, gums, etc ake it an especial object to collect the fruits of plants which cannot be preserved in the her- barium, such as pine cones, nuts and others. Get also specimens of the woods. Of stems not thick- er than 3 inches in diameter, take a whole piece 12 inches long, of larger ones only a section of the stem, showing bark, alburnum (exterior soft wood) and hard interior wood. All such specimens of fruits, woods, roots, may very conveniently be labeled and marked with the same numbers as the specimens of the plant from which they are derived. Collect also ripe seeds as many as you can get. Preserve them in their pods or fruits as they keep longer fresh in them. In wrapping them up, put if possible, a few leaves, a small branch, flowers or whatever part you can get, with them, and number them with the same number as the dried specimen of the same plant. Seeds ought to be sent as soon as possible, as many lose their power to germinate if too old. They ought to be packed dry but not too tight as they may suffocate and moulder. I have more especially studied several families to which I wish to direct your particular atten- tion. I mention the Asclepiadeae and Euphorbi- aceae, both comprising plants with milky juice, the first mostly with showy, umbellated flowers and silky appendage to the numerous seeds, the other with very inconspicuous greenish flowers and seeds. Further the Pines—then all the par- asites such as the Cuscutas (Dodder or Lovevine) and the species of Viscum which grow on trees (Mistletoe) some of them of great interest on Pines in the Pacific region. Above all others I mention the Cactus tribe, which I have not only studied (Fig. 1), but also cultivated. Specimens of Cacti in flower and in 1986] ENGELMANN— PRESERVATION OF BOTANICAL SPECIMENS fruit are important as well as entire plants, es- pecially living ones for cultivation. For the barium the flower must be preserved with a piece of the plant attached to it, which shows the ridges or tubercles and spines. In your labels do not forget to describe the shape of the entire plant, number of ribs, number, shape, direction and color of spines in each bunch. The Cacti are easily propagated by seed, but also living plants may be kept very long (from 6 to 12 months); they must be kept dry and not packed too close, nor before they have been kept for some time withering, or they will rot. Young plants are preserved better than old ones in this manner. 507 The most convenient apparatus for drying plants I have found in traveling is the following. 20 or 25 of such layers on one string, each fas- tened about one inch from the next. Put them in a strong pasteboard portfolio, and put several of these portfolios in a press of two pieces of plank, strapped together tightly. The strings of layers, when damp, are hung up in the open air and quickly dried, and put together again without loss of time. Finis. FIGURE 1. ps. Th se notes relate t an enumeration of the plant My botanical works were often baee illustrated, especially by Paulus duni. rather than ad e two pda M her y paper published with Dr. s collected in Texas, ae — to сое ” Boston J. Nat. His of the plants I studied, and these are now re accompany four pages of notes I made from Gray, “Plantae Lindheimerianae, t. 5: 210- 264. 1845. I labelled the rp at the left Cereus caespitosus to Benson’s recent mon ograph. That at the right I called pte Seinen difficult complex: it may represent any one of a number of species of Opuntia. CHANGING BOTANY IN NORTH AMERICA: 1835-1860 THE ROLE OF GEORGE ENGELMANN! ELIZABETH A. SHAW? Although the history of a subject or of a time should treat the parts played both by the “great” — or by the well-known—and by lesser figures, it is generally true that there are some few persons who stand out amongst the others and who give definition to the subject and time. Even the most cursory lonk at botany i in North th A merica shows pic- eminent the triumvirate— Asa Gray, John Tor- rey, and George Engelmann. There were other outstanding botanists at the middle of the cen- tury—one might mention W. S. Sullivant and Edward Tuckerman—but it was Gray, Torrey, and Engelmann who had the greatest effect upon their adopted field. Their combined careers as botanists cover nearly seventy years. During this time, botany in North America developed from the simple collecting of plants and amassing of data with consequent assignment of genera and species into Linnaean pigeonholes, into some- thing like modern botany. I think, too, that examination of events in any field of science cannot be usefully done unless seen against the background of contemporary po- litical, social, and intellectual realities. It can be done neither for the present nor for what might seem to have been a simpler and less complicated time. In any case, developments in botany in the United States during the middle of the nine- teenth century were pulled along by the political events, and I have chosen to consider this time by looking at the relationship of botany to ex- ploration in the American west and at the part played by one man, who was in the proverbial “right place at the right time." The subject is botany on the American frontier and for the first sixty years of the century, which is nearly equivalent to the whole of American botany. Each of the three had a well-defined role. ach was originally trained as a physician, but only Engelmann maintained a medical practice. ! [ am grateful to those people who heard, at greater a than they might have w Botany in North America”; and I owe particular thanks and suggestions. Any eccentricities of interpretation are, of course, m Torrey, the oldest of the three, earned his living by teaching chemistry—at West Point, at Prince- ton, and at the College of Physicians and Sur- geons in New York. Although Torrey had close ties with many correspondents in Europe, he was, par excellence, a botanist for North America. He was from the 1820s involved with preparing the botanical reports for government-sponsored ex- peditions and until after the Civil War plants from such expeditions went to Torrey, who had the overall responsibility for them. He prepared, for example, the final report on the botanical collections made by the Boundary Commission, which operated after the Mexican War. Torrey was much less concerned than were Gray and Engelmann in direct dealings with the private collectors active during the 1840s, but for many years it was he who would have the most direct connection with any official in Washington who might be helpful to botany. Gray I see as the organizer among the three. From his arrival at Harvard in 1842, he was at the center of botanical activities in the United States and he came to have the greatest breadth of interests. His first trip abroad in 1838-1839 allowed him to make contacts and to open chan- nels of communication that made Gray the in- ternationalist among American scientists—cer- tainly among botanists. For thirty years Gray provided a one-man abstracting service—pre- senting in the American Journal of Science and Arts, “Silliman’s Journal," information about botanical work in Europe—news and reviews and criticism of work current. Engelmann was the active physician with a busy and demanding practice in St. Louis and would seem to have had the least time free for botany. Engelmann would later prepare finely done treatments of several very difficult groups of plants, but during the 1840s he was the man at the frontier. Engelmann had the most direct ished, about ‘ ‘Changi in o Bernice Schubert and to Peter i rvard wn. a pan ined Herbaria and of the Missouri Botanical Garden for permission to quote from manuscript materials in their c ? The o Herbarium of Harvard University, 22 Divinity Avenue, Cambridge, Massachusetts 02138. ANN. MISSOURI Bor. GARD. 73: 508—519. 1986. 1986] contact with collectors, both those operating pri- vately and those in government employment or under government sponsorship; he gave advice and provided money, often his own, to men who might bring back plants. To him, the frontier was immediately at hand; writing in 1845 to Torrey [the letter quoted in Rodgers (1942)], Engelmann said, “You can have no idea how near we here consider ourselves now to Oregon & California; we mentally travel with those thousands of em- igrants, and begin to think the Rocky Mts not much further off than the Alleghanies.” And En- gelmann could interpret the frontier to the col- leagues back east, neither of whom would ever have any direct experience of the west in its fron- tier stage. It has been suggested by the historian William Goetzmann (1966) in the introduction to “Ех- ploration and Empire" that: “Exploration can be seen unfolding through major periods each char- acterized by a dominant set of objectives, par- ticular forms of exploring activity, distinctive types of explorers, and appropriate institutions which governed these other factors.” The first of Goetzmann’s periods runs from the expedition of Lewis and Clark (1804-1805) to about 1845; the second, from then to the start of the Civil War in 1861. The third is the time of the great multi-purpose, government-orga- nized surveys after the War; King’s Geological and Geographical Exploration of the Fortieth Parallel, Wheeler’s Geographical Surveys of the Territories of the United States West of the 100th Meridian; Hayden’s United States Geological Survey of the Territories; and Powell’s explora- tion of the Colorado, which opened up the last wn territory in the continental United States. The “great surveys" were done by the end of the 1870s and exploration in the west was effectively complete. Ten years later the Report for 1890 of the Bureau of the Census declared that the continuous frontier was no more. It seems, though, that so far as exploration that contributed to botany is concerned, it is more useful to think about four phases. For con тепсе I shall refer to “botanical a with the understanding that we are dealing with exploration, for whatever proximate purpose, that brought back plants. The first period I shall rec- ognize is Jeffersonian in conception and extends from the end of the Revolution to 1835; and it is not by chance that I’ve chosen to end this period the year in which George Engelmann set- tled in St. Louis to start his medical practice. SHAW —GEORGE ENGELMANN 509 Although “‘manifest destiny" as a national slo- gan dates only from the 1840s, the idea that our national destiny lay in expansion westward to the Pacific, no matter that territories claimed by France, by Spain, and by England lay in the way, goes back at least to the beginning of the nation. For Thomas Jefferson, who combined great na- tional pride with a passionate interest in natural history, the west was an area bound, someday, to become part of the United States, and during his presidency exploration west of the Mississip- pi was begun. The second phase is, if the analogy be extend- ed, Grayian in conception and continues from the 1830s through the end in 1848 ofthe Mexican War and includes the post-Mexican War bound- ary surveys; these were significant both for their scientific results and for bringing into the field the first concentration of botanical collectors — Charles Wright, John Bigelow, George Thurber, Arthur Schott, and Charles Christopher Pa each man having other duties, but collecting plants as well. The third period covers the years ofthe Pacific railroad surveys, which during the 1850s tied the newly acquired territory to the old United States; this period, for several reasons, ends at the end of the decade. The last phase is Goetzmann's period of “Great Surveys," up to the start of the twentieth century. The two middle periods, 1835 to 1860, pro- vide the background for the major changes in American botany and for the changes in the way in which exploration was carried out. It is exactly during this time that Engelmann was botanical gatekeeper, advisor, banker, and holder-of-hands for collectors setting out from St. Louis, or acting under his direction. But by the end of the 1850s the direction of Engelmann's botanical activities had changed; he had met Henry Shaw and in 1856 left St. Louis to spend more than a year in Europe visiting herbaria and buying specimens and books for Shaw's botanical garden. And the end of the 1850s marked the end of one phase of western exploration. I think that a strong case can be made for recognition of 1860 as the time of a major shift in scientific activities in this country, and spe- cifically, in botany. It is the historical divide be- tween botany that was a continuation of a tra- dition eighteenth century in origin and botany that shows clearly the first signs of growth into its modern form. There were several forces operating to change 510 botany in ead United States. For twenty years the slow movement of science into the hands o sisse i.e., people whose livelihood was gained from science, had been going on. The bot- anist as professional in this country can be dated from the appointment of Gray in 1842 to Har- vard. But the event that had the most immediate effect upon botany was the passage by Congress of the Morrill Act of 1862, which founded the system of land-grant colleges that were required to specialize in “agricultural and mechanic arts.” These colleges arose quickly after the Civil War, and following the requirement for agricultural arts, gave in their curricula a prominent place to botany. So, a crop of young botanists appeared, who were aware of, and eager to learn and use, new techniques from Europe. Indeed, some of them went to Europe for study and to acquire the degree of doctor of philosophy not then given in this country. When they returned, there were opportunities for them in the new universities, which, in contrast to the older ones, were unen- cumbered by the inertia of programs already es- tablished and sometimes a bit old-fashioned (Pauly, 1984). Quite suddenly botany diversified from collecting and taxonomy into a peacock's tail of possibilities. The nature of exploration changed, too. Dur- ing the Civil War little was done except in Cal- ifornia, but then came the four great surveys. Although two, King's exploration of the fortieth parallel and Wheeler's surveys west of the 100th meridian, were under army sponsorship, they were civilian-organized and their goal was not mere military reconnaissance but the organized acquisition of scientific information. The parties were accompanied by men who were along, not to be an “assistant computer" as was Charles Wright, or a surgeon/naturalist, as was John Big- elow, but to be scientists. And one should add to this the changes brought by the increase in population, the appearance of foci of settlement in the west —Salt Lake City was founded in 1847 and Denver in 1858—and by railroads. People interested in plants were, more and more, in situ and travel was becoming easier and safer. THE BOTANICAL FRONTIER IN 1835: PHYSICAL AND INTELLECTUAL A is tempting to offer up the west of 1835 as ul 1 be written during the next twenty-five years, bur this would be quite wrong. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 People's views of the west have been varied; it could be a place to live, a place to exploit, and a place to explore; and this last in two senses, the adventurer's desire to see what lay over the next hill, and the scientist's desire to have in his hands and under the microscope whatever lay over the next hill. The west initially offered seem- ingly unlimited for Thomas Jefferson it was a rich mine to dis- prove the claims then current and popular in Europe that America was a rude and miserable wasteland, degenerate and monstrous, notewor- thy only for the poor quality of everything placed here by the creator (see, for example, de Pauw, 1768-1769) After some unsuccessful attempts to get ex- plorers into and across the west, Jefferson, as President, was able in 1804 to send Merriwether Lewis and William Clark up the Missouri and eventually to the mouth of the Columbia River, and they would bring back the first American- made collections of natural history materials. The plants were not the first from the west—in 1786 Jean-Nicolas Collignon with La Pérouse's ex- pedition had sent back to France seeds of Abronia umbellata collected in California. Collignon and is shipmates died in the Solomon Islands, but the Abronia flowered in the Jardin des Plantes and was described as a new genus by de Jussieu. And still before 1800 Taddeus Haenke with the Malaspina expedition and Archibald Menzies with Vancouver's expedition had visited the west coast (McKelvey, 1955). The greatest contributions to the botany of the west were made homas Nuttall, of whom Gray said, “... no botanist has visited so large a portion of the United States, or made such an amount of observations in the field and forest. Probably few naturalists have ever excelled him in aptitude for such observations, in quickness of eye, tact in discrimination, and tenacity of memory" (Gray, 1844). Nuttall as a young man emigrated from York- shire to Philadelphia in 1807 or 1808, and over the next twenty-five years wandered across the United States as it then was, and far beyond, protected, it seems, by innocence and a blazing and consuming interest in plants. He made three trips into the west. In 1810 Benjamin Barton of Philadelphia enlisted Nuttall to go to the Great Lakes, then northwest to Winnipeg, and return by the Missouri and Mississippi. Nuttall traveled during 1810 and 1811, part of the time with another Englishman, John Bradbury, whom he explor ation, and 1986] had met in St. Louis, and got as far as Fort Man- dan, North Dakota. The second trip in 1818 and 1819 took Nuttall down the Ohio and up the Arkansas into eastern Oklahoma. In 1822 Nuttall was appointed cu- rator ofthe botanical garden at Harvard, but after eleven years resigned to join an expedition head- ed up the Missouri and on to the Columbia. On this trip Nuttall again passed through St. Louis, but Engelmann was not yet settled there. Once on the west coast Nuttall made two trips to Ha- wali and became the first “American” to collect in California, returning in 1836 to the east. This was his last trip; in 1842 Nuttall returned to England to take up a bequest that required him to spend most of his time at home (Graustein, 1967). Indeed, by 1835 a map of areas from which plants were known would point — and this list is far from inclusive—to the following botanically- productive forays. In 1819 and 1820 Stephen Long's “Yellow- stone Expedition," with William Baldwin (who died en route) and then Edwin James, as botanist, did not get to the Yellowstone, but did make a great loop up the Missouri and North Platte rivers into eastern Colorado, and back to Missouri through the Texas panhandle, Oklahoma, and Arkansas. During the 1820s there was also bo- tanical activity on the California coast, visited in 1826/1827 by Collie and Lay with Beechey's expedition; and in the northwest John Scouler and David Douglas (ofthe fir) collected in Wash- ington and Oregon. By the end of the 1820s Jean Louis Berlandier was working in eastern and southern Texas (see McKelvey, 1955). A bit later (1830-1832) Douglas was in Cali- fornia, as were Thomas Coulter and Nuttall, while another Thomas, Drummond, was collecting in east Texas. The northwest, still under British European explorer, Humboldtean and romantic. The botanist among these was Prince Maximil- ian zu Weid-Neuweid, who during 1833 traveled up the Missouri to Fort McKenzie in Montana. Maximilian is best known as an ethnographer, and his trip is better known for the beautiful watercolors done by Karl Bodmer, but Maxi- milian did collect plants that were written up by Nees ab Esenbeck. SHAW —GEORGE ENGELMANN 511 So the map shows that by 1835 we knew some- thing of plants of the upper Missouri, of the northwest, of coastal California, and of parts of Texas, Oklahoma, and Arkansas. What the map shows untouched is more significant—the great unknowns of the Great Basin and the southern Rockies and of the lands to east and west — from San Antonio westward to San Diego; and the desire to fill in this blank would drive Asa Gray, and through him, American botany for the next fifteen years. In 1835 anyone who wished to do serious work on American plants would have had to go to England, where so much of the material from the northwest and from California was, and to Ge- neva where de Candolle was working on the "Prodromus." North American botany was not well-known; but neither was it unknown, for there were several floristic works, some “national” in scope, others regional or local in coverage. The first “national” flora was André Michaux’ “Flora Boreali-Americana” published in France in 1803. Michaux had himself seen much of the United States; he and his son, Frangois André, had traveled and collected in the southeast; and Michaux had made a remarkable trip north to Lake Mistassini in Quebec, at the latitude of the southern end of James Bay. The next was “Flora America septentrionalis” (1814), published in London by Ferdinand Pursh, pirate and plagiar- ist, who had helped himself to plants collected by Lewis and Clark, to the plants of Nuttall and Bradbury, and to whatever else lay at hand. Four years later Nuttall’s “Genera of North American Plants” was published and it marks a change. It is, for one thing, in English. The floras by Michaux and by Pursh are in Latin, and each uses the Linnaean sexual system of classification, with no indication that anything else is possible. Nuttall’s “Genera” is also Linnaean in arrange- ment, but he notes in the introduction that he much prefers the “natural” system, but only for convenience uses the Linnaean. The natural system had been first presented in 1815 to the American botanical ity when the Abbé Correia da Serra gave a series of lectures on “elementary and philosophical botany” at the Academy of Natural Sciences in Philadelphia. To accompany the lectures, Correia offered a “re- duction” of the genera of Muhlenberg’s “Catal- ar- ranged among Jussieu’s 100 families. The lectures seem to have had little impact, and Correia was in Philadelphia, not to guide American botany 512 into the new age, but as ambassador from Por- tugal. American botanical works stuck with the Linnaean sexual system until 1826 when John Torrey turned to the natural system in publishing on plants collected by Edwin James on the “Yel- lowstone Expedition"; even he had used the Lin- naean system in his “Compendium of the Flora of the Northern and Middle States," which ap- peared earlier that year e of the natural system by the young man who rapidly was becoming America's most re- spected botanist marked the intrusion of mod- ernity into a botany that had been developing in an isolated way for several years. During the ear- ly years of the nineteenth century there were, in addition to the collectors I've earlier mentioned, many active botanists in the United States— Big- elow and Oakes in New England; Beck, Eddy, Hosack, and Mitchell in New York; Barton, Dar- lington, Muhlenberg, and von Schweinitz in Pennsylvania; Charles Short in Kentucky; Elliott in South Carolina; — and Rafinesque —are a few An amazingly long list of them is found in the “Historical Sketch" in Rafinesque's “New Flora and Botany of North America" (1837). Many of these American botanists were known and respected in Europe, but they were all Lin- naeans by training and inclination. By the 1820s upon the natural system, laid o Jussieu in “Genera Plantarum" T and built upon by Brown and Ventenant and Jaume St. Hilaire. The principles of a natural system had been lucidly described and explained by A.-P. de Candolle in “Théorie élémentaire de la bota- nique" (1813, 1819), and were being followed by de -Candolle in his “Prodromus,” but in this country botany still was Linnaean pigeon-holing and that, too, was the botany being taught. There had been earlier textbooks of botany in the United States, but from 1817 on, the text- book was Amos Eaton's “Manual of Botany for the Northern States" which by 1840 had gone through eight editions. Eaton was an unrepentant Linnaean and for twenty years this hugely pop- ular textbook inculcated the Linnaean system into students of botany. There was, in addition, AI- mira Hart Lincoln's “Familiar Lectures in Bot- any," first published in 1829 (see Rudolph, 1984). Mrs. Lincoln (later Mrs. Phelps) was Eaton's pro- tégée and the “Familiar Lectures" included the sexual system toned down for young people. Mrs. Lincoln, in particular, sounded the trum- pet of militant Americanism — “. . . although Eu- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 rope may boast of more brilliant stars than ap- pear in our firmament of letters, shining with greater lustre, contrasted with the darkness and ignorance by which they are surrounded; we may justly feel a national pride in that more general diffusion of intellectual light, which is radiating from every part, and to every part of the Amer- ican republic" (Lincoln, 1829). This commend- able wish to spread learning throughout the re- public meant, though, that Eaton and his disciples would continue to use the Linnaean system be- cause they perceived it as being much easier to learn and much less likely to bring about uncer- tainty and confusion in the tyro botanist. During the 1820s John Torrey, still a very young man, rose to the top of the botanical heap. Torrey learned his botany directly from Eaton and in his early publications followed the Lin- naean system. Torrey had ambitious plans for himself; he had started in 1819 with a catalogue of the plants growing within thirty miles of New York City; in 1824 he published the first volume of “A Flora of the Northern and Middle Sections of the United States," and two years later the * Compendium" of the flora of the same area. But then in writing up James's plants, Torrey turned to the natural system and introduced it into use in America. The second volume of the Flora of 1824 never appeared — perhaps revising it into the natural system was more than Torrey chose to do. But in 1831 he presented an Amer- ican edition of John Lindley's “Introduction to the Natural System of Botany," and American botany turned away, with some reluctance and with some exceptions, from the Linnaean sys- em. Torrey's boundaries had been much widened. He had by now set upon doing a “Нога of North America" and in 1833 went to England and to France to visit herbaria and to discuss with Wil- liam Hooker his “Flora Boreali-Americana," al- ready under way, which would cover the plants of British America. He had, too, a likely co-work- er, young Asa Gray from upstate New York, showing great promise. When Torrey returned from Europe, he was prepared to go forward with his most comprehensive work in botany, the “Flora of North America," but in 1835 the State of New York resolved to organize a survey of natural history ofthe state and Torrey was placed in charge of the botany. He was, though, reluctant to leave the “Flora” and suggested to Gray that he begin work on some families. After initial reluctance Gray agreed and the first two parts o Lnd 1986] the Flora were published in 1838. It was re- viewed by William Darlington, physician-bota- lv pointed out that, “The authors of the Flora have, of course, adopted the natural system as being the only one consistent with a truly scientific arrangement .. ." (Darlington, 1838). The natural system, the modern botany of the time, was at last in place in the United States. A textbook, Gray’s “Elements of Botany" (1836), following the natural system, rather than pre- senting it as something too complicated to learn, was available for use and would soon displace Eaton in the colleges. And an American-pro- duced flora using the natural system was at hand. A few years before, Torrey in Lindley (1831) had written, “The catalogue wok I have = ра теа ега апа speciés which are not described in EG im general Floras, but it is by no means asserted to be complete. There are extensive districts in North America which have never been visited by a Botanist, and even in the United States there are large spaces which are but little known or very imperfectly explored." American botany was now ready to tackle the last major problems of the west GEORGE ENGELMANN Although each of the three was trained as a physician, only Engelmann devoted his profes- sional life to the practice of medicine; and this makes the more amazing the quantity of botan- ical work he accomplished. He was far from being a home-grown doctor who had done the botany then required in medical curricula and had re- tained some interest in plants. There were in the 1830s dozens of such people in this country and some interest in plants was almost de rigueur for medical men. Engelmann came to this country with a а that would have allowed him to step directly into the very top level of botanical activity — had it worked out that m It is ironic, but quite appropriate, that it s Engelmann, coming with his sophisticated D back- ground out of the political ferment in post-Na- poleonic Europe, who settled on the frontier. He was born in 1809 at Frankfurt-am-Main, the oldest in a family of thirteen children whose parents were in the rather progressive profession ofrunning a school for young women. The family school came on hard times and was given up in SHAW —GEORGE ENGELMANN 513 1825 or 1826, but young Engelmann received the congregation of the family church a scholars that allowed him to enter in 1827 the university at Heidelberg. He was already at- tracted by botany for he noted in a short auto- biographical sketch written in 1880 for his son hat, *... I then began in my fifteenth year to become greatly interested in the study of plants" (Engelmann typescript at At Heidelberg he met Alexander Braun and Karl Schimper and these were men with whom he maintained life-long connections. He seems, however, to have become involved with student activists, pushing for a unified Germany, and after only one year at Heidelberg, he moved to the university at Berlin for two years, and then on to Würzburg where he graduated in 1831. The medical degree required presentation of an in- augural dissertation and Engelmann's thesis, “De Antholysi Prodromus," was a remarkable pro- duction for a young student of medicine. Engel- mann was interested in morphology and the the- sis deals with some teratological phenomena in plants; the NOME hono he explains as re- ferring to the or loosening of con- trols over— normal developmental pathways in flowers. After receiving the degree in medicine Engel- mann went to Paris in 1832 where, as he said, . in place of medical studies I found only the cholera. Still Braun, Agassiz, Constadt, and other friends were there and we led a glorious life in scientific union" (Engelmann typescript). A few years earlier some of Engelmann's cousins had been lured to settle in Belleville, Illinois, by the enthusiastic reports of a fertile midwest in Gott- fried Duden's “Report of a Journey to the West- ern States of North America"; and the family decided that young Englemann should be sent to ec world must have been appealing to young En- gelmann. Engelmann arrived in 1833 at Belleville and stayed in that area until the spring of 1835 when, restless, he set out through the then southwest, traveling in Missouri, Arkansas, and Louisiana, returning to St. Louis where he opened his med- ical practice in November 1835. Clearly this was an attractive setting for кашылап. a small Ger- d St. Louis diced oss as the * “gateway” to every- thing that lay beyond. So at 26, Engelmann with 514 the best formal training in botany—no doubt of that—on this side of the Atlantic, grounded in the modern botany of the time and with an es- tablished network of correspondents, was settled in St. Louis. Even though occupied with seeing his practice established, Engelmann continued to collect and made a point of sending plants to his correspondents in Europe For the rest of his life Engelmann carried on a medical practice that became so successful that by 1856 he could leave it for two years of study and travel in Europe. The medical practice, al- though less intensive in later years, is always there and must be regarded as background to his bot- any. Years later Asa Gray wrote, “They [Engel- mann’s publications] are the more remarkable as being the result of studies and labors aside from the preoccupations and toils of a well-filled professional life, the fruit of wnt e natu- rally have been devoted t n eedful rest” (Trelease & Gray, 1887). The Sors are those of the elderly Gray being a bit ponderous, but the point is made. It was inevitable that Engelmann, by the twin virtues of his botanical talent and his location on the frontier, soon would joint the Torrey- Gray partnership. By 1836 Gray had become an equal partner with Torrey in preparation of the Flora of North America. Two years later came the opportunity for Gray to visit England and the continent. Gray had been appointed profes- sor of botany at the new University of Michigan and his first commission was to buy books and scientific equipment for the university. This he did, but his own interests lay in seeking out bot- anists and their herbaria. So much of the available material of American plants lay, unseen by American botanists, in Eu- ropean herbaria. Gray had two goals; he would get a firm grip upon this mass of botanical in- formation that had to be mastered for the Flora, and he would open lines of botanical commu- nication between the two sides of the Atlantic (see Dupree, 1959: 74). His first botanical call was upon William Hooker in Glasgow; here he saw a richness of plants from Canada and from the northwest and became fully aware ofthe great blank on the botanical map. To fill it in would be a goal for the next ten years. And Gray “met” in Europe his future friend and partner; he spent nearly a month working in the herbarium at Ber- lin and here he saw plants from the American midwest, collected by a Dr. Engelmann of St. Louis (Gray, 1840). When Gray returned he was effectively un- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 employed for the University of Michigan was having financial problems, and the regents asked him to serve without benefit of salary, but in April of 1842 Gray was appointed Fisher Pro- fessor of Natural History at Harvard, with re- sponsibility for the botanical garden and the promise of some freedom for research. The Flora of North America was not yet finished—it still isn't —and it would always be a weight hanging over Gray, who saw completion of the Flora as a distant goal to be reached. However, there were always other attractions, responsibilities, and op- portunities not to be missed. And the partnership now was ready to go ahead. Engelmann, Gray, and Torrey met for the first time in New York in 1840 when Engelmann was returning from Germany where he had been mar- ried. The meeting seems to have been a success; soon after Engelmann returned to St. Louis, Gray asked (12/X/1840) if it would be acceptable to name as Engelmannia a genus of Compositae. Of course, there never was any kind of formal agreement among the three, but over the next thirty years there was a loose working relation- ship that functioned effectively. Each man had in it his place; my concern here is to examine the part played by George Engelmann during the years 1840 to 1860. It was clear that the west already was changing. In 1839 Frederick Wislizenus, Engelmann's good friend and fellow physician of St. Louis, traveled with a fur-trading party into the Rockies and reported, “It is perhaps only a few years until the plow upturns the virgin soil, which is now touched only by the light-footed Indian or the hoof of wild animals. Every decade will change the char- acter of the country materially, andi in a hundred years perhaps the present f mountain life may sound like fairy tales" (Wislizenus, 1912). George Catlin, active in the west during the 1830s, had seen with his painter's eye even тоге — “. . the gand and irresistible march pe civilization d beheld jugger naut i sweeping desolation” (quoted i in McCracken, 1959). However the changes came about, new op- portunities to seize the plants of the west would develop. From 1840 onwards there was a real effort made by the three to be aware of what was happening in terms of travel in the west and to see which of the travellers might be induced to collect plants. Sometimes plants arrived almost without notice. In November of 1842 Torrey wrote to Gray that he had received a letter from a Lt. Fremont in the U. S. service” who was 1986] sending a box of plants from the Rocky Moun- tains (18/XI/1842). Torrey sent the Compositae to Gray who liked them well enough and wrote back to Torrey, “... Lieut. Е. must be indoctri- nated & taught to collect both dried spec. & seeds. Tell him he shall be immortalized by having the 999th Senecio called S. Fremonti ” (S/XII 1842). Such immortality was a carrot that could always be dangled before a collector. Engelmann’s immediate “job” was to keep open an eye for likely collectors on the spot— whether rich Englishmen shooting buffalo, young Germans eager to see the west, adventurers, or members of military parties. He was to watch for people; he should persuade them that col- lecting of plants was a good thing and might even bring a little fame— perhaps a Senecio named after one; and Engelmann was to show and tell the recruits how collecting was best done. So in the spring of 1841 Gray wrote to Engelmann, “You see you may be of much use to us. Mean- while you will give me pleasure if you will tell me how I may serve you" (17/IV/1841). Engelmann was eager, too, to make his own contributions to botany and soon became in- volved with Cuscuta, first in a long line of dif- ficult groups that Engelmann would study over the next forty years. His interest in such groups seems to be the intellectual continuation of his early work on morphology. Current works on botany recognized in the United States only the Linnaean Cuscuta americana, but just in the vi- cinity of St. Louis Engelmann had observed sev- eral species. His first botanical publication in this country was “А Monography ofthe North Amer- ican Cuscutineae" (1842) followed the next year by some additions and corrections. Engelmann would always be interested in these plants, al- though by 1849 he was writing to Asa Gray that the time had come to “. . . go at Cuscuta and not cease until I finish them. The thought of them is like an incubus to me—and I am almost aston- ished that I have not yet dreamed of a Cuscuta coiling itself about my limbs, etc., etc." (26/X/ The system for encouraging collectors in the west went into full swing during 1843. Engel- mann saw that it was necessary for a private ба ie to be self-supporting by his collecting and suggested to Gray that sets of plants from two such men could be offered for sale (18/I/ 1843). Engelmann would guarantee the quality of the specimens, which would be only of the rarest plants, and would, therefore, have to sell at eight to ten dollars per hundred. The collectors SHAW —GEORGE ENGELMANN 515 were Ferdinand Lindheimer who was farming, but not too successfully, in Texas and Karl Gey- er. Geyer had been collecting i in Illinois and in rn rn country." If Gray liked the plan he should place an advertisement in Silliman's Journal. Gray did like it and in the April-June number of Silliman's it was adver- tised that three botanists (the third was Fried- erich Lüders) were preparing to explore “the most interesting parts of the far West ....” Both Lüders and Geyer would disappoint En- gelmann and Gray, but Lindheimer was the most productive of the collectors active during the 1840s; dealing with his plants occupied countless hours of Engelmann's time and he wrote dozens, quite literally, of pages of descriptions and notes about the collections for Gray. Gray was not always an colleague — he was impatient — he drove himself unceasingly — and he wanted more and more from the men in the field. Then, too, Gray did have field experience, but not in the west, and for Gray, comfortable on Garden Street in Cambridge, it sometimes was difficult to understand the realities of col- lecting on the very edge of civilization. Engel- mann had to deal with Gray's demands and he simply coped with Gray's impatience by being patient with it. Occasionally there comes a hint of asperity. Writing about Lindheimer's plants, Engelmann pointed out to Gray, “Perhaps they ought to be more pressed. But in travelling and in putting up plants in a cart it is not easy to obtain the neatness required by a closet botanist. The ДИС Pet vA at least as complete as pos- sible ...” (13/V/1845). ere was, too, vec" not here, but in urope. During 1846 Lindheimer had a visitor, Ferdinand RO from Germany, with whom he collected. Roemer returned to Germany with his own collections and a set of those made that year by Lindheimer. The plants, including Lind- heimer's, were turned over by Roemer to Adolph Scheele, a clergyman and indifferent botanist, who described the new species, one hundred and thir- ty-nine of them, in a series of papers in Linnaea. Engelmann was angry and scornful; ““And how characteristic for those persons, who seek noto- gelmann to Gray, 13/V/1849). Each article in Linnaea struck a nerve in Engelmann; he really was not interested in priority, but he did believe that, so far as practicable, new plants should be described in the country where discovered. On 516 15 February 1850, he wrote to Gray, “I think we should classify the new ones [of Lindheimer’s plants] at once, so as to prevent any ‘Scheele- ing.” Englemann had become truly an American; when Gray returned from his 1850/1851 trip to Europe, Engelmann welcomed him back to “our side of the big pond" (18/IX/1851). This loyality to the adopted country made the more objec- tionable actions such as those of Scheele, and Engelmann reacted with annoyance to unreason- able demands made by European botanists for American plants for study. In the same letter urging action to forestall Scheele, Engelmann noted that he had earlier *. . . forgot to mention that Dr. C. H. Schultz ‘Bip.’ as he calls himself .." wished the Compositae for study; “Егот his letters he must be a vain and presumptuous person " Gray replied that Schultz Bip. had done good o but своя replied Баск, that although that might be, . he is, I believe a vain and not very менн man...” (17/III/ 1850). In light of this it is surprising that Engelmann seems not to have complained, at least not to Gray, about Gray’s inability to publish beyond the Compositae. This was Gray’s favorite family, so his treatment of some collector’s material al- ways would be complete through the Composi- tae. But in the Candollean sequence which Gray used, the Compositae fall early among the sym- petalous families, so that most of the sympetal- ous groups, all of the monocotyledons, and the gymnosperms, would not be written up. Gray and the others did indeed publish on many of the new things in these groups, but the unpub- lished families still were ripe for plucking by any- one who had access to the widely distributed sets of plants. Lindheimer was a pleasing success as a collec- tor—but only in Texas—and Gray yearned for the unknown plants of the southern Rockies and the Great Basin. As early as 1843 Gray had sug- gested that Lindheimer should be sent part way up the Oregon Trail then to work his way south- ward, but Engelmann gently explained that Lin- dheimer’s home base was Texas. Gray then wanted Lindheimer to move farther westward into Texas and, perhaps, even to Santa Fe; into territory, ”... where I trust many of his plants will have no Latin names until we christen them" (Gray to Engelmann, 3/11/1845). Lindheimer stayed in east Texas and even married; Gray ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 fumed, “Send me the new Lindheimer plants as soon as you get them and stir him up. Can't you send a collector to Santa Fe?" (Gray to Engel- mann, 27/11/1846). However, within a few months' time, the sec- ond great success as a collector would be in the field. War with Mexico was declared in May 1846, and Gray saw at last an opportunity to send a collector to Santa Fe with the army. For a mo- ment Gray might have considered going himself, but Engelmann was commissioned to find in St. Louis a collector. He found Augustus Fendler who received free passage with the army and Engelmann saw this as a good sign for science in the United States (Shaw, 1982). Engelmann was always a republican, but he doubted whether a republican form of government was best suited for the promotion of science. Fendler spent several months in Santa Fe and returned with beautiful material that pleased Gray very much. But Engelmann again had to devote hours of the time left from his medical practice, to work with Fendler in arranging the sets for sale, and to deal with Fendler himself. The Texan plants from Lindheimer were de- scribed by Engelmann and Gray as "Plantae Lindheimerianae," the first part published in 1845 and the second in 1850. The third part would not appear until 1907 (Blankinship, 1907). The amount of work done by Engelmann on Lindheimer's material truly is amazing— during the middle years of the 1840s he sent to Gray dozens of closely written pages of descriptions and of notes on the plants. There was, in addi- tion, the labor involved in sorting out the ma- terial into sets and then the worst, perhaps, of all, acting as banker One goal, after all, was to sell or exchange sets of plants in the hope that the collector could be nearly self-sufficient. Gray in the east served as “distributor” of the sets and dunned the pur- chasers for payment. So Gray to Engelmann in December 1849, “I have written a very strong letter to Brown [Robert Brown] (who was in Ire- land at last accounts) to send on the cash for Fendler—I think he can neglect no longer— but he isa slow and indolent old sinner." The monies eventually received would then go to Engel- mann who saw that they reached Lindheimer or Fendler, but often Fendler would be living on what Engelmann could afford from his own pocket айне апа Fendler were the collectors 1986] who occupied most of Engelmann’s free time during the 1840s, but he was also involved with Josiah Gregg, Fremont, Gordon, and collectors of lesser significance. At times Engelmann de- spaired of doing all that he wished; he wrote to Gray, “How often have I thought as you, if I could only multiply myself to do all that I have before me—how often cursed those numberless fellows, who do not know how to kill their time— which would be so eminently valuable to oth- ers!" (11/II/1847). Fortunately, after Lindheimer and Fendler, collectors and their plants made fewer demands upon Engelmann's time—times, in fact, were changing. From the mid years of the 1840s col- lecting in the west turned from the hands of pri- vate individuals, such as Lindheimer and Fend- ler, to those of government-sponsored collectors and of military men. In 1846 Major William Emory made a reconnaissance from Fort Leav- enworth to San Diego, and although Engelmann helped with botanical preparations, the plants went to Torrey and Gray. Charles Wright col- lected in Texas during the 1840s and then went with the post-Mexican War Boundary Commis- sion, but his botanical contact was Asa Gray and the overall botany of the Boundary Survey was finally prepared by John Torrey. During the 1850s collecting in the west was done by the parties carrying out the railroad surveys and, again, those plants went to Torrey. Torrey and Gray saw that Engelmann's botan- ical abilities were being wasted by the drudgery of sorting out plants and writing lists and wrap- ping parcels; and Torrey in particular urged En- gelmann to concentrate upon some genus or few genera that he might prepare for the Flora of North America. No matter how much time was taken by the flood of material from the west the unfinished Flora still made its own demands. In 1850 Engelmann remarked to Gray, “I am glad to see that both your northern flora and textbook are in our bookstores and are bought— But you forget the flora of М. America!" (16/IV/1850). A group that appealed to Engelmann, with his keen eye for the interesting and the difficult, was the cacti now coming in great numbers from the west. From 1846 he published on the cacti of every collector and expedition and by 1856 En- gelmann had prepared a "Synopsis of the Cac- taceae of the Territory of the United States and Adjacent Regions," in which he recognized 117 species where only a few years before “‘scarce half SHAW —GEORGE ENGELMANN 517 a dozen had been known." The cacti were the best group Engelmann could have chosen— he was close to the field, so that plants, including living ones that could be grown in the family garden, could be sent directly to St. Louis. And Engelmann knew where problems lay; “The study of the Cactaceae has been too much in the hands of D. and amateurs" (Engelmann to Gray, 13/IV/1855). The 1840s were an exhausting time for En- gelmann with botany piled upon his medical practice, especially when cholera swept through the city. During the winter of 1848/1849 and into the autumn of 1849 the epidemic of cholera was а severe. In May Engelmann told Gray t "We are in a state of war here; day and ind on our legs— fighting the great destroyer, the cholera — of course no room for anything else, not even for sleeping or eating comfortably — like soldiers before the enemy. I like the excitement which it is for me" (12/V/1849). Yet the very next day, he wrote again to Gray, filled with an- noyance about the shoddy work of Pastor Scheele on the plants of Roemer and Lindheimer. Eventually, though, the strain showed. In the summer of 1851 Engelmann took off from med- icine for a few days to attend the AAAS meetings in Cincinnati; and as he told Gray (29/VII/1851) it was, after eleven years of "slavery" in St. Louis, a marvelous break — making new friends and meeting again old ones— Louis Agassiz, for one— nearly twenty years after the summer of 1832 in Paris. Gray was concerned and asked when Engelmann could retire from medicine — for the greater good of botany. The reply was, of course, that his patients needed him, but Engelmann had started to think about change. When the railroad survey parties were set up in 1853 he considered seeking an appointment with one, but decided against it. “I should probably prefer to be independent" (En- gelmann to Gray, 4/VI/1853). The thought lin- gered though, and a year later Engelmann “half thought" about going with Major Emory to So- nora, but he had heard about it too late; he told Gray though that, “I must have some relaxation and reanimation after a steady medical practice for 14 years" (Engelmann to Gray, 18/1X/1854). Fortunately the change was not far away. In May of 1856 Engelmann mentioned to Gray that he had met Henry Shaw. Shaw, intending to found a garden in St. Louis, had asked Sir Wil- liam Hooker to recommend someone who might 518 organize the garden and Hooker replied that the very best man was right at hand in St. Louis. Engelmann had already planned a visit to the east and on to Europe and in November 1856 the family sailed. During the fifteen months abroad, Engelmann bought books and the Bern- hardi herbarium for Shaw’s garden and studied Cuscuta and cacti, so that in 1859, Cuscuta was at last laid to rest with a "Systematic arrange- ment of the species of the genus Cuscuta, with critical ee on old species and descriptions of new ones.’ Engelmann practiced medicine to the end of his life, but after the trip of 1856-1858 at a more leisured pace. Exploration, too, had changed very much, and Engelmann no longer had to deal with the privately operating collectors. The conse- quence was that during his last twenty-five years Engelmann was able to carry on with the kind of carefully done and detailed work begun with Cuscuta and the cacti, and even to travel with the chance to see his plants in the field. Again, this list is not inclusive, but after 1860 Engel- mann was able to produce major works on Vitis, on which he would publish nearly to the end; on Yucca and Agave, on Isoetes, on Euphorbiaceae and Asclepidaceae, and, most elegantly done, treatments of Quercus and of conifers. He could also produce, to the tune of forty pages’ length, “A Revision of the North American Species of the Genus Juncus with a Description of New or Imperfectly known Species." The paper opens with one of the longer, but most appealing sen- tences in any systematic work, and one which tells much about the man. The difficulty I found in arranging the species of Juncus of my own herbarium, the doubts in which the authors left me by in- complete and unsatisfactory descriptions, and by confusion in names and synonyms, the want of confidence which all my corre- spondents, even such as had paid a good deal of close attention to it, seemed to place in themselves and their own judgement when this genus was under dicussion —all this in- duced me to enter upon a critical study of our Junci. Engelmann tended to be a splitter— often Gray would chide him for splitting, while admitting that he, Gray, tended to lump, but Engelmann never just poured out new species. And conspic- uous in his monographs and revisions is a careful effort to array the species in some pattern of re- lationships. Certainly he was not always suc- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 cessful, but Engelmann sought always to present a finished product. The kind of work that En- gelmann did is impressive, and the happy com- bination of quality and quantity of his work is amazing. After Engelmann's death, Henry Shaw asked Gray and William Trelease, then director of the Garden, to prepare a collection of his bo- tanical publications (Trelease & Gray, 1887). The collection, which is not quite complete, produced a book three inches thick. Engelmann arrived just as the frontier was moving beyond St. Louis, and he spent only a few more than ten years being directly involved in the “processing” of plants coming from Lind- heimer and Fendler. His contributions, though, during the 1840s toward making the west botan- ically known were substantial, and he was the perfect frontier collaborator for Gray — patient, willing to spend the hours needed for turning bundles of plants into specimens for an herbar- ium and then into publications; and a very good botanist. Valuable as was this work, though, his contributions to American botany through his monographic and revisional publications are still more important—these set for young system- atists a standard of American-produced taxo- nomic excellence. LITERATURE CITED AND OTHER REFERENCES ANON. 1942. Formative days of Mr. Shaw's Garden. Missouri Bot. Gard. Bull. 30: 100-110. —— —— |]. F. Co rreia da Serra]. 1815. Reduction of all the G ts contained in the Catalogus Plantarum Americae Septentrionalis of the late Dr. Muhlenberg to the т families of Мг. de Jussieu’s system. Philadelph 1929, 1930. George РИТЕ man of Rev. 23: 167-206, 427- 1907. Plantae Lindheimerianae, Part III. Missouri Bot. Gard. Ann. Rept. 18: 123- 3. CANLOLLE, A. -P. DE. 1813. 1819 (ed. 2). Théorie élé- décrire et d’étudier les végétaux. Paris. 1813. ed. 2, Paris A. DE CANDOLLE. 1824-1870. Prodromus systematis naturalis regni vegetabilis . . Paris. о 1838. Review of “A Flora of North Ameri " Amer. J. Sci. 35: 180-182. AVIS, R. B. 1955. The Abbé Correa in America: 12-1820. Trans. Amer. Phil. Soc. 45: 87-197. 1959. Asa Gray. Harvard University d EATON, A. 1817. A Manual of Botany for the North- ern and Middle States, 1st edition. Albany, New ork. ENGELMANN, G. . A monography of the North American Cuscutineae. Amer. J. Sci. 43: 333-345. 1986] 1856. Synopsis of the Cactaceae of the ter- ritory of the United States and adjacent regions. Proc. Amer. = Arts 3: 259-311, 345-346. seg arrangement of the species of the genus С uscuta, with critical remarks оп old species and E of new ones. Trans. Acad. Sci. St. Louis 1: 453-523. A revision of the North America species of the genus Juncus, with a description of new or х known species. Trans. St. Louis Acad. Sci 424— Жем PH. Pla ntae Lindheimerianae; n enumeration of the plants collected in Texas, пт distributed to я rs, by Е. Lindheimer, d des ogra M new species, M к E Nat Hist 5: lantae Li ndhei eimerianae, Part II. An xn of a iS E plants made by F. Lindheimer in к western part of Texas, in the years 1845-6, 1847-8, with critical re- marks, тыл к ew species, &с. By Asa Gray, J. Boston Soc. Nat. Hist. 6: 141-235 ENGELMANN, J . Sketch of the life and wor Ewan, J. 1952. Frederick Pursh, 1774-1820, and his botanical associates. Proc. Amer. Phil. Soc. 96: 599-628. ETZMANN, W. H. 1966. Exploration and Empire. Vintage Books, New York. GRAUSTEIN, J. E. 1967. Thomas Nuttall, Naturalist. 1808-1841. Harvard University Press, Cam- bridge. Gray, A. 1836. Elements of Botany. G. & C. Carvill Co., New York. . 1840. Notices of European herbaria, к ularly those most interesting to the North Am ican botanist. Amer. J. Sci. Arts 40: 1-1 44. The longevity of trees. North Amer. Rev. 59: 189-238. Hooker, W. J. 1829-1840. Flora Boreali-Ameri- cana; or, The Botany of the Northern Parts of British America... , 2 volumes. J. H. Bohm, Lon- don. Jussieu, А. L. DE. 1789. Genera Plantaru LAWTON 1968. George Engelmann, "1809. ТЯ Scientific gr of the do: Missouri Bot. Gard. LEE, 2 T. 1932. Tui Gregg and Dr. George En- mann. American Antiquarian Society, Worces- 8 Bel [PHELPS], А.Н. 1829. Familiar Lectures on otany, Ist edition. Alban LINDLEY, J. An Introduction to the Natural System of Botany. First American edition, with an Ie by John Torrey. G. & C. Carvill & York. 1959. George Catlin and the Old Frontier. New York. McKeLvey, S. D. 1955. Botanical Exploration of the Trans-Mississippi West. 1790-1850. Arnold Ar- boretum, Jamaica Plain McVAUGH, В. 1947. The travels and botanical col- lections of Dr. Melines Conkling Leavenworth. Field and Lab. 15: 57-70. SHAW —GEORGE ENGELMANN 319 d b. Eie Flora Boreali-Americana, 2 vol- 1984. The appearance of academic bi- ology in late nineteenth-century America. J. Hist. Biol. 17: 369—397. 1814. Flora Americae Septentrionalis: Or, vstematic A ription ft Plants of North America; het Beet. what have been Described by Preceding Authors, Many New and Rare Species, Collected During Twelve Years Travels and Residence in that сол, 2 volumes. White, Cochrane and Со., RAFINESQUE, C. 1837 [1836]. New к aa Botany of North America, Part II. Philadelphia. Корсев$, A. D. Ш. 1942. John Torrey. A Story of North American Botany. Princeton University Press, Princeton. RUDOLPH, E. D. 84. Almira Hart Lincoln Phelps (1793-1884) and the spread of botany in the Nine- teenth Century America. Amer. J. Bot. 71: 1161- 1167. SHAW, E. A. 1982. Augustus Fendler’s collection list: мо Mexico, 1846-1847. Contr. Gray Herb. 212: ЕЯ Р. 1909. A biographical history of botany t. Louis, Missouri III. Pop. Sci. Monthly 74: 9. . A Catalogue of Plants Growing Spontaneously within Thirty Miles of the City of Sections of the United States; or, A Systematic Arrangement and Description of All the Plants Hitherto Discovered in the United States North of Virginia, Volume 1. T. 182 Compendium of the Flora of the Northern and Middle States. S. B. Collins, New ork. . 1826, 1827. Some account of a collection of plants made during a journey to and from the Rocky Mountains in the summer of 1820, by Ed- win P. James, M.D., Assistant Surgeon, United States Army. Ann. Lyceum Nat. Hist. New York 2: 161-164; 2: 165-254 & A. GRAY. 1838-1 843. A Flora of North America. New York and London TRELEASE, W. & A. GRAY. 1887. The Botanical Works of the late George Engelmann Collected for Henry Shaw, Esq. J. Wilson and Son, Cambridge. WHITE, C. A. 86. Memoir of George Engelmann, 1809-1884. Natl. Acad. Sci. Biog. Mem. 4: 3-21. м + Е. 1912. А Journey to the Rocky Moun- s in the Year 1839. Translated from the Ger- man, with a sketch of the author’s life, by Fred- enus, Esq. St. Louis. This is a us m 9." Missouri His- torical Society, s Louis. 1840 AUGUSTUS FENDLER (1813-1883), PROFESSIONAL PLANT COLLECTOR: SELECTED CORRESPONDENCE WITH GEORGE ENGELMANN MICHAEL T. STIEBER! AND CARLA LANGET ABSTRACT Augustus Fendler, a German-American — in jte e physics, PARES and botany, rea a Santa Fe, New Mex n; the originals at the n 1846. His life can both of erma lated into English by the ond author and were used extensively to create the following biography. During the nineteenth century a small number of people who preferred to seek their livelihood far from the madding crowd chose the arduous life ofthe plant collector. Although the hardships of this modus vivendi attracted some, the explo- ration of nature's diversity — and doing so at one's own pleasure — lured most. For Augustus Fendler the quest for plants provided a haven for his *painfully diffident" personality and inquiring mind (Canby, 1885). His philosophy practically coincided with that of his more famous contem- poraries, Emerson and Thoreau, as may be seen in Fendler's privately published “Mechanism of the Universe" and in his letters. Despite his shy- ness or perhaps because of it, Fendler won the lasting affection of William Marriott Canby, Asa Gray, and George BASS АВВ. In HEN own ways these three p yto floristic botany in North America during the 1800s. Fendler enriched the herbaria of all three as well as those of others who purchased sets of plants from him. The summer following Fendler’s birth on 10 January 1813 in Gumbinnen, East Prussia, his father died, but he soon A a ерак Although for some years th he eventually had to = с his family could not afford the tuition. Apprenticed to the town clerk, he soon found this a “spirit-killing employment," and when given the opportunity to accompany a government physician on an in- spection tour of cholera quarantine stations along the Russian border, young Fendler volunteered, even though he had to clerk for the doctor (Can- by, 1885). When Fendler returned to Gumbin- nen, he learned the tanning trade, hoping that this would allow him to earn a living anywhere he might travel. Tanning, however, so exhausted and, at times, nauseated the frail youth that Au- gustus decided to enroll in the Royal Gewerbe Schule, a polytechnical school in Berlin. But after one successful year there, the daily schedule had so drained his strength that in the Fall of 1835 he withdrew. Following a period of wandering in German cities, he sailed from Bremen for Bal- timore, Maryland, in 1836. With little cash in his pockets he sought and found work in a tan- yard in Philadelphia, but he quickly abandoned in lean years ahead. However, the and panic of 1837 so depressed the business that he was forced to quit the factory in the Spring of 1838. Even before the hard times of 1837, Au- gustus had read of the opportunities to be found in St. Louis and the Far West. thirty days via canal-boat and steamer from Albany through Buffalo, Cleveland, and Portsmouth, Fendler finally reached the Missis- sippi city of 13,000 people. He worked for a time with a man “who had just commenced making spirit-gas for lighting public houses, as the man- ufacture of coal-gas had not reached so far west" (Canby, 1885). The prospect of working through the winter in an unheated room so discouraged Fendler, however, that he departed for the South shortly before Christmas in 1838, knapsack on his back. He walked through the forests of south- ern Illinois, the canebrakes of Kentucky and part of Tennesssee before meeting two other wayfar- ! Hunt Institute for Botanical Documentation, Carnegie-Mellon University, Pittsburgh, Pennsylvania 15213. ANN. MISSOURI Bor. GARD. 73: 520-531. 1986. 1986] STEIBER & LANGE—AUGUSTUS FENDLER ers. The three bought an old skiff just below the Ohio River’s mouth and managed to float a good distance before finally sighting a steamboat that had broken through the winter ice upriver. His inborn wanderlust urged him westward; so he left New Orleans for Galveston, Texas, in Jan- uary 1839. Not finding Galveston particularly inviting — he described it as “a dozen poor-look- ing houses scattered about its low and sandy sur- face" (Canby, 1885)— Fendler tried Houston. He explored the area and almost settled there, which would have been simple under the immigrant land grant law, but he had no interest in fighting Comanches for the right to secure his claim. After he returned from the then unsettled area known as Austin the ravages of yellow fever that greeted him at Houston clinched his decision. His wallet empty and his body fever-weakened, Fendler was forced to retreat iuh to Illinois, where he taught school for a short tim At this time, possessed by a particularly acute case of transcendentalism, Fendler wrote in his own “autobiography” (referring to himself in the third person) something akin to Thoreau's Wal- den: Autumn in North America, and especially in the Western States, always presented more charms to F.’s mind than any other part of the year. Hence in 1841, when autumn winds began to scatter the falling forest leaves, he was seized with an uncontrollable desire for solitary life in the wild woods, removed from the haunts of man, in short, for the inde- pendent life of a hermit. In his search for a proper place, he came upon a little village called Wellington, situated on the banks of the js eA River, three hundred apap . Louis. Here he learned that a bed island, two and a half miles joi called Wolf's Island, not very far below the village, was at his service. Without delay, F. packed his little baggage, including some bed-clothes and cooking utensils, a rifle, an axe and some books, in a canoe, also taking along some provisions, and started for his new home. This island was densely wooded with gigantic trees. On the lower part of it, farthest removed from the village, was an old, dilapidated log cabin, the former abode of some woodchoppers. The upper part of the chimney was gone, so that a tall man standing on the outside of it could look down inside upon the low fire- place, from the burrows of which wild rab- bits popped forth at the approach of man; part of the roof was gone, and the door car- ried off. There was plenty of game, however, especially wild turkeys. These latter had chosen the island as a roosting place for the night and as a place of safe retreat in daytime when chased on the mainland by hunters. In a so-called ““turkey-pen” they were easily entrapped, and thus an abundance of excel- lent food secured. To return the borrowed canoe to its owner and to make one of his own was his first aim. So he went to work at a big trunk of a prostrate tree, and with an axe shaped part of it into proper form of a light canoe eight feet long. Removed from the crowd, the hum and strife of men, his pastimes consisted alternately in trapping, hunting, reading, musing an meditating, and on mild and sunny days in paddling up a placid arm of the river, then turning round lean idly back in his canoe, thus floating home again. Occupied in this way F. lived for about six months, enjoying the sweets of solitude with a satisfaction of inward peace of mind and bliss higher than he had expected —contented and happy as ever mortal man, similarly situated, can claim to be. His feelings of content would at times culminate into feelings of thank- fulness, which then found vent in words akin to the soliloquy of Faust at his forest cave: “Spirit sublime! Thou unto me gav'st ev'ry thing I pray for." Only once he met on the island with a hu- man being, namely, with its owner, coming to see him. How long F. would have con- tinued to live here is hard to say, if the great e a short distance of his cabin he thought 'twas time to leave, and entrusting himself and baggage to his frail canoe, was hurried along at no mean speed by the precipitate rush of the foaming and rapidly swelling stream. Dodging floating logs and broken ledges of ice, he expected every moment to be swamped by the high waves caused by a stiff breeze blowing up stream. To land his tiny craft admidst eddies and whirlpools at Lex- ington, ten miles below the island, was, how- 521 522 ever, the most perilous part of the venture (Canby, 1885). While the somewhat utopian life of an en- sconced hermit always attracted Fendler, a fas- cination for travel constantly lured him away from it. In 1844 he returned to Germany for a visit home. A Koenigsberg professor of botany, rst Meyer, aroused Fendler’s interest in plant collecting. Encouraged that he could make a de- cent living by collecting plants and selling the specimens to foreign herbaria, Fendler returned to St. Louis, this time with his brother, and began in earnest to collect as the seasons changed, be- tween Chicago and New Orleans. He took his excellently prepared plants to the local expert, Dr. George Engelmann, whom he had met the year before, for identification. When in Houston Fendler had first learned of plant collecting from one of Engelmann’s collectors, another German, Ferdinand Jacob Lindheimer, but he had thought nothing more of it as a profession until Ernst Meyer had suggested it. Lindheimer warned him of the hazards of life in the untamed West, where he would have to conquer plains and mountains, drive a wagon, survive on buffalo meat or worse rations, live in the saddle or walk twenty miles a day, and occasionally fight for his life against the natives. Lindheimer, Fendler and others who accepted this vocation never had time to learn enough botany to work on their own collections, but they developed a keen sense of the new and unusual, which profited science in the end (Du- pree, 1959: 156). Plying their trade, these pi- oneers for botany steadily grew unused to the discipline and courtesies required in civilized life, so they generally could not stand the cities for long, but sought refuge in the country. This, of course, suited Fendler’s shy and retiring nature perfectly. In time, Engelmann and Fendler became fast friends. The doctor recognized Fendler’s botan- ical promise and communicated such to Asa Gray. The latter, who never lost an opportunity to get plants from the Far West when he could, secured a letter from the Secretary of War, au- thorizing the Army, then at war with Mexico, to provide the collector with free transportation and provisions to Santa Fe and back. Since his was the first botanical trip in that part of the country, it has recently been the focus of some historical attention (Shaw, 1982). Gray’s Plantae fendler- ianae novi-mexicanae (1849) appropriately ac- counted for this classical plant collection. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 во наои keenness for cacti brought ob- servations from his new field collector, which А; highlights one of the many problems aced by the field botanist in obtaining specimens of large succulent plants. On 8 November 1846, this from Santa Fe: Cactaceae can be found here in large quan- tities, however, barrels and cases to send them in are very hard to obtain even at the highest prices. You just cannot imagine the lack of wooden boards here; the volunteers have to use the boards of the wagon bodies in order to make caskets for their dead, and empty cases are sold by most of the mer- chants according to weight for 12!^ cents per pound. With great efforts I was able to obtain an empty sugar barrel from the Commissary for my collected Cactaceae. Meanwhile Gray had received some of Fend- ler's specimens from Santa Fe and wrote to En- gelmann that Fendler had to go back, because his specimens were excellent. Gray said that they would sell well, too, but lamented that if only Fendler had known more botany, he would have eschewed the “common plants" and bestowed his labor on the new ones abounding there. How- ever, Engelmann had already perceived his col- lector's shortcomings and was training him. No lack of enthusiasm for “herborizing” discour- aged Fendler from another trip to Santa Fe. Only basics deterred him, as he told Gray: When my pecuniary means at Santa Fe were nearly all exhausted, when I had to sacrifice one thing after another of my most necessary effects, to keep up a few days longer the scanty assistance as soon as I should have returned with my collections to St. Louis. It was this hope that made me bear all the difficulties most cheerfully under the happy impression that the enjoyment ofthe fruits of my labour would soon compensate for all. But alas! It was to be otherwise (McKelvey, 1955). Unfortunately Fendler had to borrow money 1986] STEIBER & LANGE—AUGUSTUS FENDLER for sustenance while sorting the nearly 17,000 plant specimens into salable sets. His brother, also ы. enlisted in the Army. As Fendler ех- plained his plight further: I have expended about 200 dollars of my own money; but I gave up a business in which I was doing well, and in which I cannot at present engage again for want of means. To this may be added the sad prospect I have now before me, that by the time I shall re- ceive some money, this money will be nearly all wanted to pay my debts which I had to contract during all this time (McKelvey, 1955) By now Fendler knew what others doing like- wise discovered, namely that plant collecting was less, Fendler resolved to try Santa Fe again, but the ensuing debacle marked his last traffic with the military. In June 1849 ten wagon-loads of people from Wisconsin en route to California joined the Army caravan at Fort Leavenworth, Kansas. On 26 July Fendler wrote Engelmann (in English) that he had been thwarted constantly by one Captain Morris of the Mounted Riflemen whose orders Fendler’s teamster had to obey. On 13 June, shortly after noon, a thunderstorm drenched the party. We proceeded nevertheless on our journey, and the wagons soon after commenced to cross a creek (a branch of the Nemahah Riv- er). After a part of the train had к the creek, the progress of the art was stopped as I understood by the E officer for more than an hour, and then or- dered to proceed again, while during all this time the rain was pouring down. My team- ster accordingly drove on into the bed of the creek in which at this time the water was only about 18 inches deep.The wagon next before us after crossing the creek stopped for some reason or other a few minutes in get- ting up the bank, and my wagon was hereby prevented from going on, and obliged to stop for 5 or 6 minutes in the creek-bed. During this short interval the water in the creek be- gan to rise very fast and all at once the back water came rushing up the stream with such fury as to carry logs and branches before it. The wagon that had detained us now started 523 ahead, but alas! it was too late for us to follow, for the mules were already swim- ming and all that my teamster, who was a most expert swimmer, could do was to jump into the water and cut loose the mules as soon as possible. But they got entangled in the harness and were fearfully struggling for their lives. At the risk of his own life the teamster succeeded at last to cut them all loose. One mule however died in the water, another one died the next morning and the remaining four had suffered so much from the swallowing of water, that they were in a very enfeebled condition. The water soon rose to the middle of our wagon cover far above the load. During all this time there did do was to holler to the teamster to come out, save his own life and let the mules be drowned, but they did not even lend me a hand in pulling out the mules. My own ef- forts to rescue anything of baggage from being soaked by the water were too feeble, and in the attempt I came near losing my life.... unloaded most of our things, and after the rain had ceased next morning I was at the doleful work of opening and unpacking all my trunks, boxes and provisions-bundles, to spread and strew everything to the open air, and to wash the settled mud from blan- kets, clothes, books, paper and a hundred minor articles. There was not a single thing that I had been able to get out before the water had soaked it, and you may judge of the loss and damage in this unfortunate event. I had laid in a stock of provisions at St. Louis for at least 6 months for two men [Augustus and his brother]. Of these provisions the sugar was nearly all dissolved, the coffee soft enough to press it flat between my fingers; into the flour the water had penetrated from 1 to 2 inches all around the sacks. I had to throw away nearly all of hard-bread, cornmeal, tea, rice, etc. Most of the medicines you kindly had put up for me were destroyed. Not to say anything about the great number of ar- ticles that were now covered by rust or oth- erwise damaged I will only mention here 14 524 reams of brown paper that I had taken along for the purpose of drying plants. Every one of these reams after being taken out of the water had its weight increased to such a de- gree, that I was barely able just to lift it off the ground. The creek was still too deep next day to ford it, and several wagons being on our side of the creek, Capt. Morris was obliged to lay by next day, which was a hap- py circumstance to me as it gave me an op- portunity at least partly to dry some of my soaked baggage. A few days later, all my books had lost their covers and the leaves moulded and partly rotten for want of suf- ficient time in drying. But what pained me most was to see the greater part of the 1000 specimens of my Santa Fe plants that I had so carefully with so much labor and patience pasted on paper and bound to book form, fall rotten from the moulded leaves [of] the book. This was a loss that could not be re- stored again by money.... All the plants (about 180 specimens) that I had collected during my stay about Weston, Fort Leavenworth and on the journey up to the day of disaster were completely decayed and had to be thrown away. On 25 June Fendler reached New Fort Kear- ney on the Platte River at the head of Grand Island, about 300 miles from Fort Leavenworth. Although he had hoped to reach the Great Salt Lake, the condition of his mules would not allow it. For some reason Capt. Morris would not per- mit an emigrant to lend two mules to Fendler so his team could manage the load. After several unsuccessful negotiations with other emigrants, Fendler reluctantly joined a small Government train of three ox-drawn wagons returning to Fort Leavenworth. The Quartermaster there offered Fendler passage to Fort Laramie whence he would have to find his own way to the Great Salt Lake, but the time for this journey would have put him at the Lake in winter. “For the present,” he in- formed Engelmann, “I only hasten to inform you of the failure of my expedition." Asa Gray, who had encouraged him to try for the Lake, sym- pathized with Fendler’s temptation to abandon plant collecting and seek his fortune with the other forty-niners heading for the gold discov- ered at Sutter's Mill, for he would certainly ‘таке more at digging gold than he ever can with plant gathering" (Dupree, 1959: 164). And Fendler was sorely tempted. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 When Augustus finally arrived in St. Louis, however, he was overwhelmed by yet another series of mishaps: All his plant collections, books, travel journals and other worldly goods had been destroyed in the “Great Fire” that had swept St. Louis during his absence. This should have been enough to send him back to the lamp factory, Wolf's Island, or San Francisco. Instead, on 4 December 1849 he boarded the steamboat Uncle Sam bound for New Orleans, and after obtaining necessary supplies at Engelmann's expense and after fitful starts, Augustus and his brother finally sailed from Louisiana to Chagres, Panama. Be- tween rainstorms they managed to collect some plants, but were finally driven back to New Or- leans on 20 April 1850. [Fendler's field notes for his Panama trip are at US (see Stieber, 1982); his other field records are evidently lost.] “I hard- ly believe that there is any other part on this earth where there is so much rain and where the rainy period is comparable to that of Panama," Fendler, now settled in Camden, Wachita Coun- ty, Arkansas, wrote Engelmann on 1 May. Since leaving St. Louis the previous year, the Fendlers had spent $274 and were debtors once more. So Augustus begged Engelmann to buy his entire set of plants and peddle them from St. Louis. Mean- while the brothers tried to raise some money from vegetable and fruit gardening. Fendler also had some insects that he wanted to send to J. L. LeConte in New York and asked Engelmann to arrange it. He had progressed in his study of botany enough to remark to Engelmann that “the south- western part of Arkansas where I am now may have much of interest for botany; and the flora of this area is probably a transition of the Texas flora to the middle states." Fendler also asked for books, Gray's “Illustrated Genera" and Mi- chaux's **work about the North American trees." The sojourn in Arkansas was aborted by unpro- ductive soil, which meant that “cash was very rare." In June the Fendlers moved back to New Or- leans, where Augustus added Gray's “Botany of the Northern United States" to his library, and it cost only seventy-five cents! In July they re- settled in Memphis, Tennessee, and prospered there for the next four years by operating a gas- lamp business. It pleased Fendler to learn that Engelmann and Gray intended to name a plant after him. (Fendlera Engelmann et A. Gray, a new genus of Saxifragaceae, was published in the Smithsonian Contr. Knowl. 3: 77, 1852.) Fendler 1986] conducted horticultural experiments in Mem- phis, sowing seeds of plants that he had collected in Chagres and of others that Engelmann sent him. Four species from Chagres grew well, of which “two beautiful mimosa species (both bog plants) are now 4—5 feet high, but probably won't last very long outside" (28 September 1851). By this time Fendler had already recorded temper- atures of 38 degrees and observed frost on the leaves. Meanwhile the chemical-oil business was succeeding so well that Fendler had to abandon collecting and also most of his reading. The latter enjoyment had so increased his proficiency in English that by the 1870s he had translated Goethe's Faust into English verse (Canby, 1885), presumably for his own pleasure. Fendler reported his exact observations on the growing plants to Engelmann and his daily me- teorological measurements to the Smithsonian Institution. “I found that the thermometric stud- ies made by the local Navy-Yard here and some- times published are very different from mine," he wrote on 1 March 1853. In fact, he continued, during a whole month not one single item agreed with mine. Only one example: 3rd February 1852 sunrise 3 p. m. according to my obser- vation Navy-Yard— There seems to be no difference between the thermometers themselves since on the 19th of January 1852 both showed at sunrise — 2°F . . So much for official meteorological state- ments of the local Navy-Yard Meanwhile his gas-lighting business began los- ing customers. By 22 February 1853 the Mem- phis Gas Company was supplying many resi- dences and businesses with natural gas, but Fendler had known of the impending introduc- tion of gas-lighting in Memphis the previous Sep- tember, at which time he had planned tentatively to convert part of his business to the distillation of alcohol. Although on 15 August 1853 Fendler expressed some interest in exploring the Great Plains with Engelmann, he had already deter- mined to move farther south to a German set- tlement in Venezuela known as Colonia Tovar. Business had declined too much for him to re- main any longer in Memphis, but he told En- gelmann that his main reasons for wanting to move to the village near Caracas were: STEIBER & LANGE—AUGUSTUS FENDLER 525 In a city with a population of 70,000 I be- lieve to be able to find more security than here, and secondly, because I believe I shall have a better climate there. Then also, the wish to live again in a mountain valley and to be in the vicinity ofa rich mountain flora! without having the difficulties of long and difficult communication as I found in Santa Fe .... I should think that the mountain area of Caracas should be rich in cactus species [15 August 1853]. Always eager to please Engelmann, Fendler also noted that the Opuntia plants that he had sown for him grew “very slowly" in his Memphis gar- By 16 December 1854 the Fendlers had relo- cated to Colonia Tovar and were preparing near- ly eight cases of specimens for Engelmann, in- cluding fifteen to twenty specimens each of 250 fern species and of six palm species. “You can imagine how much work this involves," Augus- tus explained. “However, Lindley's Vegetable Kingdom is a great help to me ... [and] apart from the ferns, the colony contains orchids, So- lanaceae, Rubiaceae, Melastomataceae in great numbers of species." Meanwhile, he continued his meteorological observations, later receiving a barometer and dry and wet-bulb thermometers from Joseph Henry (Fendler to Engelmann, 27 April 1856). The description of the conditions that Fendler left behind in the United States would make any current weather report of the same area sound humdrum: There is [at Colonia Tovar] no scorching summer's heat, no fearful winter's cold, nei- ther tornadoes to devastate the country, nor gales to blind the inhabitants with sand or dust, or penetrate their clothes and flesh with piercing frost. Lightnings are rare and rather harmless, thunders merely grumbling (Can- by, 1885). The Fendlers designed a fountain next to their house, hollowing out a palm trunk and propping it so that the water could shoot up to nine feet in the air. Besides apple trees, bananas and palms, a vegetable garden provided them with potatoes and other staple foods. But since cash for sale of plant specimens merely trickled in from his friends in North America, Fendler resumed dis- tilling alcohol to earn some income. Engelmann's letter of 27 March 1856 heartened him with the news that a natural history society and a botan- 526 ical garden (Shaw’s Garden) would soon be founded in St. Louis. “The reason I hope to end my life in the United States," Fendler wrote on | April of that year, “is the wish to follow the results of intensive scientific education. From the last I feel so far removed in Venezuela and in this due I am very isolated." During Fendler's visit to the United States lat- er that pasión Sullivant paid him for sets of mosses and liverworts; Tuckerman did likewise for lichens. However, LeConte would not buy Fendler's 600 **bugs" even though the entomol- ogist thought them admirable specimens. He bought *only North American things. ere- fore," wrote Fendler, “I shall keep them for my- $ By the time the Fendlers returned to St. Louis in 1864, Shaw’s Garden had already been thriv- ing for five years. Soon after arriving, the broth- ers established a small farm at Allenton, Mis- souri, not far from the city, and lived there for seven years. Augustus, however, did accept a short-term offer from Asa Gray to work as a curator at Cambridge, one of many abortive ef- forts by the elder botanist to alleviate his work- load (Dupree, 1959). The negotiations began nearly a month after Abraham Lincoln’s assas- sination and culminated in Fendler’s move to Cambridge. He described his routine to Engel- mann (2 November 1865): During the day, which begins at 7/ in the morning until 5 o’clock in the evening, I work in the herbarium here where there is always work to do and of the four working days of which Gray wrote me earlier, noth- ing is said any more. I did not claim any time for myself, since I well know he would like me to lose as little time as possible. I did the sorting of Ranunculaceae to Rosa- ceae inclusive, according to the partly pub- lished work by Benth. and Hook. “Genera plantarum” and numbered the genera ac- cordingly. The grasses according to Steudel; Labiatae, Scroph., Verb., Solanaceae, Com- Gray still has a great number of bunches of dried East aaa plants from the Kew her- barium more, several packages of Japanese ba coles ca by Oldham, which will still have to be poisoned and pasted on white paper. Then, the plant parts of Wilke’s expedition have to be looked at, labels to be ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 written for the duplicates. Furthermore, many plants, before being laid out for past- ing, should be compared with those in the herbarium, whether there is still room on the old sheets and whether already a suffi- cient amount of specimens of one and the same species in the herbarium, in which case Dr. Torrey receives the duplicates. Also there are other collections to sort, for instance my own and that of Wright. From time to time, I also plant something small in the garden, keep the rooms of the herbarium clean, get charcoal out of the basement, etc. ... Gray learned yesterday that the gardener de- mands $5 per week for my board, he had only counted on $4 at the most and does not want to give more in the future; and therefore I again have my meals at Gray’s table where the midday meal is served at 515, This never worked out well because Fendler felt it a hardship to have to shave and dress well in the presence of “ladies” and because he pre- ferred to eat alone. Gray accommodated Fend- ler's shyness and paid for his meals at a local boarding house. Meanwhile Fendler had been considering two scientific topics, the theory of heat and the cause of fluctuation of barometric pressure from hour to hour that he had noticed in Venezuela. His lucubrations gave rise to the following remarks to Engelmann on 4 February 1866: When I sent Prof. Henry my meteorological observations from Venezuela in June 1857, I sent them together with the half-hourly barometer readings and stated my theory about the cause of the periodical differences of the barometer .... In the report of the Smithsonian Institution of 1857 in which my observations and letters were published, this theory was left out, probably because Prof. Henry thought it too daring. Now, I find in one of the numbers (no. 114, p. 380) of the Journal of Science and Art that a Mr. Chase from Philadelphia arrived at the same result after three years of continuous hourly barometrical readings made by the request of Sabini during the years 1842-45, and proves this with a large algebraic account. The same will probably happen to my theory about the warmth [heat] if I cannot publish it soon. 1986] Dr. Gray says I should write to Henry and ask him to return my manuscript in order to prove the priority of my theory. [On 25 December 1865 Fendler had written Engelmann about the ““warmth”:] Gray ob- tained for me “Tyndall’s Lectures on Heat considered as a mode of Motion" and I spent many evening hours reading this. I found that the experiments were carried out well and with great ingenuity. But the author did not succeed in proving that heat is nothing but a “mode of motion”. I have several notes and could show direct contradictions and that he gets sometimes involved with prob- lems from which he cannot extricate him- self. I have never been able to see that warmth is something material; but neither can I be- lieve that it is nothing but motion. I believe in my own opinions which I have not yet found in any work and intend to develop them later on when I am back home and have more time. In Spring 1866 Augustus returned to help his brother farm at Allenton and continue his friend- ly collaboration with Engelmann. Unfortunately, he discovered that in his absence, his brother had developed night-blindness. In the summer of 1871, the brothers sold the farm and moved into St. Louis, an action that he later regretted because he found so few people there with whom he shared any interests. He loathed the city prin- cipally, he said, because of “‘the horrible noise, din and hurry of men adoring mammon." The wandering life of a collector certainly had turned him against city dwelling, too. St. Louis dis- agreed with him so much that he and his brother decided to visit Gumbinnen, Germany, in 1872. It was to be an extended vacation to judge from the four boxes of books they took with them. However much he disliked living in St. Louis proper, his first letter to Engelmann from Ger- many indicated that Augustus Fendler missed the United States. Although the weather in East Prussia was exceptionally mild in 1872 —so mild that he compared it to the idyllic Colonia To- з? ' he continued to Engel- mann (2 January 1873), “we don’t like the still ever-present narrowminded spirit of the bour- geois, the soldier and officials, and my capital is not sufficient to live on." So Fendler planned again to return to America and bolstered his bo- STEIBER & LANGE—AUGUSTUS FENDLER 527 tanical go by visiting the d bo- n. The for- mer boasted its fine Ауа pe aquatics, especially the attractive species of the Nym- phaeaceae, whose foremost student, Prof. Jo- hann Caspary, greeted Engelmann's friend with warmth. Professor Alexander Braun, a notable phycologist, drove Fender through the Berlin garden where the y d to see the large collection of living cac- tus plants with their strong and healthy growth.” Braun also informed him that George Engel- mann, Jr. had passed his doctoral examination. On 4 April 1873, after only eleven months in Gumbinnen, the Fendlers left for Hamburg. They sailed with some 500 other passengers, mostly Scots and Irish, on the Zsmaila, which arrived on 1 May in New York, one year to the day after they had disembarked at Hamburg. Finding that John Torrey had recently died, the Fendlers con- tinued on their way to Philadelphia where Au- gustus renewed acquaintances with LeConte and Thomas Meehan, both of the Academy of Nat- ural Sciences. Here Fendler planned to present “some small scientific papers” and so settled temporarily in the city. Again he grew restless and tired of city life and after some exploration of the environs of Philadelphia found lodgings in the rustic town of Seaford, Delaware, on the shores of the Nanticoke River. Seaford had “a bank, two hotels, b dh churches, one mill and and one other German. The oys- ter trade was dime dm there, as Fendler wrote Engelmann on 16 February 1874. The plants of the Coastal Plain, a phytogeo- graphic region new to Fendler, interested him. He was especially startled to find that **this sandy and sterile looking soil can produce such huge oak trees as I have seen myself." Besides an in- come from interest on bonds that Fendler owned, supplemental cash soon came from a botanical source. William Marriott Canby, a wealthy rail- road owner at Wilmington, Delaware, and col- laborator of many botanists including Engel- mann and Gray, soon “looked in" on Fendler at the urging of the former. Eventually Canby man- aged to persuade Fendler to move to Wilmington specimens and enjoyed his conversation about the wonderful and curious vegetation of the trop- ics. Meanwhile both Canby and Gray viewed with 528 growing dismay Fendler’s attention to his ec of the Universe.” So concerned was Gray that he wrote Engelmann on 19 June 1874: *[ wish he had let Cosmical Science alone! But now he never will and is a gone goose." Gray's own practice of eschewing the metaphysical de- bates about Darwinism informed this judgment. Canby confirmed that “nothing could persuade him that this book was not to bring him ever- lasting fame and no reasoning could discourage him from undertaking the expenses of publishing this work" (Canby, 1885). Despite his reserva- tions, Gray assessed Fendler’s work in “Cosmi- cal Science" in a more affirmative fashion later: “In the year 1874 he published at Wilmington, Delaware, where he then resided, at his own ex- pense [about $300] and, we suppose, with small returns, a well-written treatise (of 154 pages, 8vo) on ‘The Mechanism of the Universe and its pri- mary effort-exerting Powers; the Nature of Forces and the Constitution of Matter, with remarks on the Essence and Attributes of the All-Intelligent.’ He was one of the ingenious race of paradoxers, and it may be left to the future De Morgan to characterize his work. He will certainly be last- ingly and well remembered in botany” (Gray, 1884). Fendler was pleased enough that Joseph Henry bought a copy of the book. On 15 April 1875 he told Engelmann that Gray had visited him and Charles C. Parry, too. Parry had cut his teeth as a collector in the Mexican Boundary Survey and was about to join Edward Palmer at San Luis Potosi in Mexico. Fendler hadn’t seen Gray in ten years. Although he was glad to learn that Gray was now steadily churning out parts of the Synoptical flora of North Amer- ica, he was surprised at how he had aged. This prompted some assessment of his own and En- gelmann’s situation (9 November 1875): Even though Gray told me during our meet- ing last spring, that I still look as I did ten years ago, I feel the coming on of old age since next January I shall have completed my 63rd year. I believe you are five years older than I and Gray three years, Parry ten and Canby fifteen younger. [And elsewhere:] You ask whether I am now satisfied with my life. It seems to be a natural trait in a human being who has moved around so much and so constantly, that he wishes to have a quiet home at least in his old age where he can spend the last days of his life without worries. I believe I have ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 reached this aim. Whatever I was looking for, I found. In the winter I am busy with literary works I like, in the summer plant collecting and I find much satisfaction and pleasure. Some times I have the desire, as in younger years, to go far into the woods and meadows and to read the *1000 flowers' which the earth offers in such abundance even in the most distant lands. Especially, I should like to see again the jungle of the tropical mountain areas where I once wan- dered and look up the old and well known vegetation in all its hiding places. But phys- ical weakness from which I sometimes suf- fer, reminds me to stay quiet at home. How- ever, I had given some serious thoughts to find enough subscribers to Venezuelan ferns to support such an enterprise financially. The Sea of Valencia with its beautiful and pleas- ant natural scenery, its picturesque sur- rounding hills and its incomparable lovely climate is very tempting to me and I should like to enjoy again the balsamic evening and morning air. However, I would only decide to take such a trip if the rheumatism which attacks me regularly during the winters in Delaware continues and makes my life mis- erable The next spring Fendler observed flowering and leaf vernation in oaks for Engelmann (Fend- ler to Engelmann, 16 May 1876); the two com- pared meteorological information over the next few years as well. Finally, however, the rheu- matism that had plagued Fendler in recent years increased his discomfort so much that he decided to move to Trinidad where he hoped to both find relief and continue contributing to botany and meteorology in his final years. Engelmann asked him to pay close attention to the cacti there to which Fendler replied (6 May 1877): In his Flora of the British West Indian Is- lands, Grisebach only described one single species of Cactaceae growing in Trinidad, namely Rhipsalis Cassytha. From all the British West Indian Islands, he had of Mam- illaria 1, Melocactus 1, Cereus 7, Rhipsalis 2, Opuntia 5, Peiriscia [=Pereskia?] 1, all in all only 17 species. I very much doubt that this is the complete number. Once in Trinidad Fendler wasted no time and sent Gray fifty sets of about seventy-five species each of ferns, which Gray described as “‘nice and 1986] satisfactory." Unfortunately, Fendler had no luck in finding cacti. However, he happily reported to Engelmann (26 August 1878) that he had “125 species [of ferns] in my own collection and intend to get up to 200 ... [and] Euphorbiaceae are strongly represented [here]." Even in the 1870s tropical forests were being ravaged by progress, and Fendler lamented: “Неге in Trinidad, a small railroad is being built (the second one); it is a pity that I did not learn about this earlier so that I could have been present at the cutting down of the trees." Despite this and his general disgust with the “‘folly and rancor of the surging multi- tude" both in Trinidad and in Delaware (an event that reinforced this shortly before he left Wil- mington will be recounted momentarily), Fend- ler still felt that his final days just might be a little idyllic. “The time for the mangos and breadfruit has almost past," he wrote in August, “but the branches of the two orange trees are bending down already under the burden of their fruit which will not be ripe until October." He later reported (6 January 1881) that he and his brother customarily ate 5 to 6 oranges daily (over 1200 oranges from their trees in six months), numerous plantains and pure chocolate that they prepared themselves from the cocoa beans they rew. Shortly before Fendler had left Delaware the police had been alerted by some ignorant people who, presumably unaware of the profession of plant collecting, suspected the Fendlers of coun- terfeiting money. Consequently, the over-zeal- ous officers searched the house and even dug up the garden in search of contraband. Finding nothing they, nevertheless, left everything a mess. After arriving in Trinidad, Fendler wrote Canby: “That this little affair weighed heavily upon my mind and gnawed deep into my immoderately sensitive feelings, you may well imagine.’ On 26 March 1879 Fendler's compassionate heart went out to his bereaved friend in St. Louis who had informed Augustus in his 27 February letter of the death of his wife. Fendler could only: hope that the hand of your son will heal better than you think and also that your health will be improved in the meantime, so that you can still count on a long and fruitful life. It must be a comfort to you that you are surrounded by many admirers and friends and that you enjoy all the comforts a human being craves. Engelmann, in his turn, counseled his friend STEIBER & LANGE— AUGUSTUS FENDLER 529 that he was too sensitive, nervous and easily up- set. Fendler responded (6 September 1879): Why I am concerned with the rabble which surrounds me constantly and which only re- minds me too vividly of the degradation of the human race into animals or the opposite, from an animal to a human being [I do not know], but the disgusting impressions re- main the same. Fortunately, the masses have not much to say in this country. [Other things also occupied his mind, how- ever.] For the last 35 years since we became friends, I never lost track of your successful career. That your usual good health is be- coming worse is probably due to the late sad events and I hope that you will be well again soon .... From the New York Tribune I see what tremendous progress the ‘far west’ especially Colorado is making lately, espe- cially with regard to population and wealth. If one thinks back of the times and condi- tions of the regions of the territories west of the Mississippi when we came to St. Louis, I in 1838 and you even earlier, it seems that the developments made during the last years are as wonderful as the tales from *1000 and one night[s]'. But how will things look in the United States after another 40 years? In 1881 Engelmann traveled to the “far west” and reported to Fendler his observations on his trip from San Francisco to Vancouver on a steamer. In the meantime Fendler kept collecting and also occupied himself with weather data. He helped Henry Prestoe, Superintendent ofthe Port- of-Spain Botanic Gardens (1864-1886), “figure out the average numbers of his fourteen years of meteorological observations" and sometimes worked for him as a clerk, that “spirit-killing” occupation. Fendler was glad to hear from En- gelmann that George W. Letterman had “Ъе- come such an eager and energetic collector and observer, EL since, as you say, I gave him the first push" in 1879, Fendler was reminded once more of his own frail condition. Augustus even may have suffered a mild heart attack “lifting a wooden fence" that he and his brother “tried to put into the right place." He described it thus: I injured myself (apparently my chest and 530 back) and then immediately suffered from an attack of rheumatism, from which I have not yet recovered. The pains in the chest are a great nuisance sometimes, especially if they occur around the heart (6 July 1881). In the same letter he learned that Engelmann’s health was not improving either. However, none of this deterred him from exploring Trinidad. He regaled Canby (1885) with an amusing tale of his encounter with a local dealer in what, from all appearances, had to be hallucinogenic drugs: Having ascended one of the highest ridges of the Saut d’eau mountains, about ten miles from town, I took occasion to visit a man known all about as Popo Fernand (though his real name is Joseph Isidore), in order to inquire of him about a piece of land that was offered for sale in his neighborhood. On my way thither I was astonished to find that in and beyond the village of Maraval every man, woman and child knew where the man lived, though his cabin was miles away in the mountains in an out-of-the-way place. When I at last reached his premises I found no one there, but noticed, as something un- usual, a great number of beehives stuck all around his cabin and outhouses, the first beehives that I have seen in Trinidad. After ance. Neither he nor any of his neighbors could speak English and I could not speak their language .... The man seemed, how- ever, to be courteously disposed. In order to see how the land lay, I exposed my little pocket compass in his presence, when at once he seemed to become alarmed, and made me understand that he thought the instru- ment was intended to show the spot where money was hid in the ground. Of this notion I tried to disabuse him. Soon after he invited me into his room and, as is customary here . he asked me to help myself to the con- tents of a small bottle he set before me. Not to show any signs of distrust, I poured out about two thimblefuls of the liquor, mixing it with plenty of water, but became some- what suspicious after drinking it on noticing that Fernand himself had not taken any of the bottle's contents. About ten minutes lat- er, on my way back, I experienced a strange state of mind such as never before I had happened to be in. There were neither diz- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ziness, stupefaction nor exhilarating symp- toms. Visions and strange | и J 4 and аео at as they came. Any theme I made an effort to think upon slipped from my memory, and instead thereof quite a different theme presented it- self with the same futile result, until I be- came frightened at my own thoughts and ipia at my condition of mind. After a wo hours brisk and steady walk this un- ean irritation of mind gradually subsid- ed .... at would have been the result had I taken a little more of that liquor? In the summer of 1881 Fendler informed En- gelmann that he had sent J. D. Hooker five sets of his phanerogams. So Kew also was reaping part of Fendler’s harvest. But they had done so before. In fact, years earlier Hooker had written Gray praising Fendler’s Santa Fe specimens (Du- pree, 1959). Augustus also remarked that Henry Shaw seemed to be one of those long-lived people who might reach his father’s age, near ninety. He fondly recalled and inquired about Shaw’s sister and brother-in-law. A year later, 20 July 1882, his last letter to Engelmann (that we know of) states: I received your letter of May 24th.on June 20th and was glad to learn that you did not suffer from rheumatism for quite a while. What I mistakenly thought when I wrote you my last letter on January 4th, turned out to be a congestion of the liver, a dangerous one, as the doctor told me; after taking the medicine he prescribed I felt quite well several weeks later. Lately, however, I don’t have much appetite which is probably the result of the contin- uous heat and humidity, I also suffer from a weakness in all bones and especially the destroyed several years after I left there; I have not yet been able to find out whether it was rebuilt again. [He provided some weather data, promised him a specimen of 1986] Mamillaria papyracantha, and observed that] we expect to have streetcars in Port of Spain which will be a great convenience for me. Fendler may have enjoyed the trolleys for a short time, but he never returned to Colonia To- var nor did he ever find his utopian farm in the backlands of Trinidad. Death struck him down on 27 November 1883. His life-long friend En- gelmann had a memorandum of this event (how recently arrived is not known) on his desk at the time that he died, namely 4 February 1884; Fen- dler was 71, Engelmann 76 (Gray, 1885). Canby had written in 1885 that only “brief notices” about Augustus Fendler had appeared in scientific periodicals since his death, “but scarcely such as so excellent a man and one so useful to science deserved.” Perhaps these selec- tions from his letters to Engelmann will lengthen the deserved notices about this man so beloved of Canby, Gray and Engelmann. (For the most recent bibliography on Fendler’s life and works, see Stafleu and Cowan, 1976, vol. 1, pp. 822, 823, 990). LITERATURE CITED CANBY, W. M. (editor). 1885. An autobiography and some reminiscences of the late August Fendler. I, STEIBER & LANGE— AUGUSTUS FENDLER 531 Bot. Gaz. 10: 285-290; II, Bot. Gaz. 10: 301-304; III, Bot. Gaz. 10: 319- 322. Dupree, A. Н. Asa Gray, 1810-1888. The Belknap Press of Harvard Univ. Press, Cam- bridge. Gray, A. 1849. Plantae fendlerianae г mexican- ae. Mem. Amer. Aca s, ser. 2, 1: 1-116. —. 1884. General notes. lari Bot. Gaz. 9(7): ыс 112. ———. 1885. Botanical necrology for 1884. Augus- tus Же George Englemann, S. В. sa ckley, J. Williamson, J. H. Balfour, Н. К. Goe g Bentham. Amer. J. Sci., ser. 3, 29(129): ben 172. МСКЕГҮЕҮ, S. D. 1955. Botanical Exploration of the а West. 1790-1850. The Arnold um, Jamaica Plain. Augustus Fendler’s collection list: New v Mexico, 1846-1847. Contr. Gray Herb. 212: 1-70. STAFLEU, F. A. & В. S. Cowan. 1976. Taxonomic mas В 2nd Ud nume 1. Bohn, Schel- a & Holkema, Utre бш М.Т. 1982. | pr plant collectors’ field notes held in North American institutions. Huntia 4(3): 151-202. ARCHIVAL SOURCES Letters from Augustus Fendler to George Engelmann, compiled and translated = Carla ав 87 typed рр. In the Library, Missouri Botanical Garden. Copy at Hunt Institute. Original ae at Missouri Botanical Garden THE EVER-CHANGING LANDSCAPE OF CACTUS SYSTEMATICS! ARTHUR C. GIBSON,” KEVIN C. SPENCER,? RENU BAJAJ,’ AND JERRY L. MCLAUGHLIN* ABSTRACT Since Linnaeus originally described 22 species in the single genus Cactus, over 11 ,000 binomials and 400 generic names have been propo a constant state of change and turmoil. When Engelmann be names had been proposed, and the Eus was not well collected. Engelmann | described and skillfully phylogenetic information in tribe Pachycereeae, which includes the large colu о nocereinae, which reas species with funicular pigment species are often treated as two subtri arisons are ma existing phylogenetic hypotheses of the species groups. r Lemaireocereus hollianus, Neobu. xbaumia mezcalaensis, Pachycereus grandis, and sev- ne area of active phylogenetic research has been mnar cacti of Mexico and adjacent areas. These eral i of Cephalocereus, pecies of Cephalocereus and Neobuxbaumia as well as Mitrocereus. In addition, alkaloids are reported here for the first time in species of Stenocereinae. Cacti are famous for their beautiful flowers and many bizarre vegetative features, but they are equally famous— or infamous — for their nomen- clatural and systematic problems, which are, in- deed, formidable. Although most botanists are overwhelmed or greatly confused by the taxo- nomic literature on Cactaceae, some strides have been made to unravel the phylogeny of this fam- ily, which includes about 120 genera and 1,550 species (Gibson & Nobel, 1986). Moreover, sys- tematic goals for cacti are no different than those met and addressed in any large and diverse fam- ily of plants: 1) to define the limits of each species; 2) to choose the oldest valid binomial; 3) to rec- ognize monophyletic taxa; 4) to define the cri- teria to be used for erecting each genus; and 5) to produce a truly phylogenetic classification of the genera. This paper examines the problems of producing a phylogenetic classification for cactus genera. Following a brief review on the remark- able taxonomic legacy of Cactaceae, discussions will concentrate on the columnar cacti of tribe Pachycereeae to demonstrate what types of re- search are needed to solve systematic problems in this family. A BRIEF HISTORY OF CACTUS TAXONOMY Christopher Columbus and his crew were un- doubtedly the first Europeans to see cacti (How- ard & Touw, 1981). Unfortunately, cacti were not mentioned in the published logs of the Co- lumbus voyages (Morison, 1963), so this state- ment cannot be verified; but these explorers could not have missed cacti, which constitute a con- spicuous part of the vegetation in the West In- dies. Apparently the earliest New World descrip- tion of cacti by a European was contributed by Oviedo (1526), who lived in the New World. Later Hernández (1514-1578) discussed 15 species of cacti in his famous account of Mexican plants that was eventually published in 1615. By the 17th century cacti were already being cultivated in western Europe. Two unvalidated ! This study was supported by National Segen Foundation Grant Ms 81-08891. We thank Myron Kimnach ane James Dice at the Huntington Botan ? Departm cal Garden for their assistan jin e Biology, University of California, Los Angeles, California 90024. + eo ognosy dicinal C ies * De men se Purdue University, West Lafayette, Indiana 47907. ANN. MISSOURI Bor. GARD. 73: 532-555. 1986. and Medicinal Chemistry, University of Illinois, Chicago, Illinois 60612. stry and Pharmacognosy, School of Pharmacy and Pharmacal Sciences, 1986] exu names, “Cereus Peruanus” and “Ficus In- " appeared on a list of British plants by x dh (1599). The earliest authentic report of Melocactus communis Link & Otto has been credited to Clusius in Holland, who in 1605 was brought a cultivated specimen from the island of Maio in the Cape Verde Islands near West Africa, where it was first spotted by Dutch sea- men in 1601 (Heniger, 1973). How Melocactus, a New World genus, became established on Maio is a matter of speculation, but the Dutch were probably the ones hon Mei die this cactus from Curacao, presumably in the mid or late 1500s. Bauhin and о pears шо. 1623; Rowley, 1976) mentioned the presence of species of “Cereus” and “Ficus Indica” in European gar- dens. Herbarium vouchers of Opuntia from the 1660s occur in the bound herbarium (Hortus Sic- cus, L) of Gaymans, a Leiden pharmacist, who documented the plants in Leiden's Hortus Bo- tanicus, the oldest botanical garden in The Neth- erlands. Many cactus species, including seven species of cereoids, were described from gardens in The Netherlands by Hermann (1687, 1698), and Boerhaave (1720) showed an engraving of the first flowering specimen (1 August 1691) of “Cereus peruvianus” in the Leiden botanical gar- den. Whatever role other countries had in intro- ducing cacti to Europe is still unstudied, but cer- tainly Dutch horticulturists were central in promoting the early interest in cactus cultivation, which soon became an avid passion of many gardeners. The Dutch influence on cactus taxonomy ex- tended, of course, to Linnaeus, who studied plants in The Netherlands. In Hortus Cliffortianus, which is a garden catalogue of the estate of Dr. George Clifford near Haarlem, Linnaeus (1737) mentioned 16 species in the genus Cactus and one species, which has large leaves, in the genus Pereskia. Pereskia, a name borrowed from the botanical illustrator Charles Plumier, honored Nicolas Claude Fabri, seigneur de Peiresc, who was a correspondent friend of Clusius. However, when Linnaeus (1753) published Species Plan- As за ~ ES = Z5 B = О, р кә SE gz oí o E [^^ — o ро Ф © Е S =. о 5 В. р 5 “и. ec nd 0 no E ® 33 5 5 growth fi bearing cacti. ae Miller (1754)? rec- ognized three additional genera, Cereus, Opun- tia, and Pereskia, based on the earlier common GIBSON ET AL.—CACTUS SYSTEMATICS 533 names, to emphasize some major differences in the cacti that were already known. At present, the species described by Linnaeus and Miller are classified in ten or more genera, belonging to three different subfamilies; since Linnaeus about 11,000 Latin binomials and an additional 400 generic names have been published for cacti. Re- grettably, most published cactus binomials are illegitimate, invalid, or incorrect, because pres- ent-day cactus systematists recognize only 1,400- 1,700 valid species and 70-140 genera. More- over, in 1930, when Mammillaria Haworth was conserved, the name Cactus L. was declared a nomen rejiciendum because the lectotype of the family, C. mammillaris L., became a species of Mammillaria (Hunt, 1967; Shaw, 1976; Howard & Touw, 1981). Rejecting Cactus enabled tax- onomists to resolve many nomenclatural prob- ems and to reduce th the inconsistent ан аа and usage of Cactus by various au Cacti peeing before 1820 were mostly species that were collected in the West Indies and along the eastern coastline of North and South America, such as from Brazil, Venezuela, Mex- ico, Florida, and Virginia. Some of the early dis- coveries were epiphytes, which were so remark- able in vegetative appearance and diversity that new generic names were proposed for the differ- ent forms, e.g., Hariota Adanson (1763), Cassyta J. Miller (1771), Rhipsalis Gaertner (1788), Epi- phyllum Haworth (1812), Phyllocactus Link (1831), and Lepismium Pfeiffer (1835) Taxonomic knowledge of Cactaceae acceler- ated rapidly when botanical exploration pene- trated the arid and semiarid regions of the New World. New genera were proposed for low growth forms of Mexico: Mammillaria Haworth (1812), Echinocactus Link & Otto (1827), Ariocarpus Scheidweiler (1838), Astrophytum Lemaire (1839), Echinoft Lawrence (1841), and Pelecyphora Ehrenberg (1843). Also, Pfeiffer (1838) proposed Cephalocereus as the first seg- regate genus of Cereus, based on the type, Cactus — stemming from collected in South America and the West Indies were assigned to other new genera: Melocactus Link & Otto (1827), Echinopsis Zuccarini (1837), Discocactus Pfeiffer (1837), and Gymnocalycium Pfeifier (1845) George Engelmann entered the field of cactus 5 Literature citations for all generic names mentioned in the following pages can be found in Hunt (1967). 534 taxonomy in the 1840s, after his reputation as a botanical taxonomist had already been estab- lished. Cactus materials collected in the United States and its territories were generally referred to him. Soule (1970), Mitich (1974), and Benson (1982) have written accounts of Engelmann’s contributions to cactology between 1845 and 1878. During these years, Engelmann obtained live plants, herbarium specimens, first-hand de- scriptions, and illustrations of cacti from western North America and then reorganized, analyzed, and legitimately published names for hundreds of new taxa. Not all of Engelmann’s names have survived careful systematic analysis, but it is a tribute to Engelmann that a great many of them are still considered correct. e vast majority of common cacti of the southwestern United States and adjacent Mexico were first described by Engelmann, including seven of the ten cereoid species of the United eee o: 17 species and three O species and varieties of platyopuntias, and over 50 species ofthe globular, caespitose, and barrel forms. Half d by Ben- son (1969a, 1969b, 1982) in the large cactus flora of California and Arizona have Engelmann as an authority, and Wiggins (1980) accepted 24 En- gelmann taxa in Baja California, particularly species that occur in the northernmost latitudes. Specific epithets commemorate the people who collected specimens for his studies in St. Louis: Ferocactus wislizenii (Engelm.) Britt. & Rose, Opuntia parryi Engelm., O. stanlyi Engelm., O. lindheimeri Engelm., O. bigelovii Engelm., Loph- ocereus schottii (Engelm.) Britt. & Rose, Stenoce- reus thurberi (Engelm.) Buxb., Peniocereus greg- gii (Engelm.) Britt. & Rose, Echinocereus fendleri Engelm., and Mammillaria wrightii Engelm. Each taxon was very carefully described, and many were exquisitely illustrated by Paulus Roetter. Engelmann was also the person who proposed the genus Echinocereus (Engelmann, 1848), al- though he later changed his mind (Engelmann, 1849); and he recognized the distinctiveness of Coryphantha, which was later elevated to generic rank by Lemaire (1868). Lemaire, who studied cacti during the same period as Engelmann, also made many individ- ual contributions to cactus taxonomy for North and South America. He published 12 new generic names, at least six of which are still widely rec- ognized (Aporocactus, Cleistocactus, Schlumber- gera, and Tephrocactus, as well as the two Le- , about 2 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 maire genera already mentioned). At that time South American cacti were still relatively un- known, although Philippi (1860) reported new Chilean discoveries and described three distinc- tive genera, Eriosyce, Eulychnia, and Maihue- nia. Cactus studies in arid and semiarid regions of North America were progressing through the floristic studies of Coulter, Orcutt, the Brande- gees, and Weber. Weber also conducted cactus investigations on the Galapagos uc RON and the ш species of Costa Ric The onomic history of Bien was directed mostly 2 large descriptive monographs of the entire family, instead of by smaller monographs of individual genera. The first broad familial monograph was published at the turn of the cen- tury by Schumann (1898). Judged by today's standards, this classification is considered prim- itive, but Schumann made several interesting contributions. He was the first person to divide cacti into three subfamilies, which are still in use: Cereoideae, the largest taxon, which is now prop- erly renamed Cactoideae in accordance with Ar- ticle 19 of the International Code; Opuntioideae, which are the species having a white, bony aril covering the seed; and Peireskioideae (now spelled с, the large, relatively nonsucculent cacti that have large leaves (Pereskia) and the closely related Maihuenia, which are cushion plants (Gibson, 1977). Schumann recognized 21 genera, and he grouped the genera of Cactoideae into three tribes, Echinocacteae (including the cereoids), Mamillarieae, and Rhipsalideae. These divisions were not phylogenetic in any modern sense. He rejected the name Cactus before it was proper to do so, and for the low growth forms he accepted many of the valid segregate genera that were mentioned above. анн ulti- mately proposed four new gen Zygocactus (1890), Rebutia (1895), =н ыр ин, (1897), апа Wittia (1903), but oddly enough he did not use either Zygocactus or Rebutia in his 1898 mono- graph; indeed, he did not even recognize Rebutia as a subgenus of Echinocactus, and he made Zy- gocactus a synonym of Epiphyllum Haw., which he and other cactologists totally misunderstood. In 1904 Britton and Rose began their famous studies on the taxonomy of Cactaceae, which included a careful reexamination of all original descriptions and type specimens, extensive new field work in cactus areas throughout the New World, and assemblage of large living and dried collections for close study and photographing. Beginning in 1911, these studies were financed 1986] by the Carnegie Institute of Washington, and Britton and Rose (1919-1923) eventually pro- duced a four-volume taxonomic monograph, The Cactaceae, which is familiar to every systematic botanist and cactologist and is undoubtedly the most commonly cited and central cactus refer- ence. To discuss the contributions of Britton and Rose, some attention first must be paid to Berger, who was curator of the botanical gardens in La Mortola, Ventimiglia, Italy. Berger (1905) pub- lished a revision of the genus Cereus in which he recognized 18 subgenera. Many of these names were later used as genera by Britton and Rose but often in a much modified form. In Berger’s revision, the columnar cacti that were known to him were those species that occurred in Mexico, the West Indies, Costa Rica, Peru, Chile, Argen- tina, and Brazil. To assess fully the taxonomic contributions of Britton and Rose would take a very long article, half of which would consider the progressive, positive changes initiated by them, and half would still is the cornerstone of cactus systematics. Brit- ton and Rose accepted Schumann’s three sub- divisions, although they renamed these as tribes. In a massive revision of subfamily Cactoideae (Tribe Cereeae), Britton and Rose recognized 114 genera in eight subtribes and subdivided Cereus into many smaller and more homogeneous units. With this action, species of columnar cacti were placed into seven of the eight subtribes, and the name Cereus became restricted in usage to a small group of species in eastern and southern South America. Britton and Rose together described 77 Looking at this great cactus monograph with 20-20 hindsight, it is easy to find many faults. One unfortunate choice was substituting Neo- mamnmnillaria Britt. & Rose for Mammillaria Haw., which in 1930 was accepted internation- ally as a conserved name; this resulted in 186 unnecessary binomials under the new and ulti- mately illegitimate name. There were, of course, a number of genera that had improper member- ship, and binomials listed in synonymy have sometimes turned out to be “good” species that belonged in another genera. Moreover, at least 10 genera in the monograph appear now to be too narrowly defined to recognize, especially some GIBSON ET AL.—CACTUS SYSTEMATICS 535 names that were proposed as segregates of Mam- millaria and Rhipsalis. Regardless of this, at least 56 generic names of Britton and Rose (out of a total of 79) are widely regarded as sound taxa, although, as we shall discuss, some of these have been greatly redefined or are now being consid- ered as subgenera. Consequently, the generic concept in cactus systematics often corresponds fairly closely to that presented by Britton and ose. Berger (1926) published an interesting book on the evolution of cactus genera, in which he used the Britton and Rose classification. Berger (1929) soon after published his own familial monograph, in which he accepted only 41 genera. He used 54 names authored by Britton and Rose jui reduced these m subgeneric genu Overall, the cl bled that of Schu- mann, having three subfamilies, but in Cactoi- deae (as Cereoideae) he recognized only two tribes: Rhipsalideae, in which he included the rhipsaloid epiphytes plus Epiphyllum, and Cere- eae, with four sub subtribal classification of Berger differed mark- edly from that of Britton and Rose. Berger’s 1929 classification system had many unusual and inconsistent features. For example, in *subsubtribus Echinocacteinae" the genus Echinocactus was very broadly defined, much like Cereus, and included barrel and small cacti in tribe Rhipsalideae he recognized six genera and many subgenera, most of which are not ac- cepted now. Consequently, in this “conserva- tive" classification, 13-15 of the 41 genera were more narrowly defined that most current cactus pones would Оер С. 1 ^ 1 +1 the great сола of Britton and Rose, who were trained taxonomists. Bravo (1931, 1937) fol- lowed Britton and Rose in studies of the Mexican cacti, which are still being updated (Bravo, 1979); consequently, many collectors studying Mexican cacti have also followed Britton and Rose. None- theless, Schelle (1926) published a classification scheme that followed Berger (1905) 536 Working with a different generic concept were Frič, Backeberg, Ito, and other cactus horticul- turists, who published many new generic names, cactus growers began to collect extensively in the rich cactus areas of western South erica. The plants that arrived in Europe from End Bo- livia, Peru, Chile, Uruguay, and Paraguay greatly changed the data base used by Britton and Rose to evaluate the South American genera. The greatest source of new generic names was Backeberg. Slowly at first and then with a flour- ish, Backeberg (1958-1962, 1966) proposed many new genera and hundreds of species and varieties for South American materials. At the same time, he greatly subdivided and redistributed most of the Britton and Rose genera from the rest of the New World. In all about 81 generic names were published by Backeberg; however, today only 15- 20 of these are still being considered seriously in taxonomic circles as useful taxa for a phyloge- netic classification. ackeberg was just one of several cactus hor- ticulturists who increased the generic confusion of Cactaceae. For example, Frié published 52 new generic names, but of these only Obregonia has survived (Anderson & Skillman, 1984); and Ito proposed 18 new names, of which none have been widely accepted. Backeberg, Fric, Ito, and backlash by numerous cactologists, both profes- sional botanists and serious collectors, who have worked hard over the last 20 years to curb in- discriminate overnaming. Space limitations here do not permit a full review of the studies— pub- lished over the last 20 years— in which the num- ber of species and genera has been carefully re- evaluated. In North America the most remarkable part of this narrative unfolded. Cactus enthusiasts generally adopted Britton and Rose and began Е earnest to define the species and varieties and produce monophyletic genera. In contrast, жна (1940) set out to study the cacti of the United States and Canada using the careful tech- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 niques of Engelmann but the generic concept of Berger. Hence, in all cactus floras published by Benson (1969a, 1969b, 1970, 1982) and in floras and ecological studies that relied on Benson’s monographs, such as the flora of Texas (Correll & Johnston, 1970), columnar cacti were classi- fied in Cereus $.1. Benson (1969a) has briefly de- fended his reasons for using Cereus. From his viewpoint, “there is no ‘right’ or ‘wrong’ system" (p. 8), and a system should be one that conforms with the classification systems of other taxa and that is ‘practical.’ Benson felt that a conserva- tive policy, i.e., placing all columnar forms in one genus Cereus, was in harmony with the over- all conservative policy of plant systematics, and he also labeled the efforts of Britton and Rose as that prevailed in few columnar cacti north of Mexico that it would be impractical to recognize so many monotypic genera for this area. No other cactus taxonomist in recent history has agreed with Benson about the classification of columnar cacti. The reasons for this are many, but only two major points need to be discussed. First, the diversity of vegetative and reproduc- tive features in columnar cacti is as great as the as distinct genera. Second, in order to understand and describe the phylogeny of the low growth forms in Cactoideae, which originated from co- lumnar taxa, it is imperative to determine which part of the cereoid complex is the putative ances- tor or sister taxon. The goal of modern cactus systematics, like that of the rest of plant system- atics, is to make certain that each taxon above the species level is monophyletic in the strictest sense and reflects phylogeny. Because none of the currently used classifica- tion systems for Cactaceae has been able to show that all taxa used are monophyletic, botanists can studied intensively and rewritten to fit the mod- ern guidelines for presenting phylogenetic hy- potheses. PACHYCEREEAE: AN EXAMPLE OF TAXONOMIC COMPLEXITIES The Austrian botanist Franz Buxbaum de- serves full credit for initiating a movement to develop a truly phylogenetic system of classifi- 1986] cation for Cactaceae. He published many mor- phological papers on cacti (for a partial listing, see Gibson & Horak, 1978) during the same pe- riod that Backeberg was publishing his long list of segregate genera. The timing was unfortunate, because, in general, plant systematists were greatly distressed by the deluge of generic cactus names and paid little attention to new cactus publica- tions, including the phylogenetic studies of Bux- baum. Although Buxbaum accepted a number of Backeberg names, he also studied the rela- tionships of the species and provided excellent T for rejecting many of Backeberg's conclus Ali cups Berger (1926) published on the ge- neric relationships within Cactoideae, actually it was Buxbaum (1958) who proposed the first in- novative phylogenetic classification of cactus genera and therein proposed most of the contem- porary tribal names. Of these, Pachycereeae is the tribe that includes the large columnar cacti of Mexico and several other species that occur outside Mexico in the West Indies, Central ica (Gibson Buxbaum defined this tribe chiefly by listing the genera that he included in it. No synapomorphy currently defines this tribe in a strict cladistic sense; consequently, no one can categorically state which genera and species must be included. In the narrowest definition of the tribe, about 70 cluded but presently are not classified in this tribe. One derived feature that occurs in all species so far included is a wood skeleton composed of a ring of parallel, discrete, fastigiate rods (Gibson, 1978). A few columnar cacti in northern South America, e.g., Neoraimondia, have this design but are currently classified in tribe Leptocereeae Buxb. Tribe Pachycereeae can be effectively used as a model to show the systematic complexities of Cactaceae. Coincidentally, some of the species were first described by Engelmann, and even in his time there was a debate on generic names for columnar forms. In fact, numerous species of Pachycereeae have been transferred between genera, and a few have valid binomials in four or more genera. Another advantage of studying this tribe is that the alpha taxonomy of the Mex- ican species has been done (Bravo, 1979). To complement this, much structural and phyto- chemical data have been collected on these GIBSON ET AL.—CACTUS SYSTEMATICS 537 species— more than in any other cactus tribe— which has enabled workers to attempt phyloge- netic reconstructions using contemporary meth- ods. Finally, phylogenetic modeling of this tribe is now in a new, third generation, so that older phylogenetic models can be tested to determine whether the original criteria still yield a parsi- monious solution. The taxonomic history of Pachycereeae began in 1753, when two West Indian species (Cephalo- cereus s.l.) were named by Linnaeus; however, the descriptions of columnar cacti from Mexico did not begin until the 19th century and then continued at a fairly steady pace until 1973, when Stenocereus fricii Sánchez Mejorada was de- scribed. Columnar cacti were initially described as species of Cereus or Cactus, but new combi- nations appeared when three segregate genera were proposed, Cephalocereus Pfeiff., Cephalo- phorus Lem., and Pilocereus Lem., which all cit- ed the same type, Cactus senilis Haw. These seg- regates described what are broadly referred to as cephalocerei, fruticose and arborescent colum- nar cacti that have spineless flowers and fruits, relatively few bracts on the ovary and floral tube, and persistent and long, usually white hairs on the floriferous areoles The first species of Pachycereeae described by Engelmann was saguaro, Cereus giganteus (1848), and this was followed by organ pipe cactus, Ce- reus thurberi (1854), and senita, Cereus schottii (1856). In 1856 Engelmann described the sub- genera Lepidocereus, Eucereus, Pilocereus, and Echinocereus. In subgenus Lepidocereus he in- cluded C. giganteus and C. thurberi of the So- noran Desert along with C. chilensis Pfeiff. from rica. Subgenus Pilocereus included the other species of columnar cacti from Mexico, C. pecten-aboriginum Engelm. and C. gum Engelm., were defined by Engelmann but were not published until after his death. Lemaire recognized Pilocereus, in which he eventually placed all known cephalocerei as well as C. schottii and C. giganteus. Pilocereus was also accepted as a genus by Salm-Dyck, Weber, Rümpler, and Console, and it was used widely in horticultural circles. At the end of the century, Console proposed Myrtillocactus (1897) for Ce- reus geometrizans Mart., a distinctive, wide- ranging, arborescent cactus that produces two or more small, spineless flowers per areole. How- 538 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE 1. Classification of Pachycereeae into subgenera according to Berger (1905). Descriptions were ab- stracted from the key and diagnoses. After each species are listed the names used by Britton and Rose (1920) and Gibson and Horak (1978). Cephalocereus (Pfeiff.) A. Berg. Flowers produced in a distinct cephalium; ribs with isolated mammillae that re surrounded by long hairs and spines; flowers small and arising singly from each mammilla. Cereus chrysomallus Pachycereus chrysomallus (Lem.) ipsc nd (Audot) Bravo Hemsl. Britt. & Rose ex Sánc Mejorada C. columna-trajani Pachycereus columna-trajani Cephalocereus hoppenstedtii (Weber) Karw.) Britt. & Rose um. rw. С. о (We- С ephaloceres macrocephalus (Haw.) о macrocephala (Haw.) ber Britt. & Rose Buxb. C. О (Vell. JA. C epaloceens fluminensis (Miq.) Unassigned Ber Rose C. sis DC. C bem senilis (DC.) Pfeiff. Cephalocereus senilis Lophocereus A. Berg. Flowers several from each areole; flowering areoles differing from vegetative areoles by having numerous long setulose bristles; flowers reddish or yellowish; fruits scaly. Cereus schottii Engelm. Lophocereus schottii (Engelm.) Britt. Lophocereus schottii Rose ?C. аш (Poselg.) A. Cephalocereus EAS (Poselg.) Neobuxbaumia scoparia (Poselg.) Ber, Britt. & Ros Backeb ?C. патат (К. аи батат (Gürke & Tribe Hylocerceae hum.) A. Berg. Weing.) Britt. & Ros dip e (Cons.) A. Berg. Flowers very small, several per areole; ovary naked with a few minute scales; fruits small, smooth, reddish brown berrie Cereus geometrizans Mpyrtillocactus geometrizans (Mart. Myrtillocactus geometrizans (Mart.) Cons. in Pfeiff.) Cons. Pachycereus A. Berg. Flowers solitary, actinomorphic, tubular; ovary and tube covered with dense hair and thin bristles Cereus pringlei S. Wats. Pachycereus pringlei (S. Wats.) Britt. Pachycereus pringlei & Rose C. pecten-aboriginum ра pecten-aboriginum (Еп- Pachycereus pecten-aboriginum i n S. Wats.) Britt. & Rose C. thurberi Engelm. Lemaieoer thurberi (Engelm.) Stenocereus thurberi (Engelm.) Buxb. t. & Ros C. fulviceps (Weber) A. паста РТУ (Lem.) Mitrocereus i (Weber) Back- Berg. itt. & Rose e C. orcuttii K. Brandeg. E orcuttii (K. Brandeg.) x в orcuttii (K. Bran- Britt. & Rose eg.) Moran pacar (Englem.) A. Berg. Flowers solitary, actinomorphic, tubular, and large; greenish white; ovary ttle short wool and sometimes a few deciduous bristles; fruits obovoid or pear-shaped with ut e remote deltoid scales; plants very tall. C. giganteus Engelm. Carnegiea gigantea (Engelm.) Britt. Carnegiea gigantea Rose Stenocereus A. Berg. Flowers solitary, actinomorphic, tubular, small, and reddish or brown; ovary globos and with deltoid scales, naked, or with a few setulose hairs and little wool; fruit globose, brownish, ad with a reddish pulp. Cereus chiotilla Weber Escontria chiotilla (Weber) Rose Escontria chiotilla C. dumortieri Salm-Dyck Lemaireocereus dumortieri Stenocereus dumortieri (Scheidw.) (Schei Britt. ose uxb. ?C. marginatus DC. Pachycereus marginatus (DC.) Britt. Pachycereus marginatus Rose C. sonorensis Rünge Rathbunia alamosensis (Coult.) Stenocereus um (Coult.) Britt. & Rose ibson & Hor 1986] GIBSON ET AL.—CACTUS SYSTEMATICS 539 TABLE |. Continued. C. stellatus Pfeiff. Lemaireocereus stellatus (Pfeiff.) Stenocereus stellatus (Pfeiff.) Riccob. C. alamosensis Coult. Rathbunia alamosensis Stenocereus alamosensis (juvenile form of C stellatus) C. aragonii Weber Lemaireocereus aragonii (Weber) Unassigned itt. & Rose ?C. pruinosus Otto Lemaireocereus pruinosus (Otto) Stenocereus pruinosus (Otto) Buxb. Britt. & Rose ?C. eburneus Salm-Dyck Lemaireocereus griseus (Haw.) Britt. Stenocereus griseus (Haw.) Buxb. & Rose 2C. resupinatus Salm- Lemaireocereus griseus Stenocereus griseus k yc ?Pilocereus tetetzo Weber Cephalocereus? Neobuxbaumia tetetzo (Weber) Back- eb Pilocereus A. Berg. Flowers campanulate; ovary with very few scales and naked; fruit smooth and naked. Cereus chrysacanthus Cephalocereus chrysacanthus (We- Cephalocereus chrysacanthus (Weber) A. Berg. ber) Britt. & Rose C. exerens Link Cephalocereus arrabidae (Lem.) Unassigned Britt. & Rose C. hermentianus Monv. Cephalocereus un Unassigned (Mo Britt. & Ros C. Pi iE (Weber) Cephalocereus da (Weber) Cephalocereus hoppenstedtii . Ber K. Schu m. C. howl (Lem.) A. ipee dig leucocephalus (Po- Cephalocereus leucocephalus erg. ritt. & Rose C. paneer Mill. сећати lanuginosus L. Unassigned C. royeni L. Cephalocereus royenii (L.) Britt. & Cephalocereus royenii Rose C. strictus P. DC. Cephalocereus nobilis (Haw.) Britt. Cephalocereus nobilis & Rose C. ulei (K. Schum.) A. Cephalocereus ulei Gürke Berg Unassigned Eucereus (Engelm.) A. Berg. Flowers large with a long and slender tube; ovary with numerous small deltoid scales; fruit more or less roundish and reddish, covered with spines that are often deciduous in clus- ters Subsection Nyctocereus. Flowers nocturnal; stems more or less upright, cylindrical, and ribbed. Cereus bavosus Weber Lemaireocereus hollianus (Weber) Pachycereus hollianus (Weber) Buxb. Britt. ose C. candelabrum Weber | Lemaireocereus weberi (Coult.) Britt. Pachycereus weberi (Coult.) Backeb. & Rose C. cumengei Weber Machaerocereus gummosus (En- Stenocereus gummosus (Engelm.) gelm.) Britt. & Rose Gibson & Horak C. eruca Brandeg. Machaerocereus eruca (Brandeg.) Stenocereus eruca (Brandeg.) Gibson itt. & Ros & Horak C. gummosus Engelm. Machaerocereus gum Stenocereus gum C. queretaroensis Weber | Lemaireocereus о (We- Stenocereus оош (Weber) ber) Saff. Buxb. ever, according to Berger (1905), the segregate genera Myrtillocactus and Pilocereus were not accepted by professional botanists. a 1 shows how Berger (1905) classified into subgenera the species of Pachycereeae known to him. Although Berger deserves credit for hav- ing the insight to search for subdivisions of the columnar cacti, it is easy to demonstrate now that his subgenera were often artificial. In sub- genus Cephalocereus, Berger included five species 540 having “cephalia,” four of which actually have pseudocephalia and which now belong to differ- ent genera. Lophocereus was created for C. schot- tii from the Sonoran Desert, which has multiple flowers from an areole that also produces long bristles, but included here by Berger was an un- related species from Veracruz, Mexico, and another from the West Indies that is presently classified in tribe Hylocereeae. Myrtillocactus sensu Console was accepted as a subgenus, but Berger did not accept or discuss the taxon C. cochal Orcutt (1889) from Baja California. Pachycereus included species with dense wool and golden bristles on the fruit, but this subgenus also included C. thurberi Engelm., which does not fit the description of the subgenus. Steno- cereus was Berger's most interesting taxon and included many species that are closely related, but the diagnosis of this subgenus was inaccurate. For example, the flowers of Stenocereus prui- nosus are not small, but rather are 9 cm or more in length; moreover, the flowers of Escontria chiotilla have bright yellow petals, not reddish or brown, and the pulp of its fruit is purplish, not reddish. In addition, fruits of Lemaireocereus aragonii have a white pulp and those of Steno- cereus dumortieri have a pulp that is essentially colorless. Pilocereus was used in the same sense as Lemaire's except that the type, C. senilis, was excluded and placed in subgenus Cephalocereus. Finally, Berger combined species having flowers with nocturnal anthesis in subsection Nyctocere- us in subgenus Eucereus. After Berger published his 1905 version of Ce- reus, Britton and Rose studied columnar cacti the Mexican pachycereine speci e largest genus accepted was Cephalocereus, which they interpreted in the comprehensive sense of Le- maire's Pilocereus, i.e., all species with spineless reproductive structures and long hairs on the floriferous areoles. Consequently, this genus in- cluded species from South America. Myrtillo- cactus was adopted, and the monotype Escontria Rose (1906) was proposed for the only species in Mexico with large, translucent, chartaceous bracts covering the flowers and fruit. The generic name Carnegiea was proposed to replace sub- genus Lepidocereus; this substitution eliminated the need to use the polyphyletic taxon of Engel- mann and also permitted Britton and Rose to honor the Carnegie Institute for supporting cac- tus and desert research. In a landmark taxonomic paper, Britton and ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Rose (1909) proposed four more genera. Lopho- cereus was raised to generic rank but was reduced from three species in Berger's system to one, L. schottii. Pachycereus became a genus similar to the Berger subgenus and was based on the pres- ence of bristles on the fruit. The name Lemaireo- cereus was proposed to include Mexican species that have fruits and sometimes flowers with spine clusters. Species of Lemaireocereus were pulled from several subgenera of Berger (Table 1), es- pecially Stenocereus and Eucereus. Two species with short, tubular, slightly zygomorphic red flowers were also segregated as the genus Rath- bunia. Britton and Rose (1919-1923) recognized nine genera because they removed two species from Lemaireocereus to create Machaerocereus, M. gummosus (Engelm.) Britt. & Rose, and M. eruca (Brandeg.) Britt. & Rose from Baja Cali- fornia, which have nocturnal anthesis and heavy central spines. About one month after Britton and Rose pub- lished their first major generic monograph, Ric- cobono in 1909 made an important contribution when he proposed the name Stenocereus stellatus has all of the important features of the stenocerei. Since Britton and Rose, over 20 generic names have been proposed for the pachycereine plants. Most of these names were added by Backeberg, and these were often defined so narrowly that most had only one or a few species. Consequent- ly, there existed a great need to determine the phylogenetic relationships of the species so that generic decisions could be solidified. PHYLOGENETIC STUDIES OF BUXBAUM When Buxbaum (1958) proposed his first phy- logenetic hypothesis for Cactoideae, in tribe Pachycereeae Buxb. he accepted six genera, Pachycereus Britt. & Rose, Lemaireocereus Britt. & Rose, Neobuxbaumia Backeb., Carnegiea Britt. & Rose, Cephalocereus, and Mitrocereus Back- eb., with Escontria Rose and Anisocereus Back- eb. listed as ““genera incertae sedis." He adopted Pachycereus, Lemaireocereus, and Cephalocere- us of Britton and Rose with some important dele- tions: Mitrocereus was a taxon removed from Pachycereus and split into two species; Neobux- baumia consisted of some species that had been scattered throughout Cephalocereus by past au- thors; and Austrocephalocereus Backeb. consist- 1986] “GIBSON ET AL.—CACTUS SYSTEMATICS 541 Pochycereus Pseudomitro- Lophocereus Cornegieo | Neobuxboumio Cephalocereus cereus Mitrocereus Myrtillocactus Polaskia Heliabravoo Stenocereus | ет Myrtillocactinae Росћусегетае Stenocereimae Cepholocereinae Р ы Plerocereus | Prerocereinae | [| Ti 1 Leplocereae FIGUR The first phylogenetic model of Pachycereeae m Buxbaum (1961), in which he recognized five E l. айий and proposed Leptocereeae as the putative ancesto ed of South American cephalocerei and was re- moved to tribe Cereeae Buxb. Three years later Buxbaum (1961) presented a greatly modified interpretation of Pachycere- eae. For its time, this was a remarkably detailed baumia and Carnegiea were unchanged. Lemair- eocereus was reduced in size by removal of cer- tain non-Mexican species for reassignment in Armatocereus Backeb. of tribe Leptocereeae; by recognition of two monospecific genera, Helia- bravoa Backeb. and Polaskia Backeb., from out Mexico; and by exchanging several adoption of the name Stenocereus for the species left in “Lemaireocereus.” Three-ribbed species were removed from P. segregate Pterocereus Backeb.; A lepidanthus (Eichl.) Backeb., which was formerly a species of Pachycereus, was added to Escontria. Mitrocereus, which originally consisted of two species, was treated as two distinct, monospecific genera, Mitrocereus and Pseudomitrocereus Bra- vo & Buxb., which later were renamed Backe- bergia Bravo and Mitrocereus, respectively (San- chez Mejorada, 1973b). In addition, Buxbaum transferred Lophocereus Britt. & Rose and Myr- tillocactus to Pachycereeae from tribe Cereeae. Hence, this tribe began to be redefined as a Mex- ican taxon with some coi in the neighboring countries and the West Indie Buxbaum described and illustrated the flow- ers, fruits, seeds, and seedlings more carefully than most previous workers, and he discovered some features that were exceedingly important in analyzing their phylogenetic relationships. First, he documented seed structure for many species, generally magnified 10-30 times, re- vealing testal features. He also discovered that at anthesis some species of Pachycereeae had idioblastic pigment cells in the funicular epider- mis, which subsequently develop into spherical pigment cells in the fruit pulp. He termed these structures “pearl cells" (Perlzellen) boca they appeared as colored beads on a colorless string (funiculus). Moreover, Buxbaum was keenly in- terested in the presence of triterpenes in some of these species, which had been biochemically in- vestigated by Djerassi (1957; Gibson & Horak, Buxbaum’s first phylogenetic diagram of Pachycereeae (Fig. 1), which arranged the genera into five subtribes, differed in no substantial way from even the final version (Gibson & Horak, 1978), except that he eventually combined sub- tribes Stenocereinae and Cephalocereinae to form a larger d Machaero- cereus Britt. & Rose (Buxbaum, 1968) and Rath- bunia Britt. & Rose (Buxbaum, 1975) to this subtribe as derivatives of Stenocereus (A. Berger) 542 ker ber? ANNALS OF THE MISSOURI BOTANICAL GARDEN Myrtillocactus [VoL. 73 Lophocereus "n pecten-aboriginum pringlei schot ull PUNIRE cocha/ eomefrizans y ; riz standleyi griseus YY, и Кы weberi dumortieri hystrix geometrizans Polaskia chende Mitrocereus "N . fulviceps marginatus ON re eichlamii benecke; A . ongispinus aioe chrysocorpus UN hollianus ¿A schenckir Quevedonis euphorbioides martinezi. Escontria Carnegiea S chiotillo ү” ugantea thurberi fricii-- 2-. ui U — /^ gummosus 222 Stenocereus treleasei stellatus ^ mercalaensis Neobuxbaumia = scoparia PACHYCEREINKE С _— ме “о аа STENOCEREINAE Anisocereus -—- lepidanthus _ — Prerocereus gaumer: E 2. The original phylogenetic hypothesis of eon by Gibson and Horak (1978), which rec- os sie two major subtribes, Stenocereinae and Pachycere Riccob. Interestingly, Buxbaum used the pres- ence of funicular pigment cells and of stem tri- terpenes as the chief criteria for including these two genera in Pachycereeae. TESTING BUXBAUM’S MODEL For many years population biologists have studied the evolutionary genetics and ecology of Drosophila living in rotting tissues of cacti, es- pecially in the columnar cacti of the Sonoran Desert (Barker & Starmer, 1982). Some species of cactophilic Drosophila are host-specific, whereas others are found on several or many species of cacti. In 1974, the Drosophila scientists requested assistance from the senior author for information on the phylogenetic relationships of Drosophila host plants in Pachycereeae; conse- quently, the phylogenetic model of Buxbaum was examined. None of the patterns in Drosophila speciation, host-plant preference, or host-plant chemist could be logically explained by the phylogenetic model; consequently, a study was initiated to test the overall validity of Buxbaum’s phylogenetic model for Pachycereeae. To do this, species lists were prepared from the literature for selected derived characteristics, such as glycosidic triter- penes (23 species), pearl cells (13 species), and alkaloids (6 species), and these were tested against the existing phylogeny for congruence. Three ob- servations were made Buxbaum’s model ailed tests of parsimony for these sets of data. For example, in his model, species with abun- dant triterpenes were portrayed as derivatives of species with abundant alkaloids, and vice versa. (2) Species known to have abundant glycosidic triterpenes seemed to lack alkaloids, and those with abundant alkaloids generally lacked triter- penes. (3) All species known to have pearl cells also had abundant glycosidic triterpenes. Be- cause there was no apparent biological link be- tween the specialized pigment cells in the locule and colorless stem triterpenes, the coincidence of two apomorphic features in a group of species suggested that these species belong to a mono- phyletic taxon and that realignments would be required To aid in the production of new phylogenetic hypotheses, stem transections were studied from the majority of species in Pachycereeae as well as in some columnar species from other tribes. Gibson and Horak (1978) discovered silica bod- ies in the skin (epidermis plus hypodermis) of some Pachycereeae. Silica bodies are uncommon only in species with abundant glycosidic triter- 1986] penes (oleanane class) and pearl cells. Converse- ly, a different set of species in Pachycereeae that lacked pearl cells had prismatic crystals of cal- cium oxalate in the skin. Using these synapo- morphies, combined with general data on plant mprphology, anatomy, and о a new . 2). Species having or thought to have abend oleanane triterpenes and pearl cells were classified as sub- tribe Stenocereinae, whereas any taxon with cal- cium oxalate crystals or abundant stem alkaloids was classified in subtribe Pachycereinae. In Stenocereinae sensu Gibson and Horak, all species possessing silica bodies were reclassified in the emended genus Stenocereus, including Machaerocereus and Rathbunia as well as the Backeberg segregates Hertrichocereus, Isolato- cereus, Marshallocereus, Neolemaireocereus, and Ritterocereus. When Lemaireocereus hollianus as moved out of subtribe Stenocereinae, be- cause it lacked the synapomorphic features, Gib- son and Horak (1978) followed Buxbaum (1961) in recognizing this as a species of Pachycereus. Also included in subtribe Stenocereinae was Mpyrtillocactus, Escontria s.s. (excl. Anisocereus lepidanthus), and Polaskia, which was combine with the monotype Heliabravoa. In Pachycereinae, all of the genera recognized y Buxbaum were retained (Lophocereus, Pachy- cereus, Backebergia, Cephalocereus, Carnegiea, Mitrocereus, and Neobuxbaumia), but the genera were clustered to explain the occurrence of syn- apomorphic structural features. Two species of Pachycereus, P. marginatus (DC.) Britt. & Rose (= Marginatocereus Backeb.) and P. weberi (Coult.) Backeb., had been classified by many authors as either species of Lemaireocereus or Stenocereus; but because they have abundant al- kaloids and lack pearl cells, they were removed from Stenocereinae. Pterocer d Anisocereus were placed at the base of the phylogenetic tree (Fig. 2) without defining E how these are related to the two subtri Gibson and Horak pem speculated on the proposed phylogenetic relationships of the species. Briefly, they suggested that speciation of these cacti fit an allopatric model, i.e., geographic speciation, in which the northern cacti, located in the Sonoran Desert, are the most highly de- rived and probably relatively recent species of seven different clades. Moreover, these cacti have many different floral designs that are adapted to use different types of pollinators, so that ve closely related species often differ markedly in GIBSON ET AL.—CACTUS SYSTEMATICS 543 flower shape, length, color, position, and time of anthesis. The plethora of generic names in this tribe has resulted in part because previous au- thors emphasized the importance of external flo- ral features for identifying taxa. RECENT INVESTIGATIONS OF PACHYCEREEAE Although Gibson and Horak (1978) were able to provide a fairly clear justification for reorga- nizing the species into monophylene ке se evidence for intrageneric groups was, to be sure, ines an s incomplete. Any contemporary phylogeneticist viewing Figure 2, which was drawn in 1976, can see that this is not a precise statement of phy- logenetic relationships of the species, especially in Stenocereus. Beginning in 1980, new vegeta- tive and reproductive materials of Pachycereeae were collected in the field for structural and chemical analyses, with the goal that a cladistic model of the genera and species eventually could be produced. As new data were collected, some of the conclusions of Gibson and Horak were strongly reinforced, whereas others were clearly wrong or misguided and had to be discarded or modified. Seed ultrastructure. The seeds of Pachycere- eae were collected and examined with a scanning electron microscope to document testal sculp- turing. Because data on seeds were very uneven and incomplete in Gibson and Horak (1978), a thorough study of seed ultrastructure was the first broad test of their phylogenetic model. Lemair- eocereus sensu Britton and Rose (1920) exhibited a wide diversity of seed types, ranging from 4 mm to less than 0.5 mm long, brown to black, ough to smooth, and dull to glossy. As recon- obe] subtribe Stenocereinae Buxb. emend. Gibson & Horak basically included those species with relatively small, dull, rough seeds, called verrucose in recent accounts. However, the sub- tribe seemed to include several noteworthy ex- ceptions. Seeds of 22 species and one variety of the 30 species in subtribe Stenocereinae were analyzed, including any species whose seed morphology was not noticeably verrucose. Figures 3-11 show that all four genera have seeds characterized by convex cells with convex outer walls that have prominent cuticular striae, and the striae tend to traverse the cell margins (Figs. 3—9). This testal design was previously found in Pachycereeae by Leuenberger (1974) and Barthlott and Voit (1979) ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 3-11. Scanning electron photomicrographs of testa of Stenocereinae, ктщ characteristic cuticular striae.—3. Escontria chiotilla (Weber) а Gibson 3731 (RSA). Ваг = 0.1 mm.—4. Polaskia chende (Gossel.) Gibson & Horak, Gibson ш (ARIZ). Bar = 0.2 тт. — 5. Myrtillocactus ее нА (Mart.) Cons., Gibson 3754 (RSA). Bar = 0.2 mm.—6. P Pe. irs (Gossel.) Backeb., Gibson Heh ius A). Bar = 0.2 mm.—7. mm.— Stenocereus pruinosus (Otto) Buxb., Gibson 3729 (RSA). Bar = 0.2 mm . Sten reus pr иіпоѕиѕ; same seed as but different region than Figure 7, showing that cell size and prominence ice of siae: varies considerably. Bar — 0.2 mm.—9. Stenocereus quevedonis (G. Ortega) Vene Sea 3713 oe = 0.2 mm.—10. Stenocereus chrysocarpus Sanchez Mejorada, Gibson 3716 (RSA). B .2 тт.— II ЖМ thurberi (Engelm.) Buxb., Gibson 3751 (liquid-preserved voucher). Bar = 0.2 mm. and is fairly widespread in Cactoideae (Barthlott still quite evident; and in some species of Steno- & Voit, 1979; Barthlott, 1981; Behnke & Barth- cereus, the striae in the center of each cell may lott, 1983). On the very small seeds of Myrtil- be reduced so that the surface appears fairly locactus and Polaskia chende (Gossel. Gibson smooth. In Stenocereus thurberi, which superfi- & Horak (Figs. 4—5), striae are fairly low but are cially appears to have smooth, black seeds, the 1986] testal cells are somewhat flattened but still retain striae (Fig. 11). In S. alamosensis (Coult.) Gibson & Horak (Fig. 12) the seed coat is very smooth except between the cells, where faint striae can be observed to cross the cell margins. Once again, the closely related species S. kerberi (Schum.) Gibson & Horak and S. standleyi (G. Ortega) Buxb. have typical seed features of Stenocereus. Finally, in 5. beneckei (Ehrenb.) A. Berger & Buxb. (Fig. 13), which has the largest seeds in the subtribe (3 mm) and the thickest testa, the testa lacks well developed rugae and striae but is still quite rough. Stenocereus be- neckei has the vast majority of features found in the genus, including silica bodies, abundant tri- terpenes, and red areolar trichomes but no pearl cells. Each of these deviations from the standard testal design of Stenocereinae can be explained as a secondary modification. In species assigned to Pachycereinae, none of the seeds are dull or rough. On these seeds (Figs. 14—20), the cuticle is smooth, and the cells bulge outward only slightly or are flat. In some species of Pachycereus, the outlines of the cells are dif- ficult to distinguish (Figs. 15-20); but in P. mar- ginatus (Fig. 15) and P. weberi (Fig. 16), deep pits form at the cell corners. Only in certain species of Cephalocereus have striae been ob- served traversing the cell margins, in a manner similar to S. alamosensis. There does not seem to be a single seed type or testal design charac- teristic of all Pachycereeae as presently defined, and especially the seed of Pachycereus hollianus (Fig. 20) is atypical of the subtribe. Morphological distinctiveness of seeds, fruits, and stem of P. hollianus were reanalyzed when Unger et al. (1980) surveyed tet line alkaloids in eight species of Pachycereinae but reported none in P. hollianus, whereas the other species of Pachycereus had many impor- tant alkaloids. Consequently, Gibson (1982) rec- ommended that the old name, Lemaireocereus hollianus, be used to recognize a monotypic ge- nus of ex hein and presumably a sister taxon of Pachycereus. Stem o In Gibson and Horak (1978), stem triterpene data obtained earlier by Djerassi played a key role in the reorganization of the species into subtribes, and the distribution of triterpene skeletons was used to evaluate the intrageneric relationships of species in Stenoce- reinae. However, plant chemotaxonomists should compare taxa by using the glycosidic forms of secondary compounds, the actual plant com- GIBSON ET AL.—CACTUS SYSTEMATICS 545 pounds, because the glycosides have maximum information content (Giannasi, 1978; Crawford Mabry, 1978; Stuessy & Crawford, 1983; Spencer & Siegler, 1985). A fairly simplistic survey was made of triter- pene chemistry of 20 available species of Steno- cereinae and four other species of columnar cacti, first to determine whether they have abundant triterpenes and then to estimate the overall sim- ilarity of the glycosidic triterpenes of the species. Extractions were made in the field by grinding pieces of cortex with skin and spines of young, vigorous shoots in a blender with 95% ethanol. Each extract was filtered and concentrated in a flask evaporator to a thick syrup, and the syrup was partitioned between water, ethyl acetate, and ether to remove waxes. Triterpenes and carot- : 1, chloroform: methanol, 1: 1, hexane: ethyl acetate, and 100:100:1, n-hep- tane : benzene: methanol. Compounds were de- tected with Lieberman-Burchard reagent and vanillin reagent and examined with UV light. From these extractions, a total of over 50 distinct UE tested positive as triterpenes. These hytochemical studies revealed abundant а: in four species of Stenocereinae that had not been investigated by Djerassi, S. kerberi, S. fricii Sanchez Mejorada, S. chrysocarpus San- chez Mejorada, and S. standleyi, and also from the plant called Lemaireocereus thurberi (En- gelm.) Britt. & Rose var. littoralis (K. Brandeg.) Linds. We also detected large numbers of triter- = as a good marker fo the 25 species (out of 30) that have been tested, all possess abundant triterpenes. The presence of many, apparently different triterpenes in Ber- gerocactus needs intensive study. Because samples provided were small, isola- tion and chemical identification of the glycosidic triterpenes were not attempted. Instead, a smal set of intraspecific and interspecific comparisons was conducted to estimate the relative similarity of taxa within the subtribe. Three or four species were run simultaneously on each thin-layer chro- 546 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 12-20. Scanning electron photomicrographs of testa of Pachycereeae. — 12. s ll (Coult.) Gibs. & Horak, Gibson 3709 (RS Bar = 0.2 mm Pachycereus pringlei (S. Wats.) Britt. Pachycereus marginatus (DC.) Britt. & Rose voucher); deep pits form at cell junctions 3721 (RSA). Bar = 0.1 тт. — 17. slightly convex Y walls and smooth cuticle. Gibson 3741 (R (Weber) Britt. ^; Rose, Gibson 3735 (RSA). B A); testa is fairly smooth, but erg. & Buxb., Gibson lado & Rose, oe chen (liquid- pidan vou er). B ger Zucc.) B т.— 13. Stenocereus beneckei (Ehrenb.) Be Lo phocereus scho (Engel. ) Britt. & lmm.— A); this seed has a fairly sel developed raphe. Bar = .2 mm.— 20. DAL hollianus; same seed as Figure faint striae traverse the margins. m.— 14. 0.1 5. vO, rine 3722 (liquid- pomi . Pa el weberi (Coult.) Backeb., Gibson Rose, Gibson 3195 (ARIZ); seed with 18. C | ema dud hoppenstedtii (Weber) K. Schum., — 19. Lemaireocereus hollianus atus ( 19, showing that this has a different shape Td other Pachycereeae. Bar = matographic plate, and the sequence of bands on agents in a qualitative way. drawn from these tests are, of course, very crude | | but in- for estimating p formative (Fig. 21). y 1. The 14 species of Stenocereus tested shared a common pattern of triterpene glycosides 2. Stenocereus stellatus and S. treleasei (Britt. & Rose) Backeb. were essentially the same, an S. stellatus was also very similar to S. gummo- sus. 1986] 3. Stenocereus standleyi and the two rathbu- nias, S. kerberi and S. alamosensis, seemed to be identical in triterpenes. These three species were examined more closely with trial separation performed on a silica gel column using an as- cending polar solvent system. Each specimen had ands, which all appeared to be the same. This group of three species was most similar to the S. stellatus group. The two populations of S. alamosensis examined were identical. 4. The nine taxa tested that have glandular areolar trichomes, e.g., S. thurberi and S. que- retaroensis (Weber) Britt. & Rose, were more similar to each other than they were to the pre- vious two groups. 5. Within the taxon with glandular areolar tri- chomes, there were two groups, one with colum- nar cacti centered in Nueva Galicia (S. fricii to S. chrysocarpus) and another that reaches to the onoran Desert (S. thurberi and relatives). 6. Triterpene glycosides of 5. thurberi ap- peared to be identical to those of ““Lemaireocere- us thurberi var. littoralis," which has been treat- ed either as a variety of organ pipe cactus or a distinct species but does not yet have a Steno- cereus name. . Two species, S. eruca (Brandeg.) Gibson & Horak and S. dumortieri (Scheidweil.) Buxb., had a number of unusual triterpenes, and decisions on these could not be made for placing them with any of the species groups. 8. The two species of Polaskia had relatively few, simple, and similar triterpene glycosides. 9. Both Escontria chiotilla (Weber) Rose and the three species of Myrtillocactus differed mark- statements on the presumed relationships of these species. The close similarity of the three species in the rathbunia alliance based on 30 triterpene bands lends support to elimination of Rathbunia, which was previously argued on morphological grounds. Next, according to these data, S. gum- mosus is most closely related to S. stellatus. Gib- son and Horak (1978) had presumed that the sister taxon of S. gummosus (Engelm.) Gibson & Horak was 5. fricii, based solely on exami- nation of literature accounts. Another northern cactus, 5. thurberi, does appear to be related to S. quevedonis (G. Ortega) Buxb., as suggested in Gibson and Horak (1978), but 5. beneckei should GIBSON ET AL.—CACTUS SYSTEMATICS 5 alamosensis 5 kerberi 5 standleyi 5 eruca-- ? S. gummosus S. stellatus $. treleasei $ fricii $ montanus $. pruinosus $. queretaroensis $. chrysocorpus $. benecke/ $. quevedonis $. thurberi S. littoralis $. dumortierí-- ? P chende P chichipe > Ar . CHUTE M. schenckii -] M. cocha/ JJ M. 9 IGURE21. Crude dendrogram that shows the qual- itative similarities of glycosidic triterpenes found in 21 taxa of Stenocereinae. S. = Stenocereus, P. = Polaskia, E. — Escontria, M. — Myrtillocactus. be considered as a member of this group. Steno- cereus eruca has always been treated as a deriv- ies are needed to explain why the stem chemistry of these two species is so different. Stem flavonoids. Research on flavonoids in cacti has been overshadowed by that on betalain pigments, which are unique in the Centrosper- mae (Mabry, 1976). Recently, several authors have reviewed data on flavonoids from cactus flowers and stems (Clark & Parfitt, 1980; Burret et al., 1981; Miller & Bohm, 1982), but flavo- noids had not been studied in Pachycereeae. To determine the relative value of flavonoids for phylogenetic models of Pachycereeae, a very simplistic survey was conducted using the same ethanol extractions obtained for triterpene anal- ysis. A survey of flavonoids was conducted using two-dimensional paper chromatography; and К, values and standard indicator sprays viewed un- der UV light were used to analyze the results. Surprisingly, over 200 distinct flavonoids were 548 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE 2. Columnar cacti tested for alkaloids from ethanol or CHCL, extracts. All species except “Lemai- reocereus humilis,” sometimes called Armatocereus humilis, are currently classified in tribe Pachycereeae. Au- thorities for all taxa appear in Gibson and Horak (1978). of Wt. of Alco- Dry holic/ t. Plant Chloro- Fractions/mg ate- fo Taxon Voucher or Source rial/g Extract B A C n Extracts . Backebergia militaris Gibson 3717 (RSA) 19.7 4.7 85.0 246.0 290.0 : Cephalocereus chrysacanthus Gibson 3743 (RSA) 19.7 2.3 1250 32.5 99.5 3. C. collinsii Gibson 3742 (RSA) 22.0 2.89 890 420 95.0 4. C. en Gibson 3741 (RSA) 44.6 3.1 184.0 8.0 158.5 5. C. purpusii Gibson 3710 (RSA) 15.0 3.8 99.0 200.0 199.5 6. C. se nilis Gibson 3750 (RSA) 17.8 2.0 120.0 9.0 97.5 7. C. totolapensis Gibson 3740 (RSA) 26.3 2.4 129.0 7.0 Te 8. Escontria chiotilla Gibson 3751 (RSA) 28.7 12.6 341.0 78.0 300.0 9. Lemaireocereus Gibson 3735 (RSA) 6.7 0.9 30.0 39.0 310.0 10. Mitrocereus fulvice, Gibson 3744 (RSA) 11.5 1.5 180.0 8.0 115.0 11. Neobuxbaumia da Gibson 3745 (RSA) 36.2 6.1 247.0 7.0 92.5 12. N. mezcalaensis Gibson 3725 (RSA) 56.5 3.4 246.0 12.5 140.5 13. N. mezcalaensis Gibson 3756 (RSA) 23.8 3.7 98.5 5.0 94.0 14. N. tetetzo Gibson 3737 (RSA) 29.5 2.4 201.5 7.0 32.5 15. Pachycereus grandis Gibson 3749 (RSA) 21.4 1.2 4030 45.3 167.0 16. P. marginatus marginatus Gibson 3755 (RSA) 6.2 2.1 90.0 150.0 196.0 17. P. pecten-aboriginum Gibson 3748 (RSA) 15.9 1.3 128.5 44.0 93.0 18. P. weberi Gibson 3728 (RSA) 15.1 1.2 72.6 84.5 173.0 19. P. weberi Gibson 3721 (RSA) 15.9 1.2 173.0 29.0 180.5 20. Stenocereus dumortieri Gibson 3727 (RSA) 22.3 11.0 85.0 67.0 320.0 CHCL, Extracts 21. Lemaireocereus humilis Gibson 3183 10.0 0.25 21.0 13.5 14.5 ARIZ 22. Neobuxbaumia polylopha Huntington Botani- 15.0 1.5 9.0 8.0 4.5 cal Garden 23. Pachycereus grandis Gibson 3749 (RSA) 15.0 0.20 19.5 11.0 11.0 24. P. pecten-aboriginum 40 km south of 20.0 0.70 55.5 105.5 87.5 Ciudad Obregon, Sonora, Mexico 25. Polaskia chende Gibson 3180 5.0 1.75 submitted for mass (ARIZ) spec./mass spec. analysis 26. Pterocereus gaumeri Huntington Botani- 6.96 0.20 24.0 24.0 31.0 cal Garden 27. Stenocereus beneckei Huntington Botani- 5.0 0.13 submitted for mass cal Garden spec./mass spec. analysis 28. S. stellatus Huntington Botani- 5.0 0.18 submitted for mass cal Garden spec./mass spec. analysis detected from the 24 species, and on a chro- parent diversity is remarkable because the stud- matograph of Stenocereus thurberi, there were at ies of other Cactaceae have included few reports least 70 distinct spots. Most specimens had more of flavonoids in stem tissues. No attempt was than 40 distinct phenolic compounds. This ap- made to identify these phenolic compounds be- 1986] Extraction and Identification of Alkaloids from Cactus Plants by Thin Layer Chromatography 1) Extraction Procedure for Alcoholic Extracts. Scheme | Wet plant material + alcohol filtered & da with Зы alcohol and finally with abs. alcohol y y Marc. Filtrate air dried iii under reduced pressure at 45°C wt. of dry plant material (a) wt. of alcoholic concentrate (b) Extracted 3-5 X with Total wt. of plant material = (a) + (b) 25 ml portions of IN HCI Residue discard Acidic aqueous soluti | pc 2 X with 100 ml portions of h CHCl4 and ethe y y Combined ether and СНСІз Solution Acidic aqueous solution Dry over anhydrous Adjust pH to 9.5 | Na2S04 and concentrate with NaOH ( 7.5 N) Syrupy residue B Basic aqueous so | Елга ген өүү 2 x 100 ml of each CHCl4 and ethe y Basic aqueous solution Pull on rotary vacuum for 20 minutes to remove traces of organic solvent and freeze dry. r~ Combined СНСІз and ether solution Ory and concentrate бугиру residue A Residue (mostly NaCl) Extracted 3 X with 20 ml of 10% EtOH n CHCl, no 4 Insoluble salt Organic solution (discard) concentrate Syrupy residue C FIGURE 22. Flow chart of the extraction and iden- tification of alkaloids from cactus stems by thin-layer chromatography using ethanolic extracts (Scheme I). cause there were so many and because the sam- ples and sample sizes were very small. Never- theless, comparisons among S. standleyi, S. kerberi, and S. alamosensis revealed 41 visually identical spots, many of which were not present in other species. This observation reinforces the previous statement based on the visual similar- ities of 30 triterpenes, and these and morpho- logical evidence suggest that these three species are not only closely related but also fairly recent segregates. Stem alkaloids. Prior to 1977 alkaloids had ad concentrated on these unique tetrahydroisoquinoline alkaloi The phylogenetic hypothesis of Gibson and Horak (1978) placed all known alkaloid-bearing species of this tribe in Pachycereinae, and this model inferred that other alkaloid-bearing species should occur in that subtribe. The model was immediately tested for this prediction, and in- vestigators found abundant and even new alka- loids in the chemically unknown Backebergia militaris (Audot) Bravo ex Sanchez Mejorada GIBSON ET AL.—CACTUS SYSTEMATICS 2) Extraction Procedure for Powdered Plant Material. Scheme II Weighed plant material moisten with СНСІ з :Ме0н: сопс. NH40H (2:2:1) shake vigorously for 10 min with CHCI3 and filt CHC13 extract Marc. repeat extraction with CHCl, and filter condense on rotary evaporator at 40°С Marc. [rni Ck syrup] Rest of the procedure is same as for alcoholic extract. 3) TLC Systems Used (silica gel plates) a) JH (conc.) 5:3:3:0-5 b) MeOH : NH4OH (азе 4:1*5:05 с) Econ: тан HOH (con 4:3:0+5 d) eu prse ope món. ва 8:1-5:0-5 e) CHCÍz:Acetone:Diethylam 5:4:1 4) Spray Reagents Used a) UV light (plates have fluorescent indicatori b) Fluram (Fluor eS COR ne) 0.02% in aceto c) Tetrazotized benzidine reagent d) Iodoplatinate FIGURE 23. Flow chart of the extraction and iden- tification of alkaloids from cactus stems by thin layer chromatography using powdered materials obtained from fresh plants (Scheme IT) (Mata & McLaughlin, 1980a; Pummangura & seeks xd — 1) as well as in species that had t been ined very closely (Mata & Mc- a EA 1980b, 1980c). Alkaloids known from Pachycereinae before 1981 are listed in Mata and McLaughlin (1982). Extractions were made in the field for 16 species of Pachycereinae and two species of Stenocerei- nae (Table 2) by grinding pieces of cortex with skin and spine of young, vigorous shoots in a blender with 95% ethanol. These were then ana- lyzed using the isolation procedure outlined in Scheme I of Figure 22, and all recovered fractions (A-C) were tested for alkaloids. In addition, chlo- e e II in Figure 23. Methods used for Pet гаи x alkaloids have been published Qr (Ranieri & McLaughlin, ). Table 3 lists "E alkaloids identified by thin- layer chromatography from these samples, and Figures 24 and 25 show the chemical structure of each alkaloid. This survey demonstrated that alkaloids are present in some species of Cephalo- cereus, Neobuxbaumia mezcalaensis (Bravo) Buxb., and Pachycereus grandis Rose, which had not previously been tested. However, the survey made two other important findings. First, it showed that not all Pachycereinae have alka- loids, because alkaloids could not be detected in 550 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ABLE 3. Alkaloids identified from columnar cacti (Table 2) by Bajaj and McLaughlin using thin-layer chromatography. Each alkaloid cochromatographed and gave identical visualization reactions with reference alkaloids in all (at least three) solvent systems tested. Fraction/ Species Alkaloids Identified by TLC 1) Backebergine (traces) 2) Dehydroheliamine 3) 3,4-dimethoxy PEA 4) ио es) methyl-3,4- cine hens PEA 1. Backebergia militaris o 7) N-methyl-3,4-dimethoxy PEA 1) N,N-dimethyl-3,4-dimethoxy PEA 2) N-methyl-3,4-dimethoxy PEA 1) N,N-dimethyl-3,4-dimethoxy PEA 2) N-methyl-3,4-dimethoxy PEA 1) N-methyl-3,4-dimethoxy PEA (traces) 1) Tyramine 2) N-methyl-3,4-dimethoxy PEA 3) Dehydroheliamine 4) three unidentified alkaloids No alkaloid No alkaloid One unidentified polar secondary alkaloid 1) N,N-dimethyl-3,4-dimethoxy PEA (traces) 2) N-methyl-3,4-dimethoxy PEA (traces) кә . Cephalocereus chrysacanthus Ww . Cephalocereus collinsii > Cephalocereus hoppenstedtii . Cephalocereus purpusii un Р>Р>>>»>>>>»>>»>»>»>Р»>> С) . Cephalocereus senilis Cephalocereus totolapensis С) . Escontria chiotilla © о ч © . Lemaireocereus hollianus — oo © N © 21. . Mitrocereus fulviceps . Neobuxbaumia macrocephala . N. mezcalaensis . N. mezcalaensis . N. tetetzo . Pachycereus grandis . Pachycereus marginatus marginatus . Pachycereus pecten-aboriginum . P. weberi . P. weberi . Stenocereus dumortieri Lemaireocereus humilis 3) 3,4-dimethoxy PEA 4) 3-methoxytyramine 5) N-methoxy-3-methoxytyramine No alkaloid No alkaloid One unidentified alkaloid (traces) 1) Heliamine (traces) No alkaloid No alkaloid 1) Pilocereine 2-3 unidentified alkaloids 1) t 2) Salso 3) 3,4- e des (traces) 4) 3-methoxytyram 1) tada ms ) 2) Tehuanine (conc.) 3) Pellotine (traces) esie dad (traces) 5) Weberidine (tra es) 1) О: 2) Teh 3) Donc РЕР РИ 4) Тугатіпе (traces) Traces of two unidentified alkaloids No alkaloid AF Se > a >>>>>>>>>>»>»> С) ЕЕ 66 Орр > С) 1986] TABLE 3. Continued. GIBSON ET AL.—CACTUS SYSTEMATICS Fraction/ Species Alkaloids Identified by TLC 22. Neobuxbaumia polylopha No alkaloid 23. Pachycereus grandis 1) Tehuanine A 2) O-methylpellot A 3) Nailin taa (traces) A 4) Carnegine A 24. P. pecten-aboriginum 1) Tehuanine (traces) A 2) O-methylpellotine B/A 3) Carnegine (conc.) B/A/C Two unidentified alkaloids B/A 25. Polaskia chende One primary alkaloid A/C 26. Pterocereus gaumeri 1) Deglucopterocerei A/C Two unidentified is | alkaloids (high conc.) A/C 27. Stenocereus beneckei One primary alkaloid A/C 28. S. stellatus Three primary alkaloids A/C Mitrocereus fulviceps, four species of Neobux- baumia, and several species of Cephalocereus, including the type, C. senilis (Haw.) Pfeiff. Sec- ond, alkaloids were discovered in Stenocereinae. To date, primary alkaloids have been detected in Polaskia chende, Stenocereus beneckei, S. du- mortieri, S. stellatus, and S. treleasei, whereas Escontria chiotilla from Puebla contained an un- identified secondary alkaloid. Moreover, S. er- uca showed a positive test with commonly used alkaloid indicators. 1 sé 1 1 |. ha 1 11 1 cA 1 1 found in all species of Pachycereus, Lophocereus, Backebergia, and Carnegiea, in Pterocereus gau- meri, two species of Cephalocereus, and Neo- buxbaumia mezcalaensis, whereas the simpler alkaloids of tyramine and phenethylamine occur in most of these same species as well as in Le- maireocereus hollianus and numerous species ia Cephalocereus (Mata & McLaughlin, 1982). T high diversity of tetrahydroisoquinoline AT loids (about 25 different compounds) in this group is in sharp contrast to their poor representation in other cactus taxa; and the occurrence of these secondary metabolites in the genera Pachycereus, Lophocereus, and Backebergia is a fairly good indicator that these genera are closely related, especially because these taxa possess other syn- apomorphic morphological and anatomical fea- tures. In addition, fresh stems of all five species of Pachycereus and Backebergia turn red and then blacken rapidly when they are cut, which in P. weberi was shown to be caused by the hydrolysis of the glucoside lemairin (Mata & McLaughlin, 1980d). Of the species rich in tetrahydroisoquin- olines, Lophocereus and P. marginatus are chem- ically very similar and distinct from the rest of the species, which frequently share the same compounds. In contrast, Carnegiea, whic perficially appears to share many alkaloids m achycereus, can be distinguished from the other species because it has different isomers of tetra- hydroisoquinolines, suggesting that a different biosynthetic pathway may be involved (Unger et al., 1980); also, cut stems of Carnegiea blacken very slowly. Pterocereus has two unique alka- loids; however, this observation cannot be used as evidence for or against inclusion of this genus in Lg usn e abundance of primary alkaloids in Pachy- Neun is an unreliable feature for analyzing relationships within the Pachycereinae because these are not only absent in some of the species but also are widespread compounds in all three subfamilies of Cactaceae, e.g., mescaline, = peyote hallucinogen (Doetsch et al., 1980). P mary alkaloids have now been discovered in six species of Cephalocereus s.l., including C. hop- penstedtii but not in the very closely related type species, C. senilis, and not in C. totolapensis, which has been classified as the genus Neodaw- sonia Backeb. Because Mitrocereus fulviceps and the species of Neobuxbaumia are not rich in stem alkaloids and share the pattern of calcium oxa- late crystals in the skin, all of these Pachycerein nae c a > о e Pachycereus. Placement of the alkaloid-rich Car- negiea and of Lemaireocereus hollianus is still R, Ry ,В2,Вд=Н; R3=0H: tyramine Rj,R2*H; R3,R4*CH30: 3,4-dimethoxy РЕА EL 3, R47 CH 3 Ry=CHy; Rg*H; Аз,Ад=СН30: N-methyl-3,4-dimethoxy PEA R1,R2=CM35 R3»Rg=CH30: N,N-dimethyl-3,4-dimethoxy PEA R,,Ro*H; R3*0H; Rq=0CH3: 3-methoxytyramine R]=H; R2*CH3; R3*0H, Ra=UCH3 — o H,CH(CH,), | м-сн, pilocereine FiGURE 24. Alkaloids isolated from columnar cac- tus stems (Tables 2 and 3 problematic, although on morphological grounds Carnegiea is most similar to Neobuxbaumia eu- phorbioides (Haw.) Buxb., which appears to lack alkaloids entirely. Phylogenetic conclusions on Pachycere- eae. The theme of this paper has been that sys- tematic treatments of cacti have been in a con- stant state of flux since Linnaeus first assigned them Latin binomials. With each additional data set, interspecific relationships have been reana- e un g ical relationships of the еек поа hy- potheses generated by combining new and subtle parameters with traditional and conspicuous fea- tures are producing a new classification of the large columnar cacti of Mexico, which can be used to study patterns of speciation in the Pachy- cereeae and to stabilize nomenclature Some aspects of the phylogenetic studies of Pachycereeae are still unresolved, in part because morphological, anatomical, and chemical data ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Ri =CH3; R2=H; R3*H; R4*0H; Rs=CH30; Rg*M: salsoline R¡,R]=H; R3=H; Rq,R5=CH30; RgsH: heliamine RI ,R2,R3,R4*H; R5 ,RE=CH30: lemaireocereine Ri=H; R2=CH3; R3,R4,R5=CH30; R6=H: tehuanine R1,R2=CH3; R3=H; Rq,R5=CH30; Rg*0H: pellotine Rj,Ro,R3,R4*H; Ro=CH30; Rg-H: weberidine R1,R2=H; R3,R4,Re*CH30; Rg*H: nortehuanine R¡=CH3; Ro,R3=H; Ra ,R5=CH30; Вб=ОН: anhalonidine R1,R2=CH3; R3-H; Ra4,R&,Rg-CH40: O-methyl-pellotine RisH; R2*CH3; R3*H; R4 ,RS*CH30; RgzH: N-methylpellotine Rj,R2*CH3; R3*H; R4,Rg*CH30; Rg*H: сагпедіпе Rj*CH50H; R2=CH3; R3-0H; R4,R&*CH30; Вб=Н: deglucopterocereine backebergine FiGURE 25. Alkaloids isolated from columnar cac- tus stems (Tables 2 and 3). are incomplete on species residing in southern- most Mexico, and in part because we seem to be running out of “easy” synapomorphies to use in discerning cladogenesis in some of the species groups. For example, to use triterpene and fla- vonoid data for phylogenetic reconstruction of Stenocereinae, a massive project would be need- ed to identify each glycosidic compound in sam- ple populations of every species, and this project is not a trivial one because these species have 50-100 distinct compounds per sample. Tribe Pachycereeae has not yet been defined as a true monophyletic clade, because there is no way to determine at this time with our insuffi- cient data base whether taxa currently eed in other tribes should be brought into Pachycere- eae and whether other taxa, such as some of the aragonii Britt. & Rose, which grows in dry trop- ical deciduous forest and disturbed habitats in western Costa Rica. Like Pachycereus margina- tus, this species can be used as a living fence because its young, unbranched stems root very easily; in addition, this Costa Rican taxon does not appear to produce flowers and fruits very often. Britton and Rose (1919-1923) presented a very incomplete diagnosis of L. aragonii, but they were impressed by the wax chevrons pro- duced on young stems marking growth intervals. The seeds of this species were described as shiny, 1986] black, and 3 mm long, features not expected in Stenocereinae but more similar to Pachycereus. However, stem materials collected in Costa Rica by athias were examined and found to have none of the characteristic features of either taxon, and they do not blacken when cut. If L. aragonii is a species of Armatocereus, which it might be, then extensive studies of that genus are required, and then a study of columnar forms locereus Backeb., which included Stenocereus thurberi, so additional nomenclatural problems may have to be resolved. Searching for outgroups to produce a con- vincing model of Pachycereeae is also not a triv- ial matter. The normal procedure for studying phylogenetic systematics is to know the limits of the taxon and probable outgroups before analyz- ing infrataxon relationships. Certainly, this can- not be done in cacti without strongly biasing analyses, and it is safer to start reconstructing the phylogenies of tight species groups and work gradually to higher taxonomic levels. Using Stenocereinae as a model, we can see how this can be accomplished, assuming that all species have been included and making informed deci- sions on where generic lines should be drawn. For example, in Stenocereinae, Escontria, Myr- tillocactus, Polaskia, and Stenocereus are defen- sible genera because they are each sharply de- fined by several to many structural and chemical synapomorphies, and each is as distinct as many well-defined genera in other tribes of Cactoideae. For these same reasons, it makes no sense to continue to use the name Cereus for any native columnar cactus from Mexico, because Cereus s.s. is as distinctive in its own way and has not been ancestral to either Stenocereinae or Pachy- cereinae. The problems of developing a parsimonious phylogenetic model of cacti are not insurmount- able if workers begin to abandon earlier treat- ments and reanalyze synapomorphies. An ex- ample of this i is the epiphytes, which have been another l The most recent | generic classification of cacti by Hunt (1967) placed all epiphytic genera in one tribe, X i ment is unsatisfactory for three important rea- sons. First, the primarily South American rhipsalid genera appear to be most closely related to Corryocactus Britt. & Rose (incl. Erdisia Britt. GIBSON ET AL.—CACTUS SYSTEMATICS 523 & Rose; tribe Notocacteae Buxb.), and cannot be related to the ribbed epiphytes of Mesoamer- ica. Second, even among the epiphytic genera centered around Mesoamerica, there seem to be three clades, and each of these epiphytic sili nei is most closely related to a different terrestri genus, Selenicereus, Acanthocereus (A. Ber en Britt. & Rose, or Nyctocereus (A. Berger) Britt. & Rose. Therefore, epiphytism has apparently evolved independently in the family at least four times. Finally, if epiphytes have evolved re- peatedly from terrestrial forms, then a phyloge- netic classification must place terrestrial and epi- phytic sister groups in the same taxon. This is the future direction that cactus systematics must take so that the evolutionary history of this fam- ily can be properly studied. LITERATURE CITED AGURELL, S. 1969. Cactus alkaloids. I. Lloydia 32: 206- Е ANDERSON, Е. Е. & S. M. SKILLMAN. 1984. A com- parison of Aztekium and St } Syst Bot : 42-49. BACKEBERG, C. 1958-1962. Die LUE 6 vol- umes. Gustav Fischer-Verlag, | 66. Das ton. po Fischer- Verlag, Jena. BARKER, J. S. Ecoldgical Genetics and Evolution Yeas Ac . T. STARMER (editors). 1982. . The Cactus- ademic Press, E Sydnby. BARTHLOTT, W. 1981. Epidermal and seed surface dua) da of plants: systematic go and me evolutionary aspects. Nord. J. B : 34 == & С. Мот. 1979. Mikromorphologie der Sa- menschalen und Taxonomie der Cactaceae: ein raster-elektronenmikroskopischer Uberblick. Pl. Syst. die 132: 205-229. 623. Pinax Theatri Botanici. Joannis & J. CHERLER. 1619. шиа Plantarum Ge- neralis ... Prodromus. Yve BEHNKE, H.-D. & W. BART 3. New evi- dence from the ultrastructural and micromor- n. Nord. The Cac Univ. of Arizona Pre 1969b. The Native Cacti of California. Stan- ford Univ. Press, Stanfor | O. Cactaceae. Pp. 1087-1113 in D. S. orrell & M. S. Jo hnston, Manual of the Vascular Plants of Texas. Texas Research Foundation, Ren- r cti of Arizona, 3rd edition. No) m 1982. The Cacti of the United States and Canada. Stanford Univ. Press, Stanford. 554 BERGER, A. 1905. A systematic revision of the genus Cereus. Ann. Rept. Missouri Bot. Gard. 16: 57- Die en der Kakteen. Gustav Fischer- Verlag, Je 1929. Kakteen. Verlag von Eugen Ulmer, Stuttgart. BOERHAAVE, H. 1720. Index Alter Plantarum quae in Horto е Lugduno-Batavo. Lugduni Contribución al Conocimiento de Las Cactáceas de Tehuacán. Universidad Na- se de México, Instituto de Biologia, Chapul- tepe Ў 1937. Las Cactáceas de México. Universidad Nacional de México, México. Las Cactáceas de México, 2nd edition Universidad Nacional de México, Volume i, éxico. BRITTON, М. Г. & J. N. Rose. 1909. The genus Cereus and its allies in North America. Contr. U.S. Natl. Herb. 12: 413-437. —— ——. 1919-1923. The ied 4 vol- s. Publ. Carnegie Inst., Washington, 248. bn: 1. G. ; Carnegiea PA de saguaro and its uses. Econ. Bot. 25: 32 & J. LINDGREN. 1 76. Cactus alkaloids jeden Alkaloids of Pachycereus pecten-abor and Cereus jamacaru. Lloydia 39: 175- Moe кен Ed Alkaloids of Car- negiea gigan , a new tetrahydroiso- quinoline alkaloid, Lloydia 39: 197-203. . SVENSSON & S. AGURELL. 1970. Biosyn- thesis of tetrahydroisoquinoline alkaloids in C ar- negiea gigantea Br. & R. Acta Chem. Scand. 24: 3375-3377. A, P. ZANDONELLA & B. VOIRI р Biochemical contributions to the sy aL atics of the order Centrospermales. Biochem. Syst. Ecol. 9: 257-262. BUXBAUM, F. 1958. The phylogenetic M Е Ys subfamily Cereoideae, Cactaceae. Madro 177-206. 1961. Die Entwicklungslinien der Tribus Pachycere eae — F. Buxb. (Cactaceae — Cereoideae). open Studien, Heft 12. Gustav Fischer-Ver- lag, Jen 1968. The к a of the E Machaerocereus Bri . Cact. Succ. J. (Los i-a 40: 53 99. З Gattung Rathbunia, C III. In H. Krainz Ек. Die Kakteen. Franckh’sche Verlagshand- lung, Stuttgart. CLARK, W. D. & B. D. PARFITT. Flower fla- vonoids of Opuntia series Opuntiae. Phytochem- istry 19: 1856-1 ap CORRELL, D. S. & M. S. JOHNSTON. 1970. Manual of the Vascular Plants of Texas. Texas Research Foundation, CRAWFORD, 1978. Flavonoid гум ABRY. cre eel of ‘Chenopodium fremontii: infraspecific variation and systematic о at the in- terspecific level. re e E DJERASSI, C. 195 Cac iterpen Festschrift Arthur Stoll. а. sel. — ———-, М. Frick & L. E. GELLER. 1953. Alkaloid ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 studies. I. The isolation of pilocereine from the cactus Lophocereus schottii. J. Amer. Chem. Soc. 75: 3632-3635. KANOV & J. M. Воввитт. 1958. р studies. XX. Isolation and structure of two n cactus alkaloids, piloceredine and lophocerine. Tetrahedron 2: 58-63. , H. W. WER, C. CLARKE & L. J. DURHAM. 1972. Alkaloid ще XXXVIII. Pilocereine — a trimeric cactus alkaloid. J. Amer. Chem. Soc 84: 3210-3212. . FIGDOR, J. M. BOBBITT & Е. X. MARKLEY. 1956. Alkaloid studies. XIV. The pinta of the cactus alkaloid pilocereine. J. Amer. Chem. Soc 78: 3861-3862. и АО А. J. LEMIN, S. К. FI RADA. 1954. Alkaloid vid III. Isolation of сагаа апа anhalonidine from four cactus s s. J. Amer oc. 76: 3215-3217 NALD, McLAUGHLIN. i5 . Cactaceae. Plantas Fendler- : 49-53. . Am г. of the Cactaceae of the ter- el of the United States and adjacent regions. oc. Amer. Acad. 3: 259-346. йш 1599. Catalogus Arborum, Fruticum AC Sys noid biosynthesis and evolution. Bot. Rev. 44: GIBSON, A 1977. Vegetative anatomy of Mai- huenia о with some theoretical discus- sions of ontogenetic changes in xylem cell types. Bull. Torrey Bot. Club 104: 35-48. 78. Architectural designs of wood skele- tons in cacti. Cact. Succ. J. (Great Brit.) 40: 73- . 1982. Phylogenetic relationships of Pachy- cereeae. Pp. 3-16 in J. S. F. Barker & W. T. Star- mer (editors), Ecological Genetics and Evolution. The Cactus-Yeast-Drosophila Model System. Ac- ademic Press, Sydney. & K. E. HoRAK. 1978. Systematic anatomy and phylogeny of Mexican columnar cacti. Ann inr Bot. Gard. 65: 999-1057 . S. NOBEL. 1986. The Cactus Primer. ч Univ. Press, Cambridge a J. 3. De eerste nederlandse weten- schappelijke reis naar Oost-Indie, Communicat. Biohist. Ultraject. 44: 27-4 HERMANN, P. Horti Academici Lugduno— Ba- tavi Е 9 Paradisus Batavus. Lugduni Batavo- m, pees HERNANDEZ, F. ‘1615. Historia de las plantas de Nueva spaña. Reprint, Imprenta Universitaria, México (1942, 1943, 1946). HowaARD, К. A. & M. Touw. 1981. The cacti of the Lesser Antilles and the typification of the genus Opuntia Miller. Cact. Succ. J. (Los Angeles) 53: 233-237. 1986] Номт, D. В. 1967. Cactaceae. Pp. 427-467 in J. Hutchinson, The Genera of Flowering Plants, Vol- ume 2. Clarendon Press, Oxford. [1979 reprint with introductory note by D. R. Hunt.] LEUENBERGER, B. E. 1974. Testa surface characters of Cactaceae. Cact. Succ. J. (Los Angeles) 46: 175- 1737. LINNAEUS, C. Hortus Cliffortianus. Amster- 753. Species Plantarum, Ist edition. Hol- miae, о Salvius. MABRY, Т. J. 1976. Pigment dichotomy and DNA- RNA hybridization data for centrospermous fam- ilies. Plant Syst. Evol. 126: 79—94. МАТА, К. & J. L. MCLAUGHLIN. 1979. Cactus alka- loids. XLIV. Tetrahydroisoquinoline alkaloids of the Mexican columnar a EU hycereus weberi. J. Nat, Prod. m 42: 6 980a. Cactus alkaloid XLII: E 4- the ipea cereoid. Backebergia militaris. J. Pharm. Sci. i 94-95. E ee ca al- О мо of the Na olumnar cactus, Pachy- ereus weberi. Phytochemistry 19: 67 78. Cactus alkaloids XLV. Tet- rahydroi E eod from the Mexican cereoid Pachycereus м Planta Med. 38: 180-182. & mairin, a new glucoside from the Mexi xican pudet Pachycereus weberi. J. Би Prod. (Lloydia) 43: 411-413. 1982. Cactus м 50. А сот- de tabular summary. Rev. Latinoamer. uim. 12: 95-1 j4 MILLER, J. M. & B. A. BoH 1982. Flavonol and dihydroflavonol асн of Echinocereus triglo- chidiatus var. gurneyi. Phytochemistry 21: 95 9 MITICH, L. W. 74. The odyssey of Dr. George En- gelmann. Excelsa (4): 31-39. GIBSON ET AL.—CACTUS SYSTEMATICS 555 Morison, S. E. (editor). 1963. Journals and other Documents on the Life and Voyages of Christo- 1526. The Natural His- tory of the West Indies. [Translated by S. A. Stou- demire, editor.] Univ. North Carolina Press, Chapel Hill. PUMMANGURA, S. & J. L. MCLAUGHLIN. 1981. Cactus alkaloids XLVI. 3-methoxytyramine and lemair- eocereine from Backebergia militaris. J. Nat. Prod. (Lloydia) 44: 498—499. RAINERI, R. L. & J. L. MCLAUGHLIN. 1976. B-phen- ethylamine and tetrahydroisoquinoline alkaloids 2: the Mexican cactus Dolichothele longimam- . Org. Chem. 41: 319-323. Шуй, G. D. 1976. The rise and fall of the “‘suc- culentae." Cact. Succ. J. (Los Angeles) 48: 184— 189 SANCHEZ MEJORADA, H. 1973a. Nuevas cactaceas de la Nueva Galicia. Cact. Succ. Méx. 17: 8. 1973b. The correct name of the grenadier' S p. Cact. Succ. J. (Los Angeles) 45: 171-174. um E. 1926. Kakteen. A. Fischer, Tübingen. SCHUMANN, K. 1898. Gesamtbeschribung der Kak- teen. J. Neumann, Neudamm. (Suppl. 1902.) SHAW,E. A. 1976. The p Cactus Linn. Cact. Succ. J. (Los Angeles) 48: 2 SouLE, О. Н. 1970. doe P uos the first man of cacti and a ч scientist. Ann. Missouri Bot. Gard. 57: 135-14 STUESSY, T. Е. & D. J. di and phylogenetic Pu. Pl. Syst. : 83-108. 1983. Flavonoids Evol. ., R. G. Cooks, R. Mata & J. L. Mc- . 1980. — of columnar Mexican cacti by mass spectrometry/mass spec- trometry. J. Nat. Prod. (Lloydia) 43: 288-293. WiGGINS, I. L. 0. Flora of Baja California. Stan- ford Univ. Press, Stanford. YUCCAS AND YUCCA MOTHS—A HISTORICAL COMMENTARY! HERBERT G. BAKER? ABSTRACT The aida yucca-yucca moth pollination пи is presented in simplified, idealized form in iolog most biolo more а than any single account. has imp rked on n much more dy entomologist С. V. Riley and orn W. o A con- | It w was first noticed i y itself. In fact, ee interaction is probably by George En- rences sence of Tegeticula are e reported that may be due to self pollination c or to the r a es at the base of activities of other flower visitors. A provided by septal nectari e the in many species of Yucca. There dum ent among authors as to whether pla om species are self-compatible and as to the extent of geitonogamy i ous species. Tegeticula yuccasella pears to be the pollinator of all yucca species east of the Rockies and all the western species except Yucca whipplei (sensu lato) and Y. ze үче че уы is served by Tegeticula maculata, whic diurnal in operation and has to c wit pitate stigma. Yucca brevifolia also has its own species of yucca moth (T. synthetica) ort hybridization appears to be rampant in the species [| t eticula sale not for tł pollinated by T. h . AY УУЧУ There is some e , however, that 7. viden individual species of im Another genus is concern Tege. yuccas sella is a complex of taxa se ун apted to rned with the pollination in Ariz and south- eastern Mexico: Parategeticula, which has very different oviposition ш йн oh ace “Tentacle. less” Ti g I ke the “bogus yucca moths,” they are purely p parasitic, not di i polli I g g in an evolutionary context is made. It is interesting that in pollination biology and other disciplines significant biological discover- ies become increasingly simplified in the telling and retelling of the story in text books and the general biological literature, even as the actual process is discovered to be increasingly complex. Thus, the pollination biology of the 35 or 40 species of Yucca is widely quoted as a unique example of strict mutualism—the yucca being entirely dependent on the yucca moth for pol- lination and the moth being totally dependent on some of the developing yucca seeds for the nourishment of its larvae. he relation of the yucca moth to the yucca pollination of yucca plants include those made by Dr. Engelmann in 1 1872. June 13th. white moth of the alliance of Tortrix [crossed out] Tinea often two (a pair!) in one flower, which fly at dusk, but are quickly [?] hid in the flower in day time They seem to EUN the pollen into the stigmatic tube . July 16th. Cons very much constrict- ed, remain small, none full grown. Today [I] observed the first holes in them, where a larva of our [?] moth has gnawed through and escaped (into the ground?). Opening a capsule I find 4 or more larvae in it and almost all the seeds eaten In the first days of July “Mr Riley found peculiar appendage of the mandible (pecu- liar to the female, wanting in the male) gath- ers up the pollen, pushes it into the stigmatic tube and lays its egg (into it-no). [Parenthe- ses and “no” added later in pencil.] In the Bulletin of the Torrey Botanical Club for July 1872, Engelmann (18722) reported the I see many insects about the flowers, bees bumblebees and others, but principally a ! Tam grateful to B. L. Mykrantz, archivist at the Missouri Botanical Garden, for photocopies of correspondence vi etoc d . Heckar n at ey. rene om assisted in every possible way. any Department, University of California, Berkeley, California 94720. ANN. MISSOURI Bor. GARD. 73: 556-564. 1986. 1986] pollination of yucca flowers by a white moth of the genus Tortrix. Subsequently, he protested that he had written “allied to Tortrix” (Engelmann, 1872b). But it was Dr. Charles V. Riley, the Missouri State Entomologist, who worked for over 20 years on the pollination biology and systematics of the moths involved and their relatives, continuing the work he began in Missouri when he became Chief of the Entomology Division of the U.S. Department of Agriculture. His first account was delivered at a meeting of the American Associ- ation for the Advancement of Science at Du- buque, Iowa, which was picked up by the British journal “Nature” in August 1872 (Anon., 1872). Riley wrote many papers on the mutualism (e.g., Riley, 1872, 1881, 1892, 1893). A full listing of Riley’s papers is given by Davis (1967). Contri- butions were also made by Dr. William Trelease, Director of the Missouri Botanical Garden (Tre- lease, 1893, 1902). Coquillet (1893) observed the pollination of Yucca whipplei in California. In the older references the moth genus is given as Pronuba, but that name was shown to be invalid and has been replaced by Tegeticula (Davis, 1967) YUCCA-YUCCA MOTH MUTUALISM What is the marvellous story as described by Riley? The female moth emerges from a pupa in the ground near a yucca plant and mates in the yucca flower with a male (who plays no part in the pollination process). She (Fig. 1) flies nocturnally to a freshly opened Yucca flower and with spe- cially adapted mouthparts (including “tentacles” on the maxillae, Fig. 2), scrapes pollen from the anthers and forms it into a ball that she carries between the “tentacles” and her thorax (Fig. 3). Then she supposedly flies to another plant and, finding a suitably receptive flower, she enters it and, aligning herself appropriately, with her ovi- positor she penetrates the ovary wall and lays a thread-like egg in one of the locules in the su- perior ovary. Then she clambers from the base of the flower (actually often uppermost because the flowers are frequently pendulous) to a posi- tion from which she can place some or all of her pollen load in the tube that is formed by the separated ends of the fused styles (Fig. 4). She rams the pollen into this stigmatic groove (which is lined with papillae and exudes a stigmatic ex- udate). Pollen could not easily get there without BAKER—YUCCAS AND YUCCA MOTHS 557 the aid that she gives it. But the yucca moth does not feed at all. There may be repeat performances of egg-lay- ing by the moth so that several locules of the ovary will receive eggs. But, with the pollination achieved, the food supply of the larva (the de- veloping seeds) is assured, while not all of the seeds are destroyed and the survivors are avail- able to perpetuate the yucca species. There are six locules in the ovary and each may receive one or more eggs. The larvae con- sume seeds in their immediate vicinity and the fruit shows a constriction in the region ofa larva. The larva bites a hole in the fruit wall when it reaches the right state of maturity and descends to the ground either on a silken thread (Riley, 1892) or more likely simply by dropping (Davis, 1967). Entering the soil it forms a cocoon an rests until an environmental signal causes it to pupate and shortly afterwards, the adult emerges and the stage is set for the annual re-enactment of the interaction. The timing of the emergence is closely correlated with the flowering of the ucca. Correspondence on file at the Missouri Botan- ical Garden indicates that Riley sent cocoons of T. yuccasella to at least five persons in charge of ornamental yucca plantings in Europe and in Massachusetts, where the moth does not occur naturally, to see if the moths would emerge and pollinate these yuccas. Unfortunately, we do not know if they were successful experiments. Pos- sibly the timing of о would not be сог- rect in the new environ Galil (1973) has i ‘hint the apparently < deserves recognition nation as opposed to the conventional “topo- centric” pollination (depending upon the relative positions of anthers and stigma, with pollination clearly mechanical). Only Ficus provides other known cases of ethodynamic pollination. COMPLICATIONS In the text books this story is held to be a valid have been published reveals that the true situa- tion may be more complicated. Some field re- ports suggest that other potential pollinators visit Yucca and further, seed set may occur without pollinator intervention. Wiggins (1980: 834), in his flora of Baja Cal- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 FIGURES 1-4.— 1. Female yucca moth dicatae d in flower of Yucca С Photo by С. S. Webber іп не ер Herbarium, University of Cal synthetica) sh one max xillary palp moth (7. maculata) with ball of pollen. Photo by J. Pow rnia, Berkeley. — 2. Side view of head of yucca moth (T. us (mp) icr wr *tentacle" Don as well as the base p e antenna (at), eye (e), pe саа (ft), labial palpus | bristle (b), one proboscis (t). From Riley (1892).—3. Fe ll.—4. Half flower of Yucca brevifolia showing carpels separated i apex, producing the spe groove. Photo by C. S. Webber in the Jepson Herbarium, U.C. ifornia, noted that “АП of the yucca species are attractive plants. Their flowers are delicately fra- grant and attract bees, wasps, moths and bee- tles." But we do not know which of these are effective pollinators. It is frequently stated that the yucca has no possible pollinators besides the yucca moth, although reports of seed-setting by yuccas not serviced by yucca moths appeared as early as 1892 in a note to the Torrey Botanical Club by “J.W.B.” of Flushing, Long Island (probably J. W. Barstow), who mentioned that “In my own garden, the Y. filamentosa Gray, blooms and matures its seed annually. I have never been able to discover the intervention of any insect to assist fertilization, nor have I ever failed to secure the prompt germination of seed taken from any well-matured capsule." The same species, Yucca filamentosa, was cultivated in Is- rael by Jacob Galil. He reported (Galil, 1973) that “In Y. filamentosa spontaneous fruit set is never encountered in the garden" but Galil did find this to occur in Yucca aloifolia, which was also cultivated there. This he attributed to the action of honey bees. He is one ofthe few authors who have drawn attention to the fact that most yuccas have some nectar secretion from septal nectaries on the ovary. This may be quite vo- luminous in some species (e.g., Y. guatemalensis and Y. e/ata, Trelease, 1893). There is also a more or less abundant stigmatic secretion that could serve as a reward to flower- plants of both Y. aloifolia and Y. filamentosa are self-compatible, in contrast to the conclusion by East (1940) that these species are self-incompat- ible. Trelease (1893) also referred to the frequent fruiting of Y. aloifolia without Tegeticula polli- nation. 1986] BAKER — YUCCAS AND YUCCA MOTHS 559 TABLE 1. Infra-generic classification of Yucca (35-49 spp.). I SARCOCARPA fleshy fruit; lobed stigma a number of spp. II CLISTOCARPA spongy fruit; lobed stigma Y. brevifolia III HESPEROYUCCA dehiscent capsule; capitate stigma Y. whipplei (s.1.) IV CHAENOCARPA dehiscent capsule; lobed stigma a number of spp. "Spontaneous" seed-setting was also found by Webber (1953) in Yucca whipplei and, at least, self-compatibility (though not necessarily self- 4 in some populations of this species s been commented upon by several other au- т 3 (e.g., McKelvey, 1947; Wimber, 1958; Powell & Mackie, 1966; Aker, 1981, 1982a, 1982b). These authors concluded that self-pol- lination in this species was uncommon. Webber (1953: 67), based on years of collecting and ob- serving yuccas in the southwestern states, wrote: Outside of Y. brevifolia and Y. whipplei, our tł t y treproduci g to any extent by seeds. There can be little question, therefore, that the yucca moth is more dependent on the yucca for its exis- tence than the yucca is on the moth. During their long life, through vegetative reproduc- tion, the majority of yuccas would continue to exist for many years without the moth. On the other hand, it appears that the yucca moth would be completely wiped out if the yuccas failed to flower for a single year. Regardless of the fact that yuccas are about equally self- and cross-fertile and that the moth flies from flower to flower, it is doubt- ful if cross-pollinations are as prevalent as reported. It is very likely that the number of self-pollinated flowers far exceeds the number of cross- -pollinated ones, and bod in areas d flow- er at the same time interspecific crossing is but remotely possible. This view is possibly too extreme but the sit- uation does require further study. Apparently, Webber was not correct in assuming that all the moths in the soil would emerge the next year. Aker (1981, 1982a) pointed out that the emer- gence may be spread over three years. A PLURALITY OF YUCCA MOTHS Botanists too т, refer е insects by all-in- clusive names—like “ће bee" and “the ant." So it is with the yucca ined Act ik there are at least four species of pollinating yucca moth (and even two genera). Thus, Tegeticula yuccasella seems to be the pollinator of all the Yucca species east of the Rocky Mountains and of all the west- ern species except Yucca whipplei (in the wide sense) and Yucca brevifolia, which have their own species of Tegeticula (McKelvey, 1938, 1947). This fits with the generally accepted systematics of the genus Yucca (Trelease, 1902) where four sections are recognized if the genus is not split (Table 1 Yucca whipplei is so different in several mor- phological, physiological, and ecological features that it has been suggested that a separate genus Hesperoyucca be set up to accommodate it (J. G. Baker, 1892). Consequently, it is not surprising that it has a separate species of yucca moth, Teg- eticula maculata (Table 2), which can operate in the daytime and which has to smear pollen (which is stickier than in other yuccas) on a capitate stigma instead of ramming it into a stigmatic groove (Fig. 5). Several twentieth century studies of the pollination of Yucca whipplei have been Aker & Udovic, 1981) and its pollination biology is probably better known than that of any other species of Yucca Yucca hippie including Y. peninsularis and Y. newberryi, has at least six morphologically distinct subspecies (Haines, 1941; Epling & Haines, 1957). Its Tegeticula (T. maculata) has a melanic subspecies (subsp. extranea) as well as the type subspecies (subsp. maculata). Yucca brevifolia, the Joshua tree, is the only species with a “papery” or “spongy” fruit. It has a dei T Tegeticula synthetica (also known T. paradoxa) entirely to itself (Table 2) MORE. 1947; Davis, 1967). TIMING AND MOTH EMERGENCE Powell and Mackie (1966) showed that Yucca whipplei flowers may be available for two to three months, so precise timing of the emergence of the moth is not essential for this species. How- ever, in other species this does seem to be precise 560 TABLE 2. Tegeticula species associated with sec- tions of the genus Yucca SARCOCARPA | Tegeticula yuccasella CHAENOCARPA CLISTOCARPA Tegeticula synthetica HESPEROYUCCA Tegeticula maculata and Rau (1945), who made sustained observa- tions at Kirkwood (the place in Missouri where the mutualism was first investigated) found evi- dence that temperature was an important con- trolling factor in the coincidence of flowering of Y fil d th £ fthe moth. For Yucca whipplei, Powell and Mackie (1966) suggested that the rainfall pattern is crucial. HYBRIDIZATION AND SYSTEMATICS It is notable that interspecific hybridization is rampant among the species that are pollinated by Tegeticula yuccasella. Webber (1953) listed at least 15 combinations, and McKelvey (1938, 1947) found others. Webber (1953) and Galil (1973) have both obtained seed from artificial crosses. The gene-flow in nature may be respon- sible for the blurred specific boundaries between some of the species and taxonomic disagree- ments about species limits. But the two yucca species that have their own yucca moth species have remained free from hybridization, even where they have come into proximity to other species. Nevertheless, there is evidence that Tegeticula yuccasella is not an indivisible unit systemati- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 cally and ecologically. Davis (1967) seems to have been the first to suggest that all Tegeticula yuc- casella are not alike. Particularly the variation in genitalia suggested to him that there might be distinct yucca moths associated with each of the four sections of the genus Yucca. We know that Yucca brevifolia has its own Tegeticula, as does Y. whipplei; what he suggested is that there are also distinct moths for the sections Sarcocarpa and Chaenocarpa. In 1983, Miles carried the process further and suggested that on the basis of morphological and phenological characters shown by the moths that pollinate Yucca baccata and Yucca torreyi (sec- tion Sarcocarpa) and Yucca elata (section Chaenocarpa) in New Mexico, there are separate moth entities in each case, although all had been put in Tegeticula yuccasella in previous work. She hinted at the possibility that each yucca species has its own pollinating moth. But, I think that, in view of the widespread hybridization and blurring of specific characters in yucca east of the Rockies, the differentiation of the moths must be only at the race level rather than at the species level, with the opportunity for * "mistakes" to Be cur with tł tł lly dif- ferent plants than the kind that produced them. Thus, Miles (1983) found some evidence of the use of Yucca torreyi by moths of the Yucca bac- cata form when Y. baccata had finished bloom- ing. PARATEGETICULA I think it will come as a shock to most people that there is another genus of moth that polli- FIGURES 5, 6.— 5. Half flower of Yucca whipplei, with capitate, papillate-fringed stigma. Photo by C. S. W in the Jepson Herbarium, U.C.B. Davis (1967). ebber —6. Oviposition by Parategeticula pollenifera on pedicel of Y. schottii. From 1986] nates yucca, although it was described 17 years ago by Davis (1967). It was news to me when I began to assemble шонданай for this ‘paper, and none of the general p lished so far mention it. The new genus is Parategeticula and the only species known is P. pollenifera (Fig. 6). It was first collected in 1959, in southeastern Arizona and subsequently, a long way away, in Veracruz, in southeastern Mexico in 1963. Davis, whose monograph on the subfamily Prodoxidae is the standard reference (Davis, 1967), went to Ari- zona to see it in the flowers of Yucca schottii. Powell (1985) has subsequently made a detailed study of this moth both in Arizona and in Ve- racruz (where he found it in Yucca elephantipes). Parategeticula females have “tentacles” like Tegeticula and gather up pollen in a ball in the same fashion. Neither Davis nor Powell was able to see that it pollinated the stigma by actively pushing the pollen into the stigmatic cavity, but it seems reasonable to expect that it does. Parategeticula then does a surprising thing. In- stead of ovipositing in the ovary of the flower, it sometimes lays eggs in the fleshy petals or else in a carefully gouged line of shallow pits on the flower stalks (Fig. 6). Davis (1967) thought that it would not cause any loss of seeds to the yucca by having its larvae feed in the stem tissue, but Powell (1985) found that after develor t from the egg, the larva crawls up to the developing ovary and bites a hole to let itself in. In the ovary, the larva causes degradation of the tissues of the inner side of the wall and the adjacent ovules. In the cyst that is formed, the larva then eats the distintegrated ovules. This unique larval behavior is very different from that of Tegeticula whose larvae develop from eggs in the ovary and wait there until the seeds are al- most ripe before consuming them. Populations of Yucca schottii in Arizona may have only Tegeticula yuccasella or only Parateg- eticula pollenifera or both at the same time (Pow- ell, 1985). Obviously, Parategeticula must be looked for in more species of yucca, especially in the gap between Arizona and Veracruz. ALLOGAMY OR GEITONOGAMY? It is very important to know the breeding sys- tems of the various Yucca species. It seems, from the experiments of East (1940), Webber (1953), Wimber (1958), Udovic and Aker (1981), Aker (1981, 1982a, 1982b) that at least some plants BAKER—YUCCAS AND YUCCA MOTHS 561 of Yucca whipplei are self-incompatible and Webber found some self-incompatibility in six other species. Consequently, some authors have tended to take for granted that the female yucca moth regularly flies between separate plants be- fore depositing her pollen loa Jepson (1925: 246) wrote “The female Pron- uba works by night, collecting the pollen from the anthers and rolling it into a little ball: she then flies to the flower of another plant, deposits her eggs in the ovary, and then in a manner which corresponds to actions full of purpose and delib- eration climbs to the style and thrusts the pollen ball down the stigmatic tube." There is no evi- dence that Jepson actually saw this at first hand; the vehicle of publication being a California flora. In fact, none of the first-hand accounts justify this categorical statement that another plant is immediately sought by the Tegeticula moth after collecting the pollen ball. Riley (1892) wrote “After collecting all the pol- . . she usually runs about or flies to another plant; for I have often noticed that oviposition, as a rule, is accomplished in some other flower than that from which the pollen was gathered, and that cross-fertilization is thus secured" [em- phasis added]. It is not clear whether the “other flower" in this quotation is more frequently in the same inflorescence than in another plant. Trelease (in Riley, 1892: 125) wrote “Apropos of Meehan's idea that the Pronuba moth close fertilizes the flower, I have seen females when undisturbed go from flower to flower here, and several times in the mountains a female was seen, without having been disturbed, to fly off hori- zontally from a plant on the steep mountain side, with every evidence of the necessity for a long flight before finding another Yucca" [emphasis added]. This does not necessarily imply that the horizontal flight led to cross-pollination. Webber (1953) has already been quoted as holding the opinion that more selfings take place than outcrossings. Rau (1945: 374) wrote “The moth, when ready to oviposit, gathers a ball of the sticky pollen from the anthers... finds a flower which is suitable for ovipositing" [emphasis adde here must be — ИР flights by the moths; the interspecific hybrids testify to this, but the evidence is circumstantial. The nearest direct evidence is provided by Aker and Udovic (1981) in the case of Tegeticula maculata on Yucca whipplei, which is easier to observe because the 562 activity is in daylight rather than nocturnal. They wrote “In no case was a female seen ovipositing in the same inflorescence after collecting pollen. Having collected a full load of pollen the females typically crawl out on the branches or unopened flower buds, rest briefly and then fly off. In those cases where it was possible to observe them in flight, they either flew away in a straight line if there was no obstacle, or else spiralled out from the inflorescence until they were heading down wind and then flew straight. The flights were gen- erally high, well above the surrounding vegeta- tion, and the moths often ignored other inflo- rescences nearby” (Aker & Udovic, 1981: 96). Furthermore, Aker and Udovic (1981: 97) stated “— the fact that dispersing females fly rel- atively long distances (i.e. tens of meters) sug- gests that they are minimizing the likelihood that they will return to the same plant from which they have collected pollen or visit other closely related individuals in the vicinity of the pollen donor.” Powell and Mackie (1966) do not say that they saw Tegeticula maculata go from plant to plant of Yucca whipplei. However, Powell (1985) has more recently used mark/recapture methods to show that male Parategeticula (the other genus) stay on one plant of Yucca schottii while the fe- males go to other plants, at least sometimes. If a plant is self-incompatible, geitonogamy will be worse than useless. Clogging of the stigma by incompatible pollen could be one cause of the tremendous voluntary shedding of fruits in Yuc- ca, 50-90% according to Aker (1981, 1982a, 1982b) and Aker and Udovic (1981). They have shown that limited resources are an important cause of this and, in addition, point out that the large numbers of flowers increase the visibility of plants to the moth, and in an exceptionally favorable season there may be a higher propor- tion of fruits that can mature. All of these factors as well as the geitonogamy possibility may be operative. PARASITISM DERIVED FROM MUTUALISM Sometimes, Tegeticula oviposits without pol- linating. Sometimes it lays eggs in an ovary that is already developing following pollination by a previous visitor (Aker, 1981). In such cases, that particular Tegeticula individual is unequivocally parasitic. This leads us to another complication. It is assumed that all the female Tegeticula moths are ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 potentially capable of pollinating the Yucca flow- er but observation of collections and sampling of populations shows that this is not always so (Davis, 1967). An essential item in the pollina- tion system is the presence on the maxillary pal- pus of the female moth of the “tentacles,” which make possible the processing of the pollen into a ball that is carried between the underside of the mouth parts and the thorax ofthe moth (Figs. 2,3 But August Busck, who examined the many collections by Susan McKelvey (see McKelvey, 1947), pointed out that he found “20-30” out of at least a thousand specimens of female Tegetic- ula to have vestigial “tentacles” or none at all. Davis (1967) investigated this in population samples of Tegeticula yuccasella and found that the percentage of tentacle-less female moths could vary from zero to 7196. These tentacle-less moths apparently do not attempt to collect pollen, and if they oviposit they constitute parasites on the mutualism. It is clear that if the proportion of non-pollinators grows beyond a certain point the yucca will be less frequently adequately polli- nated and it will be interesting if further quan- titative analyses are made of this phenomenon and its effect on population structure in Yucca. Keeley et al. (1984) discussed seed predation of nine species of Yucca, pointing out that in all gh there were some fruits in with only a proportion consumed, there were others in which no larvae were to be found. This could be due to self pollination or pollination by some other agent than Tegeticula, or to the fail- ure of the moth to oviposit though it pollinated the flower. Most likely, the moth oviposited but the larvae died. They hint that some Yucca pop- ulations may be capable of inhibiting the hatch- ing of Tegeticula eggs and thus “‘regulating Teg- eticula densities" in their populations. However, at present, this is speculation. “BoGus” YUCCA MOTHS In addition to Tegeticula and Parategeticula, there are what Riley (1892) called “bogus yucca moths.” Three genera are involved. They also are members of the Prodoxidae and oviposit in the wall of the ovary or in some part of the stem system (Powell & Mackie, 1966; Davis, 1967). They have no tentacles, and even if they visit the flowers, they do not collect pollen. Consequently they are dependent for their larval survival on 1986] the pollinating acitvities of Tegeticula, which causes the flowers not to be abscised and seed to be set that will result in new Yucca plants. EVOLUTIONARY HISTORY What may be the evolutionary a in building the yucca—yucca moth mutualism? the Agavaceae, apart from um species, pollination is topocentric (see page 557), and it is likely that a distant ancestor had this conven- tional pollination mechanism, which is pointed to by the production of floral nectar. When as- sociation with the Prodoxidae began (probably by the greater certainty of pollination by these poi universality of its association with Yucca as the virtually exclusive pollinator. We can either pos- tulate that the “bogus yucca moths” evolved from Tegeticula with the Tegeticula providing the pol- lination for them, or some other pollinator may have been involved. The “tentacle-less” Tege- ticula females may represent a mutant form that could show the ancestral condition. Galil (1973) considers the mutualism of Yucca and Tegeticula to be an evolutionarily rather recent event (bas- ing this on the continued production of “useless” nectar) but the occurrence of the mutualism in all yuccas seems to point to a longer history. This is particularly the case in that Yucca whipplei has the rest of the genus, yet possesses Tegeticula, which has had time to produce subspecies, matching in dis- tribution the subspecies of this yucca that have disjunct populations probably relict from a more continuous distribution at some time in the past owell & Mackie, 1966). Parategeticula could be either derived from Tegeticula or ancestral to it. It has a distribution ‘ak чаан not fully revealed as yet) that is with- n the bounds of Tegeticula. But it is hard to see is oviposition behavior and the movements of its larvae making a perilous journey from pedicel to ovary as an advantage over the direct ovi- position into the ovary practiced by Tegeticula (so it might be more primitive than Tegeticula). As stated above, it is notable that none of the other American genera in the Agavaceae have developed a comparable mutualism; Agave has many different pollinators, Nolina and Dasylir- ion tend to show separation of the sexes, while Hesperaloe may be bird-pollinated. significant BAKER — YUCCAS AND YUCCA MOTHS 563 POSSIBLE PARALLELS As for the combination of lepidopteran pol- linator and seed-predator, the only suggestive case outside of Yucca apparently concerns Silene alba (also known as S. pratense and S. latifolia), where both male and female moths of Hadena bicruris visit the white flowers at night in Europe (Brantges, 1976). Both staminate and pistillate flowers of this dioecious species are visited for the nectar they produce, and cross-pollination is achieved. But the female moth oviposits on the ovary in pistillate flowers and the larva eats its way into the ovary and consumes the seeds. It then moves to other flowers on the plant and consumes their seeds before descending to the ground and pupating. Brantges (1976) has calculated that the larvae consume the contents of as many fruits as the moths had pollinated, and it is only because oth- er non-seed predatory moths also pollinate Si- lene alba that it produces enough seed to repro- duce itself (it is an annual plant). It is possible that Silene alba may evolve some means of curb- ing the appetite of the larva, in which case a mutualism of the yucca sort might be possible, but this is highly speculatory. The simple story of mutualism between a flow- er and an insect is probably basically sound, but the situations in nature are being revealed to be more complex than George Engelmann had any reason to anticipate and it is to be hoped that research (observational and experimental) into this fascinating area will be stimulated by the need for more information. LITERATURE CITED KER, C. L . Coevolution of Yucca whipplei and its pollina Tegeticula maculata (Lepidoptera Prodoxidae): relationship between pollinator be- haviora ics and the patterns of flowering and fruitin D. thesis. University of Oregon 1982a. Spatial and te ral dispersion pat- the flowering strategy of Yucca whipplei (Agava- dign Oecologia 54: 243-252. 82b. Кол of flower, fruit and seed та сам by a monocarpic perennial Yucca whipplei. J. Ecol. 72: 357—373. . Upovic. 1981. Oviposition and polli- nation behavior of the Yucca Moth, Tegeticula maculata (Lepidoptera: Prodoxidae), and the re- lation to the reproductive biology of Yucca whip- 01. 72. American Association for the vantement of Science. Nature 6: 442-4 564 . W. И 1892. Yucca. Bull. Torrey Bot. Club 3(8): BAKER, J. G. TUS Agaves and arborescent Liliaceae on the Riviera. Kew Bull. 1892: 1-1 BRANTGES, N. B. M. 1976. Riddles around the pol- lination of Melandrium album (Mill. Garcke (Caryophyllaceae) during the oviposition by Ha- dena bicruris Hufn. (Noctuidae, Lepidoptera), II. Proceedings Koninkl. Nederl. Akademie van We- tenschappen — Amsterdam, Series C, 79, 2: 127- 141. LLET, D. W. 1893. On the pollination of Yucca whipplei in California. Insect Life 5: 311—314. Davis, D. R. 7. A revision of the moths of ie subfamily n (Lepidoptera: Incurv ida e). U.S. Natl. Mus. Bull. No. 255. Smitheontan Institution, Washington D.C. East, E. M. 1940. The distribution of self-sterility in the ORE plants. Proc. Amer. Philos. Soc. 82: Coqui 4 ИО ‚С. 2a. The flower of Yucca "i its fertilization. Rn Torrey Bot. Club 3(7): 3 b. Note. Bull. Torrey Bot. Club FA 37. E C. 1957. A subspecies of Yucca whipplei Torrey. Brittonia 9: 171-172. GALIL, J. 1973. Topocentric and ethodynamic pol- "pem Pp. 85-100 in N. B. M. Brantges & H. ia n (editors), Pollination and Dispersal. Botany D t. Univ. Nijmegen, Holland. HAINES, * 1941. Variation in Yucca whipplei. Ma- drono 6: 33-45. Hoi A., J. GALIL & L. PORTNOY. 1972. Soluble sugars in the stigmatic exudate of Yucca aloifolia L. Phyton 29: 43- 1 =) Be : A California Flora. Univ. Calif. € Berkeley and s Angeles. . С. KEELEY, С. C. Swirr & J. LEE. 1984. Seed predation rid. the Yucca moth sym- biosis. Amer. Midl. Naturalist 112: 187-191. MCKELVEY, . Yuccas of the Southwestern United States, Part 1. Arnold Arboretum, Jamaica Plain, Massachusetts. 1947. Yuccas of the Southwestern United States , Part 2. сав Arboretum, Jamaica Plain, Massachus sett ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Mies, М. J. 1983. Variation and host specificity in the yucca moth, Tegeticula yuccasella (Incurva- riidae), a morphometric approach. J. Lepidop. Soc. Biological interrelationships of moths and Yucca schottii. Univ. California Publ. Entomol. 100: R 1966. Biological interrela- tionships of moths dd Yucca whipplei (Lepidop- tera: Gelechiidae, Blastobasidae, Prodoxidae). Univ. California Publ. Entomo e 42: 1-59. ucca filamentosa, sella (Riley): an ecologico- behavior study. Missouri Bot. Gar 3-39 RiLeY, C. V. 1872. The fertilization ofthe yucca plant by Pronuba Де аца Сапа 1 Further notes оп the pollination = yucca and on Pronuba and Prodoxus. Proc. Am Assoc. Adv. Sci. 1880: 617-639. 92. The yucca moth and yucca pollina- tions. Ann. Rept. Missouri Bot. Gard. 3: 99-159. 1893. Further notes on yucca insects and yucca pollination. Proc. Biol. Soc. Wash. 8: 41- 54. TRELEAsE, W. 1893. Further studies of yuccas = their pollination. Ann. Rept. Missouri Bot. Gar 41: 181-226. 1902. The Yucceae. Ann. Rept. Missouri Bot. Gard. 13: 27-133. Upovic, D. 1981. Determinants of fruit set in Yucca whipplei: reproductive expenditure vs. pollinator availability. Oecologia 48: 389-399. & C. L. 1981. Fruit abortion and the regulation of fruit number in Yucca whipplei. Oecologia 49: 245-248. cen E M. 1953. Yuccas of i. southwest. U.S. bw Agric. Monogr. и I. 80. е uj im | California. Stan- ford a а Stan WIMBER, D. R. 58. Pollination of Yucca whipplei Torr. Master of Arts Thesis. Claremont Graduate School, Claremont, California (unpublished). CENOZOIC HISTORY OF SOME WESTERN AMERICAN PINES! DANIEL I. AXELROD? ABSTRACT Since Pinus occurs in the Early Cretaceous (ca. 125-130 Ma), it probably had emerged from a Pityostrobus complex by the Late Jurassic (ca. 140-135 Ma). Initial adaptation to seasonal climate and drier sites may account for rapid evolution on several occasions. This presumably was enhanced by symbiotic association with ectotrophic mycorrhizae that gave pines an adaptive advantage i in new, spreading, e throughout its history. Pinus probably underwent major split- ting into early subsections in the Late Cretaceous-Early Tertiary as Laramide tectonism created new drier sites appeared in the Late Oligocene (28-27 Ma) as erosion increased in response to lowered base level as sea level decreased and as Drakes Passage opened and cold water was shunted northward. Further speciation no doubt pups as seasonally dry climates spread in оде to developing ice sheets (East Antarctic, 13 Ma; t Antarctic, 7-6 Ma; Arctic, 4-3 Ma). Continued volcanism, tectonism, and markedly fluctuating DIRE at the close of the Cenozoic fostered further speciation, especially in о Where pines show much intergradation owing to rampa ant hybridization i in the recent past. Species o Cembrae, Strobi, Sylvestres) in North by taxa in Eurasia. They reflect the early spread of ancestral taxa into both land areas via connections across the mid-to-north Atlantic and Beringian areas. Neogene records in Europe of taxa allied to east American pines (Australes) may be valid but need reevaluation. The fossil record suggests that six of the eight subsections indigenous to North America (Balfourianae, Cembroides, Leiophyllae, Oocar- Movement along the San Andreas rift system probably transported taxa northward from western EI as indicated by species of Oocarpae, Leiophyllae, and Strobi in Tertiary rocks of coastal California. As more extreme climates spread in the late Cenozoic, the richer Tertiary forests and woodlands lost taxa, and the survivors retreated to moister areas. Pines now increased numerically the impoverished, surviving vegetation zones. In addition, spreading new regional environments, notably drier lowlands (for piñons), drier upland slopes (for P. ponderosa, P. scopulorum, P. jeffreyi, P. flexilis), colder, wetter basins (for P. banksiana, P. contorta), and cold uplands (for P. albicaulis, P. aristata, P. contorta, P. monticola), now became are not recorded in presently-known Tertiary floras in which pines were members of rich, mixed conifer, conifer hardwood, and sclerophyllous woodland vegetation. During the past few centuries, some pines greatly increased in number as man upset ecosystems by fire, logging, and clearing. PART 1. HISTORY from the lowland broadleaved forests adapted to war i Fossil records of Pinaceae are rare in most al, climates of ample rainfall. These conditions Cretaceous to early Paleogene rocks. This is be- were the result of low, widely-flooded continents, cause they lived chiefly in sites well removed the lack of high mountain chains and plateaus, ! Many of the fossil pines on which this paper is based were secured by me during the collection of Tertiary floras over the past 50 years. This was made possible by grants from the Carnegie Institution of Washington (1935-1952), the National Science Foundation (1954— ), and, intermittently, by the Committees on Research, d iie of California, Los Angeles and Davis campuses. Acknowledgment is due W. B. Critchfield for permission to compare the remains of fossil pines with the large, excellent collection of modern pines at the Institute Of Forest Genetics, Placerville, California. Thanks are also extended to F. M. Hueber, Smithsonian Institution and to J. W. Hall, University of Minnesota, for the loan of critical pine fossils described in earlier reports. In addition, Ruth A. Kirkby, Director of the Jurupa Mountains Cultural Center, Riverside, California, C. М. Miller, Jr., and P. H. Raven reviewed the manuscript and have offered a number of valuable suggestions thet have improved it. ? Department of Botany, University of California, Davis, California 95616. ANN. MISSOURI Bor. GARD. 73: 565-641. 1986. 566 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE 1. Subdivisions of genus Pinus (after Little & Critchfield, 1969). Number of le *North Amer- Eur- ica Pinus Subgenus 1. DUCAMPOPINUS Sect. 1. Ducamptopinus — 1 Subsect. 1. Krempfianae Subgenus STROBUS Sect. 2. Strobus Subsect. 2. Cembrae 1 4 P. albicaulis Engelm. 3. ds obi 6 8 P. ayacahuite Ehrenb. P. monticola Dougl. flexilis James oo Engelm. lambertiana Dougl. stro Sect. 3. Parrya Subsect. 4. Cembroides 8 — P. cembroides Zucc. P. monophylla Torr. & Fremont culminicola Andres. & nelsonii Shaw eaman edulis Engelm. inceana Gord. maximartinezii Rzedow. quadrifolia Parl. 5. Gerardian — 2 6 и. 3 — P. aristata Engelm. balfouriana Grev. & Balf. longaeva Bailey Subgenus inb Sect. Tern Su на pu Pida 2 — me Scheide & eppe sim Robins. & Fern. 8. Canarienses — 2 9. Pineae — 1 Sect. 5. Pin Subsect. Үз ене 2 17 resinosa Ait. "Bent Morelet 11. Australes 11 — P. caribaea Morelet P. occidentalis Sw. cubensis Griseb. palustris Mill. echinata Mill pungens Lamb elliottii Engelm. rigida Mill. glabra Walt. serotina Michx. taeda L. 12. Ponderosae 14 — P. arizonica C. E. Blanco P. michoacana Martinez douglasiana Martinez montezumae Lamb. durangensis Martinez ponderosa Laws engelmannii Carr pseudostrobus Lindl hartwegii Lindl teocote Scheide & Deppe jeffreyi Grev. & Balf. washoensis Mason & Stockwell lawsonii Roezl 1986] AXELROD—WESTERN AMERICAN PINES 567 TABLE 1. Continued. Number of ie *North Amer- Eur- ica asia 13. Sabinianae 3 — P. coulteri D. Don sabiniana Dougl torreyana Parry 14. Contortae 4 — P. banksiana Lamb P. contorta Dougl. clausa (Chapm.) Vasey virginiana Mill. 15. Oocarpae 8 — P. attenuata Lemm. P. patula Schiede & Deppe greggii Engelm. pringlei muricata D. Don radiata D. Don oocarpa Schiede remorata Mason = E Totals 62 35 * Names of American species only are listed here; from Little and Critchfield, 1969, plus P. remorata in Oocarpae and the absence of polar ice caps. By contrast, Pinaceae, and Pinus in particular, adapted early to seasonal climates. This is apparent from the records of relatively abundant Pinaceae (Abies, Brit ol nold, 1955; Miller, 1973; Stockey, 1984) the Copper Basin (Axelrod, 1966a) and Bull Run floras (in Axelrod, 1968, MS), northeastern Ne- vada, and the Florissant (MacGinitie, 1953) and Creede (Knowlton, 1923; Axelrod, unpubl. data) of Colorado. The Paleogene floras of arctic and subarctic regions also have remains of Pinaceae, including Pinus (Heer, 1868-1883). In view of these circumstances, we know relatively little of the initial history of Pinus. The earliest records of presumed pine-like taxa (see Miller, 1977a) include needles and cones that are not readily referred to modern groups of pine. Most of the Cretaceous cones whose internal structure has Early Cretaceous (ca. 130-125 Ma) (Alvin, 1960), More numerous fossil pines come into the record in the Eocene, and they increased in abundance down to the present. Pinus now includes 95 species that have been grouped into three subgenera, Ducampopinus, Strobus, and Pinus (Table 1). Of the 15 subsec- tions, only three occur in both North America and Eurasia, notably Cembrae and Strobi of Sub- genus Strobus, and Sylvestres of Subgenus Pinus. The 11 subsections in North America (Subgenus Strobus with four, Subgenus Pinus with seven) include about 62 taxa, or nearly two-thirds of all pine species. These are concentrated in Mexico with 28 species, and in California with 20. Larger regions elsewhere have fewer species, notably eastern Asia (14 spp.), southern Asia (seven spp.), and southern Europe (eight spp.). To infer the modes and times of origin of these groups, and the manner in which present distri- butions may have developed, we turn to the fossil record. It is incomplete, yet it does contain the only historical facts (fossils) available for anal- ysis. Interpretations of these remains may vary, because investigators have concentrated on dif- ferent maes of analysis. Some have studied the have been available, though they are indeed few in number. Most remains of Tertiary pines occur as imprints in fine tuffaceous shale and sandstone preserved in lake beds and floodplain deposits of Cenozoic age. Both lines of evidence are re- viewed here so as to arrive at a provisional, and td died understanding of pine history n North America. EARLY ORIGIN AND EVOLUTION Some fossil ovulate cones outwardly similar to those of Pinus have been designated Pityo- strobus. They differ internally from Pinus cones in several ways (Miller, 1976). All of the pres- ently known 20-odd Cretaceous cones that have 568 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Winged seeds A adaxial Scale apex swollen bract-scale trace united at origin resin canals abaxial to vascular strands Y wee е A 9 949. © j D uo >? == 59:00 O а SOC S3 0 al ERAS can TExT-FIGURE 1.—A. Diagrammatic sketch of some principal features of a Pinus cone scale. — B. Cone scale stripped of bast tissue. — C. n imme in water and dried the cone scale flexes first toward the cone axis (a) and then away from it (Р). — D. Cross section of half ofa cone scale. The thick d 1 plate of scl hy cells (a) and a thinner ventral plate (c). Soft brown tissue (5) encloses vascular strands (d). Resin ducts (e). — E. G f m.) ro Vascular strand enlarged. (Figs. B-E from G. R. Shaw, with permission of Arnold Arboretu 1986] been examined structurally may have none or one or two of the internal features of Pinus (Mil- ler, 1976, table 2). These include an inflated scale apex, bract-scale trace united at origin, resin ca- nals abaxial to vascular tissue in the scale base, and scale strands curved on abaxial side (Text- Fig. 1). The first three features are present only in Pinus cones. The fourth is in all Pinus species and is rare and atypical in cones of other pina- ceous genera (Miller, 1976, 1977a). The reviews by Miller (1976, 1977a) make it apparent that Pityostrobus is central to pine evolution. Inas- muchas Pinus belgica Alvin (1960) is in the Early Cretaceous (ca. 120-125 Ma), and represents Subgenus Pinus (Miller, 1977a) on the basis of its internal structure, it demonstrates the antiq- uity of the subgenus. In his reviews of Mesozoic conifers, Miller (1976, 1977a) noted that leaves taceous cones reportedly allied to Subgenus Stro- bus may represent Pityostrobus, but these have not Ba examined internally. Certainly by the Eocene, species of the Strobus alliance were well established (see below). t described Cretaceous species of Pityo- dus and Pinus now occur at middle-high lat- itudes, as in Belgium, England, Massachusetts, Virginia, and New Jersey. However, the evidence of plate tectonics (Smith et al., 1981) shows that these areas were near Lat. 30-35%N in the Ju- rassic-Early Cretaceous. Climates were season- ally dry as shown by saline deposits and red beds (Hallum, 1984). By the Paleocene, pines were present at higher latitudes (70-80%N), although a cone of Pityostrobus (P. lynnii Berry) is re- corded from the Paleocene of Virginia (Miller, 1977b). Available evidence suggests that the Pit- yostrobus plexus originated over low-middle lat- itudes and that it probably gave rise to Pinus by the Late Jurassic. Pinus was then adapted to sea- sonal climate in middle latitudes, as well as in the montane subtropics, and had spread into sea- sonal climates of middle and higher latitudes by the close of the Cretaceous (ca. 65 Ma). Mirov (1967) inferred from both their mor- phology and physiology that modern pines orig- inated not in uniformly hot, humid climates, but in those with alternating seasons, either wet and dry, or warm and cold, or a combination of both. Most living pines occur where there is ample sunshine, where the soil is porous, generally poor in nutrients, and well drained. They are xero- phytes, for they withstand considerable drought AXELROD-— WESTERN AMERICAN PINES 569 and most show little tolerance of shade. Most pines inhabit exposed slopes away from rich, me- sic valley forests. They now occupy diverse cli- mates, ranging through temperate forested re- gions up to timberline, to semi-arid plateaus and desert-border environments, and to the tropics where there is a prolonged dry season, as in Cuba, Haiti, Nicaragua, Taiwan, Luzon, and Laos. Their early adaptation to seasonal conditions no doubt favored their ability to seize new en- vironmental opportunities later in their history. Diversity of terrain developed during the Lara- mide orogeny in the Late Cretaceous-Early Ter- tiary. This may well have been a time of splitting into forerunners of the principal subsections, the species of which are now extinct. By the Middle Eocene (ca. 45 Ma), there was a major shift to spreading dry climate in southwestern North America (Text-Fig. 2). This created conditions favorable for speciation in Pinus. In addition, active volcanism, extending from Wyoming- Montana southward through the Sierra Madre Occidental, Mexico, provided new, drier sites for the origination of taxa in spreading seasonal cli- mate. By the Late Eocene (35 Ma), dry climate covered much of the western interior and far south into Mexico, and l conditions were developing (Axelrod & Raven, 1985). Con- tinuing volcanism into the Oligocene (ca. 26.5 Ma) favored further diversification as indicated n species. At this time, evolution was enhanced by the spread of drier conditions. As Drakes Passage opened, colder water that developed around Ant- arctica subsided and spread northward to rise in the low-middle latitudes, bringing drier climates there. In addition, as sea level was rapidly low- ered in the Late Oligocene (ca. 27-26 Ma) (Vail et al., 1977; Vail & Hardenbol, 1979), more land area was exposed and continentality increased. With lowered base level, landscapes were reju- venated by erosion, creating new slopes for oc- cupation by pines and other plants adapted to drier, well-drained sites. Environments would be affected in both coastal and interior areas. In the coastal strip, the spread of drier sites may ac- of Ponderosae, Oocarpae, and Cembroides may have arisen, especially since tectonism and vol- canism were active there. Middle Tertiary en- vironments appear to have been similar to those now in the uplands of Mexico where active spe- ciation is occurring today, and has in the recent 570 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ps P N о Е Р Е E NE SH E p 5 S| E a Pines split 15 $ o into early be pe subsections $ S > Most modern 9 subsections Е $ established —10 T e = $ $7 4 - КЫ near-modern Epeiric seas % species n 10— g Speciation in -[ E larger ry climate spreads. x S Е Subsections Much volcanism a By in > MEXICO Widespread volcan sm] -|- E New | 5 — O ч, = H — Drakes Passage opens, circum-Antarctic l z Current develops; Turgai Sea closes; E я 5. 1. lowered; dry climate spreads Ш " E. Antarctic ice cap; Tethys Sea closes; Much volcanism; dry climate expands 2 Two polar ice caps; driest 0+ part of Tertiary; Much volcanism ol Р | Мїосепе | Oligocene | Eocene | Paleocene Cret. T T T T T T 10 20 30 40 50 60 70 TExT-FIGURE 2. Inferred correlation of | changes wi h pine evolution. Temperature curve from Savin, 1977. Early adaptation to drier, poorer sites aided rm yiii au Ак with ectotrophic mycorrhizae enabled pines to diversify as these conditions spread during the Tertiary. past (Mirov, 1967: 341—345). Speciation was fur- ther enhanced in the late Cenozoic by fluctuating uma. the pco of major Mees uplift of d the spread of drier and colder climates to which the taxa were adapted—all conditions favoring major, rapid changes in populations (Stebbins, 1974). The ages of the older Tertiary species that may be assigned to American subsections of Subgen- era Strobus and Pinus are indicated in Text-Fig- ure 3. It is evident that their origins lie in earlier times, as the preceding inferences regarding en- vironmental changes suggest. This may also be inferred from the distributions and ecological ad- aptations of taxa of Pinus subsections. 1. In Subgenus Strobus, the four species of Subsect. Cembrae have widely discontinuous distribution across Holarctica. Pinus albicaulis is in the high mountains of the western United States and adjacent Canada, P. cembra occupies the mountains of southern Europe, and P. pu- mila and P. sibirica range widely across central and eastern Siberia (Text-Fig. 4). 2. In Subgenus Strobus (Text-Fig. 5), Subsect. 1986] AXELROD—WESTERN AMERICAN PINES 571 0 10 20 30 40 50 60 m.y. SUBGENUS STROBUS T | T | T T | T | Subsections: 1 Ve 1 " i Š " erdi Vedder Ss пача ре1 Маг | Bull’ Rùn cana | Bruneau Creede Green Ri Cembroides ° -2 e n ge la | Stewart Spr. | Purple Mt. Creede Copper | Bal fourianae ee << e e- > o | 5 Chalk HillS Temblor i SUBGENUS P Ha Lao | о Subsections = . Altamira Sh. | o Leiophyllae 4 | < к Рг1псе [^2] Sylvestres s is m | o 2s кюк St. Mary's | ~ Australes ^. e .— | a Е | Princeton > T" але peer. pon Creede big: seis S E ce | > Hermosa Green R. = Mt. Eden Sabinianae y Bnavarda | бол Alvord CP. _ pull Run | | Mussel Rock ' : Oocarpae x А Martha s Vin, | Blue Mts. | TExT-FiGURE 3. Ages of fossil pines of American subsections of Genus Pinus. Strobi has 14 species widely scattered in forests across middle to low-middle latitudes in North America from southern Canada to El Salvador and Guatemala, and in Eurasia from Japan to the Himalayas and Vietnam, and to Albania in southern Europe. 3. In Sect. Parrya (Text-Fig. 6), Subsect. Cem- broides (8 spp.) occupies the drier parts of west- ern North America, whereas Subsect. Balfour- ianae (3 spp.) is in moister and/or colder montane regions. Taxa of these two subsections are very different morphologically and ecologically and also differ greatly from species of Subsect. Ger- ardianae (2 spp.) now in the Pamir region (P. gerardiana) and north-central China (P. bun- geana). 4. Sect. Ternatae includes Subsect. Leiophyl- lae (2 spp.) in Mexico and southern Arizona, Subsect. Canarienses (2 spp.) of the Canary Is- lands (P. canariensis) and the Himalayas (P. rox- burghii), and Subsect. Pineae (1 sp.) of southern Europe-Asia Minor (P. pinea) (Text-Fig. 7). 5. In Sect. Pinus, Subsect. Sylvestres with 19 taxa is chiefly Eurasian. The two American species are adapted to dissimilar climates, cold temper- ate for P. resinosa, tropical for P. tropicalis. A similar disjunction in adaptation is apparent for the 17 Eurasian taxa (Text-Fig. 8), the temperate species of which are presumably derived. The distinctness ofthese subsections (see Little & Critchfield, 1969), their disparate distributions (see Critchfield & Little, 1966), and their differ- ent climatic adaptations, all imply considerable age. This inference is supported by the fossil rec- ord, for most of these alliances occur in the Oli- gocene, and some are represented by taxa in the Eocene (Text-Fig. 3). The evidence again sug- gests that evolution of the modern subsections was well under way in the Eocene and probably had commenced earlier for some, relations con- sistent with comparative immunological and mino acid sequence studies of proteins (Prager et al., 1976) TERTIARY HISTORY OF AMERICAN SUBSECTIONS The records of Tertiary pines indicate that species of three subsections (e.g., Cembra Paleogene. By contrast, eight subsections now restricted to North America, and the four to Eur- 572 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 e о / nn PARTIM A ИН ||| НИ ҮШ! 60` Р. сетЬга Р. albicaulis TExT-FIGURE 4. In Sect. Strobus, Subsect. Cembrae has a discontinuous distribution across Holarctica (K = P. koraiensis). The Eurasian taxa have finely serulate needles, the American is entire. Was Cembrae represented in northeastern North America during the later Tertiary? 1986] AXELROD—WESTERN AMERICAN PINES 573 у ^ » 5 | // FA P. Lice S A = & 713 Wie LL ^ M le AA y ME 4 C 249 X \ Seay : И MA Ё A DA + A AY \ 1-Х : m 5 | | Д\ 04 ай \ С. a P E M | |^ pa == A. T > | ELA E: 1 AAT = = ¡HEREDA LARRA OS 6—4 хб Vg ^ * 1 re ы " Y e »1 yi Г 225. М EUER к 277. A x eS Ц 7 _ Era ES 3 ц Y n | y —— 4 N / " + че к / —- l c SN à V^ "Em. е еи а " Vue o и \ m X № AC T i € OY os J = = rhe T ree j 3 в Ts р 0% = a A o { + TexT-FIGURE 5. In Sect. Strobus, Subsect. Strobi has eight species in Eurasia (E— P. peuce, G— P. griffithii, j F— P. fenzeliana, D— P. dalatensis, M— P. morrisonicola, and W — P. wangii), whereas there are six species in North America (S— P. strobus папа, St—P. strobiformis, A— P. ayacahuite). Note that Р. strobus is also in southern Mexico (Chiapas) and emala. Guat 574 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 ( e f) À — E 2, 4 e , ' Y Subsect. и : NÓ — o. Pil TExT-FIGURE 6. In Sect. Parrya, Subsect. Cembroides has eight species in rst North America (Р. eid ide d P. monophylla, P. quadrifolia in the United States; P. culminicola, P. maximartinezii, P. pinceana, P. nelsonii in Mexico). Subsect. Gerardiana has two species (G= P. g Aranda in the Pamir region, B— dipinti in а China), and secre ны has three species (В — ЕР. balfouriana, L—P. longaeva, A — P. aristata) in the western United State AXELROD— WESTERN AMERICAN PINES 575 1986] | | | _ | Subsect. LEIOPHYLLAE ^. 1 | | | “Al Је | \ 1 | Ñ Subsect. LE, о CANÁRIENSES d IM \ М, R LOW б Qu Nem, \! С Subsect. y "СА МАВ!ЕМЗЕЗ TEXT-FIGURE 7. In Sect. Ternatae, Subsect. Leiophyllae with two species (Р. leiophylla, P. lumholtzii) ranges from southern Arizona-adjacent New xico to southern Mexico. Subsect. Pineae (one species, P— P. pi 1s on the north shore o editerranean Sea sect. Canarienses has two speci . canariensis, R— P. roxburghii) discontinuous from the Canary i long history. . Su Islands to the Himalayas. Such disjunctions clearly attest to a 576 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 к КУ | han н АШ е y WNI || | | Ht n || | FIM Ne pu: OTTO A „ЭУ || | | URL UU ҮПҮ DRE UIT emo | y | | | | | | ||| il m ИИ LN Hi | +” Р Р зы UN | || E n a | o ТЇ E. di s MI | чи SE "on " "n ^ ls My, оч UD " v E ) М e. a Ы "v. vem ` TEXT-FiGURE 8. In Sect. Pinus, Subsect. Sylvestres with 19 species is largely Eurasian. Only two species are in North America: P. resinosa in northeastern USA-adjacent Canada and P. tropicalis in Cuba. The alliance has adapted to climates ranging from tropical (Cuba, Vietnam, Sumatra, Phillipines) to Mediterranean, to cold 1986] sibirica TEXT-FIG Similar AXELROD— WESTERN AMERICAN PINES 11 "o METRIC albicaulis r cones are produced by P. albicaulis and P. sibirica. The species differ in that P. E 9. p ard жы are entire, while those of P. sibirica are finely serrulate; and P. sibirica is a tall tree, P. albicaulis ush asia, appear to be autochthonous as judged from present fossil evidence. SUBSECT. CEMBRAE Of the five species in this group, P. cembra, P. koraiensis, P. pumila, and P. sibirica are in Eur- asia, whereas P. albicaulis is chiefly at subalpine levels in northwestern North America. Cones of P. albicaulis are scarcely separable from those of P. sibirica (Text-Fig. 9). The chief differences be- tween these taxa are in the needles (serrulate or not; stomatal position), characters that may have developed following the spread of ancestral taxa into the separate land areas. The age of the ties between these taxa is not known, although Ter- tiary cones from Eurasia have been compared with P. cembra (Gaussen, 1960: 228) and other species (e.g., P. koraiensis) reportedly have allied taxa in the Paleogene (Gaussen, 1960). Since fos- sil records indicate that a cembra-sibirica-like pine was present in the Paleogene of Eurasia, an ancestral species allied to them, and to P. albi- caulis, probably ranged earlier across boreal re- gions to give rise to separate species in each area, with cembra-sibirica-pumila representing chiefly <— temperate іп northern Europe and Asia where it reaches above Lat. 60°. Its occurrence in diverse climates іп erica implies different times of entry, probably Early Paleogene-Late Cretaceous for the ancestor of P tropicalis, as compared with Miocene for P. resinosa in northeastern U.S. and Canada monticola ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 griffithii TEXT-FIGURE 10. sasi of P. griffithii from the Himalayas are quite similar to those of P. monticola of western North Ameri varietal differences across Eurasia. Whereas P. albicaulis is a small bushy tree in most of its area, the Eurasian cembra-sibirica commonly form tall forest trees. Cembrae have a wide oc- ice sheets during the Quaternary, glaciation may account for these present restrictions as well as the absence of Cembrae from eastern North America. SUBSECT. STROBI Modern species of this alliance are now Hol- rctic in occurrence. In North America are P. strobus (northeastern U.S.-Canada), P. strobus > 1986] var. chiapensis (southern Mexico-Guatemala), P. monticola (California-Washington-Idaho-Brit- ish Columbia), P. /ambertiana (Baja California Norte, California, Oregon), P. flexilis (Alberta- New Mexico-eastern California), P. strobiformis (southern Rocky Mountains-northern Mexico), P. ayacahuite (southern Mexico-Guatemala). The Eurasian taxa are widely scattered as shown by P. peuce (Balkans), P. armandii (central China), P. griffithii (Himalayas), P. parviflora (Japan), P. morrisonicola (Taiwan), P. fenzeliana (Hainan), P. wangii (S. Yunnan), and P. dalatensis (Viet- nam). Since fossil cones of the general sort pro- duced by these pines are in the Late Cretaceous (some may be Pityostrobus), the intercontinental links are old. The occurrence of numerous, pres- ently Asian taxa in North America in Paleocene to Miocene floras (e.g., Ailanthus, Cercidiphyl- lum, Ginkgo, Glyptostrobus, Cunninghamia, Metasequoia, Phoebe) suggests that in favorable montane areas, ancestral Strobi populations probably extended across both the Beringian and mid- to North Atlantic regions, thence diverging into allied species as they spread southward on each landmass. The white pines, P. strobus and P. monticola, of eastern and western North America are paired- species, and these are allied also to the Hima- layan P. griffithii (Text-Fig. 10). Both alliances may represent species of an ancestral population that spread south into each landmass and then diverged into related species east and west. This inference is supported by the internal structure of needles of P. similkameenensis (Miller, 1973; Stockey, 1984), preserved in cherts of Middle Eocene age near Princeton, British Columbia. This 5-needled pine is referable to Sect. Strobus, showing that the alliance was already in the Rocky Mountains 47 Ma ago. This implies that an early Paleogene (or older) link between allied inter- be inferred because can P. monticola has a fossil record in Siberia (Mirov, 1967; Wolfe & Leopold, 1967) are unsubstantiated. The cone of P. monticola fossilis (Sukachev, 1910) from the Pliocene is more robust and has larger, thicker cone scales, and is not so slender as P. monticola cones that are closed (from immersion in water) like the Siberian fossils. A similar, more com- le age on the Aldan River (Dorofeev, 1969). Both AXELROD-— WESTERN AMERICAN PINES 579 fossil species are more nearly allied to P. ar- mandii from the uplands of central China. Pinus flexilis from the western interior seems allied to P. armandii of central China, which also inhabits cooler montane areas (Text-Fig. 11). An ancestral population may have spread south in the interior, giving rise to paired taxa that di- verged chiefly in needle structure (entire or ser- rate; position of stomata) and seed wings (absent in P. armandii, attached to cone scale in P. flex- ilis) in each region. This would have been early, for P. florissanti Lesquereux (cf. P. flexilis) is in the Florissant flora, Colorado (MacGinitie, 1953), dated at 34 Ma (Epis & Chapin, 1975). The preceding links between North American and Eurasian white pines follow Shaw's (1914, 1924) reclassification of his Cembra Groups (Cembrae, Flexiles and Strobi) into Cembrae and Strobi. Species ofthese morphologic alliances dif- 1). This is expectable for, as suggested for the differences in needle structure, the chemical differences may also have devel- oped as the early populations spread into each landmass The Cordilleran white pines, P. flexilis-stro- biformis-ayacahuite, that stretch from the C nadian Rockies into Guatemala, seem to form a natural series. The question arises as to whether the Eocene (ca. 48-47 Ma) P. delmarensis Ax- elrod from near San Diego, which was then op- posite Guaymas, Mexico (Gastil & Jensky, 1973), may be a part of this alliance. Its present position results from displacement northward as the San Andreas rift system was activated, sep- arating P. delmarensis from any Mexican con- nection (Text-Fig. 12). Although P. delmarensis seems closely allied to P. /ambertiana of Cali- fornia, genetic studies indicate that /ambertiana crosses with Asian pines (P. armandii, P. ko- raiensis), not with American taxa (Mirov, 1967: 34). Since crossing experiments with the Mex- ican taxa are incomplete (W. B. Critchfield, pers. comm., May 1984), the data are insufficient now to resolve the problem. Structurally, P. delmar- ensis seems closely allied to P. lambertiana and may represent an early, western segregate of a group that gave rise to P. strobiformis-ayacahuite in the interior. The suggested affinity is supported by the occasional occurrence of moderately re- flexed cone scales in P. lambertiana, a feature typical of P. strobiformis and P. ayacahuite cones (Text-Fig. 13). Furthermore, all these taxa have broadly ovate seed wings with rounded tips, not 580 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 armandii TEXT-FIGURE 11. They differ in that P. armandii has finely serrulate needles, whereas those of P. flexilis Judging from their similar cones, P. armandii and P. flexilis appear to be paired species. e. Pinus armandi seeds are wingless, and those of P. flexilis have the seed wings attached to the scales. slender wings with acute tips as in P. monticola and P. strobus. SUBSECT. CEMBROIDES Records of this alliance occur in the mid- Eocene Green River flora (47 Ma) of northern ea Utah (MacGinitie, 1969), and comm.), and in the Chalk Hills Formation (5-6 Ma) near Bruneau, Idaho (Knowlton, 1901; Brown, 1940; Malde & Powers, 1962). That Cembroides do not hybridize with other pines (Mirov, 1967: 569) and have been in existence since the Eocene, demonstrates that it is a very distinct subsection. This is suggested also by the observation that the woods of nut-pines have piciform lateral ray pits and thick-walled ray cells that approach in structure those of Cretaceous “pines” (Bailey, 1910). The fossil piñons were chiefly subordinate members of rich woodland vegetation, whereas they now often dominate woodlands. The Idaho occurrence of Pinus lindgrenii Knowlton near Bruneau may be an exception. A number of cones have been recovered by local collectors, and a dozen or so are in the Jurupa Museum, River- side. Their relative abundance suggests that Pi- ñon may have чөн at least a subordinate role in the flora. This t with their worn, erod- ed nature, а indicates transport, presumably from a pinon zone that covered warmer, south- facing slopes bordering the floodplain. With respect to Pinus cembroides var. cem- broides, the question may be raised as to whether P. cembroides var. lagunae Robert-Passini (1981) may not be the most ancient living member of the species. As described further by Bailey (1983), it differs from P. cembroides var. cembroides in 1986] AXELROD— WESTERN AMERICAN PINES $AN DIEGO | PRESENT 4 ORIGINAL POSITION TExT-FIGURE 12. Disruption of western Mexico and its northward displacement following 25 Ma ago (from Gastil & Jensky, 1973). Members of several Pinus subsections were displaced northward into the developing Mediterranean climate of California and border area. its average longer needles, thinner fascicles, fewer have noted that the trees are more robust and larger than P. cembroides var. cembroides, and the cones are larger than those of most popula- tions in the Southwest and northern Mexico. The var. /agunae inhabits uplands of the Cape Region, Baja California. The limited (10 year) record at La Laguna indicates a mean annual temperature of 14.5?C and an annual range of 7.5°C (Garcia, 1973). This gives the area a very equable climate — M 65 in Bailey's (1960, 1964) classification. The pine is a component of the pinon-madrone evergreen-oak woodland at al- titudes generally from 1,600 to 2,000 m. The notion of its antiquity is consistent with the oc- currence of other marked endemics or disjunct h u (Cedros Island, Canary Islands, Revillagigedo Is- lands, Madeira Islands, Juan Fernandez Islands, and others). The Cape Region was isolated from the mainland following 25 Ma ago (Text-Fig. 12), and the Gulf opened about 5—6 Ma ago, as judged sini, 1981), involve their shorter needles, thicker fascicles, and other features that presumably re- flect more recent adaptation to increasing aridity in areas bordering the Mexican plateau. SUBSECT. BALFOURIANAE This small alliance includes three taxa, P. ar- istata Engelm. of Colorado-northern New Mex- ico and northern Arizona, P. /ongaeva Bailey of Utah and the Great Basin, and P. balfouriana 582 lambertiana TEXT-FIGURE 13. The Californ San Adreas rift system Murray, which is disjunct in California from the Klamath Mountain region to the southern Sierra Nevada, 625 km (390 mi.) distant. Their detailed distribution is charted by Bailey (1970, fig. 3). There are fossil records of two of these taxa flora (26.5 Ma), Colorado, where the branches of P. aristata overhang some of the fossil sites today. Pinus crossii, abundantly represented at Creede by cones, foliage, and winged seeds, lived adjacent to a rich pinon-juniper woodland, a community usually well removed from P. aris- tata today. Cones and foliage of P. crossii are ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 strobiformis Pinus lambertiana is allied to P. delmarensis from Middle Eocene rocks near San Diego. rnian taxa were isolated from the related Mexican P. strobiformis-ayacahuite by movement on the also in the Hillsboro flora, central New Mexico, dated at 32 Ma. Its associates included spruce and several rare, small leaflets of Mahonia. This appears to have been a high montane site with a rich, or nearly pure, stand of fossil bristlecone pine (Axelrod & Bailey, 1976). A species allied seeds and a fascicle in the Copper Basin flora (Axelrod, 19662), which represents a rich coni- fer-hardwood forest. This species also contrib- uted to pure montane conifer forests in the Late Eocene Bull Run flora north of Elko, Nevada. The distal half of a cone in the Lower Oligo- 1986] Subsect 1 | I 1 Д Р. longaeva аге not now agii cene (Chadronian) Titus Canyon Formation (Stock & Bode, 1935), Death Valley, appears to be P. crossii. Its affinity is suggested by the thin partially open cone. Its Early rence in the southern Great Basin, as well as in northeastern Nevada in the Late Eocene, sug- gests that = crossii probably originated in the western interior. Pinus EM Axelrod from the Oligocene Thunder Mountain flora, Idaho (in Brown, 1937), is represented by relatively large cones (+9 cm), as well as needles in 5s. Another Idaho record is in the Coal Creek flora (27 Ma), Lost River Range, where it is also associated with a humid conifer forest, including species of Abies, Picea, Larix, Pinus, Tsuga, Chamecyparis, and Sequoiaden- dron. Pinus balfouroides occurs in western Ne- vada in the Chalk Hills (Axelrod, 1962) and Pur- ple Mountain (Axelrod, 1976b) floras. Dated at AXELROD— WESTERN AMERICAN PINES 583 . BALFOURIANAE Texas e Mt., 6—Chalk Hills, 7—Sharktooth Hill, and 8—Death Valley. Fossils related to 12 and 13 Ma, respectively, the fossils include relatively small cones (6-7 cm long), as well as winged seeds and foliage. A cone is also in the marine Temblor Formation (15 Ma) at Sharks- tooth Hill near Bakersfield, where it was trans- ported from the Sierra Nevada into the marine basin. The few fossil cones available suggest that in whereas the smaller cones in the western Nevada floras are comparable to the southern Sierran population, P. balfouroides var. austrina (Mas- troguiseppe & Mastroguiseppe, 1980). The larg- er-coned, northern population, which has been in a more mesic climate since the Oligocene, shifted coastward during the Neogene to the hu- mid Klamath Mountain area. By contrast, the smaller-coned austrina may have originated in the subhumid western Nevada region and moved ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Subsect. Leiophyllae TEXT-FIGURE 15. & Little, 1966). The coastal southern California occurren m Mexico (see Text-Fig. 12). The record of P. coloradensis MILES | о 100 200 300 400 m т sd: " J n Occurrence of species of Subsect. а in the past and today (тар from Critchfield e of P. paucisquamosa Templeton in the Middle Knowlton at Creede reflects the wider distribution of | Leiophyllae in the Late Paleogene, a time when other fossil taxa with relatives now far to the south also ranged widely to the north. into the southern Sierra Nevada as climate be- came drier. It survived there largely south of the Pleistocene ice cap. Since the close of the Wis- consin, it has spread some 40 km (25 mi.) north to its present northern area near Onion Valley, above Independence, an area under ice during Wisconsin glaciation. SUBSECT. LEIOPHYLLAE Fossil cones of Pinus coloradensis Knowlton, and foliage presumed to represent the species, occur in the late Oligocene Creede flora, Colo- rado. The fossils appear to represent the Leio- phyllae, and show relationship to P. chihua- huana Engelm., also considered a variety of P. leiophylla. The fossil needles in 5s are shorter than those of the modern species. Pinus leio- phylla ranges today from southeastern Arizona- adjacent New Mexico into southern Mexico (Oa- xaca). The well-preserved cone on Pinus paucisqua- mosa Templeton (1953) that represents this al- Peninsula, southern California (Text-Fig. 15). 1986] Dated at 14.5 Ma, and of early Mohnian age, it was compared initially with P. chihuahuana. As judged from the broad cone scales of the fossil, it appears more nearly allied to the only other member of this alliance, P. /umholtzii of the Sier- ra Madre Occidental. As with several other taxa, this record may This would place it at the latitude of northern Sonora in the Middle Miocene, an area close to members of the Leiophyllae today and an area where they probably were also in the past (see Text-Figs. 12, 15). SUBSECT. SYLVESTRES This alliance is represented in the Eocene of British Columbia as judged from the internal structure of cones and foliage preserved in the Allenby Formation (46 Ma) near Princeton (Stockey, 1984). The evidence indicates that cones of P. princetonensis Stockey, as well as the cone of P. arnoldii Miller and needles of P. allisonii Stockey have the characteristics of Subsect. Syl- vestres. This suggests a Beringian connection with Eurasian members of the alliance in the Early Eocene or Paleocene. The cones of P. clementsii Chaney (1954) from the Late Cretaceous (85-86 Ma) of southern Minnesota were compared by Mirov (in Chaney) with the living P. resinosa Aiton. However, the preservation of the mold that represents the ho- lotype (Chaney, 1954, figs. 1, 2) is poor, and the features of the umbo are not clear. The paratype (Fig. 6) is an incomplete specimen, and neither specimen has internal structure preserved. These specimens, from localities 40 km (25 mi.) apart, appear to be different species. They may repre- sent the genus Pityostrobus. SUBSECT. AUSTRALES The 11 species of this group are now in eastern and southeastern North America. As based on genetic data resulting from numerous hybridiza- tion tries, Critchfield (1962) concluded that the roup is a natural one. The few taxa that do not cross readily are either removed from the others time. group are the Paleocene Pityostrobus (Pinus) lyn- ni (Berry) Miller (Berry, 1934; Miller, 1977b) and the Middle Miocene P. collinsi Berry (1936, 1941). Both are similar in external features to AXELROD— WESTERN AMERICAN PINES 585 cones of P. taeda L. Miller (pers. comm., 1984) pointed out that P. avonensis Miller (1969), P. buchananii Underwood & Miller (1980), and P. driftwoodensis Stockey (1983) all belong to either Australes or Ponderosae, though it is not now possible to determine which one group they may represent. (see Stockey, 1984, table 1). Species allied to Australes have been described from the Neogene of western Europe, including fossil pines similar to P. taeda, P. pungens, and P. rigida (Gaussen, 1960: 234). The illustrations generally resemble these pines, but a final deci- sion must be based on a reexamination of the fossils themselves. That the records may repre- many species allied to those now in the eastern United States occur. Among these are species of Acer, Betula, Carya, Diospyros, Fagus, Hama- melis, Juglans, Lindera, Liquidambar, Nyssa, Persea, and Quercus, as well as the conifer Tax- odium. Some of them occur in intermediate areas, notably in the Miocene of Iceland (Akhmetiev et al., 1978; Heer, 1968; Moorbath et al., 1968). If Australes pines were in western Europe, they presumably were eliminated there along with the now-American hardwoods and swamp cypress during the later Pliocene and Pleistocene as cold and then ice spread southward. The presently- American (and Asian) taxa were blocked from southward retreat by the elevated Alpine axis that stretches across southern Europe and thus became extinct there. SUBSECT. PONDEROSAE Regarding the possible antiquity of this alli- ance, the recent description of well-preserved, silicified Pinus wood from the base of the early siderable interest ven a 1984). The wood is reported most similar to that of Subsect. Pon- derosae (cf. P. ponderosa) though some of its features suggest relationship with woods of the allied Subsect. Australes (cf. P. taeda). Pinuxylon eutawense Blackwell may represent a pine an- cestral to Subsect. Australes, or possibly to Pon- derosae as well. In any event, yellow pines are clearly foreshadowed at this early date (ca. 88 a ago), as is indicated also by the cone of Pit- yostrobus (Pinus) lynni noted above. Ponderosae fossils are certainly Eocene, as judged from seed-wing records and needles in 586 the Green River (MacGinitie, 1969) and Floris- sant floras (MacGinitie, 1953). In addition, needles of P. andersonii Stockey (1984), pre- served in chert of the Allenby Formation (46 Ma) near Princeton, British Columbia, have the in- ternal structure of Ponderosae needles. Further- more, the silicified cone of P. premurrayana (Knowlton, 1899), from the east side of Yellow- stone Lake, and probably from the early mid- Eocene Langford Formation (ca. 50 Ma, Smedes & Protska, 1972), is also a member of the Pon- derosae. It is similar to cones of P. lindleyii Lou- don, now in Mexico and considered by some to be a variety of P. montezumae (i.e., Martinez, 948 Remains of several Ponderosae are in the Late Oligocene Creede flora, Colorado. They resem- ble those of species now in the Southwest and/ a This suggests that the southern Cordilleran re- gion that extended southward into Mexico was a major center for diversification of Ponderosae in the Paleogene. Its species probably were re- stricted southward gradually as colder climate spread over the central and southern Rocky Mountains, leaving P. scopulorum the chief sur- vivor, and with P. engelmannii and P. arizonica reaching their northern limits in southern Ari- zona and adjacent N figs. 5, 6). The remaining Ponderosae (e.g., P. cooperi, P. douglasiana, P. durangensis, P. hart- wegii, P. lawsonii, P. michoacana, P. montezu- mae, P. pseudostrobus, P. tecote) now occur well south of the U.S.-Mexican border, or in Califor- nia-Oregon (P. jeffreyi, P. ponderosa, P. wash- oensis). A northwestward occurrence of a presently mite quarry. The large asymmetrical cone is sim- ilar to those produced by P. pseudostrobus. During the Middle and Late Miocene, branch- lets, needles, and seed-wings similar to those pro- duced by P. ponderosa occur at a number of sites in western Nevada, including the Eastgate, Fal- lon, Fingerrock, Middlegate, Stewart Spring, Pyramid, and Verdi floras. These are chiefly for- est-border assemblages that lived for the most part under subhumid climate. By contrast, it is ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 evident that P. ponderosa-like structures are largely absent in the floras of Oregon, Idaho, and Washington that occupied more humid climates. inus ponderosa (including its varieties) has a broad range, reaching from southern British Co- lumbia to San Luis Potosí, Mexico, a distance of 3,850 km (2,350 mi.). As noted by Critchfield and Little (1966: 16), it is absent from a large part of the intermontane region, notably south- ern Idaho, western Wyoming, southwestern Montana, and a good part of the Great Basin ranges of northern Utah and much of Nevada (Text-Fig. 16). However, fossil records show that ponderosa-like structures (needles, winged seeds) occurred in the region; the general rarity of fossils in the eastern part of the area reflects the fact that the rocks 5 are chiefly of Mesozoic and older formation The present denm of P. ponder much of the intermontane region may be explained by several factors (Axelrod, 19764). In the first place, precipitation over most of the lowland region totals less than 250 cm (10 in.) and hence is insufficient to support these trees. The pine has a normal growing season from May through Au- gust. sine dii bue is low, seedlings cannot become ly because adequate soil moisture is critical js abide (Fowells, 1965). In this regard, P. ponderosa (scopulorum) does occur in the eastern Great Basin ranges, where late spring-summer rains are frequent. Pines that may have survived in the isolated gthe last glac iation probably y were eliminated there soy the post- -glacial Xerothermic riod. I rate would P s e been a critical factor in n eliminating pine seedlings. Furthermore, trees probably could not survive by retreating upslope to moister areas because climate there is too cold in summer. In addition, aspen outcompetes pine in the same zone and may partly account for its absence over much of the region. An aspen-sage-meadowland association forms the zone in which pine would occur naturally (Axelrod, 1976a). Aspen com- petes with pine for water, but aspen, with its shallow, spreading roots, outcompetes pine seed- lings. Aspen not only spreads rapidly by suckers in its early years (Fowells, 1965), it is also = tolerant of shade, as is ponderosa pine. Hen aspen competes further with pine seedlings if Hes are present. Under present conditions, the aspen- sage-meadowland zone is a relatively permanent drought-resistant vegetation type in the inter- 1986] AXELROD— WESTERN AMERICAN PINES $ | | | | | 120° 159 поо 105° 587 MILES о 100 200 300 400 M —— —— 1009 95° 30 85 TExT-FiGuRE 16. Of the 12-13 species of Subsect. Ponderosae, only three occur north of southern Arizona (i.e., P. ponderosa with two vars., P. jeffreyi, P. washoensis). Note the broad area in the intermontane region where ponderosa does not now occur ый text Юг discussion). montane region, occupying areas where P. pon- derosa might exist if precipitation was higher. he origin and relationships of Pinus wash- oensis Mason & Stockwell (1945) provide a noteworthy problem. The species occupies a con- tinental climate and occurs in the upper mixed- conifer forest and in the fir forest. It is scattered discontinuously from the east slope of the Carson Range, Nevada, northward into British Colum- bia. The species is characterized by small, com- pact, ovate cones. Haller (1959, 1965, 1984) con- cluded that it is allied to populations of P. ponderosa ranging from the Lake Tahoe region northward through eastern Oregon into interior British Columbia. He noted that in its type area, where it is quite distinct, it only crosses with P. ponderosa with reduced seed set. It becomes less distinct northward and gradually tends to merge Pacific races of P. ponderosa has played a role in 588 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 P. arizonica P. scopulorum 1986] a origin. Haller concluded that P. washoensis be a segregate from these hybrids main- Bl in isolation at the south, but only partially differentiated in the north where distribution is more continuous. However, there appears to be evidence for its relict nature. This is implied by breeding studies that show that it crosses most readily with Rocky Mountain (Colorado-Black Hills) populations of P. scopulorum Engelm. and with which it pro- duces most seeds (Critchfield, 19842). Further, in terms of its seed proteins, the pine in its type area on Mount Rose has an antigenic distance of 0.6 from the Sierran ponderosa pine (Prager et al., 1976). Fhis is greater than the distance between P. jeffreyi and P. sabiniana (0.4—0.5), although less than the values from P. coulteri and P. engelmannii (0.8—1.9) (Prager et al., 1976). The data presented by Critchfield (1984a) sug- gested that separation from an ancestral popu- lation may have occurred on the order of 20-30 Ma ago. This is possible since the remains of P. ponderosa-scopulorum-like pines occur in the Eocene and Oligocene. Critchfield also pointed out that in the ability of Washoe pine to cross with other taxa, it “behaves like a fragment of the Rocky Mountain race stranded at the west edge of the Great Basin.” To follow up this suggestion, I examined stands of P. scopulorum in eastern Nevada, western Utah, and southern to central Colorado and Ar- izona. In the last two areas, especially, the cones are generally similar to those produced by washoensis but are not so compact. They differ in that P. scopulorum cones are ovate, whereas those of P. washoensis are ovate-rotundate (i.e., rounded or globate), P. washoensis has more nu- merous scales (173 vs. 112-141); in cone color (purple vs. green for P. scopulorum); wing length т vs. longer іп Р. ѕсориоғит); and longer P. washoensis than P. scopulorum (Critehfeld, 1984a). rther evidence for a Cordilleran source for F peras is provided by P. arizonica En- gelm. of southern Arizona, adjacent New Mex- ico, and northern Mexico. This tree regularly produces small, compact, ovate-rotundate cones scarcely separable from those of P. washoensis AXELROD-— WESTERN AMERICAN PINES 589 (U.C. Herb. 335574, 334313— Santa Catalina ts., 334316 Chiricahua Mts.; also see Shaw, 1909, pl. 17, fig. 4). The long needles are in fas- cicles of 3s and 5s, adn qu in warmer regions where pines commonly hav 1914; Mirov, 1967). The similarities suggest that P. washoensis may have been derived from a population closely allied to arizonica-scopulo- rum (Text-Fig. 17). This implies that a popula- tion with generally small, globose cones ranged Miocene (ca. 13—12 Ma) as climate rapidly be- came drier and numerous exotics disappeared from that region (Axelrod, 1985). Adaptation to increasing summer drought in the west presum- ably resulted in the development of somewhat more compact cones and shorter needles com- monly in 3s, as seen in P. washoensis. The suggestion that P. washoensis may be a relict population derived from a Cordilleran source is consistent with paleobotanical evi- dence. A number of A Rocky Mountain andi reat Basin to the Sierra rada and border areas (see Little, 1971, 1976). Among these, the following have equivalent species in the Miocene floras rom the present desert area of western Nevada: Acer diffusum Alnus tenuifolia Amelanchier utahensis Betula fontinalis Celtis reticulata Cercocarpus ledifolius aeagnus velutina Paxistima myrsinites Peraphyllum ramosissimum Pinus monophylla Pinus ponderosa Populus ое Рғипиѕ ает Prunus Se ab Ribes cereum Symphoricarpos oreophilus Others with this distribution that do not now TEXT-FIGURE 17. Cones of Pinus washoensis (from Mt. Rose, Nevada, alt. 2,430 m [8,000 ft.]), P. arizonica (from Huachuca Mts., Arizona, alt. 2,130 m [7,000 ft.] a m [8,800 ft.]) are so similar they indicate ae these are closely related ta nd P. scopu a (from Creede, Colorado, alt. 2,675 a. The suggestion that P. washoensis is a relict, Cordilleran taxon is supported by paleobotanic and genetic Desi (see text). 590 have a known fossil record in Nevada include Chamaebatiaria millefolium, Pinus flexilis, and P. longaeva. Neogene floras of western Nevada, and also of central California, have a number of trees and shrubs allied to those in the southern Rocky Mountains, and some are represented by species only in Mexico. Among these are fossil species allied to: Acer grandidentatum Arbutus arizonica Bumelia lanuginosa Cercocarpus breviflorus Crataegus erythropoda Fraxinus anomala Juglans major Populus brandegeeii Populus mexicana Quercus arizonica Quercus vaseyana Robinia neomexicana Sapindus drummondii Fossil species of the above taxa contributed to mixed conifer forest, sclerophyll woodland, an a. The Cordilleran and Madrean taxa that occupied the present Great Basin during the later Miocene were largely elim- inated there as summer rain decreased further in the Pliocene (Axelrod, 1976a). Relict stands oc- cur now in the higher mountains bordering the eastern Great Basin where there is more summer precipitation, as well as along the eastern slope of the Sierra Nevada and bordering ranges where such typical “Rocky Mountain" taxa as Pinus flexilis, Populus angustifolia, and others still oc- cur. [In sum, the evidence seems compatible with the view that P. washoensis may be a relict Cor- dilleran pine, one derived from a plexus allied to P. scopulorum and P. arizonica. SUBSECT. SABINIANAE This small group of three pines, P. torreyana, P. sabiniana, and P. coulteri, occurs in Califor- nia, with P. ly into the mountains of northern Baja California (Text- Fig. 18). The only species with a limited fossil record is P. pieperi Dorf, which is allied to, if not identical with, P. sabiniana. Confined now to the Coast Ranges and foothills of the Sierra Nevada in central and northern California, all of its fossil records are in southern California (Text-Fig. 19). ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 These include the later Miocene Anaverde (Ax- flor: DeMéré, 1984) and Pico (Dorf, 1930) floras, the Early Pleistocene Seacliff flora (Axelrod, 19832), a cone from the Early Pleistocene in Lake Can- yon near Ventura (Wiggins, 1951), and a cone scale in the late Pleistocene (Wisconsinian) Car- pinteria (Chaney & Mason, 1933) and Rancho La Brea floras (Warter, 1976). Pinus sabiniana may have disappeared from southern California during the Xerothermic, which brought a hotter, semi-desert climate to the coastal strip. This cli- mate may also account for the present disjunct occurrences of P. sabiniana in central and north- ern California, where isolated populations often are surrounded by mesic, mixed-conifer forest which spread with cooler, moister climate fol- lowing the Xerothermic (Axelrod, 1966b: 50). At present, there is no known fossil record of Pinus torreyana, one of the most restricted of pines. A few thousand trees now inhabit the coastal bluffs of Torrey Pines State Park and nearby Del Mar, a few miles north of San Diego, and a subspecies occurs on Santa Rosa Island, 60 km southwest of Santa Barbara (Haller, 1986). The mainland population is near the head of La Jolla submarine canyon. Upwelling of colder water gives this local area a foggier, cooler cli- mate than the adjacent coast and may account for the persistence of P. torreyana there (Axelrod, 1982). There is also a highly marine, cooler, foggy climate on Santa Rosa Island. Cones of P. tor- reyana are similar to those of P. oaxacana Mirov of Mexico, a species that has long needles in 5s like P. torreyana (Text-Fig. 20). Significant in this regard is the occasional occurrence of strong- ly-hooked apophyses on cones of P. oaxacana that simulate those of the Sabi e (specimens in Inst. Forest Genetics, Placerville, California). Pinus torreyana may have been derived from a P. oaxacana-like pine in western Mexico, which then extended west to include the present area of Baja California. Activation of the San Andreas rift system by the Early Miocene (or earlier?) transported the area, and the pine, northward (see Axelrod, 1980, figs. 7, 8). Fossil species allied to P. sabiniana and P. coulteri may have had a similar origin at the south but were adapted to interior lowland (P. sabiniana) and montane (P. coulteri) habitats. The difference in needles, which are more slender and drooping in P. oaxacana, and thicker and more rigid in Sabinianae, may reflect adaptation to the more xeric environment 1986] AXELROD— WESTERN AMERICAN PINES A torreyana = sabiniana coulteri 4 Jii." EN torreyana Il "hr. TEXT-FIGURE 18. In Sect. Pinus, Subsect. Sabinianae has three species in California- па California, (Р. sabiniana, chiefly in the interior at lower altitudes i in central California; P. coulteri, montane R : ast may have been displaced northward from an ancestral population allied to axacana now in central e oun Mexico (see Text-Fig. 12). 592 PINUS SABINIANA ANNALS OF THE MISSOURI BOTANICAL GARDEN (VoL. 73 Ss A Bs 4 TEXT-FIGURE 19. Fossil m for P. sabiniana and its ancestor P. pieperi Dorf. си from Griffin & oe cho La hul 1972): 1 — Carpinteria, 2—Pic Vista. The species may hav e Canyon, 4 — Anaverde, 5— Ran Brea, 6 аас from southern California during the post- ИХ Xerothermic seed en, 7— which brought hotter, drier conditions to the coastal strip in which the California taxa evolved. On this basis, records of fossil allies of the latter species are to be expected in the interior. A cone scale of a species allied to P. sabiniana occurs in the Miocene Anaverde flora of the western Mohave Desert (Axelrod, 1950). Its position 0.5 km west of the San Andreas fault is consistent with this thesis. SUBSECT. CONTORTAE Members of this alliance have been recorded in the late Eocene (38 Ma) Bull Run flora, north- eastern Nevada (in Axelrod, 1968), the late Oli- gocene (26.5 Ma) Creede flora, Colorado (Ax- elrod, unpubl.), and the Early Miocene (21 Ma) Alvord Creek flora, southeastern Oregon (Axel- rod, 1944). The first site is a rich montane conifer forest, the others are mixed conifer-hardwood forests. There is no evidence that Contortae were abundant in these floras. Their rise to promi- nence in the modern flora appears to be a later Cenozoic event, with rising mountains and spreading cold providing suitable environments for them, and with logging and fires, often set by man, aiding their rapid spread and crowding out of associated trees. Pinus contorta var. contorta has a discontin- 1986] uous distribution on the California coast from near Pt. Arena northward (Text-Fig. 21). This is probably a relict late Pleistocene occurrence, for the area has cool summers suited to its survival at the south. Its distribution is paralleled by Picea sitchensis, a species that ranged to Tomales Bay, 150 km farther south in the late Pleistocene (Ma- son, 1934). The var. murrayana of the Sierra- Cascade axis is discontinuous to the high moun- tains of southern California and Baja California (Text-Fig. 21). The latter sites are relict from the Early Quaternary (2 Ma), bu mixed-conifer forest in interior southern California was fully 1,000 m lower in altitude о 19665) and continuous with that in the Sierra Nevada. Later uplift of the Peninsular-Transverse ranges dis- rupted the earlier distribution of the mixed-co- nifer forest and accounts also for the disjunct occurrence of P. contorta var. murrayana. The isolated P. clausa of Florida and nearby Alabama may have been derived from P. virgin- iana in the later Tertiary as climate cooled and a population was isolated at the south. Confined to dry (sandy) sites, it either escaped or survived competition of taxa in the rich, mixed-meso- phytic forest of the region during the climatic fluctuations of the Quaternary. SUBSECT. OOCARPAE Fossils of this group are presently known from California, Nevada, and Oregon in the west, and rom Massachusetts in the east. Some of these are certainly extinct. Pinus tiptoniana Chaney & Axelrod (1959) from eastern Oregon appears al- lied to species like P. patula and P. pringlei as judged from its umbo relationships. Pinus burtii Miller (1978) from Martha's Vineyard, Massa- chusetts, is a large-coned species that may be distantly related to the P. radiata complex. This с ES £e 3 a = © e o C o O 3 = — E 3 al Q o = 2 O = 4) o e = о = 3 C О bers of the Оосаграе now occur. Continuity with the pines of Mexico is certainly old and probably is Late Eocene-Early Oligocene. Recall that the Late Eocene (35 Ma) Barilla flora of trans-Pecos Texas (Berry, 1919) represents a rich palm-laurel forest in this now semi-desert area. Furthermore, the Appalachians already supported an Eocene temperate deciduous forest as judged from the pollen record (Gray, 1960), and the inner low- lands of the Gulf Coast were covered with a rich, warm temperate forest with many taxa now in AXELROD— WESTERN AMERICAN PINES 293 Mexico (Fredericksen, 1980, 1981). A warmer, milder climate is indicated also by the Miocene Legler flora (11 Ma) in the Cohansey Formation, New Jersey (Greller & Rachele, 1983). It has in the mountains of eastern Mexico, under wet, equable climate. At present, numerous species and varieties of trees and shrubs are common to the temperate forest of the uplands of eastern Mexico and the eastern United States (Harshberger, 1911; Ax- elrod, 1939: 78; 1960: 267-268; Graham, 1973). A corridor via the Appalachians-Ozarks-trans- Pecos Texas and into Mexico in the Late Eocene- Early Oligocene may account for these ancient, surviving links, and suggests stasis for these taxa extends back fully 30-25 Ma. Elimination of species of present temperate Mexican occurrence from the eastern United States resulted from the trend to less equable climate later in the Ceno- zoic, and especially in the Pleistocene. It also eliminated many subtropical broadleaved ever- greens, leaving the present impoverished decid- uous hardwood forest with a few evergreen di- cotyledons confined chiefly to the milder coastal strip, as exemplified by Gordonia, Ilex, Mag- nolia, Persea, and Quercus (Q. virginiana, Q. laurifolia). embers of the Oocarpae now in California include three species in the coastal area (P. re- morata, P. radiata, P. muricata) and one (P. at- tenuata) in the interior and in southwest Oregon (Vogl et al., 1977). The nearest affinities of P. radiata and P. remorata are with P. oocarpa, now in Mexico. Pinus oocarpa has cones very similar to those of P. remorata and to the small-coned P. radiata var. cedrosensis of Cedros Island and P. radiata var. binata of Guadalupe Island (Ax- elrod, 1980; Text-Fig. 22). Evidence suggests that he Oocarpae now in maritime California were displaced northward as terrain west of the San Andreas rift zone moved to its present area (Text- Fig. 23). This was from an earlier position that placed southern California at the latitude of So- nora, with the Cape Region of Baja California nestled near Cabo Corrientes, Jalisco (see Text- Fig. 12). The P. radiata populations on islands off Baja California (P. radiata vars. cedrosensis and binata) may be relicts of that distribution. In addition, a number of trees and shrubs as- sociated with the Oocarpae in coastal southern California and northern Baja, California (e.g., near Eréndira) also appear to be remnants of that movement. These include narrow endemics that find their nearest allies in the mountains of Mex- e 594 torreyana ANNALS OF THE MISSOURI BOTANICAL GARDEN sabiniana [Vor. 73 oaxacana TEXT-FiGuRE 20. Cones of Pinus torreyana are similar to those of P. oaxacana. In addition, both have long needles in 5s and large sheaths. The fossil forerunner of torreyana may have be en was displaced northward by movement on the San Andreas fault system during the Miocene and later (see Text-Fig. 12 ico where they are also associated with Oocarpae. Examples are in paired-species of Arbutus, Ce- anothus, Cercocarpus, Comarostaphylis, Myri- ca, Pinus, Prunus (Laurocerasus), Quercus, and Vaccinium (Axelrod, 1967b). With respect to evolution of the P. radiata series, the var. cedrosensis appears to simulate the most ancient member of the alliance for its cones are very similar to those of P. oocarpa. A P. radiata var. cedrosensis-like population pre- sumably gave rise to the Guadalupe Island P. radiata var. binata. The latter appears to have been ancestral to the California populations (Text- Fig. 24). Its cones range from those with apophy- ses scarcely developed (like P. radiata var. ced- rosensis and P. remorata) to those that are in- separable from cones of the Monterey population. Similar variation occurs in cones recovered from the Pliocene San Diego Formation at Chula Vis- ta, which places the insular P. radiata var. binata on the mainland (Axelrod & DeMéré, 1984). As land area west of the San A allied rifts moved north, climate there was becoming progressively drier in summer in response to cold- water upwelling along the coast. As summer rains decreased further, and the mediterranean-type 1986] AXELROD— WESTERN AMERICAN PINES We de Ra І TS — 14 E EXT-FIGURE 21. г sites for coastal Р. pus rta var. co occurr f P. contorta var. ana in southern ri / "LE S5 Sa Distribution of three varieties of Pinus contorta today (map from Little, 1971). The ntorta а hern and Baja California apparent te Wisconsin. The y dates from the early 66). were attained in s Ohana (2 Ma) when forest аң "lived aou 1,000 m ee thad they do today usd iun 19 climate became more intense, cones of the three California populations evidently increased in size and also developed larger apophyses and seeds. The population at Monterey-Carmel has smaller cones than the stands 70 km northwest at Ano Nuevo or 135 km southeast at Cambria (Text- Fig. 25). This apparently reflects survival of an older population in the cooler, foggier climate at Monterey. It results from the influence of the Monterey submarine canyon on local climate there, an area where two cypresses (Cupressus ioni в C. goveniana) also have relict oc- nces, as do a number of shrubs and forbs ева 1982). 596 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 cedrosensis binata TEXT-FIGURE 22. Cones of sped Pe related pines of Subsect. Oocarpae, P. oocarpa, P. remorata, and P. radiata vars. cedrosensis and bina 1986] Pinus radiata, P. muricata, and P. remorata all had wider occurrences in coastal California in the Late Pleistocene Pemi 1967a, 1967b). í | west of San Francisco eaa 1976a, 1980), to Chula Vista near San Diego (Text-Fig. 26), some 825 km southeast (Axelrod & DeMéré, 1984). The record of P. muricata is not so com- plete, but it certainly had a wider occurrence, for the northern population P. muricata var. borealis reached south at least to the Santa Barbara area. Pinus remorata also occurred more widely than at present. Restriction of these taxa to local areas today is chiefly the result of the warmer, drier, post-glacial climate, and especially that of the Xerothermic, which disrupted the more mesic forests. The present relict po climate with a high fog frequency. A similar re- lation exists at Eréndira, Baja California, where P. remorata and P. muricata occur in a semi- desert climate bordering an ocean that is much colder and foggier in this local area (Axelrod, 1982). The other member of the Oocarpae in Cali- fornia is P. attenuata, confined chiefly to interior regions, generally at the lower margin of mixed- conifer forest. It is most nearly allied to the Mexican P. greggii, P. patula, and P. pringlei, although the cones of P. attenuata are more strongly armed. The fossil P. pretuberculata may have spread northward from an ancestral, inte- rior plexus and then been stranded in California as developing arid climate to the south disrupted its connection with the Mexican taxa (Axelrod, 1979). In this respect, the isolated P. attenuata population. east of Ensenada hası cones somewhat | Mexican species. The fossil P. pretuberculata, which is very sim- ilar to cones of P. attenuata, was already in Cal- ifornia in the Upper Miocene (12 Ma) Table Mountain flora (Condit, 1944). It is recorded from youngest Miocene rocks (5-6 Ma) in the Mount Eden flora of southern California (Axelrod, 1937) and the Verdi of western Nevada (Axelrod, 1958). Pinus attenuata is also in Early Pleistocene rocks at Signal Hill (Mason, 1932) and Seacliff (Axel- rod, 1983), coastal southern California. At these latter sites, the cones probably were transported rom the interior. This is likely because P. at- AXELROD— WESTERN AMERICAN PINES 597 Displaced Oocarpae 7 А бо. TEXT-FIGURE 23. Distribution of species of Oocar- pae, with four species each in California- -Baja Cali- t o (se кй. Sprdiding diei climate further har these taxa. 598 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 GUADALUPE I. Pinus radiata var. binata TEXT-FIG Vari the present California popula tio Vista, southe aliforn resulted in the са to larger cones and s tenuata occurs at these sites with closed-cone mid, equa ularly inhabits drier sites away from the coastal ation in cones of P. radiata var. binata suggests that it may have been ancestral to ns. This inference is consistent with the o la rnia, similar to those of var. binata. Increasing summer drought in California evidently eeds. ccurrence of Pliocene cones at Chu lowlands, in more extreme climates of the inte- rior. The species is restricted now to isolated stands in southern California, on the crest of the Santa Ana Means and in the San Bernardino 1986] AXELROD— WESTERN AMERICAN PINES 599 R ¿Keno c Mi Shasta canyon Д4 АМО NUEVO mean length 11.5 cm Grant Forest А MONTEREY mean length 9.5 cm e Las Vegas Palm Springs Disneyland . CAMBRIA mean length 14.0 cm TEÉxT-FIGURE. 25. The California populations of P. radiata have larger cones and apophyses than do the Monterey population has survived there with other relict endemics (from Axelrod, 1982). Mountains. These local ocurrences, as well as the relict population east of Ensenada, may have re- an excavation at Oakland, on the east side of San Francisco Bay (Metcalf, 1923). Today the nearest stand is east of the Oakland Hills, some 12 km distant, where climate is hotter in summer and colder in winter. The pine evidently extended west to Oakland in a Late Pleistocene glacial age when sea level was about 100 m lower and the shoreline was fully 50 km (30 mi.) farther west (Helley & LaJoie, 1979). Climate was then more continental in the San Francisco Bay trough, which was dry land, as shown also by mammoth and other mammals recovered there during dredging. With the rise ofsea level, and the return akla still have rare populations in the Bay area prob- ably are also relicts of the drier, Xerothermic climate (Axelrod, 1981) LATE CENOZOIC EVOLUTION Into the middle Tertiary many trees (probably most) of temperate regions had comparatively wide distributions that resulted from low envi- ronmental diversity. Climates were broadly zoned, they were highly equable, and more asea- sonal in lower latitudes. Furthermore, terrain dif- ferences were not so marked as those of today. 600 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 PINUS RADIATA Present & past occurrences TEXT-FIGURE 26. Pinus radiata occurs today at: A— Ano Nuevo, B— Monterey-Carmel, C—Cambria, D— Guadalupe I., and E—Cedros I. Fossil localities for Monterey pines occur at: 1 — Tomales Bay, 2— Drakes Bay, 3— Thornton Beach, 4— Mussel Rock, 5—Spring Valley Lake, 6—Little Sur, 7—Pt. Sal, 8 —Carpinteria, 9— Seacliff, 10— Pico, 11 — Potrero Canyon, 12— Rancho La Brea, 13— Mount Eden, 14— Chula Vista, 15— Santa Rosa I. Disruption of the Late Pleistocene forest probably resulted from spreading Xerothermic climate that brought hotter, drier interior conditions to the coastal strip. The present relict stands are all confined to local areas of high fog frequency during summer. 1986] High continuous ranges like the Sierra Nevada- Cascades or the Rocky Mountains, and high pla- teaus (Colorado Plateau, Mexican Plateau) had not yet come into existence. As regional topo- climatic differences developed later in the Ce- nozoic, the stage was set, especially in Mexico, for the rapid splitting of populations into new races, varieties, and probably some species. part from geographic (spatial) and ecologic (edaphic) differences, isolation of populations may also result from seasonal climatic differ- ences that determine times of major reproduc- tive events (Stebbins, 1950: 208-209). This is seen notably in the co-occurrence at Monterey of Pinus radiata and P. muricata, which do not form hybrids because of fully two months dif- ference in maturation of microsporangia. A sim- ilar separation is seen in the populations of P. radiata and P. attenuata at Swanton and Ano Nuevo, north of Monterey. Differences in time of “flowering” that isolate populations may ac- count for different populations (or varieties) of piñon (P. cembroides) recently described from the Southwest and Mexico. Most live under sub- humid to marginally semiarid climates and are spread across 23? of latitude, or 2,800 km (1,775 mi.). Times of pollination vary greatly across this broad region depending on local climates (warmth), which are the result of recent tecton- ism that formed the ranges and basins in which these populations now occur. A similar expla- nation may also apply to Pinus ponderosa, which releases pollen at very different times of the year in British Columbia, Utah, California, and San Luis Potosí at its southern occurrence. These re- gional differences may have been a factor in the rise of varied races in these areas of very different rainfall and temperature regimes (adaptive sub- zones), and times of pollen shedding. The closely similar species of Strobi that ex- tend more or less continuously from the Cana- dian Rockies in British Columbia southward through the Sierra Madre Occidental into Gua- temala, may also be considered in this context. There is a transition zone between P. flexilis and P. strobiformis and between P. strobiformis and P. ayacahuite. The shift from one species to the next occurs in areas of major topographic-cli- mate change. These are along the Colorado-New Mexico border, and in the transition from central Arizona-New Mexico southward where climate rapidly becomes warmer (Axelrod & Raven, 1983), and in central Mexico, with warmer con- ditions farther south where P. ayacahuite occurs. AXELROD— WESTERN AMERICAN PINES 601 The climatic differences, which determine time of pollen maturation, dispersal, and fertilization, have been intensified in the later Cenozoic. Nat- ural hybrids of these species regularly occur in these transition zones. ines now in boreal regions and subalpine areas that form pure stands are subject to very low temperatures in winter. The most extreme are those in the area of P. pumila in northeast Si- beria. Mean January temperature is —47.6°C (—52°F) at Verkhoyansk and —43.2?C (—45°F) at Yakutsk. Summers are sufficiently warm in these areas for boreal forest, with a mean July temperare of 16?C (60?F) at Verkhoyansk and 6°F) at Yakutsk. These extreme condi- tions developed recently, chiefly in the Quater- Middle Miocene (ca. 15 Ma) the area — © nifer-deciduous hardwood forest much like that now on the middle mountain slopes from central Honshu to southern Hokkaido. This is apparent from the flora of Mammoth Mountain on the Aldan River (Dorofeev, 1969), about 200 km northeast of Yakutsk. It includes species of Abies, Larix, Pinus, Acer, Betula, Broussonetia, Cory- lus, Cornus, Jug ans, Morus, Padus, chatka (Menner, 1976), Sakhalin (Heer, 1878a, 1878b), Iceland (Heer, 1868), and Alaska (Heer, 1869). The shift to colder climate was largely post 5—6 Ma, as polar climates spread and mountains were elevated to alpine levels. This enabled sub- alpine and boreal conifers to form more contin- uous populations as hardwoods were eliminated by progressively increasing cold. During the later Tertiary, these northern coniferous populations 1983b: 126-128). This does not mean that pure conifer forests were not present until the later nated by conifers at seven sites distributed strati- graphically through fully 1,160 m (3,500 ft.). The conifers include species of Abies, Larix, Picea, Pinus, Tsuga, and Chamaecyparis. The only di- cotyledons are a few shrubs distributed in Cra- taegus, Holodiscus, Mahonia, Malus, Prunus, Ribes, and Salix. This Upper Bull Run flora suc- ceeded a mixed conifer-hardwood forest after 38 Ma as climates became colder (Fig. 2). If pure stands of pine (i.e., species of Cembrae, Strobi, Balfourianae) contributed to subalpine forests, 602 they must have occupied slopes fully 500 m or more above the rich (15 taxa) montane conifer forest of the Bull Run basin. A contributory role in the development of co- nifers, as well as vascular plants, has been played by mycorrhizal fungi, as discussed by Malloch et al. (1980). Pines have ectomycorrhizae, those in which the fungus surrounds the living cell of the root and does not penetrate it. The mycorrhizae extend out into the soil and serve to transfer nutrients from decaying litter into the plant from generally nutrient-poor soils. The mycobionts are very diverse in ectotropic forests, whereas the photobionts often form nearly pure, monotonous stands, as in the boreal and subalpine forests of colder regions (Malloch et al., 1980). Today, ec- totropic forests like the boreal forest may com- mence on new sites with various dicotyledonous species, but the conifers become increasingly dominant. They may reach an equilibrium state that enables the best-adapted species to form for- ests of considerable stability (Meyer, 1973: 88). However, this is a new regional relationship, for pure conifer forests are not recorded as wide- indicates that they were restricted mostly to lo- calized, higher and colder altitudes (see Axelrod, 1968; 1976a). Pure pine forests spread in re- sponse to increasing cold (P. banksiana, P. con- torta, P. cembra, P. pumila), or drought (P. cem- broides, P. monophylla, P. edulis, P. ponderosa), or to seasonality. As noted above, these are new environmental conditions in terms of forest his- tory, whether in North America or Eurasia. Thus, the mycorrhizae have given Pinaceae, and Pinus u of seasonal climate and often poor soils. Many of these sites are s ver drained sis effectively rier vol t ks and hence are favorable sites for pines. These areas largely re- sulted from relatively recent volcanism or tec- tonism, which uplifted and exposed older rock units, often carbonates. With the presence of ec- totrophic mycorrhizae in abundance, as well as seasonal climate affecting time of pollen shed- ding, fluctuating climate in the late Cenozoic, and more recently disturbance by man (fires, clearing), it is understandable that pines were ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 able to form pure stands and many new races and varieties in the Mexican uplands and border The Quaternary changes noted above are fur- ther illuminated by a current paper by Critchfield (1984b) that considers the impact of Pleistocene climatic changes on the genetic structure and dis- tribution of North American conifers, notably some species of Abies, Picea, and others. His interpretations are based on evidence from fossil pollen records, wood-rat middens, glacial and inter-glacial megafossil floras, and isoenzyme an terpene analyses of living conifers. In northern areas, the data provide evidence for population retreat during glacial times and their spread from refugia in post-glacial times. Over western areas, the record reveals the wider distribution of co- nifer species over lowlands presently desert and RM during the pluvials, as well as the oc- e of presently high montane conifers at ainda fully 900-1,000 m lower. As Critchfield concluded, the most general effect has been the o and extinction of transitory races of boreal conifers, and, in some western conifers, the rise of longer-lived races through repeated cycles of isolation and hybridization. The influ- ence of climatic fluctuations thus involved ex- tinction of some species and races, the nearly complete loss of genetic variation in others, and occasional gene enrichment through hybridiza- tion and introgression. FUTURE WORK Discovery of new fossil ovulate pinaceous cones (Pityostrobus, Pinus) will no doubt further illu- | ine his es fective if large collections can be ob pecially in older montane basins, or at higher atitudes, areas where Tertiary pines are more abundant. A most critical need, however, is the discovery of Tertiary floras in the Mexican up- lands, a prime area of pine evolution. Compar- isons of fossil taxa with species of modern sub- sections are necessary if we are to do more than merely describe new taxa. Biochemical studies (isozymes, terpenes, DNA analysis) of modern species may provide evi- dence to estimate relative age of taxa. This would be of high interest for, in contrast to fossil species that have already developed their characteristics, the data may provide indications of the time of divergence of allied taxa. This could shed light 1986] on the age of presumably paired-taxa on different continents (P. griffithii-P. monticola, P. cembra- P. albicaulis) and in different parts of one con- tinent (P. strobus-monticola, P. edulis-mono- phylla) Continuing hybridization studies of modern species of the subsections may aid in further un- derstanding species interrelationships. CONCLUSIONS The preceding review suggests that pines have en throughout their histor ey appear to have adapted early to generally ач ful areas and probably originated there. Their evolutionary success probably was favored by their symbio tic relationship with ectotrophic mycorrhizae. This enabled them to inhabit drier sites, areas of seasonal rainfall, poor edaphic con- ditions (sand, laterite, podsol), and, late in their history, very cold regions. Pines responded rap- idly on several occasions to increased topograph- ic, climatic, and edaphic diversity by splitting into alliances (“subsections”) of different adap- tive mode. They have radiated into mountains in temperate and tropical regions, they inhabit dry areas at the lower limit of tree growth (P. monophylla, P. ponderosa, P. sabiniana), they have reached into extremely cold regions at tim- berline (P. albicaulis, P. aristata), at tree-line in boreal regions (P. banksiana, P. pumila), and others have entered seasonally dry tropical areas with leached soils (P. cubensis, P. dalatensis). More recently, pines have formed pure stands in colder (P. banksiana) and drier (P. cembroides, P. ponderosa) regions as older, richer forests and woodlands lost taxa and retreated to more fa- vorable areas as climate changed. That pines are still opportunistic is evident in the uplands of Mexico for evolution is active there, as seen in the variation in P. cembroides, P. montezumae, P. oocarpa, P. pseudostrobus, and others. As on earlier occasions in pine history, Mexican pines recently responded to the rapid appearance of new environments—created there by volcanism, tectonism, new local climates, and fluctuating climate—to form new races and varieties by hy- bridization and introgression. Although pines are indeed ancient, they are highly successful. In terms of sheer numbers of individuals, there are more pines today than at any time in the past 140 million years, and they also wholly dominate vegetation over broad regions—a recent event. = = AXELROD—WESTERN AMERICAN PINES 603 PART 2. SYSTEMATICS When George Englemann’s Revision of the Ge- nus Pinus was published in 1880, scarcely any fossil pines were then known from North Amer- ica. A few had been described by Heer (1868- 1882) from the high arctic and were assembled later in his six-volume classic, Flora Fossilis Arc- tica. Secured by various expeditions, the collec- tions were small and inadequate, and the pines are represented chiefly by imprints of fragmen- tary needles, winged seeds, and a few broken cones. They are associated with taxa that con- tributed to deciduous-hardwood and conifer- hardwood forests of Paleocene, Eocene (Elles- mere I., Spitzbergen), and Miocene (Kamchatka, Sakhalin, Iceland, Alaska) age. In the volume describing the Geological Exploration of the Western Territories, Lesquereux (1878, 1883) re- ported on fossil floras from a number of localities in the western United States. Most of these rep- resent vegetation of humid, warm temperate to subtropical environments, climates in which pines are largely absent. The early explorations of western Canada disclosed several small fossil floras in British Columbia and areas to the north (Dawson, 1883, 1890, 1891; Penhallow, 1908). They also contained scrappy remains of conifers, notably needles, winged seeds, and broken cones of pines and other conifers. As more detailed geological studies of local areas proceeded during the decades following the 1920s, fossil floras were recovered from more numerous localities, and those from moderate mosa of N under Mountain of Idaho, are preserved in sedimentary deposits that accumulated in the moats of large calderas that dist аш from collapse following explosive volcani Ithough a ipud has elapsed since the early collections were made, our knowledge of pine history is still in a very preliminary state. This reflects the fact that there are relatively few good collections of fossil pines. Furthermore, since only a few investigators have been interested in the group, our understanding of pine history has been impeded further. In this review of previously- described as well as new Tertiary pines, attention is centered on the external form of cones pre- 604 TABLE 2. Tertiary ovuliferous Pinus cones with described internal structure. ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 Species Structure Occurrence and Age Apparent Affinity Pinus avonensis Miller Cone Near Avon, Montana, Oli- Subgenus Pinus (9 subsec- (1969) gocene tions) Pinus arnoldii Miller (1973; Cone Allenby Fm., B.C., Middle Sylvestres Stockey, 1984) Eocene Pinus wolfei Miller (1974) Cone Cowlitz Fm., Wash., Late Contortae, Oocarpae, Syl- Eocene vestres Pinus burtii Miller (1978) Cone Miocene Greensand, Mar- Oocarpae tha's Vineyard, Miocene Pinus buchananii Under- Cone Twin Bridges Fm., Wash., Australes, Ponderosae wood & Miller (1980) Oligocene Pinus escalentensis Banks et Cone UR Fm., B.C., Oligo- Sect. Pinus (6 subsections) al., 1981 Pinus driftwoodensis Stockey am E Beds, B.C., Australes (1983) Middle Eoc Pinus princetonensis Stockey Cone Allenby Fm., B.C. Middle Sylvestres 1984 Eocene Pinus similkameenensis Mil- Fascicle, 5s Similkameen beds, B.C., Strobus ler (1973) Middle Eocene Pinus similkameenensis Mil- Wood Similkameen beds, B.C., Parrya ler (1973), Stockey (1984) Middle Eocene Pinus allisonii Stockey Fascicle, 2s Allenby Fm., B.C., Middle Sylvestres (1984) ocene Pinus andersonii Stockey Fascicle, 3s Allenby Fm., B.C., Middle Ponderosae (1984) Eocene served as imprints 1 in fine-grained sedimentary ossil cones are similar to living pines. Less emphasis is placed on isolated winged seeds and fascicles of needles raal most cannot readily be as- signed to specie As discussed by. e (1976), and Banks et al. (1981), cones of Pinus have four main fea- tures. Species of Pityostrobus may possess some, but not all of these: 1) ovuliferous scales inflated atapex, 2) vascular supply to bract-scale complex leaves axial cylinder as a single unit, 3) resin canals in the base of the scale are abaxial to the scale trace, and 4) the vascular bundles in the distal part of the scale are rounded on their adax- ial side (Text-Fig. 1). While these differences may separate most Cretaceous and some Paleogene cones of Pityostrobus from those of Pinus, Pit- (Alvin, 1960), Pityostrobus species apparently represent surviving members of an older plexus from which Pinus had diverged by the latest Ju- rassic. Inasmuch as Pityostrobus has been re- corded only once in the Paleocene (Miller, 1977b), it presumably largely became extinct in the Cre- taceous-Tertiary transition, and was replaced by Pinus. Studies of the internal structure of perminer- alized Pinus cones from older Tertiary rocks have not grea atly clarified their relationships within the t described species have been compared with one or two subsections, yet cones of species in these alliances may differ considerably in external features (form; apophyses; umbo-mucro position, etc.). The di- versity of cone types within these groups, as well as the megafossil record of external impressions, implies that a number of species in the subsec- tions were already in existence in the Eocene (see elow). One of the problems raised by studies of internal structure of fossil cones is that the ex- prior to burial, that their external characteristics have been erased (e.g., P. wolfeii, P. avonensis), and the form of others, if preserved, was not illustrated or described prior to sectioning. In view of the preliminary stage in our knowl- edge of American Tertiary pines, it is desirable to clarify the approach to systematics followed here. Since most fossil pine taxa are based on 1986] disassociated structures (cones, needles, winged seeds) at a given locality, these are grouped into species that appear to represent one taxon. In this way, a plethora of names that might other- wise be used is eliminated, and some semblance of biotic relationship can be suggested. For ex- ample, a cone of Pinus in the Late Eocene Bull Run flora is allied to P. contorta. Also in the flora are winged seeds that fall within the range of variation shown by that alliance. Seeds similar to those produced by P. contorta are in the Creede flora, Colorado. All this material is referred to the previously-described P. alvordensis Axelrod, which is a winged seed ofthe P. contorta alliance. It is granted that the fossils from these sites may represent different species of Subsect. Contortae. However, at this stage in our understanding of pine history, and in view of the limited sample, it seems best to group them into one fossil taxon, for the fossils can scarcely be separated taxo- nomically. A second example is provided by Pi- nus crossii Knowlton, based on fascicles of 5s in the Creede flora, Colorado. My large collections from 15 sites in the Creede Formation show that short needles in 5s are abundant at most sites. There are fossil cones and winged seeds in the flora similar to those produced by the living P. aristata, a species that has short needles in 5s. Rather than describe the cones and winged seeds as two additional species in the Creede flora, all of this material is referred to Pinus crossii, a species allied to P. aristata. dentification of the fossils in this report has been based chiefly on comparisons with the large collection of modern pine material at the Insti- tute of Forest Genetics, Placerville, California. This has made it possible to suggest relationships between the fossils and modern taxa, and hence to assign the fossils to subsections. It is empha- sized that comparison ofa fossil pine with a mod- ern species does not mean that the living taxon has persisted unchanged from the Eocene or Oli- gocene down to the present. The comparison simply means that, as judged from what actually is available, a particular living species is most nearly related to the fossil pine which is given a separate name. In the following discussion of some Tertiary pines, emphasis is on the imprints of fossil cones, for this is their usual mode of occurrence. Anal- ysis of their modern relationships is aided by reference to the umbo-mucro features of the cone scale presented by Klaus (1980). Furthermore, appraisal of the affinities of cone imprints com- monly is aided by making latex casts ofthe mold. AXELROD-— WESTERN AMERICAN PINES 605 In addition, the relationships of cones may be clarified by associated fascicles and winged seeds. The Little and Critchfield (1969) system of classification followed here divides the genus into three subgenera, four sections, and 15 subsec- tions (Table 2). The distribution of the mode species is well illustrated by Critchfield Sd Little (1966). SUBGENUS STROBUS SECT. STROBUS Subsect. Strobi This subsection includes six American and eight Eurasian species that occur chiefly in rel- atively mesic, mixed-conifer and conifer-hard- wood forests. The close similarity between the American P. monticola-strobus and the Eurasian P. peuce-parviflora-griffithii pines is noteworthy. Future studies of their seed protein composition may indicate the times of separation of these allied species on each continent, and those that are now on separate land areas. Although several species of Subsect. Strobi are on each continent today, there is no clear evi- dence that presently American species were rep- resented earlier id similar species in Eurasia, or rt that Pinus monticola var. monticola (Mirov, 1967: 56; Wolfe & Le 1967: 199) is incorrect. The illustration shows that the cone differs from those of P. monticola in being more robust, is not so slender when closed (in water), and the cone scales are larger and thicker than those of P. monticola. The fossil is more nearly allied to present Asian species, notably P. armandii Franch. of central China. A similar, but more complete, cone of P. itelme- norum is in the Mammoth Mountain flora on the Aldan River (Dorofeev, 1969). Its more mas- sive size and broader, thicker scales show that it is allied to P. armandii and its relatives, not to P. monticola. Pinus anthrarivus Axelrod, sp. nov. TYPE: U. Idaho: Coal Creek. U.C. Mus. Pal., holo dirot 7165, paratypes 7166-69, 7218-20. Figures -5 Needles in 5s, 4-6 cm long, very slender, av- eraging 0.5 mm broad; tips very acute; sheat deciduous. Winged seeds 1.5-2.5 cm long, seed ovate to oval, rounded distally, apex rounded to ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 1986] blunt, wing widest near middle, distal end broad- ly rounde Discussion. The scores of fascicles in the Coal Creek flora with slender needles at first suggested relationship with P. strobus, rather than P. mon- ticola, which has thicker needles. However, the winged seeds differ markedly from the slender wings with acute tips produced by those species. In general shape they are more nearly allied to species of Groupo Ayacahuite (Martinez, 1948), and especially to the seeds of P. ayacahuite var. veitchii Shaw. Pinus anthrarivus thus appears to be an ancient member of Subsect. Strobi, one showing relationship to present Mexican species of the ayacahuite complex. A radiometric date indicates that the pine and associated flora is 29 Ma old, or Late Oligocene. Pinus delmarensis Axelrod, sp. nov. TYPE: U.S.A. California: Del Mar. U.C. Mus. Pal., holo- type 7170. Figure 6. Cone closed owing to immersion in water; cone long-cylindrical; 26 cm long (estim.), 6 cm broad; cone scales thin, 3+ cm wide in middle of cone, flat to broadly concave, no evidence of pro- nounced reflexing; scales apparently with ter- minal umbos. Discussion. This cone from the Del Mar For- mation, a few miles north of San Diego, is as- sociated with sediments that yield both a mol- luscan fauna (Hanna, 1926) and vertebrate fauna (Golz, 1976) of Middle Eocene age, approxi- mately 46 Ma old. The cone is so similar to those of P. lamber- tiana there is no ready basis for separating them. In view of the fragmentary nature and rarity of all presently-known specimens of this alliance, it seems best to tentatively consider that the species is different than P. prelambertiana Ax- that P. delmarensis may have been separate geographically from a co ex that included runners of P. strobiformis Engelm. and P dreas shifted north following the early Miocene The presence of higher relief in the region to AXELROD— WESTERN AMERICAN PINES 607 the east is indicated by coarse conglomerates in the associated section and by the spore-pollen flora (Frederiksen et al., 1983). In particular, the presence of pine pollen allied to the Ponderosae and Strobi groups indicates that terrain in the interior probably was near 1,300-1,500 mm, whereas the coastal strip was covered with a dry tropical flora. Pinus florissantii Lesquereux, Rept. U.S. Geol. rissant, Colorado. bot. Coll. holotype Ma. Figures 8, 9. Pinus T us Lesquereux. MacGinitie, д Wash. Publ. 599: 84, pl. 19, fig. 2 Pinus Simp: т U.S. Geol. E y Paper 1 . 186, pl. 41, figs. 12, 13 Pinus wheljeri Cockerell. MacGinitie, ы р. 85, pl. 18, fig. 11 only, 1953. Pinus leaves, MacGinitie, ibid., pl. 18, fig. 3, 1953. Pinus similis Knowlton was based on two fas- cicles of four needles in the Creede flora. Many additional fascicles have now been collected, and it is evident that most are in 5s, but some are in 4s and, more rarely, in 3s. The last are probably the result of needle loss during ageing or trans- port. The suite of P. similis foliage from Creede is similar to that produced by the modern P. flexilis James, and is therefore placed in synon- ymy under P. florissantii Lesquereux, based on a cone similar to those produced by /lexilis today. Supplementary description. Fascicles mostly in 5s, some in 4s; needles 4.0—5.2 cm long, 1.0 mm wide, stout and mostly curved; margins en- tire, prominent stomatal bands; tip acute; sheath absent, the proximal end forming a rounded (glo- bose) area Discussion. Knowlton made no suggestion as to the identity of the Creede foliage referred here to P. florissantii. Comparison with western members of the Strobi shows that it is most sim- ilar to foliage of P. flexilis James; fascicles of P. monticola are generally not so robust. Pinus flex- ilis is a regular member of the mixed-conifer for- est ofthe central and northern Rocky Mountains. It is discontinuous across the central Great Basin to the east slope ofthe central and southern Sierra — FIGURES 1-7.— . Pinus anthrarivus Axelrod, sp. nov. Coal Creek, Idaho. U.C. Mus. Pal., holotype 7167, paratypes 7165, 7166 (needles), р 7169 (seeds). Late Oligocene, 27 Ма. — 6. Pinus delmarensis Axelrod, sp nov. Del Mar Formation, Californ . Mus. Pal., holotype 7170. Middle Eocene, 48 Ma.—7. Pinus prelam- bertiana Axelrod. Vedder CER р Californi ia. U. C. Mus. Pal., hypotype 7171. Lower Miocene, 23 Ma. ANNALS OF THE MISSOURI BOTANICAL GARDEN (VoL. 73 608 1986] Nevada, ranging southward in the Peninsular Ranges to the Santa Rosa Mountains, southern California. The Florissant flora is well dated at 34 Ma, or earliest Oligocene (Epis & Chapin, 1975: 48). Occurrence. Florissant, Colorado: U.C. Mus. Pal., hypotypes 3727, 3775; Creede, Colorado: C Mus. Pal, hypotype 7214, homeotypes 7212, 7213. Pinus prelambertiana Axelrod, Carnegie Inst. Wash. Publ. 412: 71, pl. 6, fig. 1, 1930. Santa Clara, California. U.C. Mus. Pal., holotype 309. Figure 7. Pinus p Н йч Axelrod, Univ. Calif. Publ. Geol. 4: 127, pl. 18, figs. 7-10, 1958. The figured cone is from the Lower Miocene Vedder Sandstone at Pyramid Hill, northeast of Bakersfield, California. Vedder Sandstone is in the Turritella inezana zone and in the microfossil Zemorrian Stage, both of earliest Miocene age, 25-26 Ma (Turner, 1970) Discussion. The cone is similar to those pro- duced by P. lambertiana Douglas. The record indicates that in the Sierra Nevada to the east, climate was sufficiently moist and cool to support conifers. With ample summer rain and a more equable climate, the forest no doubt occurred at a lower altitude than the modern forest where P. lambertiana is confined now to altitudes above 1,300-1,500 mm. Occurrence. "Verdi, Nevada: U.C. Mus. Pal., hypotypes 1976-1978; Vedder Sandstone, Cal- ifornia: U.C. Mus. Pal., hypotype 7171, ho- meotype 7172. SECT. PARRYA Subsect. Cembroides Of the eight species of nut pine, four are in the United States— P. monophylla Torr. & Fremont, AXELROD— WESTERN AMERICAN PINES 609 P. edulis Engelm., and with P. quadrifolia Parl. and P. cembroides Zucc., ranging into Mexico. Four pinons are confined to Mexico, P. culmin- icola Andres. & Bearman, e Several varieties of P. cembroides have been rec- ognized in recent years (Bailey, 1979, 1983 gives references). These appear to be geographic races of relatively recent, probably Quaternary, origin. The P. cembroides population isolated in the mountains of the Cape Region, Baja California, named P. cembroides var. lagunae Robert-Pas- sini (1981), poses an interesting problem. The J dreas and allied rifts and later spreading along the mid-ocean rise, which opened the Gulf of California. This raises the question as to whether a) there are relict stands of the Cape pinon in the Sierra Madre Occidental masquerading as P. cembroides, or whether 5) the Cape population was on the mainland and became extinct here, or c) whether the pine originated in the Cape Region from a cembroides population that was isolated there as the Cape Region separated from the mainland. Pinons do not intercross with other members of the genus (Mirov, 1967: 334), although pines of other subsections may do so. This implies that Cembroides are ancient, consistent with the fossil record, which shows that piñon was already pres- ent in the Middle Eocene (46 Ma) Pinus ballii Brown, U.S. Geol. Surv. Prof. Paper 185: 53, pl. 8, fig. 5, 1934. TYPE: U.S.A. Colorado: Green River. U.C. Mus. Pal., ho- lotype 1927-20688. Figures 10, 11. Pinus dep Brown. MacGinitie, Univ. Calif. Publ. Geol. i. 83: 91, pl. 25, fig. 2, 1969. A 3-needled fascicle with deciduous sheath (Fig. 10) and an impression of a cone (Fig. 11) rep- —18.—8. Pinus те Lesquereux. Florissant, Colorado. Princeton Univ., holotype 144. Basal red Lesquereux. Creede, Colorado. Uc. Mus. Pal., orado. U. 1934.)— 11. Pinus ballii Brown. enel River, Colora otype. Late Oligoce lee "1927 (20688). Middle Eocene, 46-47 Ma. (From Brown, o. U.C. Mus. Pal., donas [1883] and pe ibo o 9. Pinus florissantii e, 26.5 Ma.—10. Pinus ballii Bro hypotype 20671. Middle Eocene, 46- 47 Ma. (From MacGinitie, 1969.)— 12. Pinus oo Axelrod, sp. nov. Creede, Colorado. Univ. Colorado ME prec d 19701. Late Oligocene, 26.5 rado. Chalk Hills, Шо, К Ма. —16. Рїп s. Pal. anie ain 7239. 7241, 7243. m Oligocene, 26.5 a Cultural Center, “California, hypotype us sanjuanensis Axelrod, sp. nov. Creede, Col- Ma.—17. Pinus Pip ada Knowlton. 11205. Late Miocene, 5-6 M Pinus lindgreniiKnowlton. Chalk. Hills, Idaho. U.S. Nat. Mus., holotype 8179. Late Miocene, 5—6 Ma. (From Knowlton, 1901.) [Vor. 73 ANNALS OF THE MISSOURI BOTANICAL GARDEN 610 1986] resent a piñon in the Middle Eocene Green River flora, dated at 46-47 Ma. Occurrence. Green River, Colorado-Utah: U.C. Mus. Pal., hypotype 20671. Pinus kelloggii Webber, Carnegie Inst. Wash. Publ. 412: 121, pl. 1, figs. 1-3, 1933. Ri- cardo, California. U.C. Mus. Pal., cotypes 156 Fossil wood from the lower part of the Ricardo Formation (12 Ma), Last Chance Gulch, on the south slope of the El Paso Mountains 40 km northeast of Mohave, California, has been iden- tified as that of piñon. The associated flora in- cludes woods of palm, oak, cypress, and locust. Pinus kinnickensis Axelrod, sp. nov. TYPE: U.S.A. California: Tehachapi. U.C. Mus. Pal. ho- lotype 1399. Figure 21a. Pinus lindgrenii Knowlton. Axelrod, Carnegie Inst. ash. Publ. 512: 85, 1939. Seed ovoid in shape with an obtuse apex and rounded base, somewhat asymmetrically swol- len, 1 cm long and 5 mm wide; testa very thin, and in one or two areas the papery nucellus still evident; endosperm with the contained embryo 7 mm long and 3 mm broad; oblong, rounded at one end and tapering at the other Discussion. During collecting of the Tehach- api flora, situated at the southeast end of the Sierra Nevada, a complete pine nut was uncov- ered in the andesite tuff that overlies the rhyolite tuff bed in which the fossil leaves occur. The specimen was identified as P. /indgrenii because at that time it was the only fossil piñon known and the pine nut seemed referable to it. Now that more numerous cones of P. lindgrenii are avail- able (see below), it is evident that that species has much larger seeds than the Tehachapi spec- imen. It therefore represents a different species and is given a new name. The complete specimen cannot now be illus- trated because over the years the tuff in which it occurs has largely disintegrated, leaving only an impression of one side of the seed. Figure 21a AXELROD-— WESTERN AMERICAN PINES 611 shows the specimen in both concave and convex views. Pinus lindgrenii Knowlton, Torreya 1: 113-115, text figs. 1-3, 1901. TYPE: U.S.A. Idaho: Chalk Hills. U.S. Nat. Mus. holotype 8179. Figures 17-21. The silicified cone collected by one of Lind- ren's associates, and described by Knowlton, comes from a site south of Bruneau, southern Idaho. It is from the Chalk Hills Formation (Malde & Powers, 1962) that has yielded diverse mammals that are judged to be Late Miocene (Hemphillian), or about 6-7 Ma. Other fossil plants in the formation include a bracket fungus and woods Pea as fir, oak, alder, poplar, and hickory (Bro 40). Additional cones Mp A by Ruth A. Kirkby add importantly to our understanding of the species and its position in the Subsect. Cem- broides. Her collections come from the north end of Chalk Hills (sec. 19) and west of Highway 51 in sec. 25, T. 7S, R. 4E Supplementary description. Ovuliferous cones ovate to ovoid, base truncate, tip acute to obtuse; 5.0-7.0 cm long and 4.0—6.0 cm broad; cone scales thick, up to 2.5 cm broad in middle of cone, smaller elsewhere, the scales with low, broadly triangular apophyses; umbo centro-par- vi-mucronate. Seeds wingless, very large, some in middle of cone 2.5 cm long, crudely tear- shaped, broader distally, larger than those of pi- non today; shell of seed very thin Discussion. Pinus lindgrenii is generally al- lied to P. cembroides and P. edulis in that they also have low, triangular apophyses. However, their cones are smaller and the seeds are not so large as those of the fossils. Pinus monophylla and P. quadrifolia both differ from P. lindgrenii in that they have more prominent, raised apoph- yses, and also smaller cones and seeds. Cones of P. pinceana of the Sierra Madre Oriental are most similar to the fossil, though P. /indgrenii may represent a species intermediate between P. pin- ceana and a P. cembroides-edulis alliance. E FIGURES 19-24.— 19-21. Pinus lindgrenii Knowlton. Chalk Hills, Idaho. Jurupa Cultural Center, Riverside, Cali fornia eee hachapi, California. U.C. M 23 n Axelrod. Thunder Mountain, Idaho. U.C. Mus. Pal., S d o 11201, 11204. Late Miocene, 5-6 Ma.— 21a. Pinus s. Pal., holotype 1399. Oriented to show concav ve and con kinnickensis Axelrod. Te- vex views. — 22-24. Pinus hypotypes 7186-7188. Middle Eocene, 45 612 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 1986] The question may be raised as to whether the larger, more prominent, upswept apophyses of P. monophylla and P. quadrifolia reflect adap- tation to a drier summer climate in the west. Some indication of this is seen in the western Arizona SOMMER of P. edulis that have larger apophyse E o Chalk Hills, Idaho: Jurupa Mus. Nat. Hist., Riverside, California, hypotypes 11201, 11203-11205, homeotypes 11202-11206. Pinus sanjuanensis Axelrod, sp. nov. TYPE: U.S.A. Colorado: Creede. U. Colo. Mus., holotype 19701; U.C. Mus. Pal., paratypes 7239- 7244, 7247. Figures 12-16 Imprint of near-proximal portion of globular cone; measures 4.0 by 4.5 cm in section; apoph- yses subdued; umbro mucronate, centro in po- sition on scales; terminal ends of scales chiefly 5-sided and asymmetrically so; fascicles in 2s; 1.5-3.2 cm long, individual needles 1 mm wide, slightly curved; tips tapered to a sharply acute tip; base is a subdued, rounded remain of the deciduous sheath. Discussion. The imprint of the cone cannot be separated from those of modern P. edulis cones. Needles in 2s occur at the Dry Creek lo- calities, but were encountered only rarely at other sites. They are also comparable to those of the living P. edulis of the southern Rocky Mountains and northern Mexico. Many of its present as- sociates have allied taxa in the Dry Creek flo- rules, including species of Cercocarpus, Fallugia, Holodiscus, Mahonia, Peraphyllum, Philadel- phus, Quercus, Ribes, Robinia, Sapindus, and others. The Dry Creek area has the best repre- sentation of vegetation of semiarid requirements in the Creede basin consistent with the occur- rence of piñon there. In her report on the Creede flora, Stewart (1940: 146) referred to fossil wood identified as a pine of the Cembroides alliance. The only other pine in the Cordilleran 1 that seems related to this species is P. ballii B rom the Green River flora, represented by a fascicle of three needles and a cone (MacGinitie, AXELROD— WESTERN AMERICAN PINES 613 1969). The 3-needled fascicle presumably is more nearly allied to P. cembroides which now ranges from southern Arizona into Mexico. The apoph- yses of the cone scales are more raised than those of P. sanjuanensis. Subsect. Balfourianae This small group includes P. aristata Engelm. of the central to southern Rocky Mountains, P. longaeva Bailey of the west-central Rocky Moun- tains and Great Basin ranges, and P. balfouriana Grev. & Balf., which is discontinuous from the Klamath Mountain area of northwestern Cali- fornia to the southern Sierra Nevada. Pinus balfouroides Axelrod, Univ. Calif. Publ. Geol. Sci. 121: 209, 1980. TYPE: U.S.A. Ne- vada: Chalk Hills. U.C. Mus. Pal., holotype 8007, paratypes 8009-8013. Figures 22-33. Pinus Lik, rien Lesquereux. Axelrod, ee Calif. Publ. 9: 227, pl. 42, fig. 9, Pinus hele Cockerell Axelrod, т ‚р. 227, pl. 42, Pinus ити ce УРИ за Missouri Bot. Gard. 63: 28, figs. 13-1 Pinus sp. Brown Rr Geol. pud Prof. Paper 186-J: 167, pl. 45, fig. 9, 1937. The above-listed material from the Purple Mountain (13 Ma) and Chalk Hills (12 Ma) flo- ras, western Nevada (Axelrod, 1976, 1962), in- cludes cones, winged seeds, and needles in 5s. They are similar to structures produced by P. balfouriana of high montane regions in Califor- nia. A new locality for the species is in the Middle Miocene Temblor Formation (15 Ma) at Shark- tooth Hill, near Bakersfield, California (Fig. 29). In addition, numerous fascicles and a cone scale of the species are in the Coal Creek flora, Idaho (Fig. 30-33) dated at 29 Ma. species of Abies, Larix, P dendron, ardwoods, including Acer, Betula, Carya, Cor- nus, alus, Sassafras, Sorbus, and ae Zelkova. Furthermore, impressions of cones and needles in 5s that represent P. balfouroides are in coarse, silicified sedimentary rocks at Dewey yp 25-33.—25-27. Pinus ү rig ru Purple Mountain, Nevada. U.C. Mus. Pal., hypotype 7177, 7178, 5501. Late Miocene, us balfouroides Axelrod. Chalk Hills, Nevada. U. C. Mus . Late Miocene, с к (Буш азе ы 1962.)—29. Pinus ике Axelrod. Sharktooth mia. U.C. Mus. Pal., hypotype 7180. Middle Miocene, 15 Ma.—3 Pinus balfouroides Axelrod. Coal Creek. Idaho. U.C. Mus. Pal. , hypotypes 7182-7185. Late Oligocene, 27 Ma. ANNALS OF THE MISSOURI BOTANICAL GARDEN 34 [Vor. 73 1986] Mine, Thunder Mountain, central Idaho (Figs. 22-24). Now at an altitude of 2,300 m, the fossils are in the Dewey Beds, a volcaniclastic subunit eroded from Sunnyside Rhyolite of the Challis Volcanics, and dated at ca. 46 Ma (Leonard & Marvin, 1984). The associated flora (Brown, 1937) is largely coniferous. It includes cones and foliage of Abies, Larix, Picea, Pinus, and Se- blage suggests a high montane environment, probably near 1,300 m Additional specimens from the Purple Moun- tain flora are also illustrated (Figs. 25-27). These occur with mixed conifer forest taxa, notably species of Abies, Picea, Pinus, а se Pseudotsuga, and S\ tliv broadleaved sclerophyll vegetation poe species of Arbutus, Cercocarpus, Chrysolepis, Heteromeles, and Quercus. Occurrence. Purple Mountain, Nevada: U.C. Mus. Pal., hypotypes 5501-5502, 7177-7179; Sharktooth Hill, California: U.C. Mus. Pal., hy- potype 7180; Coal Creek, Idaho: hypotypes 7182- 7185; Thunder Mountain, Idaho: U.C. Mus. Pal., hypotypes 7186-7188; U.S. Nat. Mus. hypotype (unnumbered) Pinus crossii Knowlton, U.S. Geol. Surv. Prof. Paper 131-G: 185, pl. 41, figs. 3, 8-10, 1923. Creede, Colorado. U.S. Nat. Mus., holotype 36514, paratypes 36511-36513. Figures 34— 42 Pinus crossii Knowlton. Axelrod, Univ. Calif. Publ. Geol. Sci. 59: 62, pl. 7, figs. 6-9, 1966. Bailey, Ann. Missouri Bot. Gard. 57: figs. 34-37, 1970. Pinus aristata crossii Cockerell, Nature 133: 573, fig. 1, 1933. Supplementary description. Cones long ovate, 8 to 10 cm long and 4.5 cm broad (closed in water); cone scales with long, sharp, upcurved prickles; needles in 5s, 1.1-3.0 cm long, with a conspicuous single groove present in the distal part; the needles on branchlets are crowded and bunched to give a fox-tail appearance; seed wings open, 1.2-2.0 cm long, the wing generally long oval, distal end rounded, 6-7 cm broad. AXELROD—WESTERN AMERICAN PINES 615 Discussion. The short, curved fascicles with needles in 5s that Knowlton named Pinus crossii are supplemented now by large suites of fascicles, branchlets with clusters of needles, winged seeds, and impressions of pine cones. All of this ma- terial is allied to P. aristata Engelm. of the south- ern Rocky Mountains in Colorado and northern New Mexico. However, there are sufficient dif- ferences to indicate that P. crossii probably is ancestral to the living bristlecone pine. The fossil cones are generally more robust and broader (as measured when closed in water) and also average larger, with most reaching 7.5 to 9.5 cm long and up to 4.5 cm broad. Furthermore, the fragile prickles, where well preserved, evidently were longer, for some are up to 8 mm long. Another difference is in the seed wing, which is broader and the seed, which is larger, than in most pop- ulations of P. aristata today. The differences may be attributed to a more genial climate at the close of the Oligocene, with a longer growing season, more moisture, an milder temperature. It is noted that the abundant remains of P. crossii are associated with wood- land taxa, notably species of Chamaebatiaria, Cercocarpus, Fallugia, Juniperus, Peraphyllum, aho d now largely to drier, rocky sites at subalpine levels, reaching down on drier sites into the middle of the mixed conifer belt, well removed from pinon- juniper woodland vegetation today. Occurrence. Creede, Colorado: U.C. Mus. Pal., hypotypes 7190-7198, homeotype 7246; U. Colo., hypotypes 18654, 19694, 19698, homeo- type 19701; Copper Basin, Nevada: U.C. Mus. Pal., hypotypes 8873-8876, 8901, 8902; Titus Canyon Formation, Death Valley, California: U.C. Mus. Pal., hypotype 7189. SUBGENUS PINUS SECT. TERNATAE Subsect. Leiophyllae Pinus coloradensis Knowlton, U.S. Geol. Surv. Prof. Paper 131-G: 186, pl. 41, fig. 6. 1923. TYPE: U.S.A. Colorado: Creede. U.S. Nat. RES 34-42. ub 7196. 7198. Late Oligocene, 26. v^ — 34-40. Pinus crossii Knowlton. Creede, Colorado. U.C. ine Pal., hypotypes 7190, 7191, Ma.—41, 42. Pinus crossii Knowlton. Titus Canyon, Death ot California. U.C. Mus. Pal., hypotype 7189. Figure 42 is a latex cast of Figure 41. Early Oligocene, са. 37 М 616 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 1986] Mus., holotype 36506. Figures 43-45, 47- 53 The type specimen was incorrectly illustrated. The cone scales are thin, not swollen as the re- touched figure suggests. The following supple- mentary description is based on the type and an additional cone in the Univ. Colorado Museum, as well as foliage and winged seeds presumed to represent this species. Supplementary description. Cones small, oval to elliptic in outline, truncate proximally, broad- ly obtuse distally, 3.0-4.5 cm long, 2.0-3.0 cm broad (open) cone scales thin, tips broadly rounded, slightly widened distally, and some- what reflexed when open; umbo dorsal, centro erecto-mucronate. Needles in 5s, quite slender, 4—5 cm long, 0.5-1.0 mm broad, tapered m to acute tip, base with small, rounded (decid ous) sheath. Winged seeds 2.0-2.1 cm long, del wing narrow, 5—6 mm broad, elliptic, distal and rounded to blunt; seed 3-4 mm long, ovate, acute tip, rounded proximally. Discussion. Knowlton (1923) suggested a general similarity between P. coloradensis and cones of P. arizonica Engelm. However, P. ari- ^ ep cones are larger, the cone scales are broad- and the cone is overall more massive. Closer EM sonsbip appears to be with P. chihuahuana Engelm. (= P. leiophylla var. chihuahuana), which ranges from southern Arizona to central Mexico, though needles of the fossil are somewhat short- er. Its occurrence at Creede is consistent with a number of Madrean taxa in the flora, including species of Arbutus, Cercocarpus, Fallugia, Ma- honia, and Populus that range from Arizona well south into e eo Durango. Pinus chi- huahuana is a common member of the middle and upper woodland bur. as well as the mixed- conifer forest zone, relations consistent with the nature of vegetation in the Creede flora. Occurrence. Ч. Colo. Mus., hypotype 27381; U.C. Mus. Pal., hypotypes 7199-7202 Pinus paucisquamosa Templeton, South. Calif. Acad. Sci. Bull. 52: 64—66, figs. 1, 2, 3b, AXELROD-— WESTERN AMERICAN PINES 617 1953. TYPE: U.S.A. California: Altamira Shale, Nat. Hist. Mus. Los Angeles Co., ho- lotype 1400. Figure 46. This small, nearly complete cone is from the upper Altamira Shale at Point Fermin, San Pe- dro, California. It occurs in the Bolivina mode- loensis foraminiferal subzone, of early Mohnian age, or about 14.5 Ma. (Woodring et al., 1946; Turner, 1970). It was initially compared by Tem- pleton (1953) with Pinus chihuahuana, but Mi- rov (1967: 35) suggested that it may be a species of Sect. Insignes, described and illustrated by Shaw (1914). Insignes are now considered a het- erogeneous group, comprising species of Syl- vestres (P. halepensis, P. pinaster), Australes (P. rigida, P. serotina, P. pungens), Contortae (P. virginiana, P. clausa, P. banksiana, P. contorta), and Oocarpae epi species, four of each in Cal- ifornia and Mexico). Comparison with cones of these species E little evidence of relation- ship with P. paucisquamosa. As suggested by Templeton, the fossil resem- bles the smaller cones of P. chihuahuana (= P. leiophylla Scheide & Deppe var. chihuahuana), which ranges from southern Arizona-adjacent New Mexico southward into central Mexico. However, in its broader cone scales the fossil shows greater relationship with cones of P. /um- holtzii Robinson & Fernald, distributed from So- nora southward into Jalisco. Pinus paucisqua- mosa probably is an extinct member of the Leiophyllae This pine occurs in the marine Altamira Shale, which includes glaucophane schist debris similar to that in the Poway Formation. Both sites de- rived this debris from the landmass Catalinia, which then stood near the present shore (Reed, 1951: 170-171; Woodford et al., 1954: 71, 74; Woodring et al., 1946). Since the basement ter- rane west of the Inglewood-Newport fault, where the fossil occurs, differs significantly from that to the east, considerable displacement is implied. This most probably was from Mexico, because Miocene rocks similar to those of the Poway oc- cur south of Ensenada and also on South Co- ronado Island (Stuart, 1974). The relationships <— FIGURES 43-53. 27381 test eng due 26.5 Ma. Histo s. coloradensi. ME о Colorado. U.C. Mus. Pal., s 7201. Latest Oligocene, 26.5 M —43. Pinus coloradensis Knowlton. Creede, Colorado. U.S. Nat. Mus., Ma. (Previously figured by Knowlton, 1923.)— 44. Pinus coloradensis Kno wlton. A latex . Pinus coloradensis Knowlton. Creede, Colorado. Univ. Colo. Mus., hypotype s Angeles Co., cast of ‘holotype A4432/PB-1400. Middle Miocene, 15 Ma. holotype 365061. hypotypes 7273, 7202, 7376, 7278, 7200, 7374, ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 618 1986] suggest that the Miocene fossil site was then op- posite western Sonora where members of Sub- sect. Leiophyllae occur today, and probably did in the Middle Miocene (ca. 15 Ma). SECT. PINUS Subsect. Sylvestres This alliance has 19 living species of which 11 are Asiatic, six European, and two American. The last include P. resinosa Aiton of the north- eastern United States and adjacent Canada, and P. tropicalis Morelet of Cuba and Isle of Pines. Several present European species reportedly have allied fossils in that region (see Mirov, 1967; Gaussen, 1960), but taxa of the alliance in North America are now known as fossil from only the Middle Eocene Allenby Formation (46 Ma) near Princeton, British Columbia (Stockey, 1984). Stockey analyzed the internal structure of cones and needles preserved in chert and showed that three taxa have the basic characters of Sylvestres. These include cones of P. princetonensis Stockey and P. arnoldii Miller and needles of P. allisonii Stockey. Whether these are to be grouped into a single species (P. arnoldii has priority) was left open by Stockey because the fossils are from dif- ferent sites. Itis desirable to note that P. clementsii Chaney (1954) from the Late Cretaceous (85-86 Ma) of southern Minnesota was compared with P. re- sinosa Aiton. Examination of the mold that rep- resents the type specimen (Chaney, 1954, figs. 1, 2) and a cast of it (Chaney, 1954, fig. 3) shows that th d and essentially con- vex. The structure of the umbo is largely erased, either by erosion or by weathering. Actually, there is insufficient detailed structure preserved to def- initely relate the pine to a subsection. Further- more, the paratype, which is from a locality some 25 miles distant, is only a small piece of a cone imprint. It certainly represents a different species as judged from the size and shape of the apoph- yses and the features of the umbo. It may rep- resent a species of Sylvestres but is too incom- AXELROD-— WESTERN AMERICAN PINES 619 plete for this to be certain. That either specimen is a pine can be disputed because their internal structure is indeterminate; both may be Pityo- strobus. From an ecological standpoint, the large di- cotyledon leaves in the associated fossil flora in- dicate a warm temperate to subtropical climate (Chaney, 1954). If relationship with Pinus resi- nosa is accepted, it would be necessary to recon- struct high relief (1,000-1,500 m) in the region to provide a cold temperate climate for the pine. This would involve miles of transport via river, and, in view of the coarse grit and conglomerate in the section, the cones should be well worn, yet this is not the case. The evidence suggests that the Pinus (or Pityostrobus) species from southern Minnesota probably were members of the low- land flora, living on well-drained, drier south slopes not far from the areas of plant accumu- lation. Occurrence. Minnesota, Cretaceous: Univ. Minnesota, Dept. Botany Paleobot. Coll., holo- type no. C 770 (from Ochs Clay pit near Spring- field), paratype no. C 711 (from near New Ulm, 25 miles east). Subsect. Australes This group of 11 species occurs in the south- eastern United States, the Bahamas, Cuba, His- panola, and Central America (Critchfield & Lit- tle, 1966, map 18). Fossil cones reportedly allied to P. taeda and P. rigida recorded from the Mio- cene of western Europe (in Gaussen, 1960) need to be reexamined before they can be accepted as valid records of Australes in Europe. One fossil species is now known for the group in the south- eastern states. Pinus collinsii Berry, Torreya 36: 124-127, text- fig. 2, 1936. TYPE: U.S.A. Maryland: Calvert Formation. U.S. Nat. Mus., holotype un- numbered and missing. Figures 57-59. КСЫ” ae Berry. Berry, Wash. Acad. Sci. Jour. 31: —508, fi gs. 1, 2, 1941. St. Marys Formation, PE we < FIGUR Ocene, ca. rado. U.C. reede, Colorado. uie collinsii Berry. Calvert Formation, Stratford Cliffs, Virginia. U.S. Nat. Mus., hypotype. )— Pinus collinsii Berry. Calvert Formation, near Plum Point, Marylan Ma. (From Berry, 1941.)—58. ES 54—59.— 54. Pd lynnii pedo Miller. Aquia Formation, Belvedere Beach, gon ale a. (From Berry, 1 s. Pal., bie d 7203. Late Oligocene, 265M Co lo. Mus., holotype 19703. Late Oligocene, 26.5 Ma. - —55. Pinus engelmannoides, a.— 56. Pinus macginites, —57. Pin Middle Miocene, 15 5 . Nat Mus., hypotype. Middle Miocene, 15 Ma. (From Berry, 1941.)—59. Pinus collinsii Berry. Calvert Formation, near Plum Point, Maryland. U.S. Nat. Mus., holotype. Middle Miocene, 15 Ma. (From Berry, 1936.) 620 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 63 р А y re 4 чї $< £ Ж x Лу (t ^r т, 1 M АА E ! le eo Ao 1986] Cones recovered from the Middle Miocene Calvert and St. Marys Formations, Maryland, are similar to those of the living Pinus taeda, a widely-distributed species along the Atlantic and Gulf coastal plain from Delaware into eastern Texas. Associates of P. collinsi in the Calvert flora (Berry, 1916) include Berchemia, Cassia, Nyssa, Platanus, Quercus, Salix, and Ulmus that occur with P. taeda today. Occurrence. St. Marys Formation, Mary- land. U.S. Nat. Mus., specimen missing from collection and unnumbered. Pityostrobus — (Berry) Miller, Torrey Bot. Club Bull. 104: 5-9, figs. 1-7, 1977. Pinus lynnii Berry, id Acad. Sci. Jour. 24: 182- 183, fig. 1, 1934. Type: U.S.A. Virginia: Aquia Formation. U.S. Nat. Mus., holotype 22786. The well-preserved cone described by Berry (1934), from the Aquia Formation, is now con- sidered Paleocene in age (in Miller, 1977b). It has the external features of a species of Australes and resembles cones of P. taeda as well as those of P. elliottii Engelm. and P. caribaea Morelet, although the last two are on average larger than the fossil. The internal structure of an additional cone studied by Miller (1977b) shows that it repre- sents the extinct genus Pityostrobus. The bract- trace is free, and the scale has a thin, not a thick, vascular strand. Apart from these differences, in its general structure and shape it certainly seems to foreshadow pines of Subsect. Australes and may well have given rise to a species that pro- duced cones similar to those of P. taeda. Occurrence. Aquia Formation, Virginia: U.S. Nat. Mus., hypotype 201950. Subsect. Ponderosae This large group of living pines includes 13 species, most of which are in Mexico. A few fossil cones are known that resemble those of different species of the alliance. Identification of fossil winged seeds and fascicles poses a major prob- lem, because those of many modern species are AXELROD— WESTERN AMERICAN PINES 621 so similar that their reference to any one fossil species seems doubtful. In spite of the uncer- tainty, there are numerous records of large seed wings (without seeds) that appear to represent those similar to the western P. ponderosa Law- son. These have been recorded at several local- ities in the Rocky Mountains, including the Green River (46 Ma), Florissant (34 Ma), Ruby and Beaverhead (ca. 30 Ma), and Creede (26.5 Ma) floras. In the Great Basin, there are similar rec- ords in the Fallon (12 Ma), Middlegate (18 Ma) and Fingerrock (16 Ma) floras. The following are records of fossil cones that appear to represent members of this subsection. Pinus engelmannoides Axelrod, sp. nov. TYPE: U.S.A. Colorado: Creede. U.C. Mus. Pal., holotype 7203a, b. Figure 55. Cone closed (in water), narrowly elliptic; 11- 12 cm long, 4.5 cm broad proximally; broadly rounded or blunt proximally, apex acute; scales with prominent apophyses, asymmetrically tri- angular and upcurved; umbo apparently centro- reflexo-mucronate. Discussion. This cone from the Creede flora resembles those produced by P. engelmannii Carr. in the Santa Rita and Chiricahua mountains, southeastern Arizona. It descends from the mixed-conifer forest zone to mingle with the middle to upper part of the oak woodland belt, an association similar to the vegetation setting at Creede. Among its associates at altitudes near 1,830 m (6,000 ft.) are Pinus leiophylla, P. pon- derosa, Pseudotsuga glauca, Juniperus dep- peana, Arbutus arizona, Cercocarpus breviflorus, and numerous oaks. From southern Arizona, Pi- nus engelmannii ranges southward in the Sierra Madre Occidental to Zacatecas, central Mexico. Pinus macginitieii Axelrod, sp. nov. MacGinitie, Carnegie Inst. Wash. Publ. 599: 84, pl. 20, figs. 1, 3, 4, 1959. Type: U.S.A. Colorado: Florissant. U.C. Mus. Pal., lectotype 3776, paratype 3778; U.S. Nat. Mus., paratype 33758. Figures 56, 60-62. — FIGURES 60—67.—60. Pinus macginitieii doge, sp. nov. Florissant, Colorado. U.C. Mus. Pal., lectotype 3776.—61, 62. Pinus macginitieii Axelrod, s Nat. Mus., paratype 33758.—63-67. Pin holotype Sta., U.C. Mus. Pal., hypotypes 4000, 4001. 75 inus е Axelrod, sp. nov. Cr 7204, paratypes 7204-7208; Fallon Nevada, U.C. Mus. Pal., hypotypes 2034, 2035; Nevada, Aldric h v. Florissant, Colorado. U.C. Mus. Pal., paratype 3778, U.S. eede, Colorado. U.C. Mus. Pal., { АЕ: * sa Rae © y М. ыл A + >; uw е ` £g í M wu e^ n. 4 " 4 я 4 2 ue Eea Z. ul Q ра < O а < Ы Z < E О a 5 > О Ф 2 = ш I E LL © M < Z 7. < 1986] Pinus Логіѕѕапіі Lesquereux. Becker, Palaeontograph- 7-B: 61, pl. 6, figs. 3, 5, 6, 9 (not figs. 1-2, 4, 7-8, 10, which may be a different species), 1969. The Florissant pine represented by long needles in 3s with a long persistent sheath, together with u^ seed wings is л of the Ponderosae al- nce. Similar seeds the Beaverhead flora E 1969) ev idently represent the same species as does a single large seed wing in the Creede flora. Needles in 3s, 17.5 cm or more long, with a large persistent sheath, needles grooved, 1-2 mm cally acute; large seeds, ovate, 8-9 mm long and 4- Discussion. The above listed fossils do not represent P. florissanti because the type speci- men is a cone similar to those of P. flexilis. It has needles in 5s, and the seed wings are attached to the scales. Pinus macginitieii is allied to species of Subsect. Ponderosae, but not to P. ponderosa, which has smaller winged seeds. Closer compar- ison may be made with P. michoacana Martinez, a species now in Mexico from Jalisco southward into Chiapas. The relation of the seeds to the large cones of P. riogrande in the Creede flora, which is compared with the living P. montezu- mae Lamb., is uncertain; they may be allied. l ed for D a tiary floras stand as excellent contributions to Tertiary paleobotany. Occurrence. Beaverhead, Montana: N.Y. Bot. Gard., hypotypes 146a-b, 147, 144, 916; Creede, Colorado: Univ. Colorado Mus., hypotype 19703. Pinus ponderosoides Axelrod, sp. nov. TYPE: U.S.A. Colorado: Creede. U.C. Mus. Pal., holotype 7204, paratypes 7206-7208. Fig- ures 63-67. Pinus crai Axelrod, Univ. Calif. Publ. Geol. Sci. 76, pl. 4, figs. 19-20; pl. 17, figs. 10-11, 1956. Cone fragment 7 cm long, scales 1.3-1.4 cm broad, 7-8 mm high, umbo centro mucronate, prickle upturned, sharp. Needles 8 cm or more AXELROD-— WESTERN AMERICAN PINES 623 long, fascicles in 3s with large persistent sheath, iid г e 1-2 mm broad, tips acute; seed ings 2.0-1.5 cm long, wing 6 mm broad, long elliptic, tip diee rounded to bluntly acute; seed ovate, 4-5 mm long, 3-4 mm broad, dehiscent. Discussion. The specimens of P. macginitieii in the Florissant flora cited above demonstrate that Subsect. Ponderosae was already in exis- tence in the Eo-Oligocene transition. An older record of Subsect. Ponderosae is that of P. pre- murrayana Knowlton from the Absaroka Vol- canics, of Middle Eocene age as noted below. Specimens in the Creede flora referred to P. pon- derosoides seem more similar to structures pro- duced by the P. scopulorum of the Rocky Moun- tains. The relation of this species to other fossils— chiefly winged seeds— that have been compared with P. ponderosa is uncertain because most of them are difficult (or impossible) to separate from other members of the Ponderosae. Occurrence. Fallon, Nevada: U.C. Mus. Pal., hypotypes 2034, 2035; Aldrich Station, Nevada: U.C. Mus. Pal., hypotypes 4000, 4001 Pinus premurrayana Knowlton, U.S. Geol. Surv. tion. U.S. Nat. Mus., E 222760. Fig- ure 71a-b. This silicified cast of a cone recovered from the east side of Yellowstone Lake was compared by Knowlton with P. murrayana. It shows no relationship to cones of that species, either in size, shape, or the nature of the cone scales. The cone, refigured here in two views, appears to be a member of Subsect. Ponderosae. It resembles the slender cones of P. /indleyii Loudon (see Shaw, 1914, pl. 25, fig. 223), considered by some to be a variety of P. montezumae Lamb. It is also as- signed to Ponderosae because woods represent- ing two p s fossil pine from the Yellpw- stone fossil fore in the hard-pine group (Read, 1930), « of llic the Ponderosae are a part. In addition, needles of P. iddingsii Knowlton from the Yellowstone <— FIGURES 68-71.— 68. Pinus riogrande Axelrod, sp. nov. Creede, Colorado. U. Colo. Mus., holotype 19704. Late Oligocene, 26.5 Ma.—69, 70. Pinus riogrande Axelrod, sp. nov. Creede, Colorado. U. Colo. Mus., paratypes 7209, 7210. Late Oligocene, 26.5 Ma.—71a, b. Pin nus premurrayana Knowlton. Langford Formation, East of Yellowstone Lake, Wyoming. U.S. Nat. Mus., holotype 222,760. Two views of specimen. Early Middle Eocene, 50 Ma. 624 flora (Knowlton, 1899: 680, pl. 82, figs. 8, 9) are in 3s, are 13+ cm long, are rounded and flat on one side and channeled on the other. They are much like needles of Ponderosae today. The precise site at which P. premurrayana was collected is unknown. However, discussion with J. Richmond of the U.S. Geological Survey and R. Baker, University of Iowa, both of whom have been actively engaged in the geology of the area, indicates that the fossil cone is from the Absa- roka Volcanics. These contain much silicified wood on the east side of the Yellowstone Lake, whereas the Pleistocene (Sangamon) sedimen- tary rocks there have plant remains but they are not silicified. The volcanic rocks on the east side of the Yellowstone Lake represent the Langford Formation of the Absaroka Volcanic Super- group, and are K/Ar dated at 50 Ma, or early Middle Eocene (in Smedes & Protska, 1972). Pinus riogrande Axelrod, sp. nov. TYPE: U.S.A. Colorado: Creede. Univ. Colo. Mus., holo- type 19704; U.C. Mus. Pal., paratypes 7209- 7211. Figures 68-70. Cones large, long elliptic to elliptic-ovate; complete specimen 12.3 cm long, 4.5 cm broad in middle; largest broken cone over 15 cm long and 6 cm broad; bluntly rounded apex and base; 9 x 11 rows of cone scales, the middle ones 1.5 cm broad, 1 cm high; umbo dorsal, centro-re- flexed, short mucronate, prickles 2-3 mm long, directed distally. Discussion. These large cones resemble those of the living P. montezumae Lambert, distrib- uted now from Durango and Nuevo Leon south- ward into Guatemala. The modern species varies in cone and needle size with increasing altitude. As documented by Shaw (1914) and Martinez (1948), these structures decrease in size with al- titude and have been described as a series of varieties. The fossils resemble cones from local- ities at moderate elevations, consistent with the ecology indicated by the Creede flora in which they occur. Pinus truckeensis Axelrod, sp. nov. U.S.A. Ne- vada: Celetom Quarry. U.C. Mus. Pal., ho- lotype 7223, paratype 7224. Figures 72p, a & 73 Cone broadly elliptic in outline; base truncate, apex obtuse; 15 by 11 cm, with proximal end deciduous on branch, so cone probably up to 17 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 or em long; adi x ]4rows of cone scales; those n side 5 аге 1.72.5 cm broad, on asymmetrical el side (1 cone only) 2 cm broad; apophyses broadly rhombohedral, symmetrical on one side (Fig. 72p), asymmetrical on the other and upswept (Fig. 72a); centro-mucronate and erect or re- flexed; prickle not preserved. Discussion. Two specimens are in the collec- tion. The strongly asymmetrical cone was pre- served as a mold diatomite. By gradually filling the cavity with latex, a cast of the entire cone was removed to display both sides (Fig. 72p, a). The other cone (Fig. 73) shows the complete face on the symmetrical (posterior) side. On its lower margin, evidence of asymmetry of the anterior side is evident. Both specimens are referred to one species. The fossils resemble cones produced by P. pseudostrobus Lindl. The chief difference is that the fossil cones are somewhat larger than those in the collection available for comparison at the Institute of Forest Genetics, Placerville. In ad- dition, the apophyses are proportionately larger. Otherwise, there appear to be no significant dif- ferences between these species. Pinus pseudo- strobus ranges from the mountains of central Mexico to Chiapas and Guatemala, inhabiting the conifer forest and the oak-madrone wood- land at altitudes from 2,300 to 3,200 m. he occurrence of a pine of southern affinity in western Nevada is not unique for the area. Other taxa in the Miocene record there that have their nearest allies in the Southwest, or in adja- cent Mexico, include species of Arbutus, Bume- lia, Cercocarpus, Fraxinus, Mahonia, Populus, Quercus, and Sapindus, and they are more nu- merous in the Miocene of southeastern Califor- a. The fossil cones occur in diatomite of the Coal Valley Formation, exposed at the Celetom Quar- ry of the Eagle-Picher Industries, 30 km east of Reno. е deposit i is about 12.5 Ma old as judged from tric dates of closely associated rocks nearby. Subsect. Sabinianae There is only a very limited record of this group of large-coned pines represented now by three living species. Pinus sabiniana Doug. inhabits the foothills of the inner Coast Ranges of central and northern California and the lower west slopes 1986] AXELROD— WESTERN AMERICAN PINES 625 72. Pinus truckeensis Axelrod, sp. nov. Celetom Quarry, 30 km E. of Reno, Nevada. U.C. Mus. Pal., pones 7223. Two views of specimen: p, left or posterior side; a, right or anterior side. Late Miocene, 12 Ma. of the Sierra Nevada. Pinus coulteri D. Don oc- curs chiefly in the mountains of southern Cali- fornia at medium altitudes, ranging discontin- uously northward in the Coast Ranges to Mount Diablo east of San Francisco Bay. Pinus torrey- ana Parry is a narrow endemic on the coast near La Jolla, and a subspecies occurs on Santa Rosa Island west of Santa Barbara (Haller, 1986). The origin of the group is obscure but may lie with P. oaxacana Mirov of Mexico (Haller, 1966). It has similar cones as well as long needles in 5s. From this standpoint, there was an increase in cone size (P. sabiniana, P. coulteri) and the de- velopment of more armed cones with hooked apophyses. Pinus oaxacana occurs now in the Sierra Madre Occidental from Durango south- ward into Nayrit. In the early Neogene, forerun- ners of P. torreyana, as well as P. sabiniana and P. coulteri, may have separated from an ancestral oaxacana-like group and were then transported northward as the San Andreas fault system was activated (Crowell, 1979). Mason (1927) identified specimens from the Bridge Creek and Cove Creek floras, Oregon, as P. torreyana Parry. However, these 5-needled fascicles with long, persistent sheaths are also similar to those of several Ponderosae taxa now in Mexico, as well as others (see Martinez, 1948). Pinus hazenii Axelrod, Carnegie Inst. Wash. Publ. 476: 165, pl. 2, fig. 4, 1937. TYPE: U.S.A. California: Mount Eden. U.C. Mus. Pal., ho- lotype 958, paratypes 7225, 7226. Figures 5, 76 Very large cone scales, with large, hooked apophyses sharply constricted above suggest that this pine may be allied to P. coulteri. Additional material is needed to ascertain more definitively its relationship to that species. P. coulteri produces one of the largest, most massive cones of living (and probably fossil) pines. Pinus pieperi Dorf, Carnegie Inst. Wash. Publ. 412: 69, pl. 5, figs. 7-10, 1930. TYPE: U.S.A. 626 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 1986] California: Pico Formation. U.C. Mus. Pal., holotype 305, paratype 304. Figure 74. Pinus e Dorf. Axelrod, Carnegie Inst. Wash. Publ. 476: 165, pl. 2, figs. 2, 3, 1937. Axelrod, пы е Inst. Wash. Publ. 590: 144, pl. 2, fig. 1, 1950. Axelrod & DeMéré, а Diego Mus. Nat. Hist. Trans. 20: 292, fig. E, The above records of fossil digger pine are from the Pliocene and Late Miocene of southern California. Pinus sabiniana does not occur in that area today except at its northern margin in Santa Ynez Valley north of Santa Barbara and at the north end of Liebre Mountain, southeast of Tejon Pass. The species is recorded also in the Pleistocene of southern California at Seacliff (Ax- elrod, 1983), Carpinteria (Chaney & Mason, 1933), and Lake Canyon east of Ventura (Wig- gins, 1951). This species may have disappeared from southern California during the post-glacial Xe- rothermic period, which brought interior and semidesert species to the coastal strip (Axelrod, 1966: 42-55). Occurrence. Anaverde Formation, Califor- nia: U.C. Mus. Pal., hypotype 3269; Mt. Eden Formation, California: U.C. Mus. Pal, hypotypes 959, 960, 7227; Diego Formation, California: San Diego Mus. Nat. Hist., hypotype 25168. Subsect. Contortae This alliance includes four living species. Two are in the eastern United States, P. clausa (Chap.) Vasey of Florida and southern Alabama, and Pinus virginiana Miller ofthe Appalachian prov- ince. Pinus banksiana Lamb. ranges from the Lake States northward across Canada to the MacKenzie River and east to Nova Scotia. Pinus contorta Dougl. (with three vars.) occupies the Mountains north of southern Colorado (var. latifolia), extends down the Cascade-Sierra evada axis into northern Baja California (var. murrayana), and has an outer Coast Range oc- currence from coastal northern California into Alaska (var. contorta). Pinus alvordensis Axelrod, Carnegie Inst. Wash. Publ. 553: 251, pl. 42, fig. 4, 1944. TYPE: AXELROD— WESTERN AMERICAN PINES 627 U.S.A. Oregon: Alvord Creek. U.C. Mus. Pal., holotype 2095. Figures 77-84. Supplementary description. Cone ovate, markedly asymmetrical and reflexed, rounded proximally, with a blunt apex; 3 cm long and 2 cm broad; peduncle short and thick, about 4 mm broad, 3 mm long (broken); 6 rows of cone scales, with sharply triangular, attenuated apophyses, slightly curved. Needles in 2s, 2.7-3.0 cm long, ] mm broad, acute tips, base with a rounded sheath, 2-3 mm long, 2.0-2.5 mm broad. Winged seeds 1.5-2.0 cm long, 0.4—0.6 mm broad, wings slender, markedly acute distally; seed ovate, 5— 6 mm long, acute distally. Discussion. The cone from the Bull Run flora, northern Nevada, falls readily within the Sub- sect. C ortortae. The tad small d бы the em- ber of the group, ai shi in the е of e a single specimen this cannot be demonstrated. The age of the deposit at this site (Loc. P 572-5) is 38 Ma, or Late Eocene. The associated flora rep- resents a pure montane conifer forest of Abies (3 spp.), Picea (3 spp.), Pinus (2 spp.), Larix, Tsuga, and Chamaecyparis as dominants, with rare small leaves of forest shrubs, notably species of Cra- taegus, Mahonia, Prunus, and Vacciniu Slender winged seeds in the Creede бога. Со1- orado, are sufficiently similar to the Alvord Creek specimen to be referred to it. The Creede spec- imens are from 1.5 to 2.0 cm long, wings nar- rowly elongate, acute to rounded distally, wing 4—5 mm broad, seed proper about one fourth size of wing; tip acute; seed permanently attached to wing. The needles in 2s are broadly similar to those of P. sanjuanensis Axelrod from the Creede flora. The latter can be distinguished because the fas- cicle sheath is smaller as compared with those of P. alvordensis. here is no clear evidence that fossils from these sites are allied to any one of the modern varieties of P. contorta, although this may be modified as larger collections become available. Occurrence. Creede, Colorado: U. Colo. Mus., hypotypes 19706, 19707; hypotype 19709; U.C. Mus. Pal. hypotypes 7174-7175, 7237, <— URE 73. Pinus truckeensis о sp. nov. Celetom Quarry, 30 km E. of Reno, Nevada. U.C. Mus. Pal., e 2224. Late Miocene, 13 628 ANNALS OF THE MISSOURI BOTANICAL GARDEN 1986] homeotypes 7228-7230, 7235-7238, 7176, 7245; Bull Run, Nevada: U.C. Mus. Pal., hypotypes 7231-7232 Subsect. Oocarpae This alliance now occurs in California, south- west Oregon, and in Mexico. The California taxa, P. muricata D. Don, P. radiata D. Don, and P. remorata Mason form discontinuous popula- tions along the central and southern California coast, whereas P. attenuata Lemmon has a scat- tered distribution in the interior. Those in Mex- ico, P. greggii Engelm., P. oocarpa Schiede, P. patula Schiede, and P. pringlei Shaw, are in the Sierra Madre Occidental and Oriental, and with P. oocarpa ranging south into Nicaragua (see maps in Critchfield and Little, 1966). Pinus burtii Miller, Torrey Bot. Club Bull. 105: 93-97, figs. 1-9, 1978. TYPE: U.S.A. Mas- sachusetts: Martha’s Vineyard. U.S. Nat. Mus., holotype 222868. Figure 85. This large cone comes from a Miocene green- sand associated with vertebrate and invertebrate fossils at the north end of cliffs at Gay Head, Martha’s Vineyard, Massachusetts (Miller, 1978). The mammals are of Early Miocene (Late Ari- kareean) age, or about 22 Ma according to D. Savage (pers. comm., 1984). Based on its external and internal structure, group in its long, cylindrical outline and large swollen apophyses. It differs from cones of the P. radiata populations in California, which are largely asymmetrical; P. oocarpa has smaller, more nearly ovate cones; P. attenuata has strong- ly hooked apophyses and is asymmetrical; and the Mexican species P. patula, P. greggii, and P. pringleii are more slender and tend to be asym- metrical, and the apophyses are subdued. There may be a relationship between P. burtii and P. O’Donnellii Teixeira from the Miocene of Lisbon (see Gaussen, 1960, fig. 365: 1). If so, AXELROD— WESTERN AMERICAN PINES 629 this would be another Miocene trans-Atlantic The occurrence of a member of Oocarpae in the eastern United States provides an additional link with the upland flora of the Sierra Madre Oriental, a distribution discussed by others (Harshberger, 1911; Axelrod, 1960: 267-269; Graham, 1973; Greller & Rachele, 1983). Pinus celetomensis Axelrod, sp. nov. TYPE: U.S.A. Nevada: Celetom Quarry 25 km east of Reno. U.C. Mus. Pal., holotype 7233. Figure 86. Cone large, elliptic-ovate in outline, fully 14— 15 cm long, 7.5 cm broad; base obtuse, tip acute; 10 and 13 rows of scales, those in middle 1.5- 1.7 cm broad and 8-9 mm high, swollen, broadly rounded, asymmetrical and upcurved; apophy- ses strongly reflexed proximally and flattened on outer (anterior) side, less so distally; umbo cen- tro-mucronate, the mucro very small, scarcely 1 mm long Discussion. This appears to be a new species of Subsect. Oocarpae. The cone is of the average size of the Ano Nuevo population of P. radiata. It differs from P. radiata cones of this size in that the fossil is elliptic-ovate, not ovate, the apoph- yses are swollen to the distal end of the cone rather than being confined chiefly to its upper third or half. In addition, the apophyses are more flattened throughout and are reflexed proximally in the fossil; in P. radiata the apophyses are more evenly rounded and hence unlike the fossil. his cone seems intermediate between those produced by P. radiata and P. attenuata. It dif- fers from cones of P. attenuata in that the apoph- yses are not sharply acute or hooked. Whether the present specimen represents a form transi- tional between P. radiata and P. attenuata can only be determined when additional fossil spec- imens become available. Pinus celetomensis was adapted to an interior, subhumid climate as judged from the composi- tion of the nearby Purple Mountain flora which <— FIGURES 74-84.—74. Pinus pieperi Dorf. Mount Eden, Californi ia. U.C. Mus. Pal., U.C. Mu s. Pal., hypotypes 7174, 7175, 7173, 7228. EM Oligocene, 26.5 Ma.—83, терте 7227. Latest . Mus u n, Pinus alvordensis Axelrod. о гадо. Sa us end won Axelrod. Creede, Colorado. U. Colo. Mus., hypotypes 19706, 19707. Latest Oligocene, 26. 630 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 1986] represents vegetation in the ecotone between broadleaved sclerophyll forest and mixed conifer forest (Axelrod, 1976). Pinus diegensis Axelrod & DeMéré. San Diego Soc. Nat. Hist. Trans. 20(15): 291, figs. 6A— B, 7A—B, 1984. TYPE: U.S.A. California: San Diego Formation. Holotype SDSNH 25135, paratypes 25110, 25136, 25137. This recently-described pine of the P. radiata alliance is from Chula Vista, near San Diego, California. The species produced cones much like the present population P. binata on Guadalupe Island. The cones show considerable variation in size, symmetry, and apophyses development, as noted earlier by Howell (1941). The occur- rence of such a pine on the mainland suggests that the insular species did not originate in iso- lation, and that it may have been ancestral to the present California populations at Monterey, Ario Nuevo, and Cambria. Pinus lawsoniana Axelrod, in R. N. Philbrick ympos. Biology of California Islands, p. 115, pl. 1, figs. 1, 2, 1967. TvPE: U.S.A. California: Old Forest Soil on Mussell Rock. U.C. Mus. Pal., holotype 20533, paratype 20534. Figures 90, 91. Pinus о Dorf., к Inst. Wash. Publ. 412: . 5, fig. 6 only, 0. Pinus pPretubereulata pudet Carnegie Inst. Wash. Publ. 476: 166, pl. 3, fig. 3 only, 1937. Cones similar to those of Pinus radiata D. Don Occur in rocks about 5—6 Ma old at Mussel Rock, on the outer coast south of San Francisco. They were recovered there by A. C. Lawson in 1893 from an old forest soil that rests on Jurassic dia- base and underlies the marine Merced Forma- tion. Similar cones are in the Mount Eden flora (Axelrod, 1937, pl. 3, fig. 3) dated at 5-6 Ma, and the Lower Pico Formation about 3 Ma old (Dorf, 1930, pl. 5, fig. 6). Cones in the Chula Vista Formation south of San Diego are allied but were named P. diegensis AXELROD— WESTERN AMERICAN PINES 631 Axelrod & DeMéré (1984) because they exhibit variation most like cones produced by the Gua- dalupe Island population. The specimens from Mussel Rock and Mount Eden seem more sim- ilar to those of the present Monterey population. Occurrence. Mt. Eden, California: Mus. Nat. Hist. Los Angeles, МКА L-1306;1014/688; Pico, California: U.C. Mus. Pal., hypotype 307. Pinus pretuberculata Axelrod, Carnegie Inst. ash. Publ. 476: 166, pl. 3, fig. 4 only, 1937. TYPE: U.S.A. California: Mount Eden. Nat. Hist. Mus., Los Angeles Co., lectotype 1014/ 696. Figures 87, 89. Pinus pretuberculata Axelrod. Condit, Carnegie Inst. ash. Publ. 553: 74, pl. 14, fig. 1, 1944; Axelrod, Univ. Calif. Publ. Geol. Sci. 34: 127, pl. 18, figs. 3-4, 11-13, 1958. Discussion. Cones of this species have been m the Mount (Axelrod, 1958). The Table Mountain flora is well dated at 12 Ma, the Mount Eden and Verdi are 5-6 Ma old. These records illustrate that the species had a wider distribution than its modern derivative, P. attenuata Lemmon, consistent with its new record at Celetom Quarry, Nevada. The Celetom record, from diatomite in the lower part of the Coal Valley Formation, consists of the upper half of a cone estimated to have been 12 cm long and 6 cm broad at its widest part. There are 9 x 6 rows of scales preserved, the apophyses are conical, swollen, upswept to re- flexed, and excentro-mucronate. The inside face (posterior) of the cone is missing, but there is evidence of smaller cone scales on the outer left margin so that it is apparent that the cone is asymmetrical. The cone scales are marked by prominent ridges on the outer sides so that in cross-section they are generally rhombic. This cone appears to be typical of populations in the Sierra Nevada to the west, and especially RES 85-89.—85. Pinus burtii Miller. Martha's Vineyard, Mass. U.S. Nat. Mus., B 222868. Early Ce IGU ре са. 22 Ма. а ipd E. of Reno, Nevada. U.C. M Pal., ER Quarry, 30 km E of Reno, Nevada. U.C Pinus tiptoniana Chane ey & Axelr od. Ert irs E ey figured by Chaney & Axelrod, 1959.)—89 U.C. Mus 1978.)—86. Pinus celetomensis Axelrod, sp. nov. Celetom Qua holotype 7233. Late Miocene, 12 Ma.—87. Pinus Я Axelrod. Mus. $e hypot Late Miocene, 12 Ma.—88 ype 7234. , lectotype 129. Middle Miocene, 15 Ma. U.C. . Pinus pretuberciata Axelrod. Table Mountain, California. 44.) . Pal., hypotype 2720. Late M 12 Ma. (From Condit, 1 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 632 Се/ 961 ‘родэху uro14) “EIA 9—6 “Bd *ouaoorjN 21€ T ‘5605 эАзетеа “Ted “sn O'N 'uoneuuo, рәоләр{ үевед MOQ [105 15210] UF 'oosroue1, URS Jo `$ “YOY [ossnjN "ролэху DUDIUOSMD] snuid `16—(е1961 ‘родэху WOII) "BW 9-6 "eo “әцәзогу NET 5505 edÁjo[ou “тед “SNA ‘O'N uoneuno;, PINIWN [ESE Moaq [IOS 15910] ш *oosrouer ues JO `$ AJO Y [Assn] "poup[oxy DUDIUOSMD] snuid '06—'16 ‘06 STANDA 1986] to those in southern California that have more prominent apophyses. The modern P. attenuata occurs in the lower part of the forest belt, often forming nearly pure stands where it has been disturbed by fire. The species ranges discontin- uously from southwestern Oregon into northern Baja California. In the south, it is regularly ad- jacent to the upper chaparral belt at the forest margin. Occurrence. Table Mt.: U.C. Mus. Pal., hy- potype 2720; Verdi, Nevada: MUC. Mus. Pal., hypotypes 1972-1975, homeotype 1971; Cele- tom Quarry, Nevada: U.C. Mus. Pal., hypotype 7234. Pinus masonii Dorf, Carnegie Inst. Wash. Publ. 41 type 308, paratype 306. Figure 94. Pinus masonii Dorf. Axelrod, їп К. N. Philbrick (ed.), mpos. Biology of California Islands, p. 117, pl. 5, figs. 1-2, 4—5; pl. 6, fig. 2, 1967. Axelrod, Univ. Calif. Publ. Geol. Sci. 120: 39, pl. 12, figs. 1-4, 1980 This pine, rather rare in pre-Pleistocene coast- al deposits of California, has cones similar to those of P. muricata D. Don., now distributed discontinuously from the north coast of Califor- nia southward into northern Baja California near Eréndira. Over this broad area the scattered pop- ulations are represented by three varieties (Ax- elrod, 1983a) Its known fossil records are in the Lower and Upper Merced Formations on the coast south of San Francisco (Dorf, 1930; Axelrod, 1967, 1980), in the upper part of the Lower Pico Formation on the coast west of Ventura (Dorf, 1930), and er Pico Formation north of Santa Pau- . The oldest of these records 5 Ma. Occurrence. Lower Pico, California: hypo- AXELROD— WESTERN AMERICAN PINES 633 types 12738, 20380; Upper Pico, California: hy- potype 20380. Pinus tiptoniana Chaney & Axelrod, Carnegie Inst. Wash. Publ. 617: 142, pl. 13, figs. 3- 6, 1959. TYPE: U.S.A. Oregon: Blue Mts. U.C. Mus. Pal., lectotype 129, paratypes 128, 643, 2866, 2867, 2869, homeotypes 642-644, 2868, 2870-2871. Figure 88. This pine from the Blue Mountains flora, east- ern Oregon, occurs in diatomaceous sediments interbedded with basalts of the Columbia River Lava Group. The fossil was compared initially with Pinus halepensis Miller, which it does re- semble. However, comparison with cones in the collection at the Institute of Forest Genetics, Pla- cerville, shows that P. tiptoniana is a member of the Oocarpae. This is indicated by the umbo, which is centro-mucronate, whereas in P. hale- pensis it is excentro-mucronate. Pinus tiptoniana appears to be an extinct species of the group, differing from those most nearly allied to it, e.g., P. patula Scheide & Deppe and P. pringlei Shaw, in having needles in 2s, and the fossil cones ap- pear to be more nearly symmetrical as judged from the incomplete specimens. These modern pines allied to P. tiptoniana occur in the Sierra Madre Oriental, Mexico, generally at middle el- evations in woodlands and forests adapted to a climate of high equability and ample summer rain. PLEISTOCENE OOCARPAE FROM COASTAL CALIFORNIA The following is a listing of younger records of Oocarpae from coastal California. Some of these deposits (e.g., Carpinteria, Pt. Sal) are (were) especially rich in cones that accumulated on floodplains. This is because the cones were not dispersed as widely by currents as in the case of ds fr ine dep ite > FIGURES 92-94.—92. Pinus radiata D. Don. Santa Barbara Formation, Veronica Springs Quarry, Santa Bar- bara, Calif. Santa Barbara Mus. Nat. Hist., hypotype 473. Early Pleistocene, ca. 1 Ma . A cluster of four cones. (Previously figured by Axelrod, 1980.)—93. Pinus muricata D. Don. Upper Merced Formation, S. of San Francisco. U.C. Mus. Pal., hypotype 159. Early Pleistocene, ca. 1 Ma. (Previously NU by Mason, 1932 and Dorf, 1930.)— 94. Pinus masonii Dorf. Lower Merced Formation, S. of San Francisco, Calif. U.C. Mus. Pal., hypotype 20532. Pliocene, ca. 3. Ma. (Previously figured by Axelrod, 1967.) FiGures 95-103. Pinus muricata D. Don, var. borealis Axelrod. Near Point Sal, Santa Barbara Co., Calif. U.C. Mus. Pal., hypotypes 20400-20408. Late Pleistocene, ca. 26,700 + 800 years В.Р. (From Axelrod, 1967.) ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 1986] AXELROD— WESTERN AMERICAN PINES 635 636 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 104-108.— 104, 105. Pinus remorata Mason. Century City, Los Angeles. U.C. Mus. Pal., hypotypes 5841, 5840. Plio-Pleistocene transition, ca. 2 Ма. — 106. Pinus remorata Mason. Santa Cruz Island, Willow Creek. U.C. Mus. Pal., hypotype 5839. Late Pleistocene, 14,200 + 250 years B.P.— 107. Pinus remorata Mason. Near Point Sal, Santa Barbara Co., Calif. U.C. Mus. Pal., hypotype 5840. Late Pleistocene, 26,700 + 800 years B.P.—108. Pinus remorata Mason. Carpinteria, Calif. Santa Barbara Mus. Nat. Hist., hypotype 474. Late Pleistocene, older than 38,000 years B.P. (All specimens previously figured by Axelrod, 1980.) 1986] AXELROD— WESTERN AMERICAN PINES 637 Pinus attenuata Lemmon. te Pleistocene Oakland (Metcalf, 1923). Early Pleistocene Seacliff (Axelrod, 1983); Signal Hill (Mason, 1932; Axelrod, 1967) Pinus muricata D. Don. Figures 93, 95-103. Late Pleistocene var. muricata. Rancho La Brea (Mason, var. borealis. Millerton (Mason, 1934); Car- pinteria (Chaney & Mason, 1933; Axel- rod, 1967, 1980). Early Pleistocene var. muricata. Seacliff (Axelrod, 1983); Wil- mington (Axelrod, var. stantonii. Seacliff (Axelrod, 1983). Pinus radiata D. Don. Figure 92. Late Pleistoce Millerton aoe, 1934); Drakes Bay (Ax- elrod, 1980, 1983); Thornton Beach (Ax- elrod, 1967); Pt. Sal (Axelrod, 1967); Car- pinteria (Chaney & Mason, 1933); Santa Rosa I. (Axelrod, 1980); Rancho La Brea (Mason, 1927; Warter, 1976); Little Sur (Langenheim & Durham, 1963). Early Pleistocene Potrero Canyon (Axelrod, 1967); Seacliff (Axelrod, 1983); Century City (Axelrod, 1980); Spring Valley Lake (Axelrod, 1967); Veronica Springs (Axelrod, 1980). Pinus remorata Mason. Figures 104-108. Late Pleistocene Carpinteria (Chaney & Mason, 1933; Ax- elrod, 1980); Santa Cruz I. (Chaney & Ma- son, 1930); ie Sal (Axelrod, 1967). Early Pleistocen Potrero с (Axelrod, 1967); Century City (Axelrod, 1980), Seacliff (Axelrod, 1983) LITERATURE CITED — M. A., G. M. BRATTSEVA, R. E. GITERMAN, V. GOLUBEVA & A. I. MoLsEEVA. 1978. Late deo stratigraphy and flora of Iceland. Trans. Acad. Sci. USSR 316: 1-188. [In Russian.] 60. Further conifers of the Pinaceae from the Wealden Formation of Belgium. Bull. Inst. Roy. Sci. Natur. Belgique 146: 1—39. ARNOLD, C. A. 1 Tertiary conifers from the Princeton coal fields of British Columbia. Univ. Michigan Mus. Paleontol. Contrib. 5(8): 245-258. AXELROD, D. I. 1937. A Pliocene flora from the Mount Eden beds, southern California. Publ. Carnegie Inst. Wash. 476: 125-183. 939. A Miocene flora from the western bor- der of the Mohave Desert. Publ. Carnegie Inst. Wash. 516: 1-128. 1944. The Alvord Creek Flora. Publ. Car- negie Inst. Wash. 553: 207-224. 0. The Anaverde flora of southern Cali- fornia. Publ. Carnegie Inst. Wash. 590: 119-158. 1956. Mio-Pliocene floras from west-central Nevada. Univ. Calif. Publ. Geol. Sci. 33: 1-321. 1958. The Pliocene Verdi flora of western Nevada. Univ. Calif. Publ. Geol. Sci. 34: 91-160. win, Volume 1. The PE of Life. Univ. Chi- cago Press, Chicago, 1962. A Pliocene a ee forest from western Nevada. Univ. Calif. Publ. Geol. Sci. 39: 195-268. 1966a. The Eocene Copper Basin flora of northeastern Nevada. Univ. Calif. Publ. Geol. Sci. 59: 1-124 1966b. The Pleistocene Soboba flora of southern California. Univ. Calif. Publ. Geol. Sci. 60: 1-109. . 1967a. Evolution of the Californian closed- cone forest. Pp. 93-149 in R. N. Philbrick (editor), Proc. Symposium on the Biology of the California Islands. Santa Barbara Botanic Garden, Califor- . 1967b. Geologic history of the California in- m а Pp. 267-316 in R. М. Philbrick (editor), Proc. Symposium on the Biology of the California Islands. Santa Barbara Botanic Garden, Califor- nia | ` 1968. Tertiary floras and topographic history of the Snake River Basin, Idaho. Bull. Geol. Soc. Amer. 79: 713-734. 1976a. History of the conifer forests, Cali- fornia and Nevada. Univ. Calif. Publ. Botany 70: 1-60. 76b. Evolution of the Santa Lucia fir кее ү ecosystem. Ann. Missouri Bot. 24- Ag e and origin of sige sine Desert vege- cad. Sci. Occas. bn s 132. "History of the е -сопе pines, Alta апа E California. "Univ. Calif. Publ. Geol. Sci. 120: 43. 1980b. te toes to the Neogene Paleo- botany of Central California. Univ. Calif. Publ. Geol. Sci. 121: 1-212. 1981. Holocene climatic changes i in relation to vegetation disjunction and speciation. Amer Naturalist 117: 847-870. 1982. Age and origin of s Monterey en- demic area. Madroño 29: 127-14 1 . New Pleistocene а records, coastal California. Univ. Calif. Publ. Geol. Sci. 127: 1-108. 1983b. An interpretation of Cretaceous and ertiary biota in polar regions. Palaeogeogr., Pa- laeoclimatol., Palaeoecol. 45: 105-147. 1985 Mi iocene floras from the Middlegate 638 Basin, wigs pai Nevada. Univ. Calif. Publ. Geol. Sci. -279. AXELROD, D. I. . P. BAILEY. 1976. Tertiary vege- tation, climate and altitude of the Rio Grande depression, New Mexico-Colorado. Paleobiology 2: 235-254. . A. DEMÉRÉ. 1984. A Pliocene flora from Chula Vista, San Diego County, California. Trans. San Diego Soc. Nat. Hist. 20: 277-300. VEN. 1985. E E the Cordil- lera n flo ora. J Biogeography 12: 21—4 BAILEY, D. K. 1970. Phytogeography ni vd of Pinus subsection ОРИК Апп. Mis: Bot. Gard. 57: 210-24 di pem segregate from and a in Pinus cembroides Zucc. at its е limit Phytologia 54: aceae of the Chihinshun Desert region. Penlon 53: 226-234 —— & Е. 1979. Pinyons of the Chihuahuan Desert region. Phytologia 44: 129- 133, & T. WENDT. 1979. New pinyon records for northern Mexico. Southw. Naturalist 24(2): 389- 3 90. ВАПЕҮ, H. Р. 1960. A method of determining the th and UM of climate. Geogra- fisker Annaler 42: 1- . 1964. Toward a ner concept of the tem- perate climate. The Geogr. Rev. (New York) 54: 516-545. BAILEY, I. W. 1910. Anatomical characters in the evolution of Pinus. Amer. Naturalist 44: 284—293. BANKS, H. P., A. ORTIZ-SOTOMAYOR & C. M. HARTMAN. . Pinus escalantensis, sp. nov., a new per- mineralized cone from the Oligocene of British Columbia. Bot. Gaz. 142: 286-293. ВЕСКЕВ, Н. Е. 1969. Fossil plants of the Tertiary Beaverhead basins in southwestern Montana. Pa- leontographica B 127: 1-142. BERRY, E. W. 1916. The physical е indi- cated by the flora of the Calvert Form . U.S. Geol. Surv. Prof. Paper 98-F: 61-73. . An Eocene flora from Trans-Pecos Texas. U.S. Geol. Surv. Prof. Paper 125-A: 1-9. . 1934. Apine from the Potomac Eocene. Wash. Acad. Sci. Jour. 24: 182-183. . 1936. Pine and cherry from the Calvert Mio- cene. Torreya 36: 124-127. ld Pinus and Quercus i in the Chesapeake Mio ene. J. Wash. Acad. Sci —508. в W. Н. 1984. Е wood from . 31: 50 ossil ma чек pine the Upper Cretaceous of Northeast Mississippi- Ann. Bot. 53: 133-1 Brown, R. W. 1934. The recognizable : species of bd Green River flora. U.S. Geol. Surv. Prof. Pape 185-C: 45-7 . 1937. eR to some fossil floras of d western Heres States. U.S. Geol. Surv. Prof. P per 186-J: 163-206. 940. x bracket fungus from the Later Ter- tiary of southwestern Idaho. J. Wash. Acad. Sci. 30: 422-424. CHANEY, R. W. 1954. A new pine from the Creta- ceous of Minnesota and its paleoecological signif- icance. Ecology 35: 145-151. ANNALS OF THE MISSOURI BOTANICAL GARDEN ——— 1981. [Vor. 73 . I. AXELROD. 1959. Miocene floras of the Columbia Plateau. Part II. Systematic consider- ations. Publ. Carnegie Inst. Wash. 617: 135-229. H. L. MASON. A Pleistocene flora from Santa Cruz Island, California. Publ. Carnegie Inst. Wash. 415: 1-24. 1933. A Pleistocoene flora from the аи deposits ч (ш California. Publ. = COCKE qo D. A. i An es foxtail pine. "ann 133: 573-574. e Table M flora. Publ. 553: as idization of t acon, a. Crossability and relationships of Washoe pine. Madrono 70. 1984b. Impact of the T uoces on the ge- netic пш of North American conifers. Pp. 70-118 in R. E. Lanner (editor), Proc. 8th North шш Forest о Workshop, July 30-Aug. 1, 1984. a ogan, Uta & E. L. LITTLE, JR. 1966. Er gm distri- bution of the pines of the world. E a Dept. Agric. Forest Service Misc. Publ. 991: CROWELL, J. C. 79. The San iiu fault system through time. J. Geol. Soc. London 136: 292-302. Dawson, J. W. 1883. On the Cretaceous and Tertiary floras of British Columbia and the Northwest Ter- ritory. Trans. Roy. Soc. Canada l(Sec. 4): 14-34. . 1890. On fossil plants collected by Mr. R. A. McConnell on MacKenzie River, and by Mr. C T. Weston on Bow River. Trans. Roy. Soc. Canada 7(Sec. 5): 69-74 1891. On fossil plants from the Similkameen Valley and other places in the southern interior of British Columbia. Trans. Roy. Soc. Canada 8(Sec. : 79-91. 1930. Pliocene floras a California. Publ. Donr, E. : The VM flora of the Mammoth Mountain on the Aldan River. Akad. Nauk CCCP. [In Russian.] ENGLEMANN, G. 1880. Revision of the genus Pinus, and description of Pinus elliottii. Trans. St. Louis Acad. Sci. 4: 161-198. Epis, R. C. & C. E. CHAPIN. 1975. Geomorphic and tectonic implication ns of the post-Laramide, Late Eocene erosion surface in the Rocky Mountains. Mem. Geol. Soc. Amer. 144: 45-74. Fowe ts, Н. А. 1965. Silvics of Forest Trees of =- United States. U.S. Dept. Agriculture, Agricultur Handbook - 1. Fox, W. W. 19 Pygmy forest: an ecologic staircase. California e 29: 1-7. FREDERICKSON, N. O. 1980. Mid-Tertiary climate of southeastern United States: the sporomorph evi- dence. J. Paleontol. 54: 728-739. Middle Eocene to Early Oligocene plant communities of the Gulf Coast. Pp. 493-549 in J. Gray et al. (editor), Communities of the Past. Hutchinson Ross Publ. Co., Stroudsburg, Penn- sylvania. — ——, D. В. Carr, С. D. Lowe & E. P. Wosika. 1986] 1983. Middle Eocene palynomorphs from San Diego, California. Amer. Assoc. Stratig. Palynol- ogists Contrib. 12. GARCIA, E. 1973. Modificaciones al sistema de clas- . Autó. de Mexico, Institutio de Geografia. GARDNER, R. A. 19 Sequence of podsolic soils alo ong the coast of Northern California. Doctoral KY. 1973. Evidence for strike- slipdisplacement beneath the Trans-Mexican vol- canic belt. Stanford Univ. Publ. Geol. Sci. 13: 171- 180 GAUSSEN, H Les Gymnospermes, Actualles et f ap. XI, Generalites, Genre nce. o Mayo plants: a study of the flora and vegetation. o the valley of the Rio Mayo, Sonora. Publ. e Inst. Wash. 527. GoLz, D. J. А Soke Artiodactyla of southern ы Nat. Hist. Mus. Los Angeles Co. Sci. Bull. 26: 1-85. GRAHAM, A. 1973. History of the arborescent tem- perature element in the northern American biota. Pp. 301-314 in A. Graham (editor), Vege- tation = Vegetational n of Northern Latin Ameri Elsevier Sci. Publ. Co., Amsterdam- London New York. GRAY, J. Temperate pollen genera in the Eocene (Claiborne) flora, Alabama. Science 132: 808-810. GRELLER, А. M. & Г. D. RAcHELE. 1983. Climatic limits of exotic genera in the Legler palynoflora, Miocene, New Jersey, U.S.A. Rev. Paleobot. Pal- s W. B. CRITCHFIELD. 1972. The Md tribution of forest trees in California. U.S. De Agriculture, For. Ser. Res. Paper PSW-82 1972. HALLER, J. R. 1959. The role of hybridization in the origin and evolution of Pinus washoensis. Proc. IX Internat. Bot. Congr. II: 149. Montreal, Can- ada . 1965. Pinus washoensis in Oregon: taxonom- ic and evolutionary implications. Amer. J. Bot. 2: 646. 1966. Systematic and evolutionary relations of. Pinus torreyana. Amer. J. Bot. 53: 635. 1 Altitudinal and morphological rela- Чопан among Ponderosa, Jeffrey and Washoe pines in the Pacific States. Amer. J. Bot. 71: 168. 1986. Taxonomy and relationshps of the mainland and island populations of Pinus torrey- ana (Pinaceae). Syst. Bot. 11: 39-50. HALLUM, A. 1984. Continental humid and arid zone during the Jurassic and Cret poe: Palae c R Palaeoclimatol., Palaeoecol. 47: 195-223. Hanna, M. A. 1926. Geology of He La Jolla Quad- rangle, с Univ. Calif. Dept. Geol. Sci. Bull. 16: 187-2 ; The identification of ina of y needle structure. New York State Coll. Forest. E 4: 1- HARSHBERGER, J. W. 1911. Phytogeographic Survey of North America. G. М Stechert & Co., New York. AXELROD— WESTERN AMERICAN PINES 639 HEER, : 1868-1883. 2 Fossilis Arctica, 6 vol- umes. J. Wurster and le Mi p бы мөн von Island. Flora Fossilis Arctica 1(4): 139-155. ps Ду fo oum Kongel. Svenske Viden Sk. Acad. m ndl. 8(4): 1—41. [Flora Fossilis Arctica д^ ——. 1878 козше flora der Insel Sachalin. Mem. Acad. Im 54 Sci. St. Petersb. 25(7): 1-61. [Flora Fossilis Arctica 5(3).] 187 Beitrage zue Miocenen Flora von Sachalin. Kongel. Svenska Vidensk. Acad. Handl. 15(4): 1-11. [Flora fossilis Arctica 5(4).] HELLEY, E. J. & К. В. Lajo. 1979. Flatland deposits of the San Francisco Bay region, California — their geology and engineering properties and their im- portance in comprehensive planning. U.S. Geol. The closed- -cone pines of insular California. Leaflets Western Bot. - JENNY, H., R. J. ARKLEY & A. M. SCHULTZ. 1969. The pygmy forest-podsol ecosystem and its dune associates of the Mendocino coast. Madroño 20: 60-74. KLAus, W. 1980. Neue Beobachatungen zur Mor- phologie des Zapfens von Pinus und ihre Bedeu- tung fur dis Systematik, Fossil-bestimmung, Arealgestaltung be Evolution der Gattung. Pl. Syst. Evol. 134: KNOWLTON, F. H. 1 Fossil flora of the Yellow- stone p Park. U.S. Geol. Surv. Monogr 32(2): 651—882. E A fossil nut pine from Idaho. Torreya 1: 113-115. 1923. Fossil plants from the pid lake beds of south-central Colorado. U.S. Geol. Surv. Prof. Paper 131G: 183-1 LANGENHEIM, J. H. & J. W. DURHAM. 1963. Quater- nary closed-cone pine flora from travertine near Little Sur, California. Madrofio 17: 33-51. LEONARD, B. F. & E. F. MARVIN. 1984. Temporal evolution of the Thunder Mountain caldera and related features, central Idaho. Idaho Bur. Mines & Geology Bull. 26: 23-41 LESQUEREUX, L. 1878. Contributions to the flora of the Western Territories. The Tertiary Flora. Geol. Surv. Terr. 7: 1-366. 1883. Contribution to the fossil flora of the western Territories. The PN and Tertiary floras. Geol. Surv. Terr : A new od variety from Tex s. Wrightia 3: 181-187. wo new pinyon varieties from Ari- zona. Tees 17: 329-34 1971. Atlas of United States Trees, Volume l. Conifers and important hardwoods. U.S. Dept. Agric. For. Ser. Misc. Publ. 1146. 1976. Atlas of United States Trees, Volume 3. Minor western poe mes U.S. Dept. Agric. For. Ser. Misc. Publ. iuba cuia Su Subdivisions ыд the genus Pinus Api sp U.S. Dept. Agric. For. S Misc. Publ. MACGINITIE, Н. E "r^ en plants of the Floris- sant beds, Colorado. Publ. Carnegie Inst. Wash. 99. 640 . 1969. The Eocene Green River flora of north- western Colorado and northeastern Utah. Univ Calif. Publ. Geol. Sci. 83. Ma pg, Н. E. & Н. A. POWERS. 1962. Upper Ce- nozoic stratigraphy of western ze River "Plain, Idaho. Geol. Soc. Amer. Bull. 73: 1197-1220. MALLOCH, D. W. . A. PIROZYNSKI T VEN. 1980. Ecological and evolutionary significance of mycorrhizal symbioses in vascular plants. [A Re- view.] Proc. Natl. Acad. Sci. 77: 2113-2118. MARTINEZ, M. 1948. Los ehh oe 2nd edi- tion. Ediciones Botas. Mex F Mason, H. L. 1927. Fossil sien of some west American conifers. Publ. Carnegie Inst. Wash. 346: 139-158. 1932. A phylogenetic series of California closed-cone pines suggested by the fossil record. Madroño 2: 49-55. . 1934. Pleistocene flora of the Tomales For- m pope Carnegie Inst. Wash. 415: 81-179. P. STOCKWELL. 1945. A new pine from ipid Rose, Nevada. Madroño 8: 61-63. MASTROGIESEPPE, R. J. & J. D. MASTROGIESEPPE. 1980. A study of Pinus balfouriana Grev. & Balf. (Pi- naceae). Syst. Bot. 5(1): 86-104. —€— V. V. (editor). 1976. The Mio-Pliocene of my est Kamtschatka. Trans. Acad. Sci. USSR. 4. (In Russian.] W. 1923. An ancient pine cone. Amer. : 172. А 1973. Distribution of ectomycorrhizae in native and man-made forests. Pp. 79-105 in G. Marks and T. T. Kozlowski (editors), Ecto- pesitos qe Their Ecology and Physiology. Ac- ademic Press, New York. MILLER, C. N., JR. 1969. Pinus avonensis, a new species of petrified cones from the Oligocene of Montana. Amer. J. Bot. 56: 972-978. ——— 1973. Silicified cones and vegetative remains 1974. Pinus wolfeii, a new petrified cone from the Eocene of Washington. Amer. J. Bot. 61: 772- 197 6. Early evolution in the аш Кеу. _Paleobotany Palynology 21: 101- 1977a. Mesozoic conifers. E. [m 43: 218- О 280. . 1977b. Pityostrobus lynni (Berry) comb. nov., a pinaceous seed cone from the Paleocene of Vir- а Bull. Torrey Bot. Club 104: 5-9. 1978. Pinus burtii, a new species of petrified cones from the Miocene of Martha’s Vineyard. Bull. Torrey Bot. Club 105: 93-97. Mirov, М. T. 53. Ta appe and chemistry of the white pines. Madroño 1967. The Genus МА Воды Press, New York. MOORBATH, S., Н. SIGURDSSON & К. GOODWIN. 1968. K-Ar ages of the oldest exposed rocks in Iceland. Earth and Plan е t Sci. Lett. 4: 197-205. PENHALLOw, D. P 08. Report on Tertiary plants of British си collected by Lawrence М. Lambe in 1906 together with a discussion of pre- a recorded Tertiary floras. Canada Dept. s, Geol. Surv. Branch Rept. 1013. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 PRAGER, E., D. F. FOWLER & A. C. WILSON. 1976. Rates of evolution in Conifers (Pinaceae). Evo- lution 30: 637-649. READ, C. B. 1930. Fossil floras of Yellowstone Park, Part 1. Coniferous woods of Lamar River flora Publ. Carnegie Inst. Wash. 416: REED, R. e bes Geology of California. Amer. As- l. Geol., Tulsa, Oklahom ROBERT- E M. F. Deux nouveaux pins а = MU Ri Bull. Mus. Hist. Nat. (Paris), c. B, Adansonia no. 1: 61-73. mE ч р peus The history of the earth's surface temperature during the past 100 million years. Ann. Rev. Earth Planetary Sci. 5: 319-355. SHAW, G. R. 1909. The pines of Mexico. Arnold Arb. Publ. 1. Cambridge, Massachusetts. . 1914. The genus Pinus. Arnold Arb. Publ. 5. Notes on the genus Pinus. J. Arnold Arboretum 5: 225-227. SMEDES, H. W. & H. J. PROTSKA. 1972. Stratigraphic framework of the Absaroka Volcanic sali up in the Yellowstone Park region. U.S. Geol. Surv. Prof. Paper 729-C: о SMITH, А. G., A. M. HURLEY & J. С. BRIDEN. 1981. Phanerozic paleocontinental maps. Cambridge Univ. Press rth Variation and Evolution Species Level. Univ. Press, Cambri STEWART, K. B. 1940. Plant ecology and paleoecology of the Creede Valley, Colorado. ш Thesis. niv. Colorado, Boulder, Color: 1935. Dou ns of lower i ne m ammali ian-bearing beds near Death Valley, California. Proc. Natl. Acads. U.S.A. 25: 270-275. SrockEv, R. А. 1983. Pinus driftwoodensis sp. n. from the early Tertiary of British Columbia. Bot. Gaz. 144: 148-156. 1984. Middle Eocene Pinus remains from British Columbia. Bot. Gaz. 145(2): 262-274, STUART, C. J. 1974. The San Onofre breccia i in north- implications. Amer. A , Geology of Peninsular California Guide- book, pp. 80-90. SUKACHEV, V. N. 1910. Nekotoryia dannya k doled- nikovoi flore severa sibiri. [New data on the per- iglacial flora of the north of Siberia.] Akad. Nauk Geologischeskii musei imeni Petra velikago Trudy 55-62. aser. R. H. & M. E. Hunter. 1984. Mioc ne-nonmarine correlations, Atlantic and Gulf Co жан Plains, North America. Palaeogeogr., Pa- laeoclimatol., Palaeoecol. 47: 129-152. 1953. TEMPLETON, B. A new record of pine cone r the Miocene epoch. Bull. S. Calif. Acad. Sci. 52: 64—66. TuRNER, D. L. 1970. Potassium-argon dating of Pa- cific Coast Miocene Foraminiferal Stages. Geol. Soc. Amer. at Paper 124: 91-129. UNDERWOOD, J. C. & C. N. MiLLER, JR. 1980. Pinus buchananii, a new species based on a petrified cone 1986] from the ое of Washington. Amer. J. Bot. 67: 1132-1 VAIL, P. R. 4 J. ae ENBOL. 1979. Sea- ee rue during the Tertiary. Oceanus 22(3): 7 ,R. M. MITCHUM, JR. &S THOMPSONTIT. hi level, Part 4. Global cycles of relative changes of sea level. Mem. Amer. Assoc. Petrol. Geol. 26: 83- 97 Уос, К. J., W. P. ARMSTRONG, К. L. WHITE & К. L. Corr. 1977. The closed-cone pines and cypress- es. Pp. 295-358 in M. G. Barbour and J. Major (editors), Terrestrial bre of California. J. Wiley & Sons, New WARTER, J. K. 1976. es Pleistocene plant com- munities— evidence from the Rancho La Brea tar pits. Pp. 32-39 in J. Latting (editor), Plant Com- munities of Southern California, Calif. Native Plant Soc. Spec. Pub. 2. AXELROD-— WESTERN AMERICAN PINES 641 WEBBER, I. E. 1933. Woods from the Ricardo Plio- cene of Last Chance Gulch, California. Publ. Car e Inst. Ls 412: 113-134. Wins, LL 1 . Anadditional specimen of Pinus eri Dorf ‘Be entura County, California. . LEOPOLD. 1967. Neogene and early Quaternary vegetation of northwestern North America and northeastern Asia. Pp. 193-206 in D. M. Hopkins (editor), The "iis Land Bridge. Stanford Univ. Press, Californ WOODFORD, A. O., J. E. SCHOELLHAMER, J. G. VEDDER . E. YERKES. 1954. Geology of the Los An- geles Basin. Pp. 65-82 in Geology of Southern California. Calif. Div. Mines Bull. 170, Chap. 2, Geology of the Natural Provinces. WOODRING, W. P., M. М. BRAMLETTE & W. S. W. КЕМ. 1946. Geology of the Palos Verdes Peninsula. U.S. Geol. Surv. Prof. Paper 207. A HISTORICAL SKETCH OF THE VEGETATION OF THE MOJAVE AND COLORADO DESERTS OF THE AMERICAN SOUTHWEST! ROBERT F. THORNE? ABSTRACT The Mojave and Colorado Deserts of the American Southwest are geologically recent in origin, resulting primarily from the rain-shadow caused by the Late Pliocene-Pleistocene elevation of к Sierra Nevada, Bede: = Peninsular ranges. Their n more recent, largely Holocene i in origin, and still evolving. During 1he d с the California deserts were well supplied with huge, deep lakes and woodland, possibly grassy, supporting numerous large mammals, now The present vegeta last, Wise onsin large streams and a rg mostly extinct. The desert flora, however, is in large sources. Most of the perennials are probably former members of the wet winters and thus able to popu available. They survived cool, A continental movement, glacial sea-level lowering, and to changes in oceanic-current patterns. pluvial, glacial periods by erican southwestern desert flora are derived fr part more ancient, having been assembled since dry summers, and d d adapt to the varie sert habitats when they migration southward or to lower elevations. Mexico, South a, ion overland im- nging ocean size due to Some desert plants, at least some of the ephemerals, may well be of very recent, even Holocene, origin in our arid Southwest Those of us without much geological training, in view of our own brief life span, tend to think of large-scale features of the landscape as rather permanent, thus very ancient. Yet geologists as- sure us that large lakes, islands, mountains, and the deserts behind them are relatively ephemeral in the geological time scale. Our southwestern deserts are no exception. The existing regional deserts or semi-deserts are geologically recent, possibly only a couple of million years old, at least as we know them today. They are essentially products of the late Pliocene-Pleistocene eleva- tion of the Sierra Nevada, Transverse, and Pen- insular ranges in southern California, which cre- ated a huge rain-shadow to the east. Thus, most of the moisture gleaned from the Pacific Ocean by the prevailing westerly winds precipitated on the western slopes ofthe ranges. As the air masses descended the transmontane slopes, they heated up adiabatically and became a drying force. REGIONAL GEOLOGY It is impossible to obtain exact dating for in- cidents of regional uplift. Often we must rely heavily upon circumstantial evidence presented by juxtaposition of strata, by radiometric dating of volcanic rocks, or by fossil floras and faunas of the region. When properly dated, fossil assem- blages of plants and animals allow us to infer much about the past vegetation and climatic con- ditions of periods in question and the changes that have taken place. Leaves that are mostly large numbers of grazing mammals presumably require rather extensive savannas or open grass- lands. Oakeshott (1971) reported that radiometric dating of volcanic rocks indicated *'that the in- tensive faulting of the eastern side of the Sierra (accompanied by uplift and westward tilting) be- gan about 2.5 million years ago." Paleoclimatic evidence from fossil floras of the region indicates that until that time the Sierra Nevada was a range with only low to moderate relief and that it be- paleobotanical evidence Axelrod estimated total ! I wish to thank T. R. Van Devender for E many helpful suggestions in his review of this paper. Also B. Han A; ner was most hel pful i in leading | me to the most appropriate geological literature. K. Tomlinson has ? Rancho thro ugh fmyc omputer. Santa Ana Botanic erai ae California 91711. ANN. MISSOURI Bor. GARD. 73: 642-651. 1986. 1986] post-Pliocene uplift in the Yosemite region at 1,980-2,135 m. Likewise he placed total post- Pliocene displacement at Donner Summit at about 1,525 m, at Carson Pass at about 2,135 m, and at Mount Whitney at about 2,745 m (Axelrod, 1962). Bachman (1978) estimated that “Relative uplift between Owens Valley and the White-Inyo Mountains may have been as much as 2,300 m during the past 2.3 m.y.” According to Foster (1980) the central Trans- verse Ranges during the late Cenozoic Era evolved from an area of low relief to a region of rugged high mountains. He estimated that the “Uplift of the San Gabriel Mountains began at least in Crowder time, 4—2 m.y.b.p., and possibly earlier. Erosion followed and uplift renewed as the up- ward coarsening Harold and Shoemaker deposits formed in Rancholabrean time, approximately 600,000-100,000? yr.b.p.” He also believed that the San Bernardino Mountains had only mod- erate relief in late Hemphillian-Blancan time [Late Miocene, 5-8 Ma], but were worn down to a pediment (an area of low relief) by Crowder time (Plio-Pleistocene). The elevation of the San Bernardino Mts. by Upper Pleistocene time is attested to by the glacial deposits on San Gor- gonio Peak. Because folded and faulted strata of Lower Pleistocene time in the Ventura Basin portion of the western Transverse Ranges are overlain un- conformably by upper Pleistocene beds, Oake- shott (1971) concluded that **one of the most violent mountain-building pulses was clearly in the middle Pleistocene." In reference to the San Gabriel Mountains, Oakeshott described the great mid-Pleistocene orogeny as “‘very widespread, involving uplift, tight folding, major faulting, and general destruction of the Tertiary basins by mountain-building processes... . Crustal move- ments have continued to the present day." Bull (1978) estimated that uplift rates along the south side of the San Gabriel Mountains appear to av- erage about | to 3 m per 1,000 years. Indeed the youthful ruggedness of the range and the all too- frequent earthquakes in "us area attest to the continued mountain buildi The fossil Soboba Flora Tu 1980a) west of the San Jacinto Mountains of southern Cali- fornia is now dated at about 1.75 Ma in Early Pleistocene time. From the plant species and plant communities represented in the fossil flora, Ax- elrod (1966) believed that the nearby San Jacinto Peak then stood at about 2,135 m compared to its 3,302 m today. Axelrod estimated that the THORNE— MOJAVE AND COLORADO DESERTS 643 major Transverse and Peninsular ranges then had elevations about half those of the present day. It would appear that the San Jacinto Mountains, at least, gained nearly 1,220 m in elevation since the Early Pleistocene. In the Colorado Desert (i.e., the Lower Colo- rado River Valley subdivision of the Sonoran Desert) ofextreme southern California the Salton Basin shows convincing evidence, according to Oakeshott (1971), of “great vertical displace- ments in late geologic history ... displacement within the Salton Trough is going on very rap- idly." Linday and Lindsay (1978), discussing the Pleistocene vertical faulting of the Peninsular Range ridges on the western side of the Salton Trough, believed that the “resulting ‘rain-sha- dow’ effect, coupled with drying of California climates in general, caused an arid climate to develop here, and a desert evolved about 20,000 years ago." Admitting the problematic accuracy of some of these dates, one must still conclude that the California regional deserts are indeed geologically a very recent phenomenon During Late Wisconsin and Holocene Tran- sition time, under a cool, moist, pluvial cli- mate, the California deserts were supplied with an extensive river and lake system. It is estimated that as much as 20 percent of the Mojave Desert was covered by fresh water to an average depth of 30 m (G. Jefferson, pers. comm.). Searles Lake, for example, was estimated at 132 m deep and 26 km long, Lake Panamint at 280 m deep and 100 km long, Lake Manly of Death Valley at 275 m deep and 175-240 km long, and Lake Manix at 60 m deep and 515-775 sq km in area (Blanc & Cleveland, 1961; Oakeshott, 1971). Until it disappeared about 500 years ago, Lake Cahuilla, a predecessor of the Salton Sea, filled the Salton Basin with about 5,180 sq km of fresh water (Lindsay & Lindsay, 1978). TERTIARY VEGETATION IN SOUTHERN CALIFORNIA It may be of some interest to examine the past vegetation of our present deserts and adjacent regions of southern California. Presently, ac- cording to my moderately liberal classification, there are 140 angiosperm families indigenous in California. The state had a much richer family representation in the Tertiary. Both mega- and microfossils indicate that at least 40 more an- giosperm families were present in California dur- ing Late Cretaceous and Tertiary time (not to 644 mention at least nine additional families of pte- ridophytes and gymnosperms) (Potbury, 1935; MacGinitie, 1937, 1941; Axelrod, 1950e, 1958, 1973; Drugg, 1967; Jarzen, 1980; Page, 1981) but have been eliminated from the flora by changed climatic conditions. Presumably other, especially herbaceous, families failed to be fos- silized or to be recognized. Many of these now exotic families are found today to the south in Baja California or east in southern Arizona, often near California’s borders, such as the Malpigh- iaceae, Menispermaceae, Moroideae, Passiflora- ceae, Sapindaceae, and Sapotaceae. Others can be found in the Cape Region of Baja California, subtropical Sonora or Sinaloa, or farther south in Mexico, as Aquifoliaceae, Bombacaceae, Bux- aceae, Chloranthaceae, Chrysobalanaceae, Clethraceae, Combretaceae, Dilleniaceae, Ebe- naceae, Flacourtiaceae, Gunneraceae, Nyssaceae, Proteaceae, Sabiaceae, Symploca- ceae, Theaceae, and Tiliaceae. Others today find refuge in eastern Asia, as Alangiaceae, Cercidi- phyllaceae, and Trapaceae, or in Australia, as Alangiaceae and Eupomatiaceae. PALEOGENE Early in the Tertiary the topography and cli- mate of California was very different from the present, and tropical seas covered much of coast- al and central California. Much of southern Cal- ifornia west of the San Andreas fault had not yet joined the state, being then part of Baja Califor- nia, which in turn was part of mainland Mexico, the Gulf of California not yet having been formed. Topography was low and rolling, and temperate (cloud) to subtropical rainforest apparently cov- ered the coast, and subtropical savanna and dry tropical (short-tree) forest probably covered much of the interior, possibly with oak woodland on the highlands above. Farther north a lush ever- green-deciduous hardwood forest similar to that found today in southeastern China, vegetated much of the state’s interior (Potbury, 1935; MacGinitie, 1937, 1941; Axelrod, 1950e, 1958, 1973, 1979; Raven & Axelrod, 1977). MIOCENE During the long Miocene Epoch the region that presently forms southern California supported in coastal areas and on moister slopes inland a gen- eralized woodland of sclerophyllous trees dom- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 inated by live oaks (Quercus) and laurels (Persea, Ocotea, Nectandra, and Umbellularia), but with a rich assemblage of other genera as well, in- cluding Acer, Annona, Bumelia, Cedrela, Cle- thra, Dioon, Ficus, Ilex, Juglans, Laurocerasus, Lithocarpus, Lyonothamnus, Magnolia, Myrica, Pinus, Platanus, Populus, Prunus, Sabal, and Sa- pindus, as found variously in the Anaverde, Mint Canyon, Mount Eden, Puente, Piru Gorge, and Temblor floras (Axelrod, 1937, 1939, 1940, 1950b, 1950c, 1950d, 1973). This vegetation thrived on ample summer rainfall and a mild climate with no frost in winter. As Miocene cli- mate became drier and cooler, many of the gen- era were forced south out of mainland California. Those genera left in Alta California are found now scattered in island oak woodland, riparian woodlands, southern mixed evergreen forest, or other more mesic communities. Closed-cone co- nifer forests (dominated by pines equivalent to г Pinus muricata D. Don and Р. radiata D. Don) occupied sheltered slopes by the ocean (Axelrod, 1980b). In the drier interior, particularly in the present desert regions, oak woodland covered the low- lands with riparian woodland following the waterways. Characteristic of the woodland-sa- vannas were Quercus, Pinus, Cupressus, Aescu- lus, Juglans, and Robinia and of the riparian woodlands Acer, Brahea, Bumelia, Celtis, Ju- glans, Lyonothamnus, Platanus, Populus, Sabal, Salix, and Sapindus. Chaparral occupied the well- drained slopes, short-tree forest the moister slopes, and subtropical thornscrub the rocky, lower, drier slopes. All of these communities, like the coastal sagescrub (Axelrod, 1978), closed-cone conifer forest (Axelrod, 1980b), and oak-laurel woodland, belong to the Madro-Tertiary flora, which had apparently evolved on dry sites dur- ing the Eocene out of the warm-temperate-sub- tropical coastal rainforests through capacity to endure dry, cool conditions. This southern flora, with more evergreen, thicker, and smaller leaves than the Arcto-Tertiary Flora to the north, soon spread in Oligocene and Miocene time over much of the drier Southwest (Axelrod, 1950a, 1950e, 1958, 1973, 1979) Most exotic to Californians would be the sub- tropical thornscrub, retained now only in part in the microphyllous woodland and mixed desert- scrub of the warmer Colorado Desert. It con- tained many trees and shrubs with small leaves or leaflets, and often with spines, of such genera as Acacia, Bursera, Castela, Celtis, Cercidium, 1986] Colubrina, Condalia, Crossosoma, Dodonaea, Euphorbia, Eysenhardtia, Fouquieria, Jatropha, Karwinskia, Leucaena, Lycium, Lysiloma, Ma- honia, Olneya, Pachycormus, Piscidia, Pithecel- lobium, Prosopis, Prunus, Psorothamnus, Ran- dia, Thouinia, Xanthoxylum, and Ziziphus, the palms Brahea, Sabal, and Washingtonia, the ar- borescent Nolina, the vines Cardiospermum and Passiflora, and such larger leaved genera of trees as Diospyros, Ficus, Quercus, and Trichilia as found in the Anaverde, Mint Canyon, Mount Eden, and Tehachapi floras (Axelrod, 1937, 1950a, 1950b, 1950c, 1973). Although it re- quired adequate summer rainfall and mild win ters, the thornscrub occupied semiarid donis This vegetation today is best developed south of the border where rain falls mostly in the warm season and frosts are rare or absent (Trichilia species, for example, require such freedom from LATE MIOCENE-PLIOCENE-EARLY PLEISTOCENE The Baja California peninsula is believed to have begun its separation from continental Mex- ico in Middle Miocene time, but with most of the rifting and spreading taking place in the last 6-4 Ma, with movement northwestward of per- haps 260 km (Larson et al., 1968; Seyfert & Sir- kin, 1979). What is now southern California west of the San Andreas Fault likewise moved north- westward in Early Pliocene about 300 km in the km also northwest- ward (Axelrod, 1979). Precipitation decreased in late Miocene with apparent elimination of the short-tree forest from the present desert areas of southern California and impoverishment of the oak-conifer wood- land and thornscrub, as indicated by the Ricardo and Anaverde floras from the western Mojave (Axelrod, 1939, 1950c, 1979). Because aridity peaked about 5-8 Ma, it is likely too that semi- desert patches increased in number and extent in drier areas of southern California toward the close of the Miocene. Previous to Pliocene time the sclerophyllous now the Mojave and Colorado deserts. With the elevation ofthe Mojave area during the Pliocene, THORNE— MOJAVE AND COLORADO DESERTS 645 the resulting colder climate eliminated most of the thornscrub and other more tropical elements from the Mojave highlands and replaced them largely with Great Basin taxa from the north. The subtropical taxa and a great diversity of life forms. For the most arid sites of the Colorado Desert a semidesert flora can be inferred (Axelrod, 1979). In late Pliocene-Pleistocene time the elevation of the Sierran, Transverse, and Peninsular ranges, as discussed above, created the regional rain- rent off the southern California coast restricted the summer rains farther south. Those floristic elements requiring summer precipitation and warmer winter temperatures were eliminated from the California deserts and found refuge in more southern divisions of the Sonoran Desert. From the various plant formations that had veg- etated the present desert or adjacent areas those Madro-Tert pockets of semidesert vegetation into the open ecological niches to form the basis of our present desert flora during the Pleistocene interglacials and subsequent Holocene time. Some of the movements of the local vegetation during the last, Wisconsin, glacial episode and the postgla- cial period are discussed below. PRESENT DESERT VEGETATION If our California deserts are largely of Late Pliocene-Pleistocene age, the desert plant com- munities are even more recent, having evolved from existing Madro-Tertiary plant formations as the desert-forming rain-shadows developed. These plant communities are of Holocene age, i.e., they acquired their present distribution and composition during the last 8,000-11,000 years since the retreat of the last, Wisconsin, episode of the numerous cycles, perhaps 19 or 20 ac- cording to Imbrie and Imbrie, (1979), of Pleis- tocene continental glaciation. The Wisconsin peaked from 21,000-1 1,000 years ago (as inter- preted by Cole, 1982). We have been treated to a relatively clear con- cept of the vegetation of the Great Basin, Mojave ugh the efforts of many botanists and e 646 Laudermilk and Munz (1938), Martin, Sabels, and Shutler (1961), and Martin and Sharrock (1964) have studied the coprolites of ground sloths, carnivores, and man. Martin (1964), Mar- tin and Gray (1962), Martin and Mehringer (1965), Mehringer (1965, 1966), and Mehringer and Haynes (1965) have used pollen analysis to interpret Pleistocene and Holocene desert envi- ronments. Especially productive has been the study of packrat (Neotoma) middens by Wells and Jorgensen (1964), Wells and Berger (1967), Mehringer and Ferguson (1961), Van Devender and King (1971), Lanner and Van Devender 1974, Phillips and Van Devender (1974), King (1976), Van Devender (1976), Van Devender and Mead (1976), Wells (1976, 1983), Wells and Hunziker (1976), Van Devender and Spaulding (1979), Cole (1982), Wells and Woodcock (1985), Cole and Webb (1985), and Cole (1986). Dendrochron- ology has also been useful. Apparently until early Holocene time (ca. 9,000 years ago), the Mojave Desert lowlands were veg- etated by a coniferous woodland of low-statured junipers and pinyons, chiefly Juniperus osteo- sperma (Torrey) Little and Pinus monophylla Torrey & Frem., 0 j panied by Yucca brevifolia Engelm. in S. Watson, Yucca whipplei Torrey, Coleogyne ramosissima Torrey, and Artemisia, Ephedra, and Opuntia spp., and other xerophytes (Wells & Berger, 1967; King, 1976; Wells, 1983; Wells & Woodcock, 1985). Absent from these woodlands were such hot desert plants as Larrea divaricata Cav. and Ambrosia dumosa (A. Gray) Payne. The first Neotoma records of Larrea in the Mojave Desert are from the north side of the Ord Mountains about 7,400 years ago (Wells 1983) and from the Lucerne Valley side of the same range to the south at about 5,800 years ago (King, 1976). Va- sek (1980) has estimated the oldest Larrea clones in the nearby Johnson Valley at about 7,800 years. Presumably the hot-desert species like Larrea found refuge during the glacials in the low, warm Colorado Desert (Wells & Berger, 1967). Larrea has been reported in Death Valley (Wells & Woodcock, 1985) no earlier than 1,990 + 160 years ago. In a study of Neotoma middens in the very arid lower Colorado River Valley about Picacho Peak, Imperial County, California, at elevations of 245-300 m, Cole (1986) documented 12,500 years to the present of creosotebush desert scrub dominated by Larrea divaricata, Encelia fari- nosa A. Gray ex Torrey, and Peucephyllum ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 schottii (A. Gray) A. Gray. He considered this hyperarid area to be a Pleistocene desert refu- gium. At this site Joshua tree, Yucca brevifolia, perhaps the most characteristic Mojave Desert species and now largely restricted to the Mojave, was present about 12,500 years ago, and asso- ciates, such as Brickellia arguta Robinson, Chry- sothamnus teretifolius (Dur. & Hilg.) Hall, Co- were present into the early Holocene until nearly 10,000 years ago. The authors also mentioned late Wisconsin and early Holocene Pinus mon- ophylla from as low as 460 m in the Tinajas Altas Mountains of Arizona and a more xeric juniper probably Juniperus californica Carr.) woodland from 600 to 240 m in the Whipple, Chemihuevi, Tinajas Altas, and Butler mountains [until about 8,900 years ago, according to Van Devender (pers. comm In contrast, Ambrosia dumosa, Fouquieria splendens Engelm., and Olneya tesota A. Gray appeared in middens only less than 600 years old. Possibly these and other Sonoran, hot-desert plants requiring warm temperatures and summer — Desert, returning for short periods in each т- terglacial. The Mojave Desert in turn served as a refuge for the cold-desert plants, as well as the pinyon- juniper woodlands, which in Holocene time re- turned to the Great Basin to become the domi- nant vegetation there. Pinus flexilis James and P. longaeva Engelm., now absent from Mojave Desert ranges, were recorded by Neotoma from the Clark Mountains about 20,000 years ago, with P. /ongaeva already absent and P. flexilis rare by 12,460 BP (Mehringer & Ferguson, 1969). Wells and Woodcock (1985) reported Yucca whipplei in Death Valley as early as 19,550 + 650 BP along with Atriplex confertifolia (Torrey & Frem.) S. Watson and Opuntia basalaris, and somewhat later (about 17,000 to 9,500 BP) Chry- sothamnus teretifolius, Purshia glandulosa Cur- ran, and Haplopappus cuneatus A. Gray. Am- brosia dumosa appeared there by 10,230 + 320 BP, but Larrea not until 1,990 BP. The presence of large, grazing mammals, such as horses, camels, bison, and mammoth, and such browsers as mastodons and ground sloths in the Mojave Desert during the Pleistocene pluvial pe- riods would seem to require that the coniferous 1986] woodlands probably were well supplied with grasses and other edible herbage. The China Lake , ade carnivores as the saat (ова cat (Smilodon), a dire wolf, and a coyote (Fortsch, 1978). Fortsch designated ita Rancholabrean fauna, and cited one pr eliminar y age of fossil i 1VOry from a mammoth at 18,600 + 4,500 years BP. Davis and Panlaqui (1978) inferred from this faunule that Lake China could be reconstructed as a “shallow, wind-stirred, fresh water lake; marshes, sloughs and sluggish streams surround- ing the lake; some gallery forests of lakeside trees. These aquatic habitats were влез at times, by a tall-grass savanna . wever, during the heights of the stadials, the ee al back- ground of Lake China was evidently a cold steppe with some xeric, desert forest such as juniper.” The PaleoIndians, who were general foragers, preyed occasionally upon the large herbivores, and were considered by Davis and Panlaqui to be culturally and linguistically complex by 15,000 BP Another site of Late Wisconsin/Holocene Transition time, in which early American man is known to have overlapped with now extinct large herbivores, is the Lehner Mammoth site in southeastern Arizona, from which the fossil pol- len was studied by Mehringer and Haynes (1965). Clovis fluted points, butchering tools, and char- coal were found there associated with the re- mains of mammoth, horse, tapir, and bison. Both grazing mammals and the pollen gave evidence of desert-grassland 11,200 years ago. Packrat midden studies in Death Valley es- tablished a 1,200-1,500 m downward displace- ment of juniper woodland 13,000-19,000 years ago (Wells & Woodcock, 1985). Similar studies in the Grand Canyon have suggested up to a 1,000 m climatic depression during the full-gla- cial period (21,000-15,000 years ago) of most plant species, particularly in the juniper and blackbush scrub communities (Phillips & Van Devender, 1974; Cole, 1982). These findings support, а! t least for Menar Sevan deserts, a model o King (1976) concluded that his packrat midden data from the Lucerne Valley of the Mojave Desert suggested a 365 m depression of Juniperus os- teosperma woodland between 12,100 and 7,800 years ago. In a study of Neoglacial vegetation changes in Greenwater Valley near Death Valley, Cole and Webb (1984) found a downward shift THORNE—MOJAVE AND COLORADO DESERTS 647 of 50 to 100 m in plant communities during the past 500 years. Thus, the latitudinal and eleva- tional movement of plant communities is still going on, along with the addition to and deletion of species from the communities. Our present desert plant communities are not only very re- cent, they are highly dynamic, indeed kaleido- scopic in content and location. ORIGINS OF THE CALIFORNIA DESERT FLORA The present California desert flora is largely composed of autochthonous elements from the Madro-Tertiary Geoflora that dominated south- ern California and adjacent areas throughout the Tertiary. Other, rarer floristic elements, also pre- adapted to long periods of drought, hot, dry sum- mers, and cool, wet winters, have arrived from other parts of America and the rest of the world to complete the desert flora. These latter ele- ments deserve some discussion. In addition to the rather obvious Sonoran ele- ments, Mexico has supplied many of the alloch- thonous floristic elements found in the California deserts. From their distribution and that of their closest relatives, it would appear that Mortonia utahensis (Cov.) Nelson (Prigge 1983), Buddleja utahensis Cov., Yucca brevifolia, Y. schidigera Roezl ex Ortgies, Yucca whippleyi, Castela emo- ryi (A. Gray) Moran & Felger, and Pilostyles thurberi A. Gray, among others, have arrived in California from the southeast, probably from the Chihuahuan Desert of northern Mexico. Fou- quieria splendens Engelm., now very conspicu- ous in the Colorado and other subdivisions of the Sonoran Desert along with several of its con- geners in Baja California, has most ofits relatives in mainland Mexico, south to Oaxaca (Henrick- son, 1972). Other possibly non-Sonoran Mexi- can immigrants are such plants as Selinocarpus nevadensis (Standley) Fowler & Turner, Pholis- ma arenarium Nutt. ex Hook., P. (Ammobroma) sonorae (Torrey ex A. Gray) Yatskievych, and Proboscidea althaeifolia (Benth.) Duchesne, all with strong relationships to the south with Mex- ican, but with no close Californian, relatives. South America has surely supplied a number of California desert plants, not the least being the creosote-bush, Larrea divaricata Cav. [or L. tri- dentata (Sessé & Mociño ex DC.) Cov. for those who consider it a distinct species], surely the most America. The North American desert plant, so 648 closely similar to an Argentine species both mor- phologically and biochemically as to be probably conspecific with it, has differentiated cytogeo- graphically since its arrival in North America (Wells & Hunziker, 1977). The Chihuahuan Des- ert race retains the ancestral diploid condition but the Sonoran and Mojave Deserts have de- rived tetraploid and hexaploid races respective- ly. The recent advent of Larrea in the Mojave Desert has been discussed above. Although L. divaricata has races in South America as far north as Peru, the North American races have closer biochemical affinity with the Argentine race. Larrea probably was carried to Mexico directly from Argentina or Chile by migrating birds, and subsequently moved from the Mexican deserts into the Californian deserts in late interglacial and Holocene tim Other distinctive aaa species with probable South American ancestry are Frankenia salina (Molina) I. M. Johnston, disjunct between west- ern Mexico-California and Chile, the cappara- ceous Koeberlinia spinosa Zucc., disjunct be- tween the North American Southwest and Bolivia, and Atamisquea emarginata Miers, dis- junct between northwestern Mexico-Arizona and Argentina- Chile. The North American desert Lycium and Nicotiana likewise have strong links with temperate South America. Menodora and Lycium have representatives as well in southern Africa, and Nicotiana also in Africa, Australia, and Polynesia. Thamnosa of the Rutaceae seems to be unrepresented ie South America but does have species in southwestern Africa, the Horn of Africa, and Arabia. Possibly it followed migra- as a species in Somalia, like Thamnosma, but no species in southern Africa or South America. Most of the desert herbaceous genera that have links to South America appear to have evolved in North America and to have been carried south to temperate South America relatively recently, no doubt also by migratory birds (Raven, 1963; Thorne, 1973). The more northern elements of the California deserts have apparently entered the warmer des- erts from the Great Basin Semidesert. The link- age of many of these genera seems to be strong ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 with the arid areas of central Asia. Among the more prominent desert genera with a possible Asian origin are Artemisia, Atriplex, Ceratoides (Eurotia), Kochia, Mahonia, Monolepis, Prunus subg. Amygdalus, Stipa, and Suaeda. These could speciated rather heavily in western North Amer- ica, suggesting that they may have been early immigrants. The linkage between the Mediterranean region with members of both the Californian and Trans- montane botanical regions also seems surpris- ingly strong considering the distant disjunction involved. Porter (1974) believed that the North American taxa of the zygophyllaceous Fagonia may be derived from North African species and that the South American species may have in- dependently reached there from South Africa. Likewise, Peganum mexicanum A. Gray, indig- enous to the Chihuahuan Desert, may be derived from a North African ancestor although the ge- nus reaches also to central Asia and Mongolia. Oligomeris linifolia (Vahl) Macbr., ranging from eastern Mexico across Texas to our southwestern deserts, also occurs in the Mediterranean region. It is the only species of the Resedaceae appar- ently indigenous in flavus (Decne.) Schultz-Bip. (S. mohavensis A. Gray) have a similar distribution. Some of these species may be recent immigrants, having en- tered with the early Spanish conquistadors. Oth- er southwestern desert genera with strong rela- tionships to the Mediterranean region are Antirrhinum phus microphyllus A. Gray, and Ribes spp., that have found refuge in the more mesic highlands of the Mojave Desert have more ancient linkage with eastern American and Eurasian represen- tatives. Thus the flora of the California deserts is of most diverse origin, having been assembled from both adjacent and far-distant sources. Many ele- ments have reached the Southwest by relatively normal overland movement of propagules over long periods of time, propelled no doubt by changing climate. Others may have crossed nar- rowed straits, seas, or even oceans when sea-level was lowered in glacial epochs, or much earlier 1986] when continents were closer to one another, al- current patterns surely were sometimes in- volved. In other instances, as especially in the amphi-tropical American disjunctions, the only logical explanation has to be long-distance car- riage by migratory birds. This assembling of the Southwestern desert flora has been going on for tens of millions of years, certainly from late Cretaceous to Holocene time. Some of the most distinctive desert taxa, mostly with no close living relatives, could have evolved in Early Tertiary or even Late Creta- ceous time. Among them can be listed such en- demic western American taxa as the Crossoso- mataceae, Fouquieriaceae, Garryaceae, Krameriaceae, Lennoaceae, and Simmondsi- aceae. It is likely that some of the herbaceous genera that have speciated so heavily in the southwestern deserts to produce the abundant ephemeral annuals have done so in the Holo- cene, some possibly in the past couple thousand years. Among them are Astragalus, Camissonia, Chamaesyce, Chorizanthe, Cryptantha, Descu- rainia, Eriogonum, Gilia, Lupinus, Mimulus, Nemacladus, and Phacelia. LITERATURE CITED ds pons T. 70. Implications of plate tectonics r the Cenozoic tectonic evolution of western North America. Geol. Soc. Amer. Bull. 81: 3513- 536 AXELROD, D. I. 1937. A Pliocene flora from the Mount Eden beds, southern California. Publ. Carnegie Inst. Wash. 476: 125-183 939. A Miocene flora from the western bor- der of the Mojave Desert. Publ. Carnegie Inst. Wash. 516: 1-128 1940. The Mint Canyon flora of southern California: a preliminary statement. Amer. J. Sci. 238: 577-585. 1950a. Classification of the Madro-Tertiary flora. Publ. Carnegie Inst. Wash. 590: 1-22. 1950b. Further studies of the aen Eden Publ. Carnegie Inst. flora, nen California. Wash. 590: 73-117 1950c. The Anaverde flora of southern Cal- ifornia. Publ. Carnegie Inst. Wash. 590: 119-158. 1950d. The Piru Gorge flora of southern Cal- ifornia. Publ. Carnegie Inst. Wash. 590: 159-214. 1950e. Evolution of desert vegetation in western North America. Publ. Carnegie Inst. Wash. 590: 215-306. . 1958. Evolution of the Madro-Tertiary Geo- flora. Bot. Rev. (Lancaster) 24: 433-509. —— —. 1962. Post-Pliocene uplift of the Sierra Ne- THORNE—MOJAVE AND COLORADO DESERTS 649 vada, California. Bull. Geol. Soc. Amer. 73: 183- 197. . 1966. The Pleistocene Soboba flora of south- ern California. Univ. Calif. Publ. Geol. Sci. 60: 1- 79 1973. History of the Mediterranean erga tem in California. Pp. 225-277 in F. di с a systems, Origin and Structure. S T Yor 19 Origin of coastal sage в, Alta and Baja California. Amer . J. Bot. 65: 1117-1131. Contributions to the Neogene Paleo- botany of = California. Univ. of California Press, Berke 1980b. Hii of the maritime closed-cone pines, Alta and Baja ie mee Univ. California Publ. ж, Sci. 120: W. S. TING. Г. ү te Pliocene floras east of the Sierra Nevada. Univ. California Publ. Geol. : 1-118. BACHMAN, S. B. 1978. Pliocene-Pleistocene break-up of the Sierra Nevada-White-Inyo Mountains block and formation of Owens Valley. Geology 6: 416- 463. BLANC, R. P. & G. B. CLEVELAND. 1961. Pise jo lakes of pore is California — 1. Mine v. California Div. Mines 14(4): un "red Front of the San кеен Кер tation a the eastern Grand Canyon. Science 217: 1142-1145. T. The lower Colorado River Valley: a Pleistocene desert. J. Quat. are 25: 392-400. R. WEBB. 19 e Holocene vege- tation changes i in Greenwater Valley, Mojave Des- rnia. J. Quat. Res. 23: 227-235. C. PANLAQUI. 1978. Present and past environments of the valley The Mojave of Los T. i cles of the Upper Mo- reno Formatio on (Late Cretaceous-Paleocene), Es- carpado Canyon, California. Palaeontographica 120: 1-71, 9 plates. FortscH, D. E. 1978. The Lake China Ranchola- brean Faunule. Pp. 173-176 in E. L. Davis (edi- tor The Ancient Californians. Rancholabrean Hunters of the Mojave Lakes Country. Natural Hist. Mus. of Los Angeles Co., Los Angeles. Foster, J. H. 80. Late Cenozoic Tectonic Evolu- tion of Cajon Valley, Southern California. Ph.D. Thesis. Univ. California, Riverside. HENRICKSON, J. S. 1972. A taxonomic revision of the Fouquieriaceae. Aliso 7: 439-53 IMBRIE, J. & К. P. IMBRIE. 1979. Ice Ages— Solving ie Mystery. eee Publ., Short Hills, New Jer- Sd D. M. 1980. The occurrence of Gunnera pol- len in the fossil record. Biotropica 12: 117-123. 650 KinG, T. J. 1976. Late Pleistocene-Early Holocene history of coniferous woodlands in the Lucerne Valley region, Mojave Desert, California. Great Basin Naturalist 36: 227-238. LANNER, R. M. & T. R. VAN DEVENDER. 1974. Mor- phology of pinyon pine needles from fossil packrat middens in Arizona. Forest Science 20: 207-211 Larson, R. L., H. W. MENARD & S. M. SMITH. 1968. Gulf of Eus a result of ocean-floor spread- ing and bw faulting. Science 161: 781-784. LAUDERMILK, J. D. & 1934. Plants in the dung of Nothrotherium from Gypsum Cave, Nevada. Publ. Carnegie Inst. Wash. 453: 29-37. LINDSAY, L. & О. LiNpsAv. 1978. The Anza-Borrego Desert Region. Wilderness Press, Berkeley. МАССІМІТІЕ, H. О. 1933. The Trout Creek flora of southeastern Oregon. Publ. Carnegie Inst. Wash. 416, II: 21-68. . 1937. The flora of the Weaverville beds of Trinity County, California. Publ. Carnegie Inst. Wash. 465, III: 83-151. . 1941. A middle Eocene flora from the von Sierra Nevada. Publ. Carnegie Inst. Wash. 534: 178 Martin, P. S. 1964. Pollen vpn. and the full- glacial landscape. Pp. 66-74 in J. J. Hester & J. Schoenwetter (editors), The Re diee of Past Environments. Fort еа Res. Center Publ. -89. Taos, New Mexi & J. Gray. 1962. Pollen E and the Cenozoic. Science 137: 103-1 P. J. MEHRINGER, JR. em Pleistocene pollen Ap ica and чш ы ЖЛ of the а h- west. 433-451 in H. E. Wright, Jr. & Frey с. The Quaternary of the ‘United t Princeton Univ. 1961. Southwestern Palynology and Prehistory, the Last 10,000 Years. Univ. Arizona Geochronology Labs., Tucson. F. W. SHARROCK. prehistoric human feces: a new approach nobotany. Amer. Antiquity 30: 168-180 MEHRINGER, P. J., JR. 1965. Late Pleistocene vege- tation in the Mohave Desert of southern Nevada. J. Arizona Acad. Sci. 3(3): 172-188. 1966. me notes on the Late Quaternary biogeography of the Mojave Desert. Geochronol- y La Uni 1: 1964. Pollen analysis of to eth- S., v. Arizona, Interim Res. Rep. | . W. FERGUSON. 1969. Pluvial occurrence of bristlecone pine (Pinus aristata) in a Mohave Desert mountain range. J. Arizona Acad. Sci. 5: о У. Haynes. 1965. The pollen evidence for the environment of early man and extinct an- imals at Lehner mammoth site, southeastern Ar- 23. . California’s Changing Land- A Gu VA ^ the Geology of the State. McGraw-Hill Book Co., New York. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 РАСЕ, V. M. 1981. Dicotyledonous wood from the Upper Cretaceous of central California. III. Con- rnold Arb >. I Я N ocene packrat middens from the lower Grand = of Arizona. Arizona Acad. Sci. 9(3): 117- pu D. 1974. Disjunct distributions in the New YT Zygophyllaceae. Taxon 23: 339-346. Ротвову, S. S. 1935. The La Porte flora of Plumas County, California. Publ. Carnegie Inst. Wash. 465: 9-81. PRIGGE, B. A. Studies on Acanthothamnus, Mortonia, sum Orthosphenia A anat- omy, ecology, and system Ph. esis. Claremont бш School, CU Califor- nia. RAVEN, P. H. 1963. Amphitropical relationships Е the floras of North an a South America. Qua Rev. Biol. 38: 151-17 & Origin and relation- ships of the California к Univ. California Publ. P SEYFERT, С. К. & L. A. Sirkin. 1979. Earth History and Plate Tectonics, 2nd edition. Harper & Row, w York. THORNE, R. F. 1973 [1972]. Major disjunctions in the geographic ranges of seed plants. Quart. Rev Biol. 47: 365-411. 1978. Plate tectonics and angiosperm distri- bution. Notes Roy. Bot. Gard. Edinburgh 36: 297- 315. VAN DEVENDER, T. R. 1976. The biota of the hot deserts of North America during the last p tion: the packrat midden — Am. Quat. Assoc Abstr. 4th Biennial Mtg., p T. ть su кәч й tational records | in western Arizona. J. Arizo Acad. Sci. 6: 240-244. J. I. MEAD. 1976. Late л апа апа Реасһ Springs Wash, lower Grand os An zona. J. Arizona Acad. Sci. 16-22. & W. G. SPAULDING. 1979. Development of vegetation and climate in the southwestern United States. Science 204: 701-710. VASEK, F. 1980. Creosote bush: longlived clones in the Mohave Desert. Amer. J. Bot. 67: 246-255. ELLS, P 1966. Late Pleistocene vegetation and degree of pluvial climatic change in the Chihua- huan Desert. Science 153: 970-975 . 1976. Macrofossil analysis of wood rat (Neot- oma) middens as a edd to history of arid America. J 1983. Paleogeography of m slands in the Great Basin eer the last parda Ecol. Monogr. 53: 34 & R. BERGER. 1967. Late Pleistocene history of coniferous woodland in the Mohave Desert. Science 155: 1640-1647. J. H. HUNZIKER. 1976. Origin of the creo- jud is (Larrea) deserts of southwestern North America. Ann. Missouri Bot. Gard. 63: 843-861. 1986] THORNE— MOJAVE AND COLORADO DESERTS 651 C. D. JORGENSEN. 1964. Pleistocene wood ——— & D. Woopcock. 1985. Full- anne vegeta- rat middens and climatic change in Mohave Des- tion of Death Valley, California: Jun od ert: a record of Juniper woodlands. Science 143: land opening to Yucca semidesert. Madroño 32: 1171-1174. 11-23. Volume 73, No. 2, pp. 225-501 of the Annals of the Missouri Botanical Garden, was published on August 21, 1986. INFORMATION _ The ANNALS publishes original manuscripts in systematic botany and related fields. Authors are asked t order tt o: editing and publication. Машка ed properly may be returned for re- vision prior to review. If an author feels that his manuscript presents special problems, he should write the editor concerning the best way to han- dle these before submitting the manuscript. Page itor will help authors to seek additional funding if necessary. Corrections on manuscripts in the proof Stage will cost $2.50 per line and are not subject to negotiation. anuscripts must be rinewiidan on one side of good quality (not erasable) 8%" х 11" paper. The manuscript should have at least one inch margin all around and be double-spaced through- out, including the abstract, notes, legends, ta- | bles, lists of specimens, and literature cited. Fig- ure legends and tables should be typed separately ~ placed at the end of the text. Legends for each group of figures should be typed as one paragraph (consecutive paragraphs on one or ated ze separate from figures themselves), ^e € style found in the current issue of the a даю Authors should indicate in the margins Pproximate places for illustrations and ta- ! A аздек to granting agenci be 5, i bl x S Include ude author's name and page number in ора | origin E tone illustrations x т of the manuscript is required, and the ~ e +. should retain a copy of the final, me E notin figure lege ed FOR AUTHORS omous, indented, and parallel. The couplets should be numbered 1a, 1b, 2a, 2b and so forth. = representative specimens as follows: Мех- . Oaxaca: Sierra San Pedro Nolesco, нае рэт 865 (ВМ, С, К, US). Use herbari acronyms as neri in Index Herbariorum sene num Veg. 106). Abbreviations should be checked for consis- tency and to make sure they are unambiguous. Periods are used after all abbreviations "e metric measures, compass directions, and h barium designations. Use authority ve tions as listed in Draft Index of Author Abbre- viations compiled at the Herbarium, Royal Botanic Gardens, Kew (1980). Except in the bib- liography, names of books are abbreviated ac- cording to Taxonomic Literature; in and out of the bibliography, periodicals are abbreviated ac- cording to Botanico-Periodicum- Huntianum. Send copies of illustrations; originals will be requested after the manuscript is accepted. All illustrative materials should be mounted on stiff cardboard. A maximum size ofa printed illus- tration is 5%” x 8⁄4”. P. either column width (2%" о or ca. 1. 68 mm) с or r full page width (5%” or ca. 140 mm). Figures are numbered consecutively with Arabic numbers, in the order they are referred to in the text. Pho- tographs should be sharp. glossy, black-and-white prints. Abut cu of photographs in composite ix line сору and continuous- on ‹ clude a metric scale and reference to latitude and longitude; where appropriate, size € rinm should be i ncluded in n^ ographs or dra 5, ends. A incipio Шшйханов» ш d be compiled with to make sure each of style, consuli a recent . The Chicago Manual of tion, University of Chi- ite to the edit ANNALS MISSOURI] BOTANICAL GARDEN кез 1986 NUMBER 4 СОМТЕМТ5 Seed morphology in North American ee Stanwyn G. Shet- ler & Nancy R. Morin i G54 Revision of the Guayana Highland E Bromeliaceae Lyman B. Smith 689 Adiciones a las Leguminosas de la Flora de Nicaragua Mario Sousa S. 722 Three New Species of Solanum Section Geminata (G. Don) Walp. (Sola- a naceae) from Panama and Western Colombia Sandra Knapp 738 | А Survey of Solanum Prickles and Marsupial Herbivory in Australia D SS } E. Symon 145 Zapoteca: A New Genus of месар Mimosoideae Héctor M. Her- __ | nandez 755 Distribution of Nonprotein Imino and Sulphur Amino Acids in Zapoteca i i John T. Romeo s i 164 € A Comparative Study of the En abyoti of Ludwigia (Onagraceae): Char- acteristics, Variation, and Relationships Hiroshi Tobe & Peter H. Raven 768 Koehneria, a New Genus of Lythraceae from Madagascar Shirley A m | Graham, Hiroshi Tobe & Pieter Baas 788 | Wood Anatomy of Lythraceae — Additional Genera (Capuronia, Galpinia, Haitia, Orias, and Pleurophora) Pieter Baas 810 = "3unfinu ) hark rover Contents continued on back cove! VOLUME 73 WINTER 1986 NUMBER 4 ANNALS MISSOURI В BOTANICAL GARDEN The ANNALs, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be ac- cepted. Authors should write the Editor for information concerning arrangements for publishing in the ANNALS. Instructions to Authors are printed on the inside back cover of the first issue of this volume. EDITORIAL COMMITTEE ANCY Morin, Editor Missouri Botanical Garden MARSHALL R. CROSBY Missouri Botanical Garden RIT DAVIDSE Missouri Botanical Garden JOHN D. DwvER Missouri Botanical Garden & Saint Louis University TER GOLDBLATT Missouri Botanical Garden For е information contact са 11; . Box. 299, St. Louis, MO 631 Sone price is $70 pe $ Airmail d delive Pri 63110. Second class dcs MASTER: Send address ch: GARDEN, P.O. Box 299, S INFORMATION _ The ANNALS publishes original manuscripts іп systematic botany and related fields. Authors are asked to follow the suggestions below in order to expedite editing and publication. Manuscripts пої prepared properly may be returned for re- | vision prior to review. If an author feels that his _ manuscript presents special problems, he should . write the editor concerning the best way to han- . dle these before submitting the manuscript. Page . costs are $60.00 per page; charges may be re- : duced under special circumstances, and the ed- itor will help authors to seek additional funding _ if necessary. Corrections on manuscripts in the . proof stage will cost $2.50 per line and are not Subject to negotiation. E Manuscripts must be typewritten on one side _ of good quality (not erasable) 82” х 11" paper. . The manuscript should have at least one inch ; К margin all around and be double-spaced through- . ош, including the abstract, notes, legends, ta- . bles, lists of specimens, and literature cited. Fig- _ ure legends and tables should be typed separately and placed at the end of the text. Legends for each group of figures should be typed as one _ paragraph (consecutive paragraphs on one or _ more pages separate from figures themselves), using the style found in the current issue of the ANNALs. Authors should indicate in the margins the approximate places for illustrations and ta- bles. Include author's name and page number in upper right-hand corner of each page. Submis- sion of the original and two carbon or xerograph- ie copies of the manuscript is required, and the author should retain a copy of the final, typed . draft A 00418 A — а ea А a ems a ae m Acl i cies, her- =- baria, illustrators, and technical assistants should bot full mailing address should appear as the second ootnote. These should be the only footnotes used n the manuscript. An abstract must accompany each paper oth- er than “Notes.” The abstract should succinctly summarize the purpose, findings, and conclu- sions ofthe paper and should be completely com- prehensible itself. E A brief Latin diagnosis for each new taxon 15 preferred to a complete Latin description. A complete description should be given in English. - Use one paragraph per basionym as follows: xon, author, literature citation, type citation, e.g., Bauhinia dipetala Hemsl., Diag. Pl. Nov. 880. Casparia dipetala (Hemsl.) Britt. & Rose, Amer. Fl. 23: 216. 1930. Type: Mexico. Ve- у | urgeau 1713 ( _ Бе placed as a footnote on page 1. The author's - FOR AUTHORS omous, indented, and parallel. The couplets should be numbered 1a, 1b, 2a, 2b and so forth. Cite representative specimens as follows: Mex- ico. Oaxaca: Sierra San Pedro Nolesco, Talea, Jurgensen 865 (BM, G, K, US). Use herbarium acronyms as listed in /ndex Herbariorum (Reg- num Veg. ). Abbreviations should be checked for consis- tency and to make sure they are unambiguous. Periods are used after all abbreviations except metric measures, compass directions, and her- barium designations. Use authority abbrevia- tions as listed in Draft Index of Author Abbre- viations compiled at the Herbarium, Royal Botanic Gardens, Kew (1980). Except in the bib- liography, names of books are abbreviated ac- cording to Taxonomic Literature, in and out of the bibliography, periodicals are abbreviated ac- cording to Botanico-Periodicum- Huntianum. Send copies of illustrations; originals will be requested after the manuscript is accepted. All illustrative materials should be mounted on stiff cardboard, A maximum size of a printed illus- tration is 5%” x 8'A". Presize illustrations to fit either column width (2%” or ca. 68 mm) or full page width (5'^" or ca. 140 mm). Figures are numbered consecutively with Arabic numbers, in the order they are referred to in the text. Pho- tographs should be sharp, glossy, black-and-white prints. Abut edges of photographs in composite plates. Do not mix line copy and continuous- tone illustrations on one plate. Maps should in- not in figure legends. All original illustrations will be returned after publication. The bibliography should be compiled with care. The author should check to make sure each entry in the bibliography is referred to in the text, and each reference to a paper in the text is entered in the bibliography. Citations of periodical lit- erature should appear as follows: author's last name, initial(s). Year. Full title of article. Name of Journal. Volume: inclusive pagination. Omit numbers for parts of volumes unless the parts are paginated separately. For more than one au- thor, follow the style: author's last name, ini- tial(s), second author's initial(s), last name & third author's initial(s), last name. C itations of books should appear as follows: Author's last name, initial(s). Year. Full title. Edition, if applicable. Publisher. Place of publication. For other aspects of style, consult a recent issue of the ANNALs; The Chicago Manual of so Style, 12th or 13th edition, University of Chi- , Chicago; or write to the editor. uscripts should be addressed to the Missouri Botanical Garden, 631056. ANNALS OF THE MISSOURI BOTANICAL CARDEN VOLUME 73 1986 NUMBER 4 SEED MORPHOLOGY IN NORTH AMERICAN CAMPANULACEAE! STANWYN G. SHETLER? AND NANCY R. MORIN? ABSTRACT Campanulaceae -ea Lobeliaceae) in North America comprise four genera and 35 species of ial annual, biennial, and pe herbs. Generic and specific circumscriptions have been treated var- iously, and studies of Menag e been few and limited. In this study, seeds of all but one of the native North American species e of selected Eurasian putative relatives were examined with the light and scanning electron microsc divaricata, and the other eastern species o . Characteristics of the seeds and their surface cells are described and relative uniformity seen within Githopsis, Triodanis, ee and the western species of Cam- panula. Seeds of the recently азаа д C. robin pear to set each of these species apart within the fepma However in highly distinctive and would a siae and of the wide-ranging C. aparinoides are iodanis colora- Triodanis, T. texana stands ы. о nmnanal doensis more than those of other m Brief cun я on the adaptive Kei sd of seed- coat sculpturing and ornamentation is giv The Campanulaceae, in the strict sense (ex- cluding Lobeliaceae), are a worldwide family of 35 to 40 genera and perhaps 800 species. Esti- mates of the number of species have ranged from 600 to 1,000 (Avetisian, 1967; Gadella, 1974; Kovanda, 1978). The species are largely peren- nial herbs, but some are annuals, biennials, or even small shrubs. The family is confined mainly to north and south temperate regions, being re- placed in tropical and subtropical regions by the Lobeliaceae. Only Wahlenbergia, with possibly 100 species or more (Thulin, 1975; Carolin, pers. comm., 1981), is well developed in, and in fact restricted to, south temperate regions, especially nk L. Lehtonen, M.-J. , S. Wiser, and W Australia and South Africa. Campanula, the northern counterpart, is the largest genus of the family, comprising some 300 species. It is con- fined to the north temperate zone, with its center of diversity in Eurasia. Only 23 species occur in North America (Shetler, 1963; Heckard, 1969; Morin, 1980). Three-fourths of these are narrow endemics. Nearly half (11 spp.) of the North American species occurs in the California Flo- ristic Province, and, of these, seven species are endemic there. The family Campanulaceae is of similar size and diversity in Europe and Soviet Eurasia. Flora Europaea (Tutin, editor, 1976) records 13 genera own for technical assistance. This study was tha fill in 1980-1981 while ien was a асова LESE Fellow, and she gratefully acknowledges the support of the Smithsonian Institution Department of Botany, Smithsonian Institution, Washington, D.C. 20560. 3 Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. ANN. MissouRi Bor. GARD. 73: 653—688. 1986. 654 and 208 species, Flora SSSR (Fedorov, 1957), 19 genera and 223 species. Of these species, the genus Campanula accounts for 69% and 67%, respectively. Even allowing for narrower generic concepts, the Russian bellflower flora shows a slightly greater generic diversity. By contrast, the North American Campanulaceae comprise only four genera and 35 species, of which 23, or 65%, belong to Campanula. Six (26%) of the latter species are clearly annuals, nearly three times the average percentage of annuals for the genus as a whole. Four of these annual species are found only in California. Campanula americana L. has long been considered to be an annual under most circumstances in the wild, but the recent exper- imental studies of its ecological life cycle by the Baskins (1984) have shown that at least in north- central Kentucky the species behaves as either a winter annual or a biennial. In the Southern Hemisphere, the family shows limited diversity and is poorly represented ex- cept in South Africa, where there are seven, mostly small genera. Two of these, Cephalostig- ma and Lightfootia, Thulin (1975) has sub- merged in Wahlenbergia. The latter, with 46 species recognized by Thulin, is the only genus showing substantial diversification in South Af- rica. It is the only genus of the family found in Australia, where it has radiated at least as much as in South Africa, and in New Zealand, where a few species occur. In South America, where the closely related Lobeliaceae achieve their greatest diversity, the Campanulaceae are represented by only a few species in three genera, including Wahlenbergia. The first and still the only worldwide mono- graph of Campanulaceae is the one published in 1830 by DeCandolle. He recognized 21 genera and 234 species. Schonland’s (1889) systematic conspectus of the family, although badly out- dated, is still a standard reference. Generic clas- sification has continued to perplex students of the family, and the question of generic limits continues to draw attention as new lines of evi- dence are introduced (Avetisian, 1948, 1967, 1973; Charadze, 1949, 1970, 1976; Heidenhain, 1953; Fedorov, 1957; Carolin, 1960, 1967; Con- tandriopoulos, 1964, 1967, 1970a, 1970b, 1972, 1976; Gadella 1964, 1966; Phitos, 1964, 1965; Damboldt, 1965, 1970, : , 1975b; Thulin, 1975; Dunbar & Wallentinus, 1976). Four genera of Campanulaceae are indigenous ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 to North America: Campanula (23 spp.), Tri- odanis (7 spp.), Githopsis (4 spp.), and Hetero- codon (1 sp.). Three of the Campanula species have sometimes been placed in other genera. Small (1903) erected the genus Campanulastrum for Campanula americana (see also Shetler & Matthews, 1967) and the genus oe (1 id for Campanula floridana S. Wats. ex A. Gra and C. robinsiae Small. MeVaugh (19454) Sd C. prenanthoides Durand in the Eurasian genus Asyneuma. Added to the native species are about a dozen exotic species, introduced as ornamen- tals, that have escaped from cultivation and in some cases (e.g., Campanula rapunculoides L.) have become widely naturalized. At least eight species of Campanula are well established, as are the following other introduced taxa: Jasione montana L., Legousia speculum-veneris (L.) Chaix, Platycodon grandiflorus (Jacq.) А. ОС. ,* and Wahlenbergia cf. marginata (Thunberg) DC. Studies of the seeds of Campanulaceae have been few and limited (Corner, 1976). Seed char- acters have been little used in the systematics of the group, and a comprehensive study of surface morphology has never been made in this family as a whole. Te advent of the iens electron the whole field of seed morphology (Brisson & Shei 1977), however. Among the spate of recent SEM studies are Geslot’s (1980) survey of 17 European species of Campanula and the two studies dealing with campanulaceous genera by Thulin (1974, 1975), notably his survey (1975) of the seed coats of the species of Wahlenbergia in South Africa and Madagascar. Geslot concentrated particularly on Campanula subsection Heterophylla, examining 11 species of this group. Another important re- cent contribution is Carolin’s (1980) SEM study of seed surfaces in Goodenia and related genera of the sister family, Goodeniaceae. Brief reports of our own study also have appeared in print (Morin & Shetler, 1981; Shetler & Morin, 1981). In our study, we examined the seeds of 38 species of Campanulaceae— 32 of the 35 indig- enous species and six other species. Seeds could * The generic name Platycodon, compounded from the Greek, i 1978, Art . 74; Stearn, 1966), not a neuter noun, as proposed originally by Alphonse DeCandolle (1830). Accordingly, the correct name for the balloon flower is P. grandiflorus, not P. grandiflorum, as published by DeCandolle and almost universally followed by others since then 1986] not be obtained for three of the native species— Campanula chamissonis Fedorov (C. dasyantha auct.), C. shetleri L. Heckard, and C. wilkinsiana Greene. The other species examined were: Asy- neuma canescens (Waldst. & Ket.) Griseb. & Schenck, A. /imonifolia (L.) Janchen, Legousia hybrida (L.) Delarbre, L. pentagonia (L.) Druce, L. speculum-veneris (L.) Chaix, and Triodanis falcata (Ten.) McVaugh. Of the 38 species ex- amined, only 36 are discussed here: Asyneuma limonifolia and Legousia speculum-veneris are not included. Together, Githopsis, Heterocodon, and Tri- odanis comprise 12 species of indigenous small annuals. Githopsis, a genus that has perplexed taxonomists over the years, is found on the west coast of North America from Baja California Norte north to British Columbia. Species of the genus have narrow leaves and clavate capsules that dehisce at the apex by a perforation where the style breaks away. Ewan (1939) reviewed Githopsis and recognized seven species. Morin (1983), who revised the genus after studying breeding systems in the group, recognized just four species, including one cleistogamous (au- togamous) species with rudimentary flowers, and several subspecies. In the monotypic Heteroco- don, which also occurs on the west coast from California to British Columbia but extends east to Idaho, Montana, and Nevada, the short, broad capsule dehisces by irregular pores or fissures near the base. Both chasmogamous and cleistog- amous flowers are produced regularly, as in Triodanis. Triodanis comprises eight species, seven in North America and one in Eurasia (McVaugh, 1945b; Tutin, 1976). Traditionally they have been included in the genus Specularia, now Legousia. Tutin (1976), in Flora Europaea, retained T. fal- cata in Legousia. The genus is characterized by narrowly spiciform inflorescences, tiny cleistog- amous early flowers in the lower axils that have vestigial corollas, chasmogamous upper flowers with rotate corollas, and slender, cylindrical or prismatic capsules that open by lateral pores. Triodanis can be distinguished from Legousia on the basis of the presence of cleistogamous flowers and the relative breadth of the capsules (Mc- Vaugh, 1945b, 1948). Triodanis species are plants largely of dry open habitats of the plains and canyons of the west and southwest, although T. perfoliata (L.) Niewl. and 7. biflora (R. & P.) Greene (both of which extend into South Amer- ica) are wide-ranging weeds of disturbed habi- SHETLER & MORIN— CAMPANULACEAE 655 tats, often growing intermixed. In addition to the seven indigenous species, we also examined the seeds of the Mediterranean 7. falcata. For com- arison, we studied Legousia hybrida and L. speculum- veneris. The genus Campanula is morphologically het- erogeneous. It ranges across North America but is concentrated in the west. It is predominantly a genus of temperate montane, especially sub- alpine, habitats in North America as well as Eur- asia. A few species inhabit truly alpine or arctic habitats. Shetler's (1963) annotated checklist is the only recent synopsis of North American species. He (1982) also has made extensive stud- ies of the Nearctic harebells, the C. rotundifolia complex. The four California annual species were treated by Morin (1980), while the annual or biennial C. americana has been studied by Shet- ler (1958, 1962) and Baskin and Baskin (1984). No one has ever produced an infrageneric clas- sification of the American campanulas, and one ofthe objects ofthe present study was to examine seed features for evidence of species clusters. METHODS AND MATERIALS Seeds for this study were taken from mature capsules of herbarium specimens on deposit in the herbarium of the University of California, Berkeley (UC), the New York Botanical Garden (NY), and the U.S. National Herbarium (US), Washington, D.C. (Table 1). Measurements were made with an ocular micrometer and stereo- scopic microscope. Size of sample per collection, N, was 10 seeds whenever possible, but in some cases only a few seeds were available. Variation was assessed qualitatively by examining seeds from three different populations for each species when possible (Table 1). Seeds were examined under the scanning electron microscope, with as many seeds of a species being mounted on one stub as available or possible. The dried seeds were prepared for SEM study by first rehydrating them with aerosol and then dehydrating them in an alcohol series. From 10096 ethanol, the seeds were placed directly into the chamber of a Den- ton DCP-1 Critical Point Drying Apparatus. Af- ter being critical-point dried, the seeds were mounted on SEM stubs with “Elmer’s glue" (a water-soluble, animal-based glue) and sputter- coated with carbon and gold-palladium. Sec- tioned seed coats were obtained by both hand fracturing. They were examined and ‘photographed in a Cambridge Stereoscan, either the Mark IIA or S410. In a 656 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 TABLE |. List of vouchers of species studied. The *S" numbers are the seed sample numbers, which are used in the figure legends. Unless otherwise indicated, all vouchers are deposited at the U.S. National Herbarium (US). Asyneuma canescens (Waldst. & Kit.) Griseb. & Schenck $072, Hungary (?), Transylvania, Olim, Richter, 12 August 1908 Asyneuma limonifolium (L.) Janchen S071, Turkey, Taurus Mts., near Gulek Bogaz (“Gulek-Boghas”) pass (= Cilician Gates), Balansa 627 Campanula americana S013, Pennsylvania, AER Co., Glenshaw, Jennings 307a S014, North Carolina, Buncombe Co., Craggy Mt., Biltmore Herbarium 293b S015, Canada, Ontario, Amherstburg, Macoun 54140 S016, Alabama, Blount Co., near Blountsville, Kral 37831 Campanula angustiflora Eastwood S029, California, Lake Co., Spring Mt., Morin 249 (UC) S055, California, Lake Co., Manning Flat, Bacigalupi & Sweeney 3346 (JEPS) S056, California, Sonoma Co., Hood Mt., Baker 11756 (JEPS S057, California, Lake Co., Mirabel Mine, Eastwood & Howell 5530 (JEPS) S058, California, Santa Cruz Co., Boulder Creek, Hesse 100a (JEPS) 5059, California, Santa Cruz Co., Boulder Creek, Hesse 100b (JEPS) Campanula aparinoides Pursh S060, Minnesota, Clearwater Co., Itasca State Park, Grant 2848 S061, New York, Crystal Lake, McCall, 15 August 1877 $062, Canada, Quebec, Ile Salaberry (Salaberry de Valleyfield, now Valleyfield), Rouleau 4256 (UC) S063, Indiana, ese ge Co., Adam Lake, Yuncker & Welch 10717 Campanula aurita Greene S086, Alaska, Walker Lake, Shetler 4962 Campanula californica (Kell.) Heller S017, California, Mendocino Co., Point Arena, Davy & Blasdale 6056 Campanula divaricata Michaux S018, Virginia, Augusta Co., Elliott’s Knob, Steele 68 S019, Kentucky, Whitley Co., Cumberland Falls, McFarland 49 S020, Virginia, Page Co., Stony Man Mt., Lehtonen & Morin, 17 October 1980 Campanula exigua Rat $064, California, Со Costa Co., Mt. Diablo, Bowerman 330 (JEPS) Campanula floridana S. Watson ex A. Gra S021, Florida, Levy Co., near Cedar Key, Godfrey 56612 S022, Florida, Manatee Co., Manatee (“Мапиее,” now Bradenton), Garber, April 1876 Campanula griffinii M S028, California, San M Co., Clear Creek, Griffin 4120 (U S084, California, Napa Co., south of Pope Valley, Breedlove ru (JEPS) S085, California, Lake Co., Mirabel Mine, Eastwood 5530 (JEPS) Campanula lasiocarpa Cham. S023, Alaska, Aleutian Is., Attu, Turner 4296 S024, Canada, Yukon Territory, Mt. Sheldon, Porsild & Breitung 11731 S025, Alaska, Cook Inlet, Шатпа Point, Gorman, 1 September 1902 Campanula parryi A. Gra $026, Colorado, Clear Creek Co., Georgetown, Patterson 97 S027, Colorado, Archuleta Co., Pagosa Springs, Wooton 2873 Campanula piperi Howell $083, Washington, Clallam Co., Hurricane Ridge, Shetler 4430 Campanula prenanthoides Durand S007, California, Siskiyou Co., Humbug Mt., Butler 1815 S008, California, Humboldt Co., Willow Creek, Tracy 3286 S009, Oregon, Jackson, Siskiyou Mts., Heller 13486 1986] SHETLER & MORIN—CAMPANULACEAE 657 TABLE 1. Continued. Campanula reverchonii A. Gray S010, Texas, Burnet Co., Granite Mt., Palmer 10266 $011, Texas, Burnet Co., Marble Falls, Biltmore Herbarium 14893 S012, Texas Nealley 149 Campanula robinsiae Small S090, Florida, Hernando Co., Chinsegut Hill, Cooley, Wood & Wilson 6029 (NY) Campanula rotundifolia L. S087, Alaska, Kodiak I., Shetler & Stone 3622 S088, Colorado, Park Co., near Lake George, Shetler & Dick 3903 S089, New York, Tompkins Co., Taughannock Gorge, Vogelmann 603 Campanula scabrella Engelm S030, California, Sieben Co, Mt. Eddy, Eastwood 2010 S031, California, Siskiyou Co., Scott Mts., Engelmann, 30 August 1880 (GH) S065, California, Siskiyou Co., Mt. Eddy, Raven 10415 (JEPS) Campanula scouleri Hook. ex S005, Washington, Skamania Co., Mt. Prindle, Suksdorf 11 S006, California, Humboldt Co., Rio Dell, Moldenke & MU 20338 Campanula sharsmithiae Mori S066, California, Stanislaus =. Red Mt., Sharsmith 3144 (UC) Campanula uniflora L. S001, Canada, Baffin I., Frobisher Bay, Collins 109 S002, Canada, Quebec, Poste de Payne Bay, Rousseau 1280 S003, Canada, Hudson Bay, Belcher Is., Abbe, Abbe & Marr 4030 S004, Canada, White I., Frozen Straits, Angel 37 Githopsis diffusa A. Gray subsp. diffusa S078, California, Santa Barbara Co., San Roque Canyon, Pollard, 31 May 1952 (UC) Githopsis diffusa subsp. candida (Ewan) Morin S079, California, San Diego Co., N of Santa Ysabel, Munz 9806 (UC) Githopsis diffusa subsp. filicaulis (Ewan) Morin 082, California, Riverside Co., 1 Morin 234 (UC) Githopsis diffusa subsp. robusta M S077, California, Lake Co., Elk ín Koch 960 (UC) Githopsis pulchella Vatke а pulchella S080, California, Amador Co., Drytown, Greene, June 1889 (UC) Githopsis pulchella subsp. serpentinicola Morin S081, California, Tuolumne Co., Yosemite Junction, Mason 11108 (UC) Githopsis specularioides Nutt. S075, California, Humboldt Co., Alder Point, Tracy 1906 (UC) S076, California, Shasta Co., near Redding, Hoover 2279 (UC) Githopsis tenella Morin 4, California, Kern Co., Greenhorn Mt., Morin 241 (UC) Heterocodon rariflorum Nutt. S052, California, Marin Co., Mt. Tamalpais, Herman 17388 S053, California, Trinity Co., Stuart Fork of Trinity R., Alexander & Kellogg 5546 S054, Oregon, Washington Co., Hillsboro, Howell 1036; A = chasmogamous flowers, B — cleistogamous flowers жоо hybrida (L.) Delarbre S067, England, Devonshire = ex herb. Hore, 13 July 1826 (**Prismatocarpus hybridus”) Legousia pentagonia (L.) D S070, Greece, Aegean Sea, CM nes cu I. (“Isola di Rodi"), Vaccari 513 Legousia speculum-veneris (L.) C S069, Italy, Bay of Naples, deer L ‚ Guadagno, May 1905 (“*Specularia hirta") 658 TABLE |. Continued. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 и Jide (R. & P.) Greene sas, Wood P Aen Baxter, Bush 1 Г; A = chas son Co., jl Hee Center, Lathrop 545 mogamous flowers, B — cleistogamous flowers S048, California, Humboldt se “еа R., Chandler 1415 Triodanis coloradoensis (Buckley) McVau gh S033, Texas, Gillespie Co., Crab Apple, Jermy 468; cleistogamous flowers sampled (“ $реси/ағіа lindheimeri") S034, Texas, Kerr Co., Kerrville, Heller 1731; А = chasmogamous flowers, В = cleistogamous flowers (““Specu- laria lindheimeri") 5035, Texas, Bexar Co., Jermy 29 (“‘Specularia lindheimeri") Triodanis falcata (Ten.) McVaugh 68, Cyprus, Palaeokhora, Sintenis & Rigo, 24 September 1880 Triodanis holzingeri McVaugh S036, Colorado, Las Animas Co., near Troy, Rogers 4646 S037, Arizona, Pima Co., near Sells, Lehto, Brown, Nash & Pinkava 10738 S038, Oklahoma, Kiowa Co., near Snyder, Waterfall 13103 Triodanis lamprosperma McV S039, Arkansas, Garland Co. pon Springs National Park, 4dams, 6 May 1960 Triodanis cla (Nutt.) Nieuwl. S040, Oklahoma, Comanche Co., Mt. 2 Waterfall 13075 S041, e on Co., Lathro S042, Texas, ordin Co., Fort arta aan Ruth 140; A = chasmogamous flowers, B = cleistog- amous flower о pert (L.) Nieuwl. ucky, Lewis, Big Sulphur Creek, Braun 4021 а Sevier Co., near Gatlinburg, Miller 2300 S045, Maryland, Prince Georges Co., College Park, McVaugh 1483 Triodanis texana McVaugh S049, Texas, Denton Co., Denton, Whitehouse 15839; A = chasmogamous flowers, B = cleistogamous flowers S050, Texas, Dallas Co., Dallas, Bush 680 S051, Texas, San Patricio Co., near Aransas Pass, Whitehouse 18204 few cases, where shriveling was minimal, the dried seeds were mounted, coated, and examined di- rectly from the herbarium specimens. Seed coats also were examined and photo- graphed under the light microscope. The seeds were soaked briefly in “aerosol OT," and the coats were teased off, stained with 0.1% Tolu- idine Blue O, and mounted on a glass side in Hoyer's solution. Camera lucida drawings of these peels were made for many of the species, and a selection of them is included here. RESULTS GENERAL DESCRIPTION The seeds of North American Campanulaceae are generally quite smooth, except in some cases under relatively high magnification (500x ог more). Ornamentation of the seed coat is mini- mal, and only a few species stand out as having noteworthy features. Based on features of seed surfaces, the genera or species-group form rela- tively homogenous assemblages. In our study, size, shape, and surface pattern tended to be con- sistent within a given species. The seeds of all but a few species are elliptical, oblong, or more or less round in outline (Table a). 106 pen of the seed coat (testa) usually are form) with thickened, lig- nified radial walls (Table 3). The surface pattern of the seeds under low magnification is rugose (shallow furrows), reticulate, or striate in most species, depending on the degree of tangential wall collapse (Table 2). When the outer tangen- tial walls have collapsed only slightly, the pattern might be described by Murley’s (1951) term fa- vulariate, used in her light-microscope studies of crucifer seeds. When these walls have collapsed completely, the seed coat is conspicuously striate or even ribbed (Fig. 26). Tubercles or ridges (en- hanced striation) may result from differential thickening of the radial walls (Figs. 35, 38—40, 1986] 113, 117). The degree of verrucosity may vary from seed to seed even within the same capsule. The inner and perhaps the outer tangential walls often are differentially thickened in rings, spirals, or more irregular patterns (Figs. 100, 120, 121). The radial walls also may have thickenings on their inner faces (Figs. 101, 110, 112). The outer or top edges of the radial walls may be smooth or thickened more or less regularly in a pattern of ornamentation that gives the surface of the seed coat a beaded or pebbled appearance (Figs. 35, 38, 39, 113, 117). On the surface, the radial walls may appear to be beaded on the long axis and ridged or upturned at the ends, which gives the surface a wavy or pebbled appearance. In some species there are superficial, noncellular tuberculae or excrescences. In cross-section, seed shape ranges from terete or oval (Figs. 88-92, 95) to lenticular (Figs. 93, 94, 96, 99) or even triangular (Fig. 97) or quad- rangular (Fig. 98; Table 2). Rounded seeds may be flattened asymmetrically toward one rib (Fig. 91). Most of the seeds are truncated at least somewhat at the hilum end and are more or less rounded at the other end. Even in the fusiform seeds, which taper more or less to a point at each end, the hilum end tends to be squared off. In some species, the hilum end is an organized, well- marked, symmetrical region (e.g., Campanula americana, Fig. 94), while in many other species this end is little more than an asymmetrical slit or pinched-off zone (Figs. 88-93). The radial walls of testa cells may be thin and the lumen large and open (Figs. 102, 103), or thick and the lumen narrow and slit-like (Figs. 109, 111) or hour-glass-shaped in cross-section, as when the radial walls are rod-like (Figs. 106, 107, 112). In some cases (not depicted), the lu- men appears to be widest in the middle and the radial walls somewhat hour-glass-shaped. Some species have pitted radial walls (Figs. 102-104, 120c), and at least one species, C. floridana (Fig. 105), has large, irregularly shaped foramina in the walls. Seed coats with cells that have thin radial walls and large lumina tend to have cells that are almost isodiametric in surface view and are deeper than long or wide (Figs. 102, 103, 120c). The area where the cells meet appears either as a trough or a ridge in some species, but in other species it can be observed only with difficulty if at all (cf. e.g., Figs. 14, 66, 108). The surface of the seed coat may have a relatively thick layer of non-cellular material or virtually none (cf. Figs. 44, 100-112). A distinct cuticle is SHETLER & MORIN—CAMPANULACEAE 659 detectable in some species, as in Triodanis tex- ana (Fig. 116 The seed-coat cells of North American Cam- panulaceae can be divided into the three basic types described below. Cell type is indicated in Table 3 for every species studied. Type 1. Cells elongate, fusiform or fibriform, with relatively thick radial walls and narrow, often indistinct lumen. The cells form a continuous weave without clearly squared, abutting end-walls and often without clear indication of where one cell ends and another begins (e.g., Campanula griffinii, Figs. 4, 120a Type 2. Cells trapezoidal or irregularly hex- agonal with thick to moderately thick radial walls and narrow to relatively wide lumen. The cells form a distinct reticulum with obvious cell out- lines and cell ends, even when the outer tangen- tial walls have not collapsed. When the tangential walls collapse, the surface appears ribbed (e.g., C. divaricata, Figs. 34, 120b). Type 3. Cells more or less isodiametric in face view, hexagonal or lobed, with large deep lumen and relatively thin radial walls having pits or foramina. The cells form a jigsaw-puzzle or quilted pattern (e.g., C. aparinoides, Figs. 31, 102, CAMPANULA The mature seeds range in length from 0.4 mm to 1.6 mm and in width from 0.2 mm to 0.9 mm, with the length/width ratio ranging from 1.0 to 3.3 (Table 2). The color ranges from buff to rich chestnut pov. РАЯ surface pattern is more or iate), the degree depending on ‘the thickness of the radial walls, relative dimensions of the cells, and the extent of the tangential-wall collapse. The wider the cells, thinner the radial walls, and more collapsed the outer tangential walls, the more reticulate the surface pattern. The individual cells are the units of the reticulum. The narrower the cells, thicker the radial walls, and the more intact the outer tangential walls, the less reticulate and more striate or rugose the seed coat appears. The Cordilleran campanulas, C. scabrella (Figs. 9, 10), C. parryi (Figs. 11, 12), C. piperi (Figs. 13, 14), and C. aurita (Figs. 15, 16), form a rel- atively homogeneous group based on seed char- acteristics. The widest-ranging of these, C. par- ryi, has the smallest seeds—0.5-0.6 mm long— distinguishing it from the larger-seeded (0.7—1.0 mm) C. scabrella, with which it sometimes is confused (Table 2). Closely similar are C. uni- 660 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 TABLE 2. Comparison of seed shape, size, surface, and presence and placement of wings of seeds of North American Campanulaceae and possible relatives Shape Approx. Length! Taxon Longitudinal Cross-section L/W (mm) Surface Wings? Asyneuma canescens long-ovate ovate 1.6 2.2 shallow furrows ) Campanula americana oblong lenticular 1.3-1.6 1.5-1.7 pebbled () angustiflora ovate broadly ovate 0.7 2 faintly striate ) aparinoides oblong terete to triangu- 0.7 1.3-2.3 shallowly collicu- — late aurita oblong narrowly ovate 1.2 interrupted striate () californica elliptical to ob- broadly ovate 0.9 1.7 interrupted striate — ovate divaricata oblong terete 0.7 2 reticulate ) exigua oblong ovate 0.7 2 faintly striate ) floridana к elliptical terete ovate 0.6 1.5 rugose ) quadrangu- ar griffinii oblong vate 0.6-1.1 2.2 faintly striate ) lasiocarpa fusiform to ovate terete 0.7 2 striate — rryi elliptical erete 0.6 1.8 striate V piperi narrowly ellipti- terete 0.7 2 striat A prenanthoides oblong ovate-flattened 0.9 2.2 striate ) reverchonii broadly elliptical- lenticular 0.9 1.3 striate — roundish robinsiae roundish triangular 0.4 1 margins tubercu- — late rotundifolia oblong terete to ovate 0.8 striate ) oblong ovate 0.7-1.0 1.5-3.3 striate ) scouleri oblong 1.0 striate-reticulate ) sharsmithiae oblong terete 0.8 2.6 smooth — uniflora oblong terete 1.0 2 striate ) Githopsis diffusa fusiform terete 0.6-0.8 3.0-3.8 striate ) pulchella fusiform terete 0.7-1.0 2.4 striate ) specularioides fusiform terete 0.8 2.6 iate ) tenella fusiform terete 0.7 1.8 flattened reticu- ) lum Heterocodon rariflorum elliptical terete 0.5 1.7-2.1 smooth A Legousia hybrida elliptical erete 1.4 1.6 shallow furrows — pentagonia elliptical narrowly elliptical 1.5 1.6 shallow furrows — Triodanis ¡fora broadly elliptical lenticular 0.5 1.3 shallow furrows () coloradoensis broadly elliptical lenticular 0.8 1.4 shallow furrows () alcata broadly elliptical nticular 0.8 1.4 shallow furrows () holzingeri broadly elliptical enticular 0.5 1.5 scattered tubercles () lamprosperma broadly elliptical lenticular 0.9 1.2 shallow furrows () to roundish 1986] TABLE 2. Continued. SHETLER & MORIN—CAMPANULACEAE Shape Approx. Length' Taxon Longitudinal Cross-section L/W (mm) Surface Wings? leptocarpa broadly elliptical lenticular 0.8 1.6 shallow furrows () perfoliata broadly elliptical lenticular 0.5 1.3 smooth or tuber- () to roundis culate texana broadly elliptical quadrangular 0.4 1.4 () ! See text for ranges. Approximate range given here only for exceptionally variable y gens 2) = peripheral ridge on one side only; ( ) V = ridge on end opposite hilum. flora (Figs. 17, 18), C. rotundifolia (Figs. 19, 20), and C. lasiocarpa (Figs. 21, These seven compose what might be ecd as the most common or typical campanula seed type in North America. The cellular pattern is Type 1 with gra- dation to Type The California annuals, Campanula angusti- flora, C. exigua, C. griffinii, and C. sharsmithiae (Figs. 1-8, 93), have seeds that are extremely smooth, almost featureless, except under high magnification, and even at 500 x the surface may be flat and smooth, as C. sharsmithiae (Fig. 8). The latter species, only recently discovered (Morin, 1980), is a narrow serpentine endemic of the Mt. Hamilton endemic area of California. Its cucumber-shaped seeds tend to be larger, smoother, and more nearly terete than the seeds of the other three annual к The four species have the Type | cellular p Campanula hen pow 23, 24) and C. scouleri (Figs. 25-27), also west coast species, share a similar seed-coat р gy witha Ту 2, or Type 1 grading to Type 2, pattern, and they stand somewhat by themselves. The seeds of C. scouleri are large, about 1 mm long and 0.5 mm wide. The surface pattern is reticulate and ribbed when the outer tangential walls are collapsed (Fig. 26), but quite smooth when they have not col- lapsed (Fig. 27). The C. prenanthoides seed is distinctive in cross-section, being round on one side and compressed to a margin or flange on the other side (Fig. 91). The seeds of Asyneuma ca- nescens (Figs. 28, 29) are similar to those of C. prenanthoides and C. scouleri in shape and out- line. In addition, all three species have thick- walled, fibriform cells. The three marsh species of North America, C. aparinoides, C. floridana, and C. californica, have distinctive seeds. The seeds of C. californica are — ridge essentially on entire periphery; ^ = ridge on hilum end; generally similar to those of C. prenanthoides and C. scouleri. Campanula californica (Figs. 32, 33) also has large seeds (0.9 mm long) with a shape similar to these species, but these are less conspicuously biconvex and have more typically a Type 1 surface pattern. The surface cells are quite deep (Fig. 109), however, possibly resulting in enhanced bouyancy. The seed coat of C. aparinoides (Figs. 30, 31, 120c) is unlike any other in the genus. The cells are Type 3, essentially isodiametric in surface view, resembling tiny "oyster crackin vi п together in a quilted pattern. The i (Figs. 102, 103) have a large deep lumen and thin radial walls that have small pits visible under high magnification (2,000 x ). In recent studies of the C. aparinoides-C. uliginosa complex, Coch- rane (1981; pers. comm.) has found seed differ- ences and other taxonomic characters that, in her opinion, justify treating the two as distinct sub- species of C. aparinoides, instead of lumping them as a single taxon, as is now customary. The seeds of C. floridana (Figs. 36, 37) are small (0.6 mm long) with a rugose surface pat- tern. The cells of the seed coat are trapezoidal or roughly hexagonal with a large lumen and can be classed as Type 2 (Fig. 121e). The seeds are more or less terete in cross-section (Fig. 89) and have radial walls that are thin and have large, irregular openings that impart a “Swiss cheese" appearance (Fig. 105). In seed-coat pattern, C. divaricata (Figs. 34, 35, 120b), the тан р bellflower, shows по clear affinity to any other species, although it approaches the pattern in C. floridana. The seed coat is reticulate with trapezoidal or hexagonal cells that perhaps best exemplify Type 2. The radial walls are moderately thick to thin and the lumen is large. The most distinctive feature is 662 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 TABLE 3. Comparison of surface seed-coat cells. Taxon Type Shape Lumen Wall Thickness Asyneuma 2 fibriform narrow thick 80 x 10 um Campanula americana 1 fibriform narrow thick 72 x 11 um angustiflora 1 oblong medium variable 43 x 10 um aparinoides 3 isodiametric large thin aurita 1-2 oblong narrow thick californica 1 oblong medium medium divaricata 2 broad-fibriform large thin 71 x 18 exigua 1 broad-fibriform narrow thick 42 x 4.5 floridana 2-3 isodiametric large thin 62 x 29 um griffini 1 fibriform narrow thick 66 x 10 um lasiocarpa 1-2 fibriform narrow thick 94 x 10 um parryi 1 broad-fibriform medium thick 5 piperi 1 broad-fibriform narrow thick 7 um prenanthoides 2 fibriform medium thick 60 x 7 um reverchonii 1 fibriform narrow medium robinsiae 3 isodiametric cell structure not visible unknown in peels rotundifolia 1-2 fibriform medium thick 70 x 10 um scabrella 1-2 fibriform medium medium 56 x 14 um scouleri 1-2 fibriform narrow thick 72 x 7 um sharsmithiae 1 irregular medium thick 52 x 11 um uniflora 1-2 fibriform narrow thick 59 x 11 um Githopsis diffusa 1-2 long-fibriform medium thin 109 x 6.9 um pulchella 1-2 fibriform broad medium 79 x 10 um specularioides 1 fibriform medium medium 89 x 8 um tenella 2 short-fibriform broad medium 49 x 9 um Heterocodon rariflorum 1 fibriform narrow thick 62 x 8.4 um TABLE 3. Continued. SHETLER & MORIN—CAMPANULACEAE 663 Taxon Type Shape Lumen Wall Thickness Legousia hybrida 1 fibriform medium thick 74 x 9.2 um Triodanis biflora 1 fibriform narrow thick 96 x 6.4 um coloradoensis 1 fibriform wide medium 51 x 8.6 um falcata 2 fibriform narrow thick 55 x 9.2 um holzingeri 1 fibriform narrow thick 60 x 10 um lamprosperma 1 fibriform narrow medium 75 x 9 um leptocarpa 1 fibriform narrow thick 150 x 9.4 um perfoliata 1 fibriform narrow thick 71 x 8.85 um texana 2-3 short-fusiform narrow medium the € beaded outer edge ofthe radial walls (Figs. 35, 113). E. ee non-Californian annual, winter-an- nual, or biennial species— C. americana, C. re- verchonii, and C. robinsiae—have their own dis- tinct seed-coat patterns. The seeds of Campanula americana (Figs. 38-40, 94, 121a), a wide-rang- ing species of woodland borders in eastern North America, have fibriform cells with thick radial walls and a narrow lumen. The radial walls have dense spirals of apparent secondary thickenings. The end walls are thickened into ridges that give the surface a pebbled or verrucose ornamenta- tion under a thick cuticle (Figs. 39, 40). The seed is lenticular in cross-section (Fig. 94) and has a ganized than in other species, and outlined by the heavy, symmetrical flange as an oval area. Campanula robinsiae (Figs. 41—44, 97, 118) has the smallest seeds recorded for the North American members of the genus (Table 2). The surface is very smooth, and the seed is trigonous in cross-section; the angles are distinctly ridged and each has irregularly placed, smoothly round- ed tubercles along it (Fig. 97). The seed-coat cells (Fig. 118) are of Type 3, digitate or lobed, with the lobes uiis Im other cells like the pieces of a Jigsaw puz Campanula ipie (Figs. 67, 68, 96), the basin bellflower of the Edwards Plateau of Texas, has seeds that are quite similar to the seeds of Triodanis species. The seed is elliptical in outline and more or less lenticular in cross-section (Fig. 96). The seed coat has Type | cells with a rugose or striate appearance. HETEROCODON Heterocodon rariflorum (Figs. 84, 85) has small (0.5—0.6 mm long) elliptical seeds that are terete in cross-section. The surface is very finely striate, being composed of long, narrow cells of Type 1 TRIODANIS Triodanis seeds are lenticular to elliptical in outline and lenticular (biconvex) in cross-section (Fig. 99). Excepting 7. texana (Figs. 54—56), the seed coats have Type 1 cells, and the surfaces striate, or smooth. Triodanis per- rugose pi (Figs. 45-48, 104, 119), T. holzingeri (Figs. 49, 50, 99), T. biflora (Figs. 51-53, 112), and T. texana (Figs. 54—56, 98) have seeds that are 0.4— 0.6 mm long. Triodanis coloradoensis (Figs. 57, 58), and T. lamprosperma (Figs. 61, 62), as well as the European 7. falcata (Figs. 59, 60), have seeds that are 0.8—0.9 mm long. The seeds of 7. leptocarpa (Figs. 63—66), which are more elon- gate, are about 0.8 mm long. 7riodanis falcata, the European species, apart from a somewhat 664 smoother seed coat, fits right into the seed pat- tern of this largely North American genus. The radial walls of Triodanis are thick and rodlike (Fig. 112) and may have pits (Fig. 104). The surface may be verrucose or pebbled (Fig. 119) from irregular thickenings. In 7. perfoliata, some populations have smooth seeds, while other pop- ulations have seed coats that are finely tuber- culate with the tubercles in lines (cf. Figs. 45- 48). McVaugh (1945b) studied this variation quantitatively with the light microscope. Triodanis texana (Figs. 54-56, 98, 121b) clear- ly stands alone, not just within the genus but in the North American Campanulaceae generally. In outline, the seed is similar in shape to other seeds of Triodanis, but in cross-section (Fig. 98) it is quadrangular rather than lenticular. The cell type is intermediate between 2 and 3, although many cells are almost isodiametric, if irregularly so. The outer cell surfaces, presumably the tan- gential walls, are strongly convex and covered by a definite cuticle, which can be seen peeling away intact in Figure 116. The cells form a braid- ed surface pattern like the surface of a cord or rope with its interwoven strands. Seeds of cleistogamous and chasmogamous flowers were examined in 7. /eptocarpa (Figs. 63— 66) and 7. texana, but no consistent differences were observed in seed-coat morphology between seeds of the chasmogamous and cleistogamous flowers. GITHOPSIS The mature seeds of Githopsis (Figs. 69-81, 110, 115) range in length from 0.6 to 1.0 mm and in width from 0.2 to 0.4 mm, with the length/ width ratio ranging from 1.8 to 3.8 (Table 2). The seeds are fusiform, more or less tapering to both ends and are terete in cross-section (Fig. 92). The surface pattern is fibriform-rugose to fibriform-reticulate in G. pulchella (Figs. 76-79, 115). The pattern appears striate in low magni- fication owing to the longitudinal ridges of the radial walls and the troughs of the lumina (Fig. 115). The cells are Type 1, grading to Type 2 particularly in G. pulchella (Table 3). Githopsis tenella (Figs. 74, 75) has the most fibriform pat- tern with the smoothest surface, owing to the very narrow lumina. The subspecies of G. diffusa (Figs. 69-73) are all similar in their surface pat- terns, as are the subspecies of G. pulchella (Figs. 76—79). The radial walls are thick and rod-like — ropy in surface view. In general, they are wider ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 than the almost slit-like lumina, although the width of the lumen is variable in G. diffusa and relatively large in G. pulchella. The seeds of the four species of Githopsis and their subspecies, as recognized by Morin (1983), are distinguishable primarily on the basis of size. Githopsis tenella seeds are the most distinctive of the four species because of the density and small size of the seed-coat cells. DISCUSSION Seed-coat morphology in the North American Campanulaceae is relatively uniform, but there are recognizable generic patterns and a number of distinctive patterns in individual species. Githopsis seeds, although clearly similar to seeds of the majority of Campanula species, stand to- gether with their fusiform shape and similar sur- face patterns of cells that have thick, rounded radial walls that look inflated, almost sausage- like. Individual cells of the testa look like min- lature inflatable rafts (Fig. 115) e more or less lenticular seed with Type 1 cells and a fibriform surface pattern is consistent within Triodanis, although the seeds of 7. /ep- tocarpa (Figs. 63, 65) are more elongate and oval than in the other species. Mediterranean Tri- odanis falcata (Figs. 59, 60), which in Flora Eu- ropaea (Tutin, 1976) is treated as a species of Legousia, fits in well with the larger-seeded American Triodanis species, although in cross- section the seeds are more rounded than typical Triodanis seeds. The one species that does not conform, except in the oval outline, is 7. texana (Figs. 54—56, 98, 116). Its quadrangular seeds with an undulating surface pattern have no coun- terpart either in the genus or the family in North America. The seed-coat cells are irregular, thick- walled sclereids (Fig. 121b). Triodanis often is included in Legousia (Spec- ularia). Indeed, the seeds of Legousia pentagonia (Figs. 86, 87) and L. speculum-veneris (not shown) are similar in shape (outline and cross-section), especially to the seeds of the large-seeded 7ri- odanis species (Figs. 57-66). The large seeds of Legousia hybrida — largest of the seeds studied — are rounded rather than lenticular in cross-sec- tion, and the cells of the testa are quite different from those of Triodanis (cf. Figs. 111, 112). These two genera clearly cannot be separated easily on the basis of their seeds, although more study par- ticularly of Legousia species is needed. The seeds of Campanula reverchonii (Figs. 67, 1986] 96) show definite similarities to the seeds of Triodanis. Campanula reverchonii is endemic to the granitic Central Basin or Llano Area of the Edwards Plateau in central Texas (Correll & Johnston, 1970). All American species of Tri- odanis are found in Texas; two of the seven species, 7. coloradoensis and T. texana, are en- demic to Texas, the former to the Edwards Pla- teau. Further study is needed of the relationships of Triodanis to C. reverchonii, which is not unlike some Triodanis species in habit. The seed sim- ilarity may be the result of convergence through environmental selection. The relationship of Heterocodon to other gen- era is uncertain. McVaugh transferred it to Spec- ularia in 1941 but subsequently (1945b) divided this latter genus into Triodanis and Specularia (Legousia) and reinstated the genus Heteroco- don. In this later paper McVaugh suggested that he was maintaining it as separate from Cam- panula primarily because of tradition. The seeds of Heterocodon, although at the other end of the size range, are more similar to those of L. hybrida than to those of the other two species of Legousia examined or to any of the other taxa studied by us. The similarity of Heterocodon seeds to Le- gousia seeds and the fact that Heterocodon and Triodanis regulan] produce CMA as well eory that Heterocodon belongs in the pruina Triodanis complex. Within Campanula, some weak groupings can be made, but for the most part the individual exceptions are more notable than the common rns. The California annuals form a small roup of four species. However, the seed of C. sharsmithiae (Figs. 7, 8) has a more terete shape and a smoother surface than the other three, ow- ing, apparently, to a thicker cuticular layer. The position of Campanula californica in the genus is uncertain. On the basis of seed mor- phology, it does not belong with the other two species of marshes, swamps, and bogs— C. apa- rinoides and C. floridana. The seeds suggest that its affinities might be much closer to C. scouleri and perhaps C. prenanthoides. McVaugh (1945a) transferred C. prenanthoides to the Eurasian ge- nus Asyneuma because of its deeply lobed co- rolla. The similarity between the seeds of A. ca- nescens, C. prenanthoides, C. scouleri, and C. californica suggests to us that these species may be related to each other. The western American perennial alpine species of Campanula have a more or less common see SHETLER & MORIN— CAMPANULACEAE 665 pattern, although with variation. Although they differ markedly in size, the seeds of C. aurita and C. parryi have a very similar surface pattern. These species have other morphological simi- larities and to some extent are ecological coun- terparts, with C. aurita occurring in the far north- ern Rocky Mountains (Brooks Range) and Yukon Tablelands and C. parryi occurring in the central and southern Rockies. The two Pacific North- west species, C. piperi and C. scabrella, ecolog- ical counterparts in the Olympic Mountains and Cascade Mountains, respectively, have quite similar seed-coat patterns (Figs. 10, 14, 121c), which also are generally like the patterns of C. aurita and C. parryi (Figs. 12, 16, 121d). The Olympic Mountain bellflower, a local endemic, resembles a smaller version of C. aurita, but there also are morphological grounds for regarding it as closely related to the downy alpine bellflower (C. scabrella). The latter forms almost a mor- phological continuum with C. parryi in the Northwest, although seed size seems to be a re- liable character for distinguishing between the species. On the basis of the seeds, Campanula uniflora (Figs. 17, 18), which otherwise appears to have no close relatives, does not stand apart notice- ably from the other perennial bellflowers of west- ern North America. In eastern North America, Campanula is het- erogeneous with respect to seed-coat pattern. Each species stands more or less alone, although C. divaricata and C. floridana have the same type of epidermal cell pattern (Figs. 34-37, 120b, 121e). It is possible that these two species have been derived from a common ancestor, but this possibility has never been raised before. Both, but particularly C. divaricata, show some simi- larity in cell pattern to the more northern and wide-ranging C. rotundifolia (Figs. 19, 20, 108, 114). The distinctive beaded thickenings on the radial walls of C. divaricata (Figs. 35, 113) are encountered sometimes also in the western C. parryi (Fig. 117), but on general morphological grounds these two species appear to be unrelated. Like the tall bellflower, Campanula apari- noides and C. robinsiae each stand apart from other taxa in the genus in North America. Cam- panula aparinoides does not appear to be closely related to C. floridana (Cochrane, pers. comm.), contrary to Shetler's (1963) earlier speculation on the basis of ecology and general morphology. The two have altogether different seed types (Figs. 30, 31, 36, 37, 89, 95, 120c, 121e), including 666 quite different epidermal cells, although in both species these cells are relatively thin-walled with large lumina and pitted or fenestrate radial walls (Figs. 102, 103, 105; see also Cochrane, 1981). Small (1926) initially considered C. robinsiae to be most closely related to C. reverchonii but later (1933) placed it, with C. floridana, in his genus Rotantha (Figs. 41—44, 67, 68, 96, 97). The seeds of these three species bear little resemblance to each other and certainly would not support this view. Of all North American Campanulaceae, C. ro- binsiae has the most unusual and distinctive seeds. This raises new questions about the origin of this elusive species, thought possibly to be extinct until its recent rediscovery. It is a small annual of puzzling origin (Shetler, 1963), known only from its type locality at Chinsegut Hill, near Brooksville, Florida. It had not been seen in the wild since 1958 until the spring of 1983 when it was found in several vernal pools in the Chin- segut Hill area first by Steven Hill and subse- quently by others, including the authors. Sur- prisingly, it was not mentioned in Ward's recent book (1979) on rare and endangered plants of Florida, presumably because it was thought to be extinct or of doubtful origin in the first place. As an annual, this species may go through cycles of abundance and sparsity, and, being so small and inconspicuous, it is easily overlooked even when fully developed. The unusual seed mor- phology deepens the mystery of this plant's af- finities. If there is any justification for Small's genus Rotantha, it is the suite of characters that seems to isolate C. robinsiae from the other species. Further studies are being conducted by the authors in collaboration with Steven Hill. The affinities of C. americana, the tall bell- flower, also remain obscure. This highly distinc- tive species with dense spicate racemes of star- shaped, rotate flowers; tall, often much-branched stems; and annual or biennial habit was given its own genus, Campanulastrum, by Small in 1903. Although the basic floral morphology is campanuloid, it does not fit well into the typical concept of Campanula. Small’s view appears to have increasing justification from palynological, cytological, and now seed evidence. The pollen grains are pantaporate, a condition unique to only a few species among many species with 3-4-por- ate grains in Campanula (Avetisian, 1967). Chromosome number is n = 29, unique in the genus largely comprising species in the n = 17- series (Gadella, 1964). Surely this is a derived ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 number, probably through hybridization. Seed- coat morphology also is unique. The upturned ends of the radial walls, which seem to be re- sponsible largely for the pebbled surface, occur to a limited extent in C. parryi (Figs. 38-40, 117), but on other grounds these species do not appear to be related. The isolated position of C. amer- icana, further underscored by Gadella's (1964) inability to cross it with other campanulas, led Shetler (Shetler & Mathews, panulastrum. Gadella, puzzled by C. americana, suggested that a search should be made among species of Asyneuma, which also have deeply cut corollas, for the origins or relationships of C. americana. Interestingly, the seeds of A. canes- cens (Figs. 28, 29) and C. americana (Figs. 38- 40, 121a) appear to have similar epidermal cell patterns beneath their cuticular layers. The adaptive significance of seed-coat sculp- turing and ornamentation has been little studied. The seeds of bellflowers (Campanulaceae) ap- parently are dispersed mainly by water, with gravity playing an important role among those species that grow on rock ledges and cliff faces (e.g., C. rotundifolia). The relatively smooth, rounded seeds seem to be streamlined to facili- tate water dispersal, and the seeds have no ob- vious adaptations for animal or bird dispersal. They are small and insignificant as a source of food, and, with the exception of Campanula ro- binsiae (Figs. 41—44) and Triodanis perfoliata (Figs. 45-48), lack superficial processes or or- namentation that might enhance their chances of being carried by animals. In the case of these two species, the tubercles are so minute that it is doubtful they have any dispersal role with re- spect to animals. Possibly, ants play a role in the dispersal of some or all bellflowers, and one can hypothesize that such tiny processes as these might facilitate transporting the seeds when ants are the dispersal agents. Triodanis perfoliata is wide-ranging, and perhaps tubercles confer a se- lective advantage in dispersal at least in some environments. This species is variable with re- spect to seed-coat tubercles, and the ratio of tu- berculate to non-tuberculate seeds varies not only from population to population but also from re- gion to region (McVaugh, 1945b). What selective advantage, 1f any, is conferred by the tubercles or lack of them, is not known. Two of the marsh/swamp species, Campanula aparinoides and C. floridana, have epidermal cells with relatively large lumina and thin walls that 1986] have many pits or, in the case of C. floridana, large foramina (Figs. 102, 103, 105). Perhaps these cell characteristics give the seeds of these two species greater buoyancy in the hydric en- vironment. At the same time, pits are not re- stricted to these species (cf. Triodanis perfoliata, Fig. 104). The seeds of the annual species have a “tight weave" of fibriform cells with thick radial walls that nearly fill the lumina and have a definite cuticle or layer of relatively smooth, non-cellular material on the outside of the testa. Perhaps such features have selective advantages in dry or dis- turbed habitats in preventing easy water-loss or water-logging by the seeds (Figs. 1-8, 38-40, 45- 87). Little is known about general trends of evo- lutionary specialization in seed-coat sculpturing, and the relatively few species of bellflowers in North America do not provide an adequate basis in themselves for reliable speculation about seed- coat phylogeny in the Campanulaceae. If there SHETLER & MORIN—CAMPANULACEAE 667 is a typical seed coat in North America, it would appear to be the type found among the western American perennial species. The annuals of the genus Campanula and of the family at large in North America, as well as the perennials in east- specialization of the seed coat. It is not clear whether these tendencies have evolved in North America in response to ecological or other se- lective factors, or represent tendencies to be found elsewhere in the family in Eurasia or perhaps even in the southern hemisphere in Wahlenber- gia. Thulin’s (1975) study of South African Wah- lenbergia species shows some intriguing similar- ities in seed-coat sculpturing and seed shape. For instance, the pictured seed of W. virgata is very much like Githopsis seeds, and the seed of W. subaphylla resembles those of Campanula pre- nanthoides. Clearly, more study of the family worldwide is needed before the results presented here can be fully interpreted. 668 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 1-8 (bar = 100 um). Seeds of annual Campanula. — Figures 1, 2. Campanula angustiflora. — 1. Seed (S59).—2. Seed coat (S59).—Figures 3, 4. C. griffinii.—3. Seed (S85).—4. Seed coat (S28).— Figures 5, 6. C. exigua. —5. Seed (S64). — 6. Seed coat (S64). — Figures 7, 8. C. sharsmithiae. — 7. Seed (S66). — 8. Seed coat (S66). SHETLER & MORIN—CAMPANULACEAE FIGURES 9-16 (bar = 100 um). Seeds of perennial C m S 9. 10. Campanula scabrella. —9. T ee —10. Seed coat (S65).—Figures 11, 12. C. parryi.—11. Seed (S27).—12. Seed coat (S27).— Figures 4. C. piperi. — 13. Seed (S83).— 14. Seed coat (S83)— Figures 15, 16. C. aurita. — 15. Seed (S86). —16. Seed d po 670 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 17-22 (Баг = 100 um). Seeds of perennial Campanula. — Figures 17, 18. Campanula uniflora. — Seed (S02). — 18. Seed coat (504). — Figures 19, 20. C. rotundifolia s.1.—19. Seed (S89). — 20. Seed coat hea Figures 21, 22. C. lasiocarpa.—21. Seed (S25).—22. Seed coat (S25). SHETLER & MORIN—CAMPANULACEAE FIGURES 23-29 (bar = 100 um). Seeds of Campanula and Asyneuma. — Figures 23, 24. Campanula prenan- thoides (Asyneuma prenanthoides).—23. Seed (S08). — 24. Seed coat (S08). — Figures 25-27. C. scouleri.—25. Seed (S05). — 26. Seed coat with tangential walls collapsed (S05). — 27. Seed coat with tangential walls uncollapsed ($05). — Figures 28, 29. Asyneuma canescens. — 28. Seed (S72). — 29. Seed coat (S72). 672 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 30-37 (bar = 100 um). Seeds of Campanula. —Figures 30, 31. Campanula aparinoides. —30. Seed (S62).—31. Seed coat (S60). — Figures 32, 33. C. californica.—32. Seed (S17).—33. Seed coat (S17).— Figures 34, 35. C. divaricata. — 34. Seed (S19). —35. Seed coat (S18). — Figures 36, 37. С. floridana. — 36. Seed (S21).— 37. Seed coat (S21). 1986] SHETLER & MORIN—CAMPANULACEAE 673 FIGURES 38-44. Seeds of Campanula. г _ Е с = 100 um). С. americana. — 38. Seed (514). — 39. Seed coat (S14). — 40. Seed coat n showing framework of cells (S13). — Figures 41—44. C. robin- siae. —41. Seed (S90) (bar = 100 um).— 2. Seed coat (S090) (Баг = 100 um). — 43. Seed coat (S090) (bar = 10 um).—44. Seed coat showing 1 papilla Paes (bar = 10 um). 674 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 FIGURES 45-56. Seeds of Triodanis. — Figures 45-48. Triodanis oi user —45. Seed of papillate type (S44) (Баг = 100 um).— 46. Seed coat of papillate type (S44) (Баг = 100 um).— 47. Seed coat of рае type (S44) (bar = 10 um).— 48. Seed of smooth type (S43) (bar = 100 um). — Figures A 50 (bar = 100 um). T. holzingeri. — 49. Seed (S36). — 50. Seed coat (S36). — Figures 51-53. Т. biflor us ө (S46) (bar = "100 y m). — 52. See coat (S46) (bar = 100 um).— 53. Seed coat (S47) (bar = 10 um).— Fi s 54-56. T. texana. бе Seed pei (Баг = 100 um).—55. Seed coat (S51) (bar = 100 um).— 56. Seed coat (648) (Баг = 10 um). 1986] SHETLER & MORIN—CAMPANULACEAE 675 FIGURES 57-62 (bar = 100 um). Seeds of Triodanis. —Figures 57, 58. T. ean ig Satyr Seed (S34).— 58. Seed coat (S34). — Figures 59, 60. T. falcata i ine falcata (Ten.) Fritsch). — 59. Seed (S68). — 60. Seed coat (S68). — Figures 61, 62. T. lamprosperma. —61. (S39). — 62. Seed coat (S3 ^u ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 GURES 63-68 (Баг = 100 ит). Seeds of Triodanis and Campanula. —Figures 63-66. T. leptocarpa.—63. Sio (S42A, ВО $). — 64. Seed coat (S42A). — 65. Seed (S42B, cleistogamous). — 66. Seed coat (S42B). — Figures 67, 68. Campanula reverchoni. — 67. Seed (512). — 68. Seed coat (S12 1986] SHETLER & MORIN—CAMPANULACEAE FIGURES 69-75 (bar = 100 um). Seeds of Githopsis. — Figures 69-73. Githopsis diffusa. — 69. Subsp. diffusa, seed (S78).—70. Subsp. robusta, seed (S77).—71. Subsp. filicaulis, seed (S82).—72, 73. Subsp. candida. — 72. Seed (S79). — 73. Seed coat (S79). — Figures 74, 75. G. tenella Morin.—74. Seed (S74).—75. Seed coat (S74). ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 2 FIGURES 76-81 (bar = 100 um). Seeds of Githopsis. — Figures 76, 77. G. pulchella subsp. pulchella. — 76. Seed (S80). — 77. Seed coat (S80). — Figures 78, 79. G. pulchella subsp. serpentinicola. — 78. Seed (S81). — 79. Seed coat (S81).— Figures 80, 81. G. specularioides. — 80. Seed (S79). — 81. Seed coat (S79). SHETLER & MORIN—CAMPANULACEAE FIGURES 82-87 (bar = 100 um). Seeds of Legousia Heterocodon. — Figures 82, 83. Legousia hybrida. — 82. Seed (S67). — 83. o coat (S67). — Figures 84, 85. Heterocodon rariflorum. —84. Seed (S53). — 85. Seed coat (S53). — Figures 86, 87. Legousia pentagonia. — 86. Seed (870) 87. Seed coat (S70). 680 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 88-93. Face views of e (bar = 100 um). — 88. Campanula rotundifolia (S88).—89. C. floridana (S21).—90. C. lasiocarpa (S24).—91. C. a (S08). —92. Githopsis specularioides (S76). —93. Cam- panula exigua (S64). 1986] SHETLER & MORIN—CAMPANULACEAE 681 FIGURES 94-99. Face views of seeds (bar = 100 um). —94. Campanula americana (S14).—95. C. aparinoides (S60).—96. C. reverchoni (S10).—97. C. robinsiae (S90).—98. Triodanis texana (S51).—99. T. holzingeri (S37). 682 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 100-105. Seed coats in section (bar = 10 um). — 100. Campanula uniflora (S04).— 101. C. scabrella (S65). — 102. C. aparinoides (S61).— 103. С. peines showing pits in cell walls (S63). — 104. Triodanis per- foliata (S45). — 105. Campanula floridana, showing wall pits (openings) (S21). 1986] SHETLER & MORIN—CAMPANULACEAE 683 4 FIGURES 106-112 (bar = 10 um). Seed coats in section.— 106. Campanula griffinii, seed coat in section (S85).— 107. C. rotundifolia, seed coat in section (S88).— 108. C. i olay ee ne surface (S88).— 109. C. californica, seed coat in section (S17).— 110. Githopsis diffusa su ro seed-coat surface and section S77).—111. Legousia hybrida, seed coat in section (S67). — 112. edP Ma yo seed coat in section (S47). 684 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 113-119. Seed-coat surfaces (Баг = 10 um).— 113. Campanula divaricata (S18).—114. C. rotun- difolia (S89). — 115. Githopsis pulchella (S81).—116. Triodanis texana, cuticle (S50).—117. Campanula parryi (S26).— 118. C. robinsiae (S90). — 119. Triodanis perfoliata (S43). SHETLER & MORIN—CAMPANULACEAE 1986] Ww TAX FiGURE 120. Camera lucida drawings of seed-coat peels (bar = 10 um). Long axis of cells corresponds to rn).—b. C. divaricata (S20) (Type 2 pattern). —c. C. aparinoides (S60) (Type 3 pattern). 685 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 a 1525. 2—5 42 АА: be AL VA NVI NV NA Sas UV WI WU vif, \ FIGURE 121. == ===> = — TJ GAVEL TU / Y ИДИТ, se Camera lucida drawings of seed-coat peels (bar = 10 um). Long axis of cells corresponds to ( long axis of the seed. Dark lines represent thickenings of radial walls. —a. Campanula americana (S15).—b. Triodanis texana (S49B).—c. C. piperi (S83).—d. C. parryi (S27).—e. C. floridana (S21). LITERATURE CITED AVETISIAN, E. 1948. Palynologica caucasica III. “Pol- len of the Caucasian representatives of the family Campanulaceae.” Trudy . Inst. Akad. Nauk Armjansk. SSR 5: 199-206. [In Russian.] . 1967. “Morphology of the pollen of the fam- ily Campanulaceae and closely related families (Sphenocleaceae, Lobeliaceae, Cyphiaceae) in connection with questions of their systematics and phylogeny.” Trudy Bot. Inst. Akad. Nauk Armjansk. SSR 16: 5-41. [In Russian.] . 1973. Palynology of the order Campanulales s.l. Pp. 90-93 in L. A. Kuprianova (editor), “Мог- phology of the Pollen and Spores of Recent Plants." International Conference on Palynology, 3rd, No- vosibirsk, 1971. [In Russian.] BAsKIN, J. M. & C. C. BAskiN. 1984. The ecological life cycle of Campanula americana in northcentral Kentucky. Bull. Torrey Bot. Club 111: 329-337. BRISSON, J. D. & R. L. PETERSON. 1977 electron microscope and X-ray mic the study of seeds: a bibliography covering the period of 1967-1976. Scanning Electron Micros- copy 2: 697-712. CAROLIN, R. D. 1960. The structures involved in the presentation of pollen to visiting insects in the order Campanales. Proc. Linn. Soc. New South Wales 85: 197-207. 1967. The concept of the inflorescence in the order Campanulales. Proc. Linn. Soc. New South Wales 92: 7-26. 1980. Pattern of the seed surface of Goodenia and related genera. Austral. J. Bot. 28: 123-137. CHARADZE, A. 1949 1970. “On the florogenesis of the Caucasian campanulas.” Notulae Systematicae ac Geogra- phicae Instituti Botanici Thbilissiensis 28: 89-102. [In Russian.] . 1976. “The genus Campanula L. s.l. in the Caucasus (Conspectus).” Notulae Systematicae ac Geographicae Instituti Botanici Thbilissiensis 32: 45-56. [In Russian. COCHRANE, B. 1981. Campanula aparinoides Pursh and C. uliginosa Rydb.: pattern and origin of vari- ability. [Abstract.] Botanical Society of America, Miscellaneous Series 160: 65. CONTANDRIOPOULOS, J. 1964. Contribution à l'étude caryologique des Campanulacées de Gréce. Bull. de la Soc. Bot. France 111: 222-235. 1967. Contribution à l'étude cytotaxino- mique des Campanulacées de Gréce. II. Bull. Soc. Bot. France 113: 453-474. 1970a. Contribution à l'étude cytotaxino- 1986] SHETLER & MORIN mique des сеи du Proche Orient. Bull. Soc. Bot. France 117: 55-7 1 . Contribution а | l'étude cytotaxino- mique des Campanulacées du Proche-Orient, II. — s Asyneuma de Turqui. Bull. Soc. Bot. France 117: 209-220. 1972. Contribution à l'étude cytotaxino- mique des Campanulacées du Proche-Orient, III. Bull. Soc. Bot. France 119: 75-94. 1976. Contribution à l'étude cytotaxino- mique des Campanulacées du Proche-Orient, IV Bull. Soc. Bot. France 123: 33-46. Corner, E. J. H. 1976. The Seeds of Dicotyledons, 2 volumes. Cambridge Univ. Press, Cambridge. CorRELL, D. S. & M. C. JOHNSTON. 1970. Manual of the Vascular Plants of Texas. Texas Research Foundation, Renner, DAMBOLDT, J. 1965. Z totaxonomische Revision der isophyllen Campanulae in Europa. Bot. Jahrb. Syst. 84: 302-358 1970. Revision der Gattung Asyneuma. Bois- siera 17: 1-128. 1975. Differentiation and classification in the genus Campan 1u gi J [Abst tract. ] Tezisy pd komo он 3-10 Iiulia 1975 pe grad) 1: . 1976. Materials За i of Turkey XXXII: Campanulaceae. Notes Roy. Bot. Gard. Edin- burgh 35: 39-52. В ^ 1830. Monographie des Campanu- lées. P. DUNBAR, ^X 1973. Pollen ontogeny in some species of Campanulaceae. A study by electron micros- copy. Bot. Not. 126: 277-315. 5a. On pollen of Campanulaceae and re- lated families bow special reference to the surface ultrastructure. I. Campanulaceae subfam. Cam- panuloideae. Bot. Not. 128: 73-101. 5b. On pollen of Campanulaceae and re- lated families with special reference to the surface ultrastructure. II. Campanulaceae subfam. Cy- phioidae and subfam. Lobelioidae; Goodeniaceae; PANA Bot. Not. 128 102- 118. . WALLENTINUS. 1976. On pollen of Cimpangliéede. III. A numerical taxonomic in- vestigation. Bot. Not. 129: 69-7 Ewan, J. 9. Areview of the genus Githopsis. Rho- 3. . A. 1957. “Campanulaceae.” Pp. 126- , 459-475 in B. K. Shishkin (editor), Flora SSSR, Volume 24. Akademii Nauk SSR, Moscow sacaceae, Cucurbitaceae, Campa nulaceae. Israel Program for Scientific КАШЫ ск. Jerusalem.] GADELLA, T. W Cytotaxonomic studies in the genus Campanula. Wentia 11: 1-104. Some notes on the delimitation of gen- era in the Campanulaceae. I, II. Proceedings of the Royal Netherlands Academy of Science, series C 69: 502-5 Ca mpanulales. Pp. 704—708 in H. H. Benton (publisher), The New Encyclopedia Bri- tannica, 15th edition, Macropaedia, Volume 3. Encyclopedia Britannica, Incorporated, Chicago. —CAMPANULACEAE 687 GEsLOT, A. 1980. Le tegument séminal de quelques à balayage. Adansonia, series 2, 19: 307-318. HECKARD, L. R. 1969 [1970]. A new Campanula from northern California. Madroño 20: 231-235. HEIDENHAIN, B. . Uber die Blutenstande der mpanulaceen. Akademie der Wissenschaften und der Literatur. Abhandlungen der Mathema- tisch-Naturwissenschaftlichen Klasse 1952: 3-32. KovANDA, M. 1970a. Polyploidy and variation in the Campanula rotundifolia complex. Part I. (Gen- eral). Rozpravy Ceskoslovenské Akademie Ved, Rada Matematickych a Prirodnich Ved 80(2): 1- d and сла in the Сат rt II. (T tanica & Phytotaxomica, 3 Я Polyplo idy and variation in the Cam panula rotundifolia complex. Part 4 n he a eobotanica & Phytotaxonomica, mpanulaceae. Pp. 254-256. H. itor), Flowering Plants f. the World. Mayflower ri Ne McVAUGH, R l. w name for Heterocodon merican Campan- ulaceae. Bartonia 5b. The genus Triodanis Rafinesque, and its relationships to Specularia and Campanula. htia 1: 13- —— Generic status of Triodanis and Specu- laria. Rhodora 5 : 38-4 MoRIN, N. 1980. Sy stematics of the annual California campanulas (Campanulaceae). Madroño 27: 149- ——. 1983. Systematics E 6 (Campanu- m Syst. Bot. 8: 436- ——— & S. G. SHETLER. T V" morphology in No EA American Campanulaceae. [Abstract.] Bo- tanical Society of America, Miscellaneous Series 60: 75 MuRLEY, M. В. 1951. Seeds of the Cruciferae of lpr ast North America. Amer. Midl. Natu- ralist 46: 1-81. Рнито$, D. (PR Trilokulare Campanula — Arten der Agais. Osterr. Bot. Z. 111: 208-230. . ES Die о. дали Campanula— n. Osterr. Bot. Z. 112: 449-498. S. D. 65. Revision x согори und nordafrikanischen Vertreter der Subsect. Hetero- phylla (Wit.) Fed. der Gattung Campanula L. Feddes Repert. 71: 50-187 T i ddr Campanulaceae. Pp. 40-70 in Engler and K. Prantl (editors), Die naturlichen Pilanzenfamilien, Volume IV, part 5 (1894). Leip- SHETLER, S.G. 1958. The Taxonomy and Ecology of anula americana L. in the Laurel Hill Re- ion of Pennsylvania. M.S. Thesis. Cornell Uni- versity, Ithaca, New York 688 1 . Notes on the life history of d americana, the tall bellflower. Michigan Bot. 963. A рн ien key to the species of Campa ыр — or commonly naturalized in 2. Rhodo ora ipo 319-337. —, Variation and Evolution of the Nea tic Harebll (C SE Subsect. Heterophylla). J. Cramer, Vaduz ——— EWS. Generic position of Ca ampanu ula american 2r ү ер t.] ASB [Asso- ciation of sera Biologists) Bulletin 14(2): 40. & М. К. Morin. 1981. Seed morphology in North i Campanulaceae. XIII Interna- tional Botanical Congress Abstracts, 21-28 August 1981 (Sydney, Australia), p. 136. SMALL, J. K. 1903. Flora of the a United States. Published by the author, New Y A new bellflower from Florida, Torreya 26: 35-36 Я 33. Manual of the Southeastern Flora. Published by the author, New York. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 STAFLEU, F. A. (editor). 1978. International Code of Botanical и Bohn, Scheltema & Hol- kema, Utre STEARN, W. T. 1966. Botanical Latin. Thomas Nel- son and Sons, London. THULIN, M. 1974. Gunillaea and Namacodon. Two new genera of Campanulaceae in Africa. Bot. Not. 127: 165-182. 1975. The genus Wahlenbergia s.lat. (Cam- panulaceae) in ra Africa and Madagascar. Symb. Bot. Upsal. 21: TUTIN, x a (editor). TU ‘Campaniiacess: Pp. 74- T. G. Tutin, V. H. Heywood, N. A. Burges, D. M. Moore, D. H. Valentine, S. M. Walters, and D. A. Webb (editors), Flora Europaea, Volume 4 Cambridge Univ. Press, Cambridge. Legousia Durande. P. 94 in T. G. Tu- . W (editors), Flora Europaea, Volume 4. Cambridge Univ. Press, Cambridge. WARD, D. B. (editor). [1979.] Pee Rare and En- dangered Biota of Florida 5: 75. REVISION OF THE GUAYANA HIGHLAND BROMELIACEAE! LYMAN B. SMITH? ABSTRACT Genera and species of Bromeliaceae that have been described from the Guayana Highland since Downs in 1974 and 26 are here described as new, 24 being Pitcairnioideae. Lindma 1977 are distinguished by amended or completely bis die eys. ania is resurrected from the synonym of Cottendorfia, and an error in the validation of Steyerbromelia i is colada. At the time of the monographs of the first two subfamilies of the Bromeliaceae in the Flora Neotropica (Smith £ Downs, 1974, 1977), the area from which the most future additions were expected was the Guayana Highland. Recent collecting in the area has confirmed the suspi- cion. Among the new species added are 26 de- scribed in the present paper. In addition to the new species, mostly in the Pitcairnioideae, two Pitcairnioid genera, Brewcaria and Steyerbro- melia, have been described (in Steyermark et al., 1984) since the Flora Neotropica monograph, and more recent studies have shown that Gua- yanian Lindmania must be resurrected from the synonymy of the genus Cottendorfia of eastern € The present paper, in addition to describ- ew species, summarizes by use of keys the ios of all species and genera described from the area since the monograph. Also, an error in the validation of Steyerbromelia is corrected. The area covered in the present paper is the same as in the Bromeliaceae of the Guayana Highland (Smith, 1967), which should be re- ferred to for a more complete presentation of the family in the area, especially in the subfamilies Tillandsioideae and Bromelioideae. The num- bers preceding the species and genera follow those in the Flora Neotropica. New entries are given a decimal number under the number of the genus or species with which they would appear in the keys of the Flora. In view of the many additions, keys are provided to all pid species of the , Lindmania, amended keys are provided for many parts of Navia. PITCAIRNIOIDEAE KEY TO GENERA la. Seeds appendage 2a. P etal-blades p spiraled after anthesis; ovary superior or slightly inferior. и ae of Puya pes and summits from Costa Ric 2b. a blades remaining separate after anthesi eis Guyana to Chile and Argentin . Ovary wholly superior; ovules S m ‘petals n regular 4a. Sepals convolute with the left side of each overlapping the right of the next one; petals ighlan naked; inflorescence simple or compoun brightly pe more or less — together after anthesis but not twisted; sepals large 4. ite or rose, сог after anthesis; — not over 10 m Sa. Anthers subbasifixed; petals 5b. Anthers equitant; petals w thin at least margina . Guayana Connellia ©. 690) — i 690) long. 4b. эмдер imbricate; petals en at with 2 vertically attached scales; d scen 5a. o (p. 699) 3b. Э ib to wholly inferior. 6a. Sepals convolute with the left side of each overlapping the right of the next one; си b usually zygomorphic and forming a hood over the anthers. Mexico and the il 8. Indies to Argentina and Braz Pitcairnia - 700) ! In this paper the new species are coauthored by Julian Steyermark, who collected and made initial obser- vations of most of the specimens, and in the Pitcairnioideae by Harold Robinson who provided the anatomical observations. Ca ANN. MISSOURI Bor. GARD. 73: 689—721. 1986. reful editing of the manuscript by Robert Read is greatly appreciated. ? United States National Museum, Washington, D.C. 690 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 6b. Sepals imbricate with both posterior ones overlapping the anterior; petals small, regular. Guayana Highland. 7a. Epigynous tube lacking; inflorescence open and definitely branched 9. Brocchinia rA 702) 0. Epigynous ш well developed; ПР р ае capita ensua 7b. 16. Seeds naked or with a very n гоо 8а. Petals naked; i cuca sini pound o ntial р» nm end Highland iu Navia (p. 703) 8b. Petals appendaged with 2 decas ee ae вни simple, cents cylindric ....... 11а. Brewcaria (р. 714) 4. Катта М. E. Brown, Trans. Linn. Soc. I. 6: 66, pl. 13. 1901. la. Inflorescence compound although the short branches often covered by ample primary bracts at anthesis. 2a. Sepals ecarinate, 12 mm long; branches not elongating in fruit. Roraima ........ . C. augustae . Sepals sharply carinate, 14-21 mm long; N c ate at base; inflorescence glabrous; sepals m long. elliptic, obtuse, 14 m tari- tepui, Cerro Venamo ...... : . nutans 3b. Leaf-blades narrowly pale-margined, serrulate throughout; inflorescence villose or S sepals lanceo- late, acuminate, 21 vee ong. Auyan- tepui C. al Ib. Inflorescence sim 4a. Leaf-blades nee tomentose- ipm above and on the margins. Roraima .... 4b. Leaf-blades glabrous above, green w y a prominent white margins. Ilu- da Gra Sabana . . С. caricifolia 1.1. Connellia varadarajanii L. B. Smith & Steyermark, Journ. Bromel. Soc. 35: 52, fig. 2. 1985 (fig. 1 was actually C. smithiana Steyermark & Luteyn The leaf anatomy of the recently described species has been examined to allow comparison with other members of the genus. The vascular bundles are recessed in chlorenchyma adaxially but exposed to water storage tissue, they are nar- rowly covered by chlorenchyma abaxially; water storage cells 3-4-stratose adaxially, ca. 3-stratose in distinct canals abaxially; subepidermal cells with one subdistinct layer adaxially with outer walls slightly thickened, with one distinct mod- erately thickened layer abaxially; substomatal pore not occluded, oval or rounded; abaxial sur- face with minute subpeltate or shortly strap- good oid bis oral oblong. pattern ol fsome water storage tissue and a narrow layer of chlo- renchyma abaxial to the vascular bundles similar to the pattern in the four species of Connellia known at the time of the anatomical study of Robinson (1969). о the tour, Du С; quelchii has larger bundles | None of the other species has as much abaxial water storage tissue, but C. nutans seems closest in this regard. Connellia nutans and C. augustae do not have the vascular bundles recessed in the chlorenchyma adaxially as in the new species, C. quelchii, and C. caricifolia. The general form of the leaf cross-section is the type seen again in such species of Lindmania as L. brachyphylla, L. tillandsioides, L. riparia and L. sessilis, and the genus Connellia may re- late to that element of Lindmania. The general type of leaf cross-section seen in Connellia is not found in the recently described C. smithiana in which the larger vascular bun- dles all have fiber sheaths continuous with the abaxial epidermis. This reenforces other evi- dence that C. smithiana should be transferred from Connellia to Lindmania. The abaxially buttressed bundles are similar to those seen in Lindmania minor, L. thyrsoidea, and the newly described L. huberi and L. imitans. 2. Connellia augustae N. E. Brown is endemic to Roraima. Other citations in Fl. Neotrop. no. 14(1): 211, 1974, are C. nutans L. B. Smith. 5.2. LINDMANIA Mez, DC. Monogr. Phan. 9: 535. 1896 The resurrection of the Guayanian Lindmania comes as a result of further study that puts greater emphasis on the aestivation or overlapping of the sepals. There have been two previous indi- cations of generic difference noted between Lind- mania and the Bahian Cottendorfia to which I had reduced it (Smith & Downs, 1974: 212). Robinson (1969: 8) pointed out significant dif- ferences in the cell walls of the abaxial epidermis and in the stomata. Recently I verified by spec- imens the unisexual nature of the flowers in Cot- 1986] tendorfia. Mez (1894, pl. 93) had illustrated it as unisexual although he described the genus as having “Flores hermaphroditi” (p. 502). Finally, I now find that the sepals of Lindmania are con- volute with the left side of each overlapping the right of the next one while those of Cottendorfia are cochlear with the anterior sepal overlapping the two posterior ones. la. ав simple ог subsimple. 2 SMITH—GUAYANA HIGHLAND BROMELIACEAE 691 The following key, although still largely arti- ficial, eliminates weak points in that of the Flora Neotropica but keeps its enumeration. Robin- son’s findings are included and in some cases show correlation with the macroscopic charac- ers e a. Scape very short or lacking; inflorescence dense, ш, in the leaves. Chimanta. 3a. gun blades straight; sepals broadly lanceolate, 9 mm long 24. L. navioides Leaf-blades strongly recurved; sepals suborbicular, 1 mm | long 24.1. L. huberi 2b. Scape evident; inflorescence lax or sublax. 4a. Flowers 1 in m inflorescence otherwise very lax. Chimantá „u 20. L. subsimplex 4b. Flowers s 5a. Sc bipinnate). No 5b. Scape-bracts exceeding the Melon а ts much shorter than the internodes; scape slender (inflorescence normally rthwestern sa eae and adjacent Venezuela 13.25, guianensis rnodes 6a. Plant arachnoid- ae pisa relatively stout, 4 mm thick, to 36 cm long. Marahuac me a L. arach s 6b. n obscurely lepidote; scape 1.5 mm thick, to 10 cm long. Chim . Sca ape-bracts very narrow, completely exposing the scape; cat blades 5 mm wide 20. L. aurea Scape-bracts ample, densely imbricate; leaf-blades 10 mm wide .... 20.3. L. imitans subsimple wi 7b. Ib. pr bipinnate or rarely subtripinnate or s Pri ith a distinct basal branch. ry bracts equaling or exceeding the sterile bases of a at least the lower branches. 9a. "Floral bracts about equaling to exceeding the pedicels. 10а. Pedicels 8 mm long; inflorescence tomentose-lepidote. Neblina ........................ 16. L. maguirei 10b. Pedicels not more than 3.5 mm long or lacking. 11a. Branches 1.5-4 cm long, inflorescence narrowly cylindric, at least at apex. 12a. Floral ione за e dh Jeaf-blades entire, much Pew e th wide. Chim pe ylla 12b. Floral ina pues to broadly e lpi, leaf blades toothed near the base or not much exceeding the inflor 13a. Inflorescence dense only near a Scape- -bracts shorter than the upper internodes; leaf blades entire, ca. 1 cm wide. АБасара-ери!........... 19. L. tillandsioides af 13b. Inflorescence dense throughout; scape-bracts — imbricate: le blades minutely toothed basally, 35 mm wide. Auyan 116. Branches more than 4 cm long, divergent; inflorescence lax or sublax. 1 owers 3-3.5 mm pedicellate. 15a. Inflorescence subdensely vestite with microscopic glanduliform fer- nous scales; primary bracts entire; leaf-blades serrulate only е base. 11 Paru L. phelpsiae 15b. Inflorescence flocculose with fine linear white scales; prim ary bracts serrulate; leaf-blades serrulate throughout. Apacara-tepui, Chimanta 2. L. serrulata 14b. Flowers sessile. Chim 16a. Leaf-blades salas 35 mm wide; plant flowering over 7 dm hi 2.1 gh .. . L. sessilis 16b. Leaf-blades entire, 12 mm wide; plant flowering 46 cm high ......... 2.2. L. saxicola 9b. Floral bracts distinctly shorter than the pedicels. 17a — A base longer than the fertile part of branch, about equaling primary bract. pe 13. L. longipes 17b. Sterile base shorter than fertile part of branch, much shorter than the basal primary bra 182. Inflorescence арі in leaf-axils, subtripinnate. Neblin rimary 18b. Inflorescences term 19a. . L. lateralis bracts equaling or a than the шш. Branches imis inflorescence densely cylindric. 692 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 20a. Primary bracts very narrow, wholly exposing the чш piden cence erect. Serrania Yutaje L. thyrsoidea 20b. iacu bracts d covering most of each branch; о. 2 rate. Auyan-tepui 1.1. L. smithiana 19b. Branches үсүе inflo orescence lax at least at base. ш a spreading, about as long as the flow 22a. Inflore nce glabrous; pedicels 10-15 mm long; kat blades laxly serrate oa Duida, Abacapá-tepui, Marahuaca ......... L. wurdackii 22b. Inflorescence vestite; pedicels 3.5-7 mm lon 23a. Pedic els 3.5 mm long; inflorescence vestite with pale nar- row spreading scales. Cerro Venamo, Gran Savanna ......... 3. L. gracillima 23b. Pedicels 7 mm long; inflorescence ноза gine culose. Neblina L. dendritica 21b. Pedicels cylindric or obconic, longer than the flowe 24 af-blades densely flocculose beneath, obscurely denticulate vu SOME 23. L. minor r out; inflorescence occu lose. C 24b. Leaf-blades glabrescent, serrulate at bas 25a. Inflorescence vestite, scape and p i m bracts with dis- tinct ovate base and long very narrow blade. Serra Aracá l. L. piresii Inflorescence glabrous; scape and primary bracts narrowly triangular with little distinction between base and blade. Marahuaca. 26a. Leaf-blades verruculose; flowering shoot obviously very tall (ca. 2 m estimate) and straight .................. 23.2. L. terramarae 2 [27 B 26b. ovis 175! 4 1 г пег ; g tall and бау Rer 23.3. L. marahuacae 8b. Primary bracts all shorter than the sterile bases of t the branches. 27a. Floral bracts exceeding the pedicels. Leaf-blades linear- cin with ы convex sides, 33 mm wide, plane ч narrowly revolute. Sarven-tepui, Auyan-tepu , brachyphylla 28b. e narrowly triangular de straight sides, 15 mm wide, хоу. involut . L. таш Chim 27b. Floral bracts ы equaling or shorter than the pedic 29a els. pe-bracts much shorter than all but the lowest internodes. 30a. a blades serrulate toward base; margins recurved; axis 5 mm thick; racemes 14 nse. Western Venezuela a an x adjacent Guyana and Brazil ............... . L. geniculata 30b. Midas Ie Seir о axis 2.5 mm thick. racemes lax. Roraima, 15. Kaieteur Plateau, Gra 29b. Scape- aedi, и all ог an but the highest internodes. 3la L. guianensis PM . Indument on the upper side of the leaf кош > recurved. Central Guayana Highlan 5. L. paludosa 31b. Indument mostly or wholly on "e underside of the leaf-blade; a us or vestite. 32a. Racemes lax; leaf-blades 15 mm wide. Central Guayana Highland .. 10. L. argentea 32b. Racemes dense or subdense; leaf blades 24-40 mm wide. 33a. Pedicels Е ш m long, much exceeding the floral bracts; leaf-mar- 3 ы gins геси 34а. Sepals retuse, verruculose; leaf-blades 40 mm wide, le in- dument persistent; pedicels 5 mm long. Serrania Bisa: TOS uS 34b. Sepals d elliptic, even; leaf-blades largely е 25- wide. 35a. Вгапсһев curved ascending. Neblina ....... 7. L. savannensis 35b. Branches straight, spreading. Cerro ipud ima ........ . L. atrorosea . Pedicels 1.5-3 mm long, about Wagen jd ps bract 36a. Sterile bases of the branches to long, are twice the length of the primary bracts. nou ИНИН ЖЕНЕ 8. L. nubigena 36b. Sterile bases of the branches not over E cm long, only slightly exceeding the primary bracts. Auyan-tepui ...... 9. L. dyckioides 1986] 2. Lindmania serrulata L. B. Smith, Contr. U.S. Nat. Herb. 29: 283, fig. 8. 1949. Cottendorfia serrulata (L. B. Smith) L. B. Smith, Phyto- logia 7: 170. 1960 2b. Lindmania serrulata var. reducta L. B. Smith, Mem. N.Y. Bot. Gard. 9: 414. 1957. Cottendorfia serrulata var. reducta (L. B. Smith) L. B. Smith, Phytologia 7: 170. 1960. 2.1. dris sessilis L. B. Smith, Steyermark nson, sp. nov. Figures 1, 27a-d. TYPE: o Piar, Macizo Chimantá-tepui, cabeceras orientales del Cano Chimantá, ca. 2,000 m alt., Lat. 5?18'N, Long. 62?09"W, 26-29 Enero 1983, Julian A. Steyermark, Otto Huber & Victor Carrefio E. 128104 (holotype, US; isotype, VEN) Planta e fragmentis perveteribus solum cognita, flo- rigera ultra 7 dm alta. Folia 5 dm longa; vaginis bre- vibus amplis; laminis anguste triangularibus, mm latis, omnino serrulatis, ad basim subtus dissite lepi- dotis. Scapus rectus, ca. 15 mm diametro; scapi bracteis erectis, subfoliaceis, internodia multo superantibus. Inflorescentia laxe bipinnatim panona setifera; asibu sterilibus brevissimis longioribus; ramis rectis diver- longa, base setifera. Capsula ovoidea; seminibus bi- caudatis. Leaf anatomy. Vascular bundles not or little recessed in chlorenchyma and exposed to water storage tissue adaxially, narrowly covered by chlorenchyma abaxially; water storage cells 3-4- stratose adaxially, in distinct canals with firm- walled cells abaxially, subepidermal cells adaxially and abaxially bistratose, smaller cells thickened on the exterior part of their walls; sub- stomatal pores small, oval, with walls subin- crassate; abaxial scales small, peltate, with cells of the disk indistinct. This leaf anatomy is seen also in Lindmania in L. tillandsioides, L. в and the newly described L. riparia, and it is probably most closely related to these in spite of the shorter primary bracts or longer and laxer branching of the inflorescence that separate the species in the key. The new species seems to have the vascular bundles adaxially less recessed in the chloren- chyma than the other species. Known only from the type. SMITH—GUAYANA HIGHLAND BROMELIACEAE 693 2.2. Lindmania saxicola L. B. Smith, Steyer- mark & Robinson, sp. nov. Figures 2, 27е- h. TYPE: Venezuela. Bolivar: Distrito Piar, Macizo del Chimanta-tepui, cabeceras del afluente derecho superior del rio Tirica (“Caño del Grillo"), ca. 2,450 m alt., Lat. A Long. 62?03'W, 7-8 Feb. 1983, Ju- lian A. Steyermark, Otto Huber & Victor E 128945 (holotype, US; isotype, VEN). Planta e fragmentis perveteribus solum cognita, flo- rigera plus quam 46 cm alta. Folia superiora 13 cm longa; vaginis brevibus amplis; laminis anguste trian- gularibus, ad 12 mm latis, integris, utrinque minutis- sime obscureque lepidotis. Scapus rectus, 5 mm diametro; scap: зонани erectis verisimilite Pon. paniculata; bracteis primariis not pe Sepala lat seminibus bicaudatis. Leaf anatomy. Vascular bundles not re- cessed in or covered by chlorenchyma adaxially, broadly covered by chlorenchyma abaxially; adaxial water storage cells multi-stratose, the outermost layer resiniferous; adaxial subepider- mal cells uni-stratose, vertically elongate, in- crassate, abaxial subepidermal cells tri-stratose, isodiametric, strongly incrassate; substomatal pores constricted between the oblong, resinifer- ous lateral subsidiary cells. The species can be distinguished in Lindmania by the very thick subepidermal layering on the abaxial surface of the leaf and the constriction of the substomatal pore between the prominent lateral subsidiary cells. The massive abaxia chlorenchyma is like that previously noted in L. subsimplex which is possibly closely related. Known only from the type. 3. Lindmania gracillima (L. B. Smith) L. B. Smith, comb. nov. Cottendorfia gracillima L. B. Smith, Phytologia 7: 418, pl. 1, figs. 3-5. 1961. 4. Lindmania wurdackii L. B. Smith, Mem. N.Y. Bot. Gard. 9: 284, fig. 13. 1957. Cottendorfia wurdackii (L. B. Smith) L. B. Smith, Phy- tologia 7: 171. 1960. 4.1. Lindmania lateralis (L. B. Smith & R. W. Read) L. B. Smith & H. Robinson, comb. 694 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 2959876 оны == - - IGURES 1-4.—1. Lindmania sessilis, Smith, Steyermark & Robinson.—2. Lindmania saxicola Smith, Stey- ermark & Robinson.—3. Lindmania aurea Smith, Steyermark & Robinson.—4. Lindmania imitans Smith, Steyermark & Robinson. 1986] nov. Cottendorfia lateralis L. B. Smith & R. W. Read, Phytologia 30: 289, pl. 1. 1975. The species description is emended by the fol- lowing observations of leaf anatomy: Vascular bundles not covered with chlorenchyma adaxi- ally or abaxially; adaxially water storage cells mostly 2—3-stratose, abaxially in distinct narrow canals; subepidermal cells adaxially 1—2-stratose, often resiniferous, not or scarcely incrassate; sub- stomatal pores not occluded, oval or rounded; abaxial scales rudimentary, rarely with a few pro- jecting cells. The leaf cross-section is of the most typical Lindmania type as seen in L. guianensis and L. argentea. 5. Lindmania paludosa L. B. Smith, Mem. N.Y. Bot. Gard. 9: 284, fig. 14. 1957. Cottendorfia paludosa (L. B. Smith) L. B. Smith, Phy- tologia 7: 170. 1960 6. Lindmania cylindrostachya L. B. Smith, Mem. .Y. Bot. Gard. 9: 286, fig. 15. 1957. Cot- tendorfia cylindrostachya (L. B. Smith) L. B. Smith, Phytologia 7: 169. 1960. 7. Lindmania savannensis (L. B. Smith) L. B. Smith, comb. nov. Cottendorfia savannensis L. B. Smith, Mem. N.Y. Bot. Gard. 10(2): 16, fig. 3. 1960. 7.1. Lindmania atrorosea (L. B. Smith, Steyer- mark & Robinson) L. B. Smith, comb. nov. Cottendorfia atrorosea L. B. Smith, Steyer- mark & Robinson in Steyermark et al., Brit- tonia 33: 28, figs. 1, 2E, F. 1981. 8. Lindmania nubigena (L. B. Smith) L. B. Smith, comb. nov. Cottendorfia nubigena L. B. Smith, Mem. N.Y. Bot. Gard. 10(2): 16, fig. 2. 1960. 9. Lindmania dyckioides (L. B. Smith) L. B. mith, comb. nov. Cottendorfia dyckioides L. B. Smith, Mem. N.Y. Bot. Gard. 14(3): 21, fig. 1. 1967. 10. Lindmania argentea L. B. Smith, Mem. N.Y. Bot. Gard. 9: 414, fig. 78. 1957. Cottendorfia argentea (L. B. Smith) L. B. Smith, Phyto- logia 7: 169. 1960. 11. Lindmania phelpsiae L. B. Smith, Mem. N.Y. Bot. Gard. 9: 286, fig. 16. 1957. Cot- tendorfia phelpsiae (L. B. Smith) L. B. Smith, Phytologia 7: 170. 1960. 12. Lindmania dendritica (L. B. Smith) L. B. Smith, comb. nov. Cottendorfia dendritica L. B. Smith, Mem. N.Y. Bot. Gard. 18(2): 31, fig. 5c. 1969. SMITH—GUAYANA HIGHLAND BROMELIACEAE 695 13. Lindmania longipes (L. B. Smith) L. B. Smith, comb. nov. Cottendorfia longipes L. B. Smith, Mem. N.Y. Bot. Gard. 14(3): 22, fig. 2. 1967. 14. Lindmania geniculata L. B. Smith, Mem. N.Y. Bot. Gard. 9: 414, fig. 79. 1957. Cot- tendorfia geniculata (L. B. Smith) L. B. Smith, i gii 170. 1960. 14b. Lindmania geniculata var. minor (L. B. Smith, Steyermark & Robinson) L. B. Smith, comb. nov. Cottendorfia geniculata var. mi- nor L. B. Smith, Steyermark & Robinson in Steyermark et al., Acta Bot. Venez. 14(3): 15. 1984. . Lindmania guianensis (Beer) Mez, DC. Monogr. Phan. 9: 537. 1896. Anoplophytum guianense Beer, Bromel 44. 1857. Cotten- dorfia guianensis (Beer) Baker, Handb. Bro- mel. 129. 1889 15b. Lindmania guianensis var. vestita (L. B. Smith) L. B. Smith, comb. nov. Cottendorfia guianensis var. vestita L. B. Smith, Phyto- logia 7: 419. 1961. . Lindmania maguirei (L. B. Smith) L. B. Smith, comb. nov. Cottendorfia maguirei L. B. Smith, Mem. N.Y. Bot. Gard. 18(2): 31, fig. 5d, e. 1969. . Lindmania brachyphylla L. B. Smith, Mem. N.Y. Bot. Gard. 9: 416, fig. 80. 1957. Cot- tendorfia brachyphylla (L. B. Smith) L. B. Smith, Phytologia 7: 169. 1960. 17b. Lindmania brachyphylla var. angustior (L. B. Smith, Steyermark & Robinson) R. W. Read, comb. nov. Cottendorfia brachyphylla var. angustior L. B. Smith, Steyermark & Robinson in Steyermark et al. Brittonia 33(1): 28. 1981 18. Lindmania steyermarkii L. B. Smith, Mem. N Gard. 9: 416, fig. 81. 1957. Cot- tendorfia steyermarkii (L. B. Smith) L. B. Smith, Phytologia 7: 170. 1960 19. Lindmania tillandsioides L. B. Smith, Mem. N.Y. Bot. Gard. 9: 416, fig. 82. 1957. Cot- tendorfia tillandsioides (L. B. Smith) L. B. Smith, Phytologia 9: 171. 1960. — un ON ~ 19.1. Lindmania riparia 1. В. Smith, Steyer- mark & Robinson, sp. nov. Figures 5, 27q- t. TYPE: Venezuela. Bolivar: Auyan-tepui, along margins of stream, cumbre de la parte sur, meseta de piedra arenisca, entre “Oso Woods Camp” y “Libertador”, 2,050-2,300 ., 15 May 1964, Julian A. Steyermark 93895 (holotype, US; isotypes, MAC, NY) 696 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 5-8.—5. Lindmania riparia Smith, Steyermark & Robinson.— 6. Lindmania piresii Smith, Steyer- mark & Robinson.—7. Lindmania terramarae Smith, Steyermark & Robinson.— 8. Lindmania huberi Smith, Steyermark & Robinson. 1986] Caulescens, florigera 2 dm alta, caule (apice) 1 cm diametro. Folia superiora 4 dm longa; vaginis vix dis- tinctis; laminis lineari- lanceolatis, 35 mm latis, subtus marginatis basi minutissime denticulatis Scapus brevis, crassus; scapi bracteis foliaceis, erectis, dense imbricatis. Inflorescentia dense bipinnatim pan- iculata, 15 cm longa, trichomatibus subulatis fragilibus basil vestita; bracteis чй 115 subfoliaceis, ramorum bases steriles super- ntibus. Bra florigerae lanceolatae, pedicellos crassos s то sepalis ellipticis, 5 mm longis, late albido-marginatis; petalis unguiculatis, albis (Steyer- mark). Leaf anatomy. Vascular bundles adaxially deeply recessed in but not covered by chloren- chyma, abaxially narrowly covered by chloren- chyma; water storage cells adaxially 3-4-stratose, abaxially ca. 3-stratose in distinct canals; sub- epidermal cells in single subdistinct layers with slightly thickened walls; substomatal pores not occluded, oval to rounded with slightly thickened walls; abaxially with scales minute, irregularly peltate or shortly strap-shaped, with cells trans- versely oblong. The leaf cross-section is like that seen in Lind- mania in L. tillandsioides, L. brachyphylla, and the bis described L. sessilis. Both the key and the anatomy indicate closest relationship to L. En imd In spite of the close proximity in the key, L. stenophylla lacks the chlorenchyma abaxial to the vascular bundles, and it is probably not closely related. Known only from the type. 20. Lindmania subsimplex L. B. Smith, Mem. .Y. Bot. Gard. 9: 417, fig. 83. 1957. Cot- tendorfia subsimplex (L. B. Smith) L. B. Smith, Phytologia 7: 170. 1960. 20.1. Lindmania arachnoidea (L. B. Smith, Steyermark & Robinson) L. B. Smith, comb. nov. Cottendorfia arachnoidea L. B. Smith, Steyermark & Robinson in Steyermark et al., Acta Bot. Venez. 14(3): 9, figs. 7, 101-п. 1984 20.2. Lindmania aurea L. B. Smith, dan & Robinson, sp. nov. Figures 3, 27i-m. TYPE: Venezuela. Bolivar: Distrito Piar, 25 del Chimantá, pequeñas altiplanicies en la base septentrional de las farallones supe- riores del Amuri-tepui (Sector Oeste del Acopán-tepui), por escarpado encima de una cascada al oeste del acampamento, ca. 1,850 m alt., Lat. 5°10’N, Long. 62?07' Oeste, 2- SMITH—GUAYANA HIGHLAND BROMELIACEAE 697 5 Feb. 1982, Julian A. Steyermark, Otto Huber & Victor Carreño E. 128611 (holo- type, US; isotypes, NY, VEN). Breviter caulescens, florigera ca. 20 cm alta, leviter curvata, ascendens. Folia 1 cm longa, sq Е triangularibus, longe attenuatis, 5 mm latis, egri ron s gracillimus, glaber; scapi bracteis erectis, subfoliaceis, angustissimis quam internodis longic ori- plex, sublaxa, 7 cm longa, glabra. Bracteae florigerae eis scapi simulantes, pedicellos inferiores su n antes; pedicellis ou gracilibus, ad ongis. Sepala ovata, obtusa, 3 mm longa, tenuia; petalis spa- thulatis, 10 mm longis aureis (Steyermark); ovario su- pero; ovulis caudatis Leafanatomy. Vascular bundles not covered by chlorenchyma adaxially, narrowly covered abaxially; adaxial water storage cells mostly 2- 3-stratose, abaxially cells smaller and bistratose in distinct but nearly c contiguous canals; subepi- scales rudimentary, eccentrically peltate, some- times strap-shaped, cells transversely oblong. Leaf anatomy is notable for the rounded, non- recessed vascular bundles with a narrow abaxial covering of chlorenchyma and for the nearly con- tinuous layer of abaxial water storage tissue. As such, the species does not seem closely related to most of the species with simple inflorescences with which it falls in the key. Closest relationship may be to L. imitans described below Known only from the type. 20.3. Lindmania imitans L. B. Smith, Steyer- ar bi Acopán-tepui), on vertical small bluff east of camp, Lat. 5?10'N, Long. 62%07'W, ca. 1,850 m alt., 2-5 February 1983, Julian A. Steyermark, Otto Huber & Victor Carreno E. 128474 (holotype, US; isotype, VEN). Breviter caulescens, florigera ca. 9 cm alta, leviter q sed obscure vestita; vaginis brevibus, latis; laminis an- guste triangularibus, attenuatis, 10 mm latis, integris. с р gi sage gl у. e Ф as : dense 698 imbricatis, subfoliaceis. opa hee iis D laxa, 5 cm longa, glabra. Bracteae florigerae eis scapi simu- m longis. Sepala sub- orbicularia, 3. onga; ovario supero Leafanatomy. Vascular bundles not covered with chlorenchyma adaxially or abaxially, often with fiber-sheath extending to the epidermis; adaxial water storage cells 3-5-stratose; subepi- dermal cells smaller, unistratose, slightly incras- sate; substomatal pores not occluded, oval or rounded; points of attachment of abaxial scales not seen, evidently very sparse. The species has some resemblance to L. aurea with which it falls in the key, and it may be comparatively closely related. Nevertheless, the leaf cross-sections are strikingly different. The vascular bundles of the present species tend to be connected by fibers to the abaxial epidermis while the equivalent area in L. aurea is occupied by well developed water storage canals. Known only from the type. 21. Lindmania thyrsoidea L. B. Smith, Mem. N.Y. Bot. Gard. 9: 287, fig. 17. 1957. Cot- tendorfia thyrsoide bees B. Smith) L. B. mith, Phytologia 7 170. 1960. 21.1. Er ото о & Lu- teyn) L. B , comb. nov. For leaf anat- omy see note Ende: Connellia varadaraja- nii. Connellia smithiana Steyermark & Luteyn, Journ. Brom. Soc. 35: 152, fig. 3. 1985. Error in C. varadarajanii 35: fig. 1 on p. 52. 1985. 22. Lindmania stenophylla L. B. Smith, Mem. N.Y. Bot. Gard. 9: 417, fig. 84. 1957. Cot- tendorfia stenophylla (L. B. Smith) L. B. Smith, Phytologia 7: 170. 1960. 23. Lindmania minor L. B. Smith, Mem. N.Y. Bot. Gard. 9: 419, fig. 85. 1957. Cottendorfia minor (L. B. Smith) L. B. Smith, Phytologia 7: 170. 1960. 23.1. Lindmania piresii L. B. Smith, Steyer- mark & Robinson, sp. nov. Figures 6, 27u- x. TYPE: Brazil. Amazonas: Serra Araca, 10 February 1975, J. M. Pires 15.010(33) (ho- lotype, US; isotype, IPEAN) Caulescens, florifera 1-1.5 m alta (Pires), caule 30- 50 cm alto. Folia d albis vestitis, glabresce , basi laxe serrul Scapus rectus, ie glabrescens; о рае ех ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 basi brevi ovata longe angustissimeque laminatis, in- s, 4-5 mm long basi sulcata, apice scariosa; petalis 7 mm longis, albis Leafanatomy. Vascular bundles not covered adaxially or abaxially by chlorenchyma; water storage cells forming distinct abaxial canals; sub- epidermal cells adaxially and abaxially 1-2- stratose, often resiniferous, not or scarcely in- crassate; substomatal pores not occluded, oval or rounded; abaxial scales peltate, with a fringe of elongate, contorted cells. The leaf cross-section is of the most typical Lindmania type as seen in L. guianensis, L. ar- gentea, and relationship is probably to that group. Most of the related species can be distinguished by either having primary bracts in the inflores- cence shorter than the sterile bases of the branch- es or having floral bracts equalling or exceeding the length of the pedicels. Known only from the type. 23.2. Lindmania terramarae L. B. Smith, Stey- po, sureste, alrededor de una lagunita, altura 2,800 metros, Lat. 3?35'N, Long. 65?20' Oeste, 26 March 1983, Julian A. Steyer- mark & Francisco Delascio 129067 (holo- type, US; isotypes, NY, VEN). (Q4 LY it A verisimiliter 2 m alta. Folia c ca. 4 dm lo brevibus amplis; laminis anguste triangulari ad 3 em latis, base laxe spinoso- тента шыш lepidotis. robustus glaber; inferioribus ignotis sed о subfoliaceis dense odia pau onga; Marr зь гатіѕ suberectis. vel divergentibus, ad 12 cm ngis клен гым 4mm longae; pedicellis patentibus, eni libus, 6 mm longis. Sepala late ovata, acuta, 5 m longa, tenuia; petalis 12 mm longis. Capsula oles Leafanatomy. Vascular bundles recessed in chlorenchyma and reached by narrow extensions of water storage tissue adaxially and abaxially, 1986] only smaller bundles sometimes covered abax- ially; water storage cells adaxially and abaxially 2—3-stratose with tenuous walls, abaxially form- ing broad often contiguous canals; subepidermal cells 2—3-stratose adaxially and abaxially, firm and often resiniferous, walls thickened in only the exterior parts of the outermost cells; substo- matal pores not occluded, oval; abaxially with numerous rudimentary scales that do not bear a distinct disc. Lindmania terramarae shows the type of leaf anatomy common in the large typical group of Lindmania. It is probably most closely related to L. marahuacae, from the same mountain, with which it shares a comparatively glabrous con- dition. The new species differs by the taller, straighter inflorescence and the more verrucose eaves. Anatomically the leaves seem to have more highly developed layers of subepidermis, that layer being scarcely distinct in L. mara- huacae. 23.3. Lindmania marahuacae (L. B. Smith Steyermark & Robinson) L. B. Smith, comb. nov. Cottendorfia marahuacae L. B. Smith, Steyermark & Robinson in Steyermark et al., Acta Bot. Venez. 14, no. 3: 9, fig. 1, k- r. 1984. 24. Lindmania navioides L. B. Smith, Mem. N.Y. Bot. Gard. 9: 419, fig. 86. 1957. Cot- tendorfia navioides (L. B. Smith) L. B. Smith, Phytologia 7: 170. 1960. 24.1. Lindmania huberi L. B. Smith, Steyer- mark & Robinson, sp. nov. Figures 8, 28a- d. TYPE: Venezuela. Bolívar: Distrito Pilar, Macizo del Chimantá, pequenas altiplani- cies en la base septentrional de los farallones superiores del Amuri-tepui (Sector Oeste del деорап- "арии еп Іа base de canon pequeno ajo nde el agua corre, ca. 1,850 m alt., Lat. ca. 5?10'N, Long. 62?07' Oeste, 2-5 February 1983, Julian A. Stey- & Victor Carreño E. y Tirepón (Torono)-tepui, Lat. 5?22'N 61°58’ Oeste, 2,540 m alt., 24 February 1978, Julian A. Steyermark, Victor Carreño E., Roy McDiarmid & Charles Brewer-Carias 115846 (US, VEN). Caulescens, florigera 13 cm alta, caule robustissimo 10 cm longo. Folia multa, in apicibus ramorum SMITH—GUAYANA HIGHLAND BROMELIACEAE 699 dense rosulata, 9 cm longa; vaginis — omnino t urvatis, eis inferioribus recurvatis, 10 mm latis, pungentibus, integris, glabris, glaucis (Steyermark), vix cretaceis Inflorescentia sessilis, _corymbi formis, multiflora. eu. gracilibus. Sepala suborbicularia, 35m ibera, inflata, nervata, squamis он dense vestita: petalis Sos 6 mm longis, cremeis = yermark); antheris ОЕ equitantibus; ario supero; ovulis caudat © Leafanatomy. Vascular bundles not covered by chlorenchyma adaxially or abaxially, not re- cessed in chlor enchyma Seay, vit fiber sheath extending to i er storage cells inflated, ca. 4-stratose; subepider- mal cells on both surfaces in one layer, with walls slightly incrassate; substomatal pores not oc- cluded, rounded; pits bearing abaxial scales not seen, evidently very sparse. Abaxially buttressed vascular bundles have described L. imitans, all of which fall in separate places in the key, and none of which seem par- ticularly closely related. Because of the sessile inflorescence, the new species keys out with L. navioides, but the latter represents a totally dis- tinct element in the genus with elongate flexuous tems and d ly imbricate, spreading, thin-tex- tured leaves. The thicker leaves of L. huberi with the non-contoured upper epidermis indicate re- lationship to the more typical element of Lind- nania. Known only from the type and paratype. ~ 1. STEYERBROMELIA L. B. Smith in Stey- ermark et al., Acta Bot. Venez. 14(3): 8. 1984. TYPE: Steyerbromelia discolor L. B. Smith & Robinson in Steyermark et al., sp. nov. Acta Bot. Venez. 14(8): 8. 1984. Plants rosulate; leaves spinose-serrate, anato- my similar to Lindmania except substomatal pores usually partially occluded; inflorescence compound; flowers sessile, not or scarcely nar- rowed below; sepals imbricate; petals regular, ap- pendaged with two vertically attached scales; sta- mens equal; ovary wholly superior, style branches often broadly lamelliform or lobed; seeds bicau- ate. The genus is known only from central Ama- zonas in Venezuela. Closest relationship is evi- dently to Lindmania from which it differs by the imbricated sepals and the appendaged petals. 700 Nomenclatural note: Steyerbromelia was orig- inally prepared as a monotypic genus with S. discolor as the only species. Steyerbromelia de- flexa was then added to the manuscript, but the resulting need to indicate a type species was over- looked. Cf. International Code Articles 37, 43. la. Branches suberect. Marahuaca ..... 1. 5. discolor 16. Branches decurved. 2a. Leaf-blades glabrous. Duida ... S. deflexa 2b. Leaf-blades densely cneeousepdot beneath. Aratitiyope 0. diffusa 1. Steyerbromelia discolor L. B. Smith & Rob- inson in Steyermark et al., Acta Bot. Venez. 14(3): 8, figs. 2a-i, 6i-k. 1984. 2. Steyerbromelia deflexa L. B. Smith & Rob- inson in Steyermark et al., Acta Bot. Venez. 14(3): 12, figs. 3a-f, 10i-k. 1984 3. Steyerbromelia diffusa L. B. Smith, Steyer- mark & Robinson, sp. nov. Figures 9, 28e— i. TYPE: Venezuela. Amazonas: Dept. Rio Negro, Cerro Aratitiyope, piedra ignea, 990— 1,670 m alt., Lat. 2°10'N, Long. 65?34' Oeste, ca. 70 km SSO de Ocamo, con riachuelos afluente al rio Manipitare, 24 Feb. 1984, Julian A. Steyermark, Paul Berry & Fran- cisco Delascio 130072 (holotype, US; iso- types, NY, VEN). _ Verisimiliter caulescens, florifera ultra 3 m alta. Fo- mis per anthesin div m longis, ramulis gracillimis, senectis 2 mm longae, с apiculata: ‚ floribus subsessilibus, paten- tibus. Sepala ovata, acuta, rioribu median mm PA albis (Steyermark); ovario supero; ovulis cau- dat Leafanatomy. Vascular bundles not covered by chlorenchyma adaxially or abaxially, abaxi ial cells sporadically resiniferous; substomatal ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 cells with unthickened walls, bearing pairs of pa- pillae protruding into pore; abaxially with sub- rotund scales, with smaller cells toward the den- tate and denticulate margin. Available material of flowers shows that the present species differs from the other two by the more elongate bases of its petals and its less broadened style branches. This third species of Steyerbromelia shows yet a third variation in the leaf cross-section with the vascular bundles ex- tending abaxially to the epidermis. In 5. deflexa the area abaxial to the bundles has a layer of chlorenchyma and one to three layers of small hyaline cells. In 5. discolor the same area has well developed water storage canals. Neverthe- less, evidence seems to show that the genus may be phyletically sound in spite of such differences. The new species and S. discolor have essentially webs structure of their abaxial leaf scales. Also, l pores hat is unlike any ne The third species, deflexa, has only rudimentary abaxial scales s cannot be compared while its substomatal pores tend to be somewhat irregular with some short intrusions. Known only from the type. 8. PITCAIRNIA L'Héritier, Sert. Angl. 7. 1788; nom. conserv REVISED SEGMENTS OF KEY TO SUBGENUS PEPINIA 9. Inflorescence bipinnate; sepals ac a. bracts 10 mm long; sepals 35 Pa long. PI Venezuela P. bulbosa cts broadly ovate, тм pedicels estes to 18 mm long; Aaral bracts 5 mm long; sepals 16 mm long. Brazil; Amazonas . . 6.1. P. кы 9. Inflorescence ишн ‘sepals dn 20m long. Colombi ‚Р, heliophila 30a. Scapes short, recurved; inflorescence dense- н ered, secund; sepals 19.2. 30a. Scapes erect; inflorescence lax, not secund; pals acu Ob. Leaves we quibua s beneath; petals ca. 50 mm longa Dn Amazonas of Brazil a ae Ela 30b. Leaves €—Ó altos on both surfaces; peta Humaque, Bad. ue- . P. sastrei 3013920 матом FIGURES 9-12.—9. и diffusa Smith, Steyermark з Robinson.—10. Navia plowmanii Smith, Steyermark & Robinso obinson. — 12. Navia polyglomerata Smith, Steyermark & Spe. 702 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 6.1. Pitcairnia rondonicola L. B. Smith & R.W. 19.2. Pitcairnia nematophora L. B. Smith & Read, J. Brom. Soc. 36(1): 10, fig. 4. 1986. Read, Brittonia 33: 31. 1981. 19. Pitcairnia platypetala Mez in Mart., Fl. Bras. 3(3): 438. 1894, emend. L. B. Smith & Read, Brittonia 33: 33. 1981. 9. BROCCHINIA Schult. f. in Roem. & Schult. Syst. 7(2): Ixx, 1250. 1830. la. Leaves and scape-bracts serrate; ovary only one-third inferior. е Venezuela ........ 1. B. serrata lb. Leaves and bracts always entire; ovary wholly or almost а inferio 2a. Leaves even above, 14-30 cm long. Venezuela, Amazo 3a. Inflorescence glabrous or subglabrous, petals DRE Cerro Sipapo ее ИН ОТА, . B. maguirei 3b. Inflorescence lepidote; petals elliptic or oblong-ellipt 4 ades narrowly triangular; upper scape- dus shorter than the internodes; sepals cucu ullat te; inflorescence tripinnate. Duida 3. B. vestita 4b. Leaf-blades ligulate; scape-bracts all much exceeding the internodes. 5a. Inflorescence tripinnate; sepals and petals obtuse. Cerro Parú, Río TE о - hitchcockii #7 cowanii 5b. Inflorescence bipinnate; sepals and petals acute. Cerro Moriche, Río та 2b. Leaves prominently nerved on dan juri 16-120 cm long. 6a. Blades ligulate, rounded and a 7a. о glabrous; ме goi plant to 8 m high. Eastern Bolívar and adjacent 11. B. micrantha 7b. "dv pees lepidote ranches strict, the pola with the basal flowers covered by the large primary bract a very narr m : Cerro Yapacana 13. B. jou РЕ 8b. Branches divergent to cre much exceeding the primary bracts; inflorescence broad. 9a. шр and secondary axes straight; leaves divergent to spreadin a. Leaf-blades very narrowly c astan eous- margined; sepals pre petals un- guiculate; styles free, thick. S Colombia and adjacent Venezuela B. paniculata 10b. Leaf-blades concolorous; sepals iem petals subequal, elliptic; styles basally nate, slender; plant highly variable. Bolí mazona 15. B. tatei l c about 3 m 1-2 cm long, doni inflorescence rarely more than | Шып. Bolivar го adjacent Guy 17. B. reducta 116. Leaves a separate for most of their length; scape much stouter with large bracts; inflorescence distinctly tripinnate. Southeastern Colombia uel 18. ea женш Venezuela B. о 6b. Blades acute ог acuminate. 12a. Blades aia linear-lanceolate, 3 cm tri ; pri shorter than the sterile bases of the bra io. prae Uriman icon B. bernardii 12b. Blades not constricted at base; inflorescence bipinnate to quadripinnate UE bracts shorter than to slightly longer than the saad deis of the branc 13a. Blades with hard, thickened, pungent 14a. "eddie densely lepidote; pes ole blades subligulate. Neblina, Gra Sabana, Guyana genia na 14b. ice a glabrous or subglabrous; ovary pedicellate; blades narrowly riangular 15a. Leaf-sheaths and apices dark castaneous. Southern Venezuela |... 10. B. melanacra 15b. Leaf-sheaths and apices green. 16a. Ca s inim sharply 3-angled; inflorescence bipinnate, the ultimate branches elongate and laxly flowered. Gran Savanna, Cerro Ya- 9. B. prismatica 165. Capsules terete; inflorescence amply tripinnate, the ultimate аа о ~ densely flowered. Southeastern Colombia and 8. B. acuminata uthern uela 13b. riage of the same EI ue ee ligulate or subligulate. 7a. Flowers secund; plant ca. 1.5 m high. Bolívar: Ptari-tepui ..... 16. B. secunda is Flowers polystichous; plant ca. 0.5 m high. 1986] SMITH—GUAYANA HIGHLAND BROMELIACEAE 703 18a. Blades linear, 4 mm wide; inflorescence bipinnate. Neblina .......... 7. B. delicatula 18b. N — . Brocchinia amazonica L. B. Smith, Jour. Brom. Soc. 34: 106, fig. 4. 1984 o ligulate, 20 mm wide; inflorescence tripinnate. I s: Rio da Serra Aracá Ama- . B. amazonica 11. NAVIA Schult. f. in Roem. & Schult. Syst. 7(2): Ixv, 1195. 1830. TO SUBKEYS OF NAVIA la. Inflorescence laxly racemose or paniculate Subkey I 16. Inflorescence glomerate or moniliform-glomerate. 2a. Ovary superior. 3a. Sepals free. 4a. Sepals acute or acuminate or attenuate, sometimes incurved but never cucullate .. Subkey II 4b. Sepals obtuse or subacute, Wee cucullate Subkey III 3b. Sepals connate posteriorly or equa Subkey IV 2b. Ovary partly or almost wholly SP ql Subkey V Revised section of Subkey I Spikes not strobilate, laxer 7. Sepals 2-5 mm long, alate. 8. Sepals ecarinate; plant little if any over 1 m high. Colombia ooo 6. N. garcia-barrigae 8a. Inflorescence amply 9 0.8-5 m high. Venezuela. 6.1. N. plowmanii 3-4-pinnate, plants 2-5 m high В Spikes digitate in threes, to 26 ст long. Amazonas: Cerro Sipapo erg . М. brocchinioides e Spikes pinnate. Neblin 9a. Eros e axis m leaves to 4 dm long; blades 27 mm wide; inflorescence iffusa glabro 9a. Central peer axis slender; leaves 6 dm long; blades 10 mm wide; inflorescence lepidot 8.1. N. thomasii 6.1. Navia plowmanii L. B. Smith, Steyermark & Robinson, sp. nov. Figures 10, 28j-o. TYPE: Venezuela. Amazonas: Dept. Rio Negro, open rocky meadow on N facing slope with large granite outcrops, forming a ridge meadow at base of a large peak, vicinity of Camp VI, on a ridge on Venezuelan-Brazil- ian border, 3.5 km W of Pico Zuloaga, 2,000 m alt., Lat. 0°53’N, Long. 65°56’W, 13-15 April 1984, W. W. Thomas & T. Plowman 3030 (holotype, US; isotypes, NY, VEN). Planta я solum cognita, caulescens, florigera 8 dm alta. Folia ad 14 cm longa; vaginis inconspicuis; laminis msn triangularibus, 16 mm latis, supra gla- Im subtus per aetatem glabrescentibus, laxe oe "Sca apus 1 cm diametro, per aetatem glaber; scapi bracteis Bos subfoliaceis, internodia superantibus nti mm longa; petalis ignotis; ovario supero; seminibus bicaudatis. Leaf anatom y. Vase bundles not covered lly; adaxial water storage della multi- stratose, highly inflated, smaller out- d innermost layer resiniferous; ed, transv scales ima eccentric, with transversely ob- long cells. This species is placed in Navia because of its imbricate sepals, although its appendaged seeds might suggest inclusion in Lindmania. Persis- tence of the ovule appendages into the mature state has been noted before in at least one other species of Navia, N. scirpiflora. As in that pre- vious case, the presence of occluded substomatal pores confirms the generic placement in Navia. Known only from the type. 8.1. Navia thomasii L. B. Smith, Steyermark & Robinson, sp. nov. Figures 11, 28p—s. TYPE: Venezuela. Amazonas: Dept. Rio Negro, dry 704 forest and scrub vegetation on ESE facing slope above Rio Marawinuma E of “Puerto Chimo" camp, dominant on open slope, ca. 600 m alt., Lat. 0%50'N, Long. 66%07'W, 26 April 1984, W. W. Thomas 3245 (holotype, US; isotypes, NY, VEN). Planta e fragmentis solum cognita, florigera verisi- militer ca. 2 m alta. Folia superiora ad 6 dm longa; vaginis inconspicuis; laminis angustissime triangula- ribus, 10 mm latis, supra glabris, subtus dense albido- lepidotis, laxe serrulatis. Scapus 8 mm diametro, glabrescens; scapi bracteis erectis, foliaceis. Inflorescentia laxissime 4-pinnatim paniculata, lepidota; bracteis primariis verisimiliter late dios et quam ramorum basibus des multo bre- mis ad 25 cm longis. Bracteae florigerae шоо atae, quam sepalis duplo ео floribus bar ч ‚ Зер ala ovata, acuta, libera, 4 mm longa, ea poste petalis 6 mm lon- gis, albis (Thomas): ovario supero. lata; Leafanatomy. Vascular bundles not covered by chlorenchyma adaxially or abaxially, abaxi- ally with fiber sheath reaching the epidermis; water storage cells multi-stratose, sporadically resiniferous, interior cells more tenuous, inner- most less Vis cpi en subepidermal cells adax- ssate, abaxially 1—2-stratose, nearly occluded, transversely elongate, halteri- form; abaxially scales small, peltate, lobed, with indistinct cells. AMENDED SECTIONS OF SUBKEY II 3. Omit lead to 13. Navia lopezii, which proves to be a member of the Xyridaceae. 9. Sepals 12-18 mm long; floral bracts straight; leaf-blades serrulate. 10. Sepals glabrous or glabrescent, acute, in- florescences as in 10a. ae 15 mm in diamet een. Duida 19. N. Ser 10a. Dude ene 40-50 mm in dia am- eter, pink to red. Guanay, Yuta 20. N.p "muris Sepals laxly filamentous- jepidoie. broadly acute, паноу alate- carinate; Dui © da- Huachamacari s 20.1. № A. аа 16. Sepals 5-8 mm lon 16a. Inflorescence compound, digitate-glo- ose from several ellipsoid spikes, many-flowered. 16b. Inflorescence sessile; floral iip broadly ovate, acute, abou ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 equaling the sepals. Amazonas- adjacent Brazil _ 5. N. crispa Inflorescence scapose; flor bracts ovate, long- attenuate, ex- тн sepals. Amazonas: Rio Marawinuma 16b. с о М. polyglomerata 16a. ж aed simple, я са], ssile. Amazonas: Duida-Rio Огіпо- 25.2 . N. huberiana Navia lopezii should be transferred from the Bromeliaceae and placed in the Xyridaceae. The distinctive epidermis and lack of trichomes in the species have previously been noted by Rob- inson (1969). The species and its variety are dis- posed as follows: 13, жаг. ун lopezii (L. В. Smith) Steyer- k & nn. Missouri Bot. Gard. 71: ed fig. 1. 1984. Navia lopezii L. B. Smith, Bot. Mus. Leafl. Harvard 15: 40. 1951; 16: 195, pl. 28. 1954; Fl. Neotrop. no. 14(1): 465, fig. 163L, M. 1974. 13b. баен Sti lopezii var. colombiana (L. B. th) Steyermark & Berry, Ann. Missouri m. Gard. 71(1): 299. 1984. Navia lopezii var. colombiana L. B. Smith, Bot. Mus. Leafl. Harvard 16: 195. 1954; Fl. Neotrop. no. 14(1): 465. 1974. 20.1. Navia aliciae L. B. Smith, Steyermark & Robinson in Steyermark et al., Acta Bot. Venez. 14 (3): 13, figs. 9a-f, 10а-е. 1984. 25.1. Navia polyglomerata L. B. Smith, Stey- ermark & Robinson, sp. nov. Figures 12, 281—7. TYPE: Venezuela. Amazonas: On banks of Río Marawinuma, 2-6 km east of Base Camp, 160 m alt., Lat. 0°50'N, Long. forest and scrub vegetation on ESE facing slope on Río Marawinuma, east of “Puerto Chimo” camp, Lat. 0°50'N, Long. 66°07'W, 26 April 1984, W. W. Thomas 3257 (NY, US, VEN). Dept. Río Negro: on moss-cov- ered bank, 5-10 km east of Cerro de Neblina Base Camp which is on Río Marawinuma, Puerto Chimo Camp, 150 m alt., Lat. 0950'N, Long. 66?07'W, 9 Feb. 1984, К. Liesner & Charles Brewer 15807 (MO, US). Puerto Chimo Camp and up north slope of canyon, 5 km east of La Neblina Base Camp by air, 150-1,800 m, Lat. 0*50'N, Long. 66°07'W, 1986] SMITH—GUAYANA HIGHLAND BROMELIACEAE 705 FIGURES 13-16.—13. Navia huberiana Smith, Steyermark & Robinson.—14. Navia pedemontana Smith, Steyermark & Robinson.—15. Navia crassicaulis Smith, Steyermark & Robinson.— 16. Navia linearis Smith, Steyermark & Robinson. 706 11 Feb. 1984, R. Liesner & Charles Brewer 15875 (MO, US) evar ets dE erecto, robusto (ad 1 m alto). Folia ad 3(-4) d onga; vaginis parvis, inconspicuis; la- minis Re po triangularibus, 7(-15) mm latis, utrinque trichomatibus i 35 ШЕ: zr Же as). Capsula subglob seminibus in loculis uns exappendiculatis. Leaf anatomy. Vascular bundles narrowly covered by chlorenchyma adaxially and abaxi- ally, narrower abaxially; water storage cells adax- undles; adaxial and abaxial subepidermal cells mostly indistinct, not incras- sate, multistratose and incrassate adaxially in the middle of the leaf; substomatal pores not oc- cluded, oval; abaxial surface with rudimentary uniseriate hairs. The species could be close to N. crispa and other non-scapose species on the basis of leaf anatomy. The more strongly scapose N. caules- cens of Colombia also seems related in spite of its multiseriate trichomes. Known only from the Neblina area. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 25.2. Navia huberiana L. B. Smith, Steyermark Robinson, sp. nov. Figures 13, 29e-h. TYPE: Venezuela. Amazonas: Dept. Ataba- po, terrestre sobre los taludes de un cano, frecuente, bancos arenosos de un caño a unos el Rio Orinoco, preparado de ma- terial vivo entregado al Dr. Steyermark en Caracas, са. 125 m alt., Lat. 3°11'N, Long. 65°37'W, 7 March 1980, O. Huber 5058 (holotype, US; isotype, VEN). aulescens; caule erecto, brevi. Folia ad 13cm longa; fra о serrulatis, р i margine incurvatis, siccis еа mediana pallida a nia semiglobosa; bracteis е "V | = A Bra teae florigerae ovatae, attenuatae, ag кол. virides, albo-tomentosae. Sepala libera, ovata, acuta, 7 mm longa, castanea, albo-tomentosa; ovario supero. Leaf anatomy. Vascular bundles broadly covered by chlorenchyma adaxially and abaxi- ally; subepidermal cells adaxially distinct, abax- ially indistinct; substomatal cells with wall not incrassate, with paired simple papillae projecting into pore from each side; abaxially with short, uniseriate bulbous-based hairs. he leaf section and the uniseriate hairs would place the new species in the general relationship of Navia crispa, but N. huberiana differs in the broader abaxial chlorenchyma over the veins and the paired simple papillae projecting into the substomatal pores. Known only from the type. Revised Subkey III 1. Inflorescence or its branches digitate-compound. 2. Inflo отрава slende rly = Bolivar: Jaua. Clu of spikes si 39.3. N. luzuloides of spikes a “ш spaced оп a long axis lu n 2: Inflorescence sess sile or subsessi Inner | са P 39.4. N. scirpiflora Is 10 mm long, much exceeding the floral bracts. Bolivar: 9. . №. jauana 4. Inner leaf-blades like those below. 5. Sepals 13-14 mm lon 6. Floral bracts ovate, attenuate; leaves 10 cm long, the blades 9 mm wide, involu Bolivar N robinsoni 6. oe bracts d rounded ais ees leaves over 30 cm long, the bado 1 1 39.1. e, flat. Amazonas: e ш! 5. Sepals 7:3. 5 mm long. Bolív 7. Floral bracts broadly Е about equaling the sepals; leaves 30 cm long, the blades n no 9 fii agi 7. Floral bracts eee distinctly shorter than the sepals; leaves 3 cm long, the blade 41. N. ned 0 mm wide. 1986] SMITH—GUAYANA HIGHLAND BROMELIACEAE 707 1. Inflorescence simple. 8. Leaf-blades 8-20 mm wide. Keel of the posterior sepals dilated below the apex; inflorescence few-flowered. Amazona 5. N. 7788 9. Keel of the posterior sepals linear or broader at base or middle. 10. Sides of the posterior sepals 1 mm wide, their keels very n Leaf-blades densely cinereous-lepidote beneath, their sides contrasting. Leaves and bracts finely subulate-attenuate; posterior sepals incurved but n cucullate. Amazonas: Marahuaca 6. N. bs 12. Leaves e bracts apically thickened and obtuse; sepals cucullate. Bolívar: Jáua 37. N. incrassata 11. и sparsely lepidote on both sides, becoming glabrous 13. Floral bracts oblong, broadly acute. Bolivar: Pauo, Rio Caura — 38. N. caurensis 13. Floral bracts broad, broadly rounded. 14. oral bracts suborbicular, entire. Bolívar: Guaiquinima ...... 38.1. N. ovoidea 14. Floral bracts oblong, serrulate, cucullate. Amazonas: Duida . _ 38.2. М. albiflora Sides of the posterior sepals 2 mm wide, ud = narrow ог broad. 15. Leaf-blades cretaceous beneath. Amazon 16. Floral bracts serrulate; sepals Е nerved, lepidote: leaf-blades 20 mm Pig; Duida 44 N. latifolia 16. uei bracts entire; sepals even, glabrous; leaf-blades 13 mm wide. Huacha- 45. N. cretacea — o 15. Leatblaes sparsely a minutely lepidote on both sides, becoming glabrous. orescence subglobose. oral bracts pandurate, broadly rounded and apiculate, serrulate, lepidote; es glabrous. Amazonas: Duida 42. N. octopoides 18. Floral bracts acute, the inner ones entire, glabrous; leaf-blades white-bar- bellate in La i of the spines. Guyana 43. N. barbellata 17. Inflorescence narrow, largely covered by the subfoliaceous bracts. puerta. as. 19. em — des narrowly triangular, 18 mm wide; caudex stout, 30 cm tall. 43.1. N. crassicaulis 19. ie blades linear, contracted toward base, 9 mm wide; caudex short. Ma- 43. ОМ. linearis 8. Leaf-blades not over 6 mm wide. 20. Leaf-blades entire or serrulate only near base, 6-20 cm long, most species over 12 cm long. 1. е bracts obtuse. 22. реален covered beneath with appressed cinereous scales; leaf blades 5 mm wide. mazonas: Yutaje 51. N. umbratilis 22: Leaf-blades glabrescent except for the densely white-ciliate margins. 23. s numerous; blades mostly 3 mm, rarely to 5 mm wide; inflorescence етн owered, ovoid. Amazonas: Cerro Vinilla ooo 51.1. N. culcitaria 3. ek relatively few; кк эв uniformly 5 mm wide; inflorescence few- к sciculate. Brazil: a nas 51.2. N. piresii 21. Floral Gace acute. Amazona 24. Leaf-blades 20 cm long, 3 mm wide, the inner ones white at base. Cerro Vinilla ... 51.3. N. berryana 24. Leaf-blades 6-12 cm long, 3-6 mm wide. Blades covered beneath with appressed white scales. Moriche ... 52. М. M 25. ipai laxly vestite beneath and on the margins with filamentous и mes. Neblina 52. SN "hlifera 20. Leaf-blades Mages throughout, 4-10 cm lon 26. Blades covered beneath with white a scales. Bolívar: Jáua ................. 50. N. lasiantha 26. Blades soon glabrous on both sides Floral bracts obtuse or apiculate; E sepals ==. or obtuse Blades with a median white stripe when dry, uniform; floral bracts and sepals mbranaceous, transparent n azonas: Neblina E S 50.1. N. liesneri ic, inn p ee bracts Ану аї арех; posterior sepals cucullate. Bolivar: Guai- 47. N. cucullata 29. Floral bracts uniformly membranaceous; posterior a Ag Am ma- as: C . N. delascionis 27. Floral 5 acute; posterior pe рве 30. oral bracts thickened а x. Bo Z Leaf-blades abrupt b i e. Лам. .. 48. N. intermedia 31. Leaf-blades evenly uem: dorem Marutani 5ууоу.о.05060606060656565656— 48.1. N. geaster ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 oral bracts serrulate near apex, red-brown; inner leaves not pale at Ain leaf-blades 2.5 mm wide. Bolivar: Rio Caura, Jaua ... 46. N. cardonae 32. Lowest floral bracts and all others entire, green; inner leaves pale at base. 33. Floral bracts broadly elliptic, acute and mucronulate, «it аин the sepals; leaf-blades 3-5 mm wide. Amazonas: Neblin abysmophila 49. 33. Floral bracts ovate, acuminate, PT than the sepals; ds blades 49.1. 708 30. Floral = uniformly thin 32. .5 mm wide. Bolivar: Guaiquine 38.1. Navia ovoidea L. B. Smith, Steyermark & Robinson in Steyermark et al., Brittonia 33: 31, fig. 2C, D. 1981. 38.2. Navia albiflora L. B. Smith, Steyermark & Robinson in Steyermark et al., Acta Bo Venez. 14(3): 11, figs. 8, 10f-h. 1984 39.1. Navia pedemontana L. B. Smith, Steyer- mark & Robinson, sp. nov. Figures 14, 29i- l. TYPE: Venezuela. Amazonas: Dept. Ata- bapo, orilla de una quebrada, vecinidad de la cominidad de Culebra, rio Cunucunuma, 200-220 m alt., Lat. 3°40'N, Long. 65?45' W, 22-23, 28-29 March, 1-4 April 1983, Julian A. Steyermark & Francisco Delascio 129063 (holotype, US; isotype, VEN) Caulescens; caule sursum curvato, 10 cm longo. Fo- lia ultra 3 dm longa; vaginis parvis, atris; laminis li- nearibus, 17 mm latis, apice attenuatis, ad basin versus paulo contractis, planis, mox glabris, serrulatis, siccis linea mediana pallida auctis. In flo rescentia uic e spicis s 3 dense digitata; brac- riis subfoliaceis, inflorescentiam multo su- A idei ; onga, poste ibo latere 1 mm latis, obtusis; petalis albis (Stey- ark & Delascio); ovario supero eaf anatomy. Vascular bundles mostly nar- rowly covered with chlorenchyma adaxially; adaxial water storage cells ca. 3-stratose; adaxial and abaxial subepidermal cells indistinct, not in- crassate; substomatal cells with walls not thick- ened, with pairs of simple papillae projecting into pore on each side; scales peltate, slightly lobed, walls of cells indistinct. On the basis of leaf-anatomy the new species falls closer to N. semiserrata than any near which it is placed in the key. None of the other species with complex i have such open sub- stomatal pores with simple pairs of papillae. Na- via navicularis and N. breweri differ further by having hairs rather than scales on the leaves. Known only from the type. N. emergens 39.2. Navia jauana L. B. Smith, Steyermark & Robinson in Steyermark & Brewer-Carias, Bol. Soc. Venez. Cienc. Nat. 22(132-133); 287, fig. 5. 1976. 39.3. Navia luzuloides L. B. Smith, е & В wer-Ca- nez. Cienc. МА. 2132. 133): 289, ба. 39.4. Navia scirpiflora L. B. Smith, Steyermark & Robinson in Steyermark & Brewer-Ca- rias, Bol. Soc. Venez. Cienc. Nat. 22(132- 133): 307, fig. 7. 1976. 43.1. Navia crassicaulis L. B. Smith, Steyer- N Plateau (arm), 13.5 km ENE of Cerro de La Neblina Base Camp, 1,750-1,850 m alt., Lat. *54'N, Long. 66%04'W, 16-18 February 1984, Ronald Liesner 16026 (holotype, US; isotype, MO). pra caule erecto, robusto, 30 cm alto. Folia 20 cm longa; vaginis brevibus, atris; laminis anguste triangularibus, 18 mm latis, utrinque trichomatibus vestitis, laxe : serrulatis. Inflore scentia terminalis, sessilis; bracteis jen foliace eis acteae h 4 4 Sepala libera, cun E lanceolatis obtusis, cucu- latis, 10 mm longis; ovario s Leaf anatomy. Vascular bundles adaxially maximally and abaxially minimally covered by chlorenchyma; water storage cells forming dis- tinct abaxial canals; adaxial subepidermal cells in 3-4 layers, resiniferous, exterior cells in part with thickened walls; abaxial subepidermal cells resiniferous, unistratose; substomatal pores not occluded, oval; abaxial scales minute, peltate. On the basis of leaf anatomy the species falls closest to such Colombian species as Navia acau- lis, N. bicolor, and N. heliophila, which have connate sepals. The species placed close in the 1986] above key all differ by having variously occluded substomatal pores, and most have no abaxial water storage canals. Known only from the type. 43.2. Navia linearis L. B. Smith, Steyermark & a ro Магази, shaded face of Wet sandstone bluff, soi о is edil southern : sector of Meseta Sur- e, 1,560 m alt., Lat. 3°36’00”N, Long. =, 10"W, 13-14 October 1983, Julian A. Steyermark 129646 (holotype, US; isotype, VEN) Caulescens; caule erecto, 10 cm alto. Folia 12 cm longa; vaginis brevibus; laminis linearibus, 9 mm latis, apice attenuatis, basi paulo contractis, utrinque tri- chomatibus subulatis fragilibus basi semiglobosis per- sistentibus laxe vestitis, laxe serrulatis. sessilis, pauciflora; bracteis primariis subfoliaceis, inflorescentiam multo super- antibus. Bracteae ие oblongae, obtusae, mem- branaceae, inte i ri 1 nearibus, obtusis, сае 10 mm longis; petalis albis (Steyermark); ovario supero Leaf anatomy. Vascular bundles narrowly covered by chlorenchyma adaxially and abaxi- ally; subepidermal cells slightly differentiated, not incrassate; substomatal cells with walls not thick- ened, with pairs of commonly lobed or flexuous papillae protruding into pore on each side; scales minute, obsoletely peltate. The combination of geography, floral charac- ters in the key above, and the commonly lobed papillae of the substomatal pores would indicate closest relationship of the new species is to Navia latifolia and N. cretacea. The last two are distin- guished by the cretaceous leaf undersurfaces. Known only from the type. 47.1. Navia delascionis L. B. Smith, Steyermark & Robinson, sp. nov. Figures 21, 31h-n TYPE: Venezuela. Amazonas: Dept. Río Ne- gro, in exposed drier places of savanna, on outcrops of quartzite, vicinity and north- ward from Cerro Vinilla, 440—600 m alt., Lat. 2%31'N, Long. 65?23'W, 1-2 March 1984, Steyermark, Berry & Delascio 130410 (holotype, US; isotype, VEN) Caulescens; caule erecto, 8 cm alto. Folia 10 c nis anguste ыш. ribus, 3 е laxe serrulatis, interioribus ad basin versus albis SMITH—GUAYANA HIGHLAND BROMELIACEAE 709 Inflorescentia terminalis, sessilis, anguste ovoidea, 2 ticae, quam sepalis paulo brev an rubro-brunneae. Sepala libera, posteri iori mu line bus, obtusis, 10 mm longis; ovario super Leaf anatomy. Vascular bundles covered by chlorenchyma adaxially and abaxially; the in- nermost adaxial water storage cells sporadically resiniferous; subepidermal cells mostly uni- stratose, abaxially mostly resiniferous; substo- matal pores nearly occluded, transversely elon- gate, halteriform; abaxial scales not evident. The new species seems rather distinct among those with nearly occluded substomatal pores by the broad covering of chlorenchyma adaxially over the vascular bundles. Of the species placed close in the above key, relationship is perhaps closest to some from western Bolivar such as Navia lasiantha and N. cardonae. Known only from the type. 48.1. Navia geaster L. B. Smith, Steyermark & Robinson in Steyermark et al., Acta Bot. Venez. 14(3): 14, fig. 2j-p. 1984. Navia emergens L. B. Smith, Steyermark & Robinson in Steyermark et al., Brittonia 33: 31, fig. 2A, B. 1981. 49.1. 50.1. Navia liesneri L. B. Smith, Steyermark & Robinson, sp. nov. Figures 21, 3 la—g. TYPE: Venezuela. Amazonas: Dept. Rio Negro, Cerro de Neblina, Puerto Chimo Camp on Rio Mawarinuma and up north slope of can- yon, 150-1,800 m alt., Lat. 00*50'N, Long. 66°07'W, 11 February 1984, Ronald Liesner & Charles Brewer 15866 (holotype, US; iso- type, MO). Caulescens; pde Fish ca. 11 cm alto. Folia 7 cm longa; vaginis ini minis ard pie ribus, 4 mm latis, mediana alba auctis aliter uniformibus, alto-ciliatis, alibi терс, laxe serrulatis. Inflorescentia terminalis, sessilis, subglobosa, 12 mm diametro; е teis involucrantibus foliaceis, inflores- centiam antibus. Bracteae florigerae ellipticae, obtusae, ам subaequantes, membranaceae, trans- pare entes. Sepala libera, posterioribus linearibus, ob- tusis, 9 mm longis; ovario supero. Leaf anatomy. Vascular bundles not or scarcely covered by ehiorenchynia adaxially and abaxially; g t y dis- tinct ты ps абера “cells incras- sate, unistratose; substomatal pores not occlud- \ FLORA OF VFMNETUTLA opg ctm ы РТО Smith, Ste у $4 y type Dase of contrat а "ite rocks slang VR ыйл amazonas meo me- Agr sents peer re puras vacio y ÓN bere o м LI oder -——— UN _ rt Pho ету WA de Mero de UA о өнүттө stares UNITED STATES [EN nR $ NET FiGURES 17-20.— 17. Navia piresii Smith, Steyermark & Robinson.— m o, Camp 111, Мем ERA Ws Plabeau (Arm) 19.5 met бео de La Hebi tne a bape Cap: onthe etta ev. 17 ament: 10 ke 30 cn dee. Flowers white. mosa)d Liesse: 18025 19-18 Feb, 194 rar omen 18. Navia culcitaria Smith, Steyermark & е — 19. Navia berryana Smith, Steyermark & Robinson.— 20. Navia filifera Smith, Steyermark & Robinso 1986] SMITH—GUAYANA HIGHLAND BROMELIACEAE 711 ES de E as DEAS me AA 2949404 MI. um wm B2 A FIGURES 21-24.—21. Navia liesneri Smith, Steyermark & Robinson. — 22. Navia delascionis Smith, Steyermark & Robinson.—23. Navia igneosicola Smith, Steyermark & Robinson.—24. Brewcaria marahuacae Smith, Stey- ermark & Robinson. 712 ed, rounded or oval; abaxial scales small, peltate, lobed, with cells indistinctly oblong. The leaf anatomy, with the vascular bundles exposed both dorsally and ventrally to water storage tissue, and substomatal pores without in- trusions, is a combination commonly seen in Lindmania, but such a combination distinguish- es the new species from all previously known ne es, d independently in various species of Navia and the combination was to be expected in the genus. The non-oc- cluded substomatal pores alone distinguish the species from all others placed with it in the above key on the basis of having the leaf-blades ser- rulate throughout. Known only from the type. 51.1. Navia culcitaria L. B. Smith, Steyermark & Robinson, sp. nov. Figures 18, 301-0. TYPE: Venezuela. Amazonas: Dept. Río Negro, sa- banas abiertas sobre altiplanicie en la Ser- ranía del Vinilla (ca. 20 km al SW de Ma- vaca) hacia el borde SW de la meseta, terrestre; muy frecuente entre las rocas, for- mando extensos cojines, 760 m alt., Lat. 2°20', Long. 65?22', 13 June 1981, O. Huber 6184 (holotype, US; isotype, VEN). Herba pulvinos extensos formans (Huber), caules- cens caule brevissimo. Folia plurima, ad 17 cm longa; laminis sublinearibus, 3 mm raro 5 mm latis, integris, margine albo-ciliatis, alibi gla Infl entia sessilis, ovoidea, multiflora; bracteis sce involucrantibus ipee Bracteae florigerae ellipticae, apiculatae, quam alis breviora. Sepala libera, an- guste oblonga, bue cucullata, 11 mm longa. Leaf anatomy. Vascular bundles broadly 4-stratose; subepidermal cells of both surfaces narrow, scarcely incrassate; substomatal pores nearly occluded, transversely elongate, halteri- form, with walls strongly thickened. In its leaf anatomy the species seems close to Navia cardonae and the above described N. de- lascionis both of which differ by their shorter more serrulate leaves. Known only from the type. 51.2. Navia piresii L. B. Smith, rail - Robinson, sp. nov. Figures 18, 30f. T Brazil. Amazonas: Serra Aracá, 10 о ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 1975, J. M. Pires 15.014 (37) (holotype, US; isotype, IPEAN). Caulescens; caule qui adest brevissimo. Folia ad 15 cm longa; vaginis s brevibus, ovatis, atro-castaneis; la- ula ribus, 5 mm latis, integris, Infl orescentia sessilis, pauciflora; bracteis involucran- ti apicu ati $, tenuibus, Siccis s rubro- brunneis. Bracteae florigerae an- n ellipticae, sepala aequantes, wing naceae. Se- a libera, oblonga, obtusa, 10 mm lon Leaf anatomy. Vascular bundles broadly covered by a sd and abaxi- ally; water p to 5-stratose, abaxially loss inflated, inde demarcated, ca. 3-stratose; subepidermal cells on both sur- faces narrow, scarcely incrassate; substomatal pores broadly oval, with slight paired protrusions on each side at junctures of cells, with walls not incrassate; abaxially only obconical basal parts of scales seen. In size and anatomy the leaf of the species seems close to the Colombian species, Navia bi- color, which belongs to a distinct group having connate sepals. The abaxial water storage tissue is not as prominent, being more like a pale, more enlarged abaxial zone of the chlorenchyma. Per- haps closest relationship is to the above de- scribed N. crassicaulis, which is distinguished by its broader leaves and generally more robust stat- ure. Known only from the type. 51.3. Navia berryana L. B. Smith, Steyermark & Robinson, sp. nov. Figures 19, 30g-k. TYPE: Venezuela. Amazonas: Dept. Río Ne- gro, on rocks along stream, on outcrops of quartzite in savannas and gallery forests near and north of Cerro Vinilla, ca. 30 km SSW from Ocamo, along streams tributary to the Berry & Delascio 130318-A (holo- type, US; isotype, VEN) Caulescens; caule erecto, 4 cm alto. Folia 20 cm Inflorescentia terminalis, breviter scaposa; birds pice apiculatae incrassataeque. Sepala libera, 9 mm ur lateralibus linearibus, apice apiculata incrassataque. Leaf anatomy. Vascular bundles covered by 1986] chlorenchyma adaxially and abaxially; inner- most adaxial water storage cells sporadically nearest leaf surface thickened, resiniferous; sub- stomatal pores nearly occluded, transversely elongate, halteriform; abaxial scales sparse, ru- dimentary In the combination of its leaf anatomy and geography, the new species may be most closely related to Navia delascionis described above. The latter differs by its longer more serrulate leaves. Known only from the type. 52.1. Navia filifera L. B. Smith, Steyermark & Robinson, sp. nov. Figures 20, 30p—u. TYPE: Venezuela. Amazonas: Dept. Río Negro, Camp III, Neblina and Massif, NW Plateau (Arm), 13.5 km ENE of Cerro de La Neblina Base Camp, 1,750-1,850 m alt., Lat. 0°54'N, Long. 66%04'W, 16-18 February 1984, Ron- ald Liesner 16025 (holotype, US; isotype, MO) Caulescens; poni erecto, ca. 9 cm alto. Folia 12 cm longa; vaginis eu. atris; pue anguste trian- gularibus, 6 mm a subtus marginibus tri- chomatibus filiformibus "albis ne Sd. laxe ser- rulatis. Inflorescentia terminalis, sessilis, subglobo Sepala libera, 7 mm longa; poste cullatisque; petalis albis (Liesner); ovario super Leaf anatomy. Vascular bundles adaxially maximally and abaxially minimally covered by chlorenchyma; и storage selle dong ranale- a rss adaxial subepidermal cells 1-3- stra- tose, resiniferous, outermost with cell walls near- est lea tal pores not од. rounded; abaxial hairs uniseriate, long, from bulbous base. The leaf cross-section and the uniseriate hairs suggest closest relationship of the new species to Navia crispa, N. viridis, N. lanigera, and N. my- riantha, all of which fall under Subkey II because of their more pointed, non-cucullate sepals. Known only from the type. REVISED SECTION OF SUBKEY IV 9. Leaf-blades not narrowed toward base; stems simple. SMITH—GUAYANA HIGHLAND BROMELIACEAE 713 9a. Posterior sepals 10 mm long, '2 connate; leaf-blades 1-2 mm wide. Cerro Sipapo 60. N. ocellata 9a. Posterior sepals 15 mm long, over Y con- nate, leaf-blades 4 mm wide. Cerro Si- papo 1. N. lactea 60.1. Robinson in Steyermark et al., Venez. 14(3): 14, fig. 5. 1984. Navia lactea L. B. Smith, Steyermark & Acta B At the time of the original description, the species was noted as having separate sepals, but they were correctly illustrated as connate. REVISED SECTION OF SUBKEY V 1. quen — и арна mm wide; petals yel- or Biss inferior. eaf-blades uniform, slightly diae a Raa base, pa orm- ly spinose-serrate. Bolívar, Guy . arida la. Stem over 17 cm long; leaf-blades di- О cm long, ly triangular, 6 cm long, uniformly nar- rowly triangular. Amazonas co - 68.1. N. igneosicola © 8.1. Navia igneosicola L. B. Smith, Steyer- mark & Robinson, sp. nov. Figures 23, 310- t. TYPE: Venezuela. Amazonas: Dept. Atures, area de selva y lajas igneas á lo largo del Río Coromoto, á Tobogán de la Selva, 35 km sureste de Puerto Ayacucho, 150 m alt., Lat. 5°22'N, Long. 67°33'W, 14 Mayo 1980, Ju- lian A. Steyermark, Gerrit Davidse & Fran- cisco Guanchez 122478 (holotype, US; iso- types, NY, VEN). Caulescens; e curvato, ultra 17 cm longa. Fo- liorum vagina cm lo ongis, 35 mm latis, flexibilibus et umbras a aman- tibus ad basin versus = contractis, omnino spinoso- -ser- rulatis, 6 c o m od sessilis, pauciflora; bracteis exterioribus ellipticis, tenuibus, foliaceo laminatis. Bracteae florigerae yma triangulares, 27 mm longae ae lepidotae. Sepala libera "add sime triangularia, a 45 mm longa, tenuia; petalis aurantiacis deiade dn ovario fere vel omnino infero. Leaf anatomy. Vascular bundles lacking a cover of chlorenchyma on all adaxial surfaces 714 and most abaxial surfaces, largest bundles with fiber sheaths extending nearly or completely to the epidermis adaxially and abaxially; water stor- age cells with only 2-3 layers; subepidermal cells not thickened on жыры surface; substomatal pores t occluded, rounded or oval; abaxial dern , pelt E ih cells distinct, subis al irregularly disposed, peripheral Bes smaller and more tenuous. Navia igneosicola is evidently closely related to N. arida, showing the same type of leaf-sec- tion, substomatal pores, and distinctive scales on the leaves 11.1. BREWCARIA L. B. Smith, Steyermark & Robinson in н et al., Acta Bot. Venez. 14(3): 1 984. Plants with leaves fasciculate-rosulate, rigid, spinose-serrate, in cross-section with vascular bundles not covered with chlorenchyma adaxi- ally, with chlorenchyma continuous and unusu- ally dense abaxially, substomatal as nearly occluded; inflorescence erect, simple, densely spicate; flowers sessile, spreading; E s imbri- cate, ecarinate; petals regular, appendaged with 2 horizontally attached scales; stamens equal; ovary partly inferior, style branches narrowly ob- long; seeds exappendiculate or very narrowly winged The genus is known only from central Ama- zonas in Venezuela. Closest relationship is evi- dently to Navia, from which it differs by its sim- ple, densely spicate inflorescence and its appendaged petals. The appendages differ from those in the more distantly related Steyerbro- melia by being transversely inserted. Plant 2.5-3 m high; leaves more than ] m 5, P y чы idote, 25 mm wide; inflorescence үз cm long B. duidensis Plant not over 0.6 m high; leaves over 0.5 m long; blades densely appressed lepidote be- neath, 15 mm wide; inflorescence 16 cm long 2. B. marahuacae — 1. Brewcaria duidensis L. В. Smith, Steyermark & Robinson in Steyermark et al., Acta Bot Venez. 14(3): 10, figs. 1a-j, 6a-e, photo 2. 1984 2. Brewcaria marahuacae L. B. Smith, Steyer- mark & Robinson, sp. nov. Figures 24, 31u- y. TYPE: Venezuela. Amazonas: Dept. Ata- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 bapo, Cerro Marahuaca-Atuhua-Shiho, cumbre, parte aislada al Sur-Oeste del Cer- ro, vegetacion no arbolada en terreno pen- diente terrestrial, common, 2,480 m alt., Lat. 3°30'N, Long. 65?20'W, 9-10 Feb. вл J. A. Steyermark, M. Guariglia, N. Holm J. Luteyn & Scott Mori 126328 E i US; isotypes, NY, VEN). PARATYPES: Dept. Atabapo: Cerro Marahuaca-Huha, 31 Jan. 1982, Steyermark, Guariglia, Holmgren, Lu- teyn & Mori 125909 (NY, US, VEN). Plant over 0.6 m high. Cerro Marahuaca, 10-12 Oct. 1983, Steyermark 129519 (NY, US, VEN). Stem varnished as well as young in- florescence. Cerro Marahuaca, March 1984, Ostos in Tamayo 6022 (VEN). Petals ellip- tic, cucullate, bearing 2 truncate scales. Acaulis florigera usque 0.5 m alta. Folia multa, s ape usque 0.5 m longis; vagini perder ys fart bus, ca. 2 cm latis, subtus lepidotis; laminis angustis- si me pore ularibus, basi 15 mm latis, planis, rigidis, spini: s sursum curvatis | mm longis laxe serratis, supra vestitis. Scap rectus, 13 mm diametro; scapi bracteis P ее foliaceis, ee шо superioribus ovatis, acum ina tis, 1 ran- tibus sedscapum angustissime alata Leaf anatomy. Scarcely distinct from Brew- caria duidensis, but with less enlarged chloren- chyma cells between the vascular strands, and with raphid-bearing idioblasts between most of the strands. The cited differences in the leaf anatomy are based on a limited sample and should be used with caution. TILLANDSIOIDEAE 14. TILLANDSIA L. Sp. Pl. 286. 1753. REVISED SECTION OF SUBKEY IX 46a. Plant flowering 1-2 m high; primary bracts lanceolate, acute. Colombia, Ecuador, Peru 9. T. pyramidata 46a. Plant flowering ca. 0.6 m high; iid = mary bracts linear-laminate. Venezue 9.1. T. р 9.1. Tillandsia abysmophila L. В. Smith & Stey- ermark, sp. nov. Figures 25, 32a-d. TYPE: 1986] SMITH—GUAYANA HIGHLAND BROMELIACEAE 715 FIGURES 25, 26.—25. Tillandsia abysmophila Smith & Steyermark.—26. Guzmania terrestris Smith & Stey- Venezuela. Amazonas: Dept. Rio Negro, Cerro Aratitiyope, epifita en selva alta, a lo largo de un riachuelo afluente a Rio Mani- pitare, 990-1,670 m alt., Lat. 2210'N, Julian A. Steyermark, Paul Berry & Francisco De- lascio 130135 (holotype, US; isotypes, NY, VEN) qm acaulis, florifera ca. 6 dm alta. Folia ca. 5 dm longa, subtus squamis adpressis cinereis in centro a ene vestita; vaginis ellipticis, 10 cm lon- gis, supra lepidotis; laminis ligulatis, acuminatis, 3 cm latis, supra glabris. Scapus erectus, gracilis; scapi bracteis subfoliaceis, is. Inflorescentia laxe bipinnatim pa- . =. y eo | оа е laevigatae. maa linearia, acuta, 16. GUZMANIA Ruiz & Pavon, Fl. Peruv. Chil. 3: 37, pl. 266. 1802. REVISED SECTION OF KEY 36a. Floral bracts 15 mm long, their margins broad, pale, strongly crisped; sepals 31 long, similar to the floral bracts. Colombia 28. G. radiata 36a. Floral bracts 10 mm long, concolorous, their margins scarcely if at all crisped. Venezue 28.1. G. terrestris 28.1. Guzmania terrestris L. B. Smith & Stey- ark sp. nov. Figures 26, 32e-l. TYPE: Venezuela. Amazonas: Dept. Atabapo, Cer- ro Marahuaca, Cumbre, parte central de la Meseta Sur-Este, al lado de una grieta, a lo largo de la Quebrada Yekuana, afluente del Río Negro, 2,560 m alt., Lat. 3?40'30"N, Long. 65?26'20"O, 10-12 Oct. 1983, Julian A. Steyermark d oo US; iso- types, NY, VEN). P : Cerro huaca, 13-14 Oct., a 129526 pad US, VEN) оша acaulis, florigera ultra 47 cm alta. Folia 5 cm longa, utrinque squamis minimis dissite lepi- 716 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 A.RTANGERIN| FIGURE 27.—a. Lindmania sessilis flower. —b. Sepals.—c. Fruit. —d. Seed.—e. Lindmania saxicola branch of inflorescence. — f. Flower.—g. Sepals. — h. Seed.—i. Lindmania aurea flower.—j. Sepals.—k. Petals, stamens and sepals. —1. Petals. — m. Pistil. —n. Lindmania imitans flower.—o. Sepals. — p. Seed. —q. Lindmania riparia flower.— г. Sepals. —s. Petals and stamens. — t. Pistil.—u. Lindmania piresii flower. — v. Sepals. — у. Petal and stamens.— x. Pistil. 1986] SMITH—GUAYANA HIGHLAND BROMELIACEAE 717 A.RTANGERIN| FIGURE 28.—a. Lindmania huberi flower.—b. Sepals.—c. Petals and stamens.—d. Pistil.—e. Steyerbromelia diffusa flower.—f. Sepals.—g. Appendaged petals and stamens.—h. Pistil and stamens.—i. Bract.—j. Navia plowmanii flower.—k. Bract.—1. Sepals. —m. Ovary.— n. Seed cluster.—o. Seeds.—p. Navia thomasii flower. — а. Sepals. —г. Pistil and stamens.— s. Petal and stamen.—t. Navia polyglomerata flower.—u. Bract.—v. Sepals. — w. Pistil.—x. Petal.—y. Stamen.—z. Seeds. 718 ANNALS OF THE MISSOURI BOTANICAL GARDEN (VoL. 73 5j ЖИЛ VEA q p A.R.TANGERIN/ FIGURE 29.—a. Lindmania terramarae flower and bract.—b. Sepals and petals.—c. Fruit and seeds.—d. Seeds.—e. Navia huberiana flower and bract.—f. Sepals. —g. Fruit.—h. Seeds. i. Navia pedemontana sepals a ruit.—j. Fruit.—k. Fruit showing seeds. —1. Seeds.—m. Navia crassicaulis flower.—n. Bract.—o. Sepals.— Petal and filament. —q. Fruit showing seeds. —r. Seed.— s. Section of ovary (orange part). 1986] SMITH—GUAYANA HIGHLAND BROMELIACEAE 719 А.Ё TANGERIN| FIGURE 30. —a. Navia linearis flower.—b. Sepals.—c. Bract.—d. Petal and stamen.—e. Pistil.—f. Navia piresii sepals and pistil.—g. Navia berryana flower and bract.—h. Bract.—i. Sepals.—j. Petal.—k. Pistil.—l. Navia culcitaria flower and bract.— m. Pistil.—n. Fruit and seed.— o. Seeds. — p. Navia filifera flower and bract.—q. Sepals.— г. Petals.—s. Petal.—t. Stamen.—u. Pistil. FIGURE 31.—a. Navia е flower.—b. Bract.—c. Sepals.—d. Ovary and seeds.—e. Seeds.—f. Petal.—g. Pistil and filaments.—h. Navia delascionis sepals "s bract. —1. Bract. —j. Posterior sepal.—k. Anterior sepal.— l. Petal.—m. Filament.— п. “Pistil, —o. Navia igneosicola flower and bract.—p. Bract.—q. Flower.—r. Sepals. — s. Petal.—t. Pistil and stamens. и. Brewcaria ia flower.—v. Sepal and stamen. — ж. Sepals. — x. Sepal, petal and stamen. — y. Petal and pistil. 1986] SMITH—GUAYANA HIGHLAND BROMELIACEAE 721 FIGURE 32.—a. Tillandsia abysmophila bract and flower.—b. Sepals closed.—c. Anterior sepal. e. Guzmania terrestris flower and bract.—f. Flow sepals. — AR een! —d. Posterior er.—g. Sepals.—h. Sepals opened.—i. Petals, stamens and pistil —j. Bract.—k. Petal and stamen. —1. Pistil and ovules. ey vaginis brevibus iat laminis anguste oblon- m latis, apice attenuat г ар cylindricus, sie epidotus; scapi bracteis erectis, imbricatis, late ovatis, attenuatis. Inflorescentia m longa, i. lepi- infimis ra- s, ca. 10- . Bracteae florigerae late obovatae vel suborbiculares, : cm longae, pad у эме инан cucullataeque, rubra (Steyermark), t Sepala 17 mm cae a, ad 3 mm c connata lobis liberis ec obtusis, forte nervatis; petalis luteis (Stey- mark), laminis ellipticis, 7 mm longis; staminibus rab en antheris 4 mm longis. LITERATURE CITED Mez, C. 1 Bromeliaceae. Jn C. Е. P. Martius, Flora brasiliensis 3(3): (1891) 173-280, pl. 51-61; (1892) 281-424, pl. 62-80; (1894) 425-634, pl. 81-114 ROBINSON, H. . Amonograph on foliar anatomy a Hig hland. Mem. New York Bot. Gard. 14(3): 15- 68. . Downs. 1974. Pitcairnioideae (Bro- meliaceae). Flora Neotropica & 1 е (Bromeli- aceae). Flora Neotropica 14(2): -14 STEYERMARK, J. A MAGUI pope В ES. 1 Nuevos taxa de la ы Venezolana. Acta Bot. Venez. 14(3): 5- ADICIONES A LAS LEGUMINOSAS DE LA FLORA DE NICARAGUA! MARIO Sousa S.? RESUMEN т: riben nueve prets nuevas para la ciencia y se hace una nueva combinación en la familia las Se des лш para la Flora de М canthu olus, L. morenoi, L. p tic a, y C. paucifoliolata). La nuev. s) y las otras a las Pap ER de ellas cinco son Millettieae (Lonchocarpus ilos d género Lonchocarpus (L. acuminatus) de una especie hasta ahora mal EE pero de amplia distribución. ABSTRACT Nine new species are described and one new combination is made in the family Leguminosae for the Flora de Nicaragua. One of the new species belongs to the subfamily Mimosoideae (Pithecellobium Robineae (Coursetia apantensis, C. elliptica, and C. paucifoliolata). The single new combination is made in the genus Lonchocarpus (L. acuminatus) and represents a poorly understood but widespread species. Delos grupos de leguminosas que se nos invitó a participar para el proyecto Flora de Nicaragua, hasta la fecha han resultado novedades, parti- en este género el problema es contar con material completo, es decir, flores, frutos y de preferencia plántulas, lo cual, en general no es posible, ya que su diversidad es grande y entre más carac- teres se cuente, su ententimiento y relaciones son representadas en los herbarios, por lo que, es aquí onde en un futuro deben aparecer nuevos re- gistros. El género Coursetia es muy interesante en Nicaragua, de las tres especies disponibles, las tres son nuevas para la ciencia, dos endémicas y otra compartida con Costa Rica, las tres perte- necen al mismo grupo: Madrensis Rydberg. De la especie nueva de Pithecellobium, en Nicaragua se encuentra su limite sur y en México en el Itsmo de Tehuantepec, es donde mejor se le conoce. MIMOSOIDEAE Pithecellobium campylacanthus L. Rico & M. Sousa, sp. nov. TIPO: 7 km al O-NO de Te- apo don distrito de Tehuantepec, Oaxa- ca, , 17 marzo de 1981, M. Sousa 11 938 ово, MEXU; isotipos, BM, MO). Figure | Frutices vel arbores usque ad 14 m; caulis ramique spinis stipularibus curvis, persistentibus armati; len- ticellis numerosis albidis praediti. Folia 5.5-11 cm lon- ga; petiolus prope basem glandula singulari instructus; rhachis (2.2-)3.5-5 cm longa, villosa vel pilosa, (6—)9— 13 paribus pinnarum praedita, quaque pinna 2.5-4 cm longa, in regie ultimis 1-2 glandulis instructa; Aog in (11-)18-2 Vice per pinnam disposita, 2.5-6(-7) mm longa, (1-)1. mm lata, lineari- redd villosa vel glabrescentia, entier па. RE obscu- riori ia, fasciculos e2 2.3-4 cm longis, villosis; capitula 2.5-3 cm diametro q tuli pedunculis sae; tub minialis in corolla insertus; ovarium ses- sile eres q gumen secus ambas suturas dehis- ens, ] longu 1 ст latum, .4 c cm m .6 1 mm crassum, stipite usque ad 2 cm longo, apice api- ' Por aceptar colaborar con el autor a la Dra. V. E. Rudd y a la Biól. Г. Rico; al M. en C. M. Lavin por su revisión crítica en Coursetia; por facilitar todo lo necesario para llevar a cabo este trabajo, al Dr. W. D. Stevens, 4 о Nacional, Instituto de ‘Biologia, U.N.A.M D.F. Méxi ANN. MISSOURI Bor. GARD. 73: 722-737. 1986. Dra. con el latín; a la Sra. P.M B. G. Schubert por su apoyo académico; al Dr. F. M. Eckel por traducir las diagnosis al zadas partado Postal 70-367, Coyoacan, 04510, México, 1986] zo M Q - К 3 EN SS. ewe 224222 aee V p оа» KS OBS \ \ 0 1 2 3 Аст. FIGURE 1. Cáliz. —d. Corola.— e. Ginec culatum. d late ellipsoidea, 1.3-1.4 cm longa, 0.9-1.0 c a, 3 mm crassa “Uña de gato", “Guamuchil” (Oaxaca, Mé- xico). Arbustos o árboles, hasta de 14 m de alto; tallos y ramas armados con espinas estipulares curvas, persistentes; ramas con numerosas len- ticelas blanquecinas. Hojas 5.5-11 cm de largo; pecíolo 2.5-3 cm de largo, de viloso a piloso con la edad, con una glándula cercana a la base o por SOUSA —LEGUMINOSAS 723 IN R all x UY “ss es, I" о иа —a. Вата con inflorescencias y frutos.—b. Flor completa. — с. —f. Androceo. Tomado de Sousa 11938. debajo de la mitad de éste; raquis (2.2-)3.5-5 cm de largo, de viloso a piloso, con (6—)9-13 pares de pinnas, 2.5-4 cm de largo, con 1 o 2 glándulas entre los ültimos pares de pinnas; folíolos de (1 1-)18-26 pares pon pinna, 2.5—6(—7) mm de lar- go, (1-)1.5-2 mm de ancho, linear oblongos de vilosos a glabrescentes, más densamente en el envés, brillantes y más oscuros en el haz que en el envés, base oblicuamente truncada, ápice agu- do, venación reticulada, marcada en ambas su- perficies. Fascículos de 2 capítulos, pedünculos 724 2.5-4 cm de largo, vilosos; capítulos 2.5-3 cm de diametro en las antesis; bractea floral de 1.5 cm de largo, claviforme, pubescente en el apice. Flores blancas; caliz 5-lobulado, estriguloso en los lóbulos; corola el doble del tamaño del cáliz, 5-lobulada en un cuarto de su largo, estrigulosa en los lóbulos, tubo estaminal inserto en la co- rola; ovario sésil y glabro. Legumbre dehiscente, 11.2-16.5 cm de largo, 1.8-2.6 cm de ancho, 1- 6.4 mm de grueso, plana, recta, los márgenes evidentes cuando inmadura, las valvas coriáceas, de hírtulas a glabras, base aguda y con un estípite hasta de 2 cm de largo, ápice apiculado a veces rostrado, el rostro llega a medir hasta 8 mm de largo; semillas anchamente elipsoides, 1.3-1.4 cm de largo, 0.9-1 cm de ancho y 3 mm de grueso, pardo oscuras, la línea fisural un poco más clara Distribución. En México en la costa del Pa- cífico en los estados de Guerrero y Oaxaca, abun- dante en el área del Istmo de Tehuantepec, en Centro América en Honduras y Nicaragua. En selva baja caducifolia, es frecuente como riparia en arroyos de temporal. Altitudes desde el nivel del mar hasta 200 m en México, y de 120 a 700 m en Centro América. Florece de febrero a mayo e inicia su fructificación en mayo, pudiéndose encontrar frutos en diferentes estados de ma- duración durante todo el año. ICO. GUERRERO: Zihuatanejo, Col. Vicente . S. Blanco O de Jalapa del Marqués, R. Cedillo T 1079 (МЕХ; 14 km al O de Salina Cruz, F. González rau ч: ied (MEXU); Tequisistlán, R. Hernández M. 2b (MEXU); 10 a 15 km al NO de Tehua i = M. King 752 (MEXU); 3 millas al SO de Tehuantepec, E. Lathrop 5944 (MEXU); 3 km al N de la carr. en el camino a la Presa Benito Juárez, A. S. Male 215 (MEXU): Tehuantepec, E. Matuda 28566, 28567 (MEXU); a 21 m al O de Tehuantepec, M. Sousa 5026 (МЕХ); Crucero Sta. Elena, desv. a Cozoaltepec, a 26 km al E-SE de Pto. Escondido, Sousa 6427 (MEXU); a 3 km al E de Marilú, a 17 km al NE de La Reforma, carr. Oaxaca-Tehuantepec, Sousa 6578 (MEXU), a 22 km al SO de Salina Cruz, Sousa 7455 (MEXU); a 11 km al O-NO de Tehuantepec, Sousa 7472 (MEXU); a 1 km a N de Playa Chipehua, Sousa 8663 (MEXU); a 27 km al SO del Morro, al E de San Pedro Huamelula, Sousa 8674 (MEXU); desv. a Playa Chipehua, carr. Salina Cruz-Pochutla, Sousa 9138, 9139 (MEXU); a 8 km al NO de Tehuantepec, т 10093 (MEXU), desv. a Playa Chipehua, М al SO del Morro de Mazatan, Sousa 1013 MEXU): a 13 km al NO de Santiago CO et Sousa 10181 (ME ES а 17 km al e Tehuantepec, Sousa 11944 (M al SO de San Pedro Huilotepec, O. Téllez U); a 3 km al NE de Santiago Laolloga, Téllez 2: 1 (MEXU); ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 a 9 km al NE de Tehuantepec, Téllez 247 (MEXU); Paso Alicia, 8 km al N-NO de Tehuantepec, R. Torres 502 (MEXU); a 7 km al NO de Tehuantepec, Torres 688 (MEXU); Rancho Lizet, 3.5 km al SO de la entrada a Buenos Aires, Torres 4789 (MEXU). Distr. de Ju- chitán: a 4 km al N de Matías Romero, R. M. King 683 (MEXU); a g km al NE de Juchitán, M. Sousa 6596 (MEXU); Ixtepec, Sousa 8720 (MEXU); La Ven- ta, Sousa 9566 (MEXU); a 6 km al N de La Ventosa, Sousa 9608 (MEXU). OND . COMAYAHUA: El Banco, J. V. Rodríguez 2439 (GH); Río Selguapa, Rodríguez 2560 (GH); Valle entre Comayahua y San Antonio de las Flores, Williams & A. Molina 12602 (GH, MEXU). NICARAGUA. BOACO: Mpio. de Teustepe, km 60 carr al Rama, P. P. ties 3748 (MEXU, MO); Hacie nda El Sitio, km 61 carr. al Rama (No. 7), Quebrada San Bartolo, Moreno 10041 (MEXU, MO). Pithecellobium campylacanthus pertenece a la Seccion Ortholobium Benth., la que se caracte- riza por tener estipulas espinescentes, varios pa- res de pinnas y foliolos, frutos planos y semillas sin arilo. Especies afines a él son: P. sonorae y P. leptophyllum los que tienen menor nümero de pinnas, las flores y los frutos son más pequenos y el habitat es matorral xerófilo; de P. elachis- tophyllum difiere principalmente en la forma de los folíolos y el fruto curvo. El nombre de esta especie hace referencia a que sus espinas esti- pulares son curvas. Es una especie que en el área del Istmo de Tehuantepec sus poblaciones son muy variables tanto en la coloración de los foliolos, desde con- coloros hasta discoloros brillantes, como en la pelosidad, desde vilosa a glabrescente; se ha tra- tado de ver si hay una correlación de estos ca- racteres tan variables con respecto a la época del ano, siendo que es una especie que se localiza en selva baja caducifolia donde hay una marcada estacionalidad de lluvias, o de acuerdo a la dis- tribución altitudinal, pero parece ser que esta variación es independiente a dichos factores. PAPILIONOIDEAE MILLETTIEAE Lonchocarpus acuminatus (Schldl. M. Sousa, comb. nov. Robinia acuminata Schlech- tendal, Linnaea 12: 306, 307. 1838. TIPO: Méx- ico. Veracruz: C. J. W. Schiede (holotipo, B, destruido; lectotipo aquí designado, HAL, foto MEXU)). ка purpusii Brandegee, Univ. Calif. Publ. Bot. 6 1919. Hermann, J. Wash. Acad. Sci. 38: 13. 194 8. TIPO: México, Veracruz: C. A. Pur- pus 7849 (holotipo, UC; isotipos, A!, GH!, MO!, US”). 1986] Lonchocarpus se Lundell, Wrightia 1: 152, 153. 1946. ann, J. Wash. Acad. Sci. 38: 310. 0. 1948. TIPO: México. Chiapas: E. Matuda 5008 (holoti- L; isotipos, MEXU!, Lonchocarpus bain dera Lundell, Wrightia 1: 154, 1946. Hermann, J. Wash. Acad. Sci. 38: 311. 1948. TIPO: Nicaragua. Mis agua: S. S. White & C. E: Gilly (holotipo, LL; isotipo, ENCB!). Esta especie ha sido confundida con L. penin- sularis (J. D. Smith) Pittier (ver Hermann, 1965), L. nicoyensis (J. D. Smith) Pittier (ver Hermann, 1965), L. cochleatus Pittier (ver Hermann, woe y L. kerberi Harms (ver Hermann, 1948). Lon- chocarpus acuminatus difiere de L. и у su sinónimo L. nicoyensis, en que en esta ül- tima el fruto es sublefioso. De L. cochleatus, ade- más de tener L. cochleatus un fruto leñoso y más ancho, sus pedünculos y pedicelos florales son más largos, así como sus plántulas llevan eófilos trifoliolados en vez de unifoliolados. En el caso e L. kerberi, es más problemático, ya que, sólo he visto fragmentos (en F), los cuales muestran afinidades con L. caudatus Pittier. Lonchocarpus acuminatus se trata de una especie de amplia distribución, de Veracruz a Panamá, en bajas altitudes y en climas cálidos secos. Lonchocarpus ferrugineus M. Sousa, sp. nov. TIPO: Costa Rica. Heredia: A lo largo de un arroyo, como a 5 km al N de San Miguel, 20 enero de 1971, G. S. Hartshorn 985 (ho- lotipo, MEXU; isotipo, CR). Ilustración: Holdridge y Poveda A. (1975), determinado como Lonchocarpus velutinus Benth. Arbores; cortex interior ubi incisus succum resina- ceum emittens; rami juniores fistulosi, angulati, ferru- gineo-tomentosi; petiolus 8-14 cm longus, profunde canaliculatus; rhachis foliaris tomentosa, ut in petiolo canaliculata; folia 7-9-foliolata, foliolis epunctatis, vulgo ellipticis vel oblongis, (7-)11-17(-22) cm longis, 5- culi florales usque a mm lo ongi. Legumen in ndehiscent e, coriaceum, ellipticum vel lanceolatum, CE neral compressum, 11-20 cm longum dim m latum, dense ferrugineo- -veluti- um vel tom qus. semina 1-3 per fructum; sutura vexillaris ae alata, carinalis carinata. “Zopilote” (Nicaragua); ““Comenegro” (Costa Rica). Arboles de 8-25 m de alto; la corteza interior con fluido resinoso al corte; ramas jóvenes fis- tulosas, anguladas, densamente ferrugineo to- mentosas, lentícelas poco visibles por el tomen- SOUSA — LEGUMINOSAS 725 to. Hojas estipuladas, estípulas pronto deciduas, no vistas; pecíolo 8-14 cm de largo, profunda- mente canaliculado, ferrugineo tomentoso; ra- quis foliar tomentoso y canaliculado como el pe- cíolo, 8-13 cm de largo; hojas 7-9-folioladas, la terminal en ocasiones más grande que las late- rales; folíolos cartáceos, epunteados, elípticos a oblongos, en ocasiones ovados a obovados, (7-)11-17(222) ст de largo, 5-7.5(-11) cm de ancho, base cuneada a redondeada, ápice agudo a corto acuminado, en el haz rugosos y densa- mente ferrugíneo tomentosos sobre las venas media y secundarias, o esparcidamente tomen- tosas, en el envés las nervaduras de primero a tercer orden marcadamente realzadas, tomen- tosos en las nervaduras y velutinos en las áreas intervenosas, nervaduras secundarias 9-14. In- florescencias (sólo fueron vistas en fruto) pedun- culadas, (9-)15-22 cm de largo; bráctea pedun- cular ensiforme, 2-2.5 mm de largo; pedúnculos florales hasta 1.5 mm de largo; pedicelos 2-3 mm de largo. Legumbre indehiscente, coriácea, elíp- tica a lanceolada, largo atenuada en la base, ate- nuada a redondeada en el ápice, plana, lateral- mente compresa, 1-3 semillas por fruto, ligeramente constricta entre las semillas sobre las suturas, 11-20 cm de largo, 2.5-4 cm de ancho, densamente ferrugíneo velutina a tomentosa, su- tura vexilar angostamente alada, sutura carinal angostamente aquillada; semillas maduras no fueron vistas. Distribución. De Nicaragua a Panama, en cli- mas muy húmedos, en altitudes de 450 a 625 m. Fructifica de septiembre a abril. CARAGUA. ZELAYA: Mpio. de Nueva Guinea, Bocas de Piedra, A. Laguna 135 (MO). CosrA RICA. HEREDIA: San Miguel de Sarapiqui, L. R. Holdridge 5129 (F). CARTAGO: Instituto Turrialba, J. León 2264 (OR, F); Puente Cajón, Turrialba, L. J. Poveda 26 pes PANAMA DEL CANAL: Isla de Barro Colorado, Van Tyne 13, v Foster 2901 (F). Sus relaciones no son muy claras por care- cerse de la información de material en flor, s embargo, por sus largas brácteas florales, Su cuya sutura vexilar es alada, además de la ve- nación de los foliolos y tipo de pelosidad, la es- pecie más afin parece ser L. orizabensis Lundell, de la cual difiere por tener menos foliolos por hoja, pecíolo y raquis profundamente canalicu- lados y frutos más grandes. Su nombre específico aduce al color de tallos jóvenes, pecíolos, envés de las hojas, inflorescencias y frutos 726 Lonchocarpus monticolus M. Sousa, sp. nov. TIPO: Nicaragua. Dept. Esteli: Cerro Tisey, a 8 km al SE de e 5 noviembre de 1981, O. Téllez 4842 con P. P. Moreno (holotipo, MEXU; o BM, MO). Frutices vel arbusculae, foliis 5-foliolatis, о pellucido-punctatis, ellipticis, ovatis vel obovatis, (1.5-)2.5-5(-7.5) cm longis, (1-)2-3(-3.5) cm ars ventraliter sparsim sericeis demum glabris. Inflores- centiae laxae, 2-3 cm longae; florescentia о pedunculi florales 4-6 mm longi, pedic ellis 2— did base superficies e denen supra suturas inter semina, 4-10.5 cm longum, 1.8-2.2 cm latum, glabrum, suturis anguste carinatis. Arbustos a arbolitos, hasta 5 m de alto; la cor- teza interior sin fluido resinoso al corte; ramas rcidamente seríceas, pronto glabras, con len- ticelas orbiculares a elípticas, cremas. Hojas es- tipuladas, estípulas pronto deciduas, orbiculares a anchamente ovadas, 0.8-1 mm de largo, 0.9- 1.1 mm de ancho; pecíolo (0.7-)1.5-2.5 cm de largo; raquis foliar seríceo como las ramas, 2.5- 4 cm de largo, hojas 5-folioladas, la terminal más grande que las laterales; folíolos cartáceos, pelúcido punteados, elípticos, ovados a obova- dos, (1.5-)2.5-5(-7.5) cm de largo, (1-)2-3(23.5) cm de ancho, base cuneada a redondeada, ápice obtuso a acuminado, opacos y esparcidamente canescente seríceos a pronto glabros en el haz, en el envés seríceos a glabrescentes, nervadura primaria y secundarias realzadas en el envés, ner- vaduras secundarias 4-6. Inflorescencias corto pedunculadas, laxas, 2-3 cm de largo; floración coetánea; pedúnculos florales 4-6 mm de largo, pedicelos 2-3 mm de largo, bracteolas opuestas, en el ápice del pedicelo, oblongas, 0.7-0.9 mm de largo. Flores 10 mm de largo; cáliz ciatiforme, glandular pelúcido punteado, 3 mm de largo, 6 mm de ancho, esparcidamente canescente serí- ceo, más densamente en el margen, casi trunco; corola azul (Moreno 8728) [probablemente púr- e ancho, casi glabra excepto en la base de la superficie externa; anteras glabras; ovario seríceo, 6-ovulado. Legumbre indehiscente, sub- ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 coriácea, elíptica a lanceolada, plana, atenuada en la base, ápice rostrado, lateralmente compre- sa, 1-3 semillas por fruto, constricta entre las semillas sobre las suturas, 4-10.5 cm de largo, 1.8-2.2 cm de ancho, glabras, suturas angosta- mente aquilladas; semillas 13 mm de largo, 8 mm de ancho, 3.5 mm de grueso, morenas. Distribución. Conocida solamente en el sur del departamento de Estelí en Nicaragua, en al- titudes entre 1,200 y 1,300 m. Habita en bosque de pino-encino. Florece en mayo y fructifica de fines de septiembre a noviembre. ICARAGUA. ESTELÍ: San José de La Laguna, P. P. Moreno 8096 (MO); a 8 km de la Carretera Paname- ricana, camino a San Nicolás, 12?59'N, 85?21'O, Mo- reno 8728 & J. Henrich (MEXU, МО); 1.5 km al М del Valle de San José de La Laguna, camino a San Nicolás, 12°58'N, 86?21'O, Moreno 11353 (МО); fal- das del рУ Тіѕеу, 12°58'N, 85?22'O, Moreno 12737 (MO); San km SE de Esteli, O. Téllez 4800 con a (MEX Esta especie quedaria en la Serie 2. Planinervi, Seccion 1. Punctati de Pittier, cercana a L. acu- minatus, del cual difiere: por el ápice obtuso de sus folíolos; sus inflorescencias más cortas; pe- dünculos florales más largos; fruto sin ala en la sutura peus y por tratarse de una especie más . El nombre especifico fué dado por su pads ola Lonchocarpus morenoi M. Sousa, sp. nov. TIPO: Nicaragua. Dept. Estelí: Cerro Quiniento, Hda. La Grecia, Mpio. San Juan de Limay, 2 septiembre de 1980, P. P. Moreno 2155 (holotipo, MEXU; isotipo MO). Figure 2. Frutices vel arbusculae, foliis 5—7-foliolatis, foliolis v S ellipticis, interdum ovatis vel obovatis, (2- J4- 6.5(- -8. 3) cm longis, (1 .4-)2. 2-3. 5 cm latis, ‚ уеп- 4-1 1c cm longae; florescentia coaetanea; pedunculi fo- rales e minus quam 1 mm longi, pedic ellis 2-3 m longis, bracteolis oppositis, parvis, in \ distali pedi. celli, oblongis vel ellipticis; flos 11-12 mm lo calyx 3 mm longus, 6 mm latus, canescenti-sericeus, paene longostipitatum, apice rostratum, lateraliter compres- S minibus per fructum gaudens, profunde inter semina constrictum, 8-11 cm longum, 2-2.4 cm iis glabrum vel subglabrum, sutura vexillari 8-9.5 crassa, concava, costa media infirme effecta. Arbustos a árboles de 2-5 m de alto; la corteza sin fluido resinoso al corte; ramas esparcida- mente кен о Jóvenes, раа glabras, con , cre- 1986] SOUSA—LEGUMINOSAS 727 GURE 2. Lonchocarpus morenoi.—a. Rama con inflorescencias.—b. Estipula.—c. Botones florales, mos- trandose las bir oper —d. Frutos, uno enseña el márgen vexilar.—e. Cáliz con pedünculo y pedicelo. — f. Estandarte. —g. Ala.—h. Quilla.—i. Tubo estaminal.—j. Gineceo. Las hojas y flores fueron tomadas de Moreno 8356 y los frutos de Maraha 2133. 728 mas. Hojas estipuladas, estipulas pronto de- ciduas, elipticas a anchamente ovadas, 0.6-1 mm de largo, 0.6-1.1 mm de ancho; pecíolo 2.3-3.5 cm de largo; raquis foliar seríceo como las ramas, 2.5-4 cm de largo; hojas 5-7-folioladas, la ter- minal más grande que las laterales; folíolos car- táceos o coriáceos, epunteados, elípticos, algunas veces ovados и obovados, (2-)4-6.5(-8.5) cm de largo, (1.4-)2.2-3.5 cm de ancho, Dads cuneada, ápice obtuso, en ocasiones corto acuminado, opacos y glabros en el haz, en el envés canescente seríceos en las nervaduras y velutinos y seríceos en las áreas internervaduras, nervadura primaria y secundarias realzadas en el envés, nervaduras tercio superior; floración coetánea; pedünculos florales de menos de | mm de largo; pedicelos 2-3 mm de largo; bracteolas opuestas, en el tercio superior del pedicelo, oblongas a elípticas, 1-1.2 mm de largo. Flores 11-12 mm de largo; cáliz ciatiforme, epunteado, 3 mm de largo, 6 mm de ancho, canescente seríceo, casi trunco, sólo pre- sentes los tres dientes carinales, el medio 0.2 mm de largo; corola morada (Moreno 8356), epun- teada, todos los pétalos canescente seríceos; es- tandarte reflexo, su lámina profundamente cón- cava, transversalmente ancho elíptica, 10-11 mm de ancho, moderadamente pelosa en la superficie externa; anteras glabras; ovario canescente se- u profundamente constricta entre las semillas so- bre las suturas y lateralmente, 8-11 cm de largo, 2-2.4 cm de ancho, glabra o casi, sutura vexilar 8-9.5 mm de gruesa, cóncava, con una costilla media poco desarrollada, sutura carinal angos- tamente aquillada; semillas maduras no fueron vistas. Distribución. Unicamente conocida de Ni- caragua, en el departamento de Estelí, en alti- tudes entre 500 y 870 m. Florece en abril y fruc- tifica en septiembre. NICARAGUA. ESTELÍ: Valle Canarias, entre San Juan de Limay y Pueblo epe E an Juan de Limay, P. P. Moreno 2413 (MEXU, MO); Escu t de Agri- cultura, 2 km sobre la — La Laguna de Mi- raflores, “Llano el Duaque”, Moreno 8356 (MEXU, MO). Esta especie es cercana a L. constrictus Pittier, de la que difiere fundamentalmente por su fruto, que es más ancho y grueso y su sutura vexilar ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 cuenta con una costilla media. Está dedicada a Pedro Pablo Moreno Castillo (1953-), oriundo de Esteli, Nicaragua, prolífico colector (26,000 nümeros) de la Flora de Nicaragua. Lonchocarpus pilosus M. Sousa, sp. nov. TIPO: icaragua. Dept. Matagalpa: Santa María de Gana: 20 julio de 1975, H. B. L. Evans s.n. (holotipo, MEXU; isotipo, FHO). Fi- gure 3. Arbores; rami juvenales ferrugineo-pilosi. Folia 9— 11-foliolata, stipulata, stipulis deciduis, erectis, lan- ceolatis, 13-15 mm longis, 2-3 mm latis; petiolus et rhachis pilosa, pilositate simili ramorum; foliola epunctata, , elliptica, oblonga vel obo- vata, 3. 5-8 cm longa, (1-)1. 5-2. 263. 2) cm lata, supra et subtus laxe pilosa, — dies craspedodrom ma pro- minenti; nervi secundarii 1 . Legum is- base t apicem attenuatu , compressum, 1-3-seminalis, 8.5-12(-21) cm longum, atu dense ferrugineo-velutinum, suturis anguste carinatis. Arboles de 4-25 m de alto, probablemente pe- rennifolios; la corteza interior sin fluido resinoso al corte; ramas jóvenes laxamente ferrugineo pi- losas, pronto glabras, con lenticelas orbiculares, cremas. Hojas estipuladas, estipulas deciduas, erectas, lanceoladas, 13-15 mm de largo, 2-3 mm de ancho; pecíolo 3-5.5 cm de largo; raquis foliar piloso como las ramas, 5-8 cm de largo, hojas 9-11-folioladas, la terminal ligeramente más grande que las laterales; folíolos cartáceos, epunteados, 3.5-8 cm de largo, (1—)1.5-2.7(-3.2) cm de ancho, elípticos, oblongos a ligeramente obovados, con los márgenes generalmente re- volutos, base cuneada, en ocasiones redondeada, ápice obtuso a redondeado, emarginado, a mu- cronado, mucrón hasta 3 mm de largo, opacos y laxamente pilosos pronto glabros en el haz, en el envés pilosos particularmente sobre las ner- vaduras, nervación craspe largo, Pedicelos 4—6 mm de largo; Legumbre i in- dehiscente, coriácea a : elip- tica, И en la base y en el apice, plana, lateralmente compresa, 1(—3) semillas por fruto, 8.5-12(-21) cm de largo, 4—5 cm de ancho, den- samente ferrugineo velutina, suturas angosta- mente aquilladas; semillas 16.5 mm de largo, 10 mm de ancho, 4 mm de grueso, moreno-rojizo oscuras. 1986] FIGURE 3. Lonchocarpus pilos sus. —a. Rama con infrutescencia. — b. Detalle nervación.—c. tescencia y detalle de la nervación tomada de H. B. L. Evans s.n. y el fruto de W. D. Stevens 23155. Distribución. Sólo de Nicaragua en Matagal- pa y Chontales, en bosque premontano muy hú- medo, en ocasiones ripario, entre los 350 y 1,200 m de altitud. Fructifica de mayo a diciembre. GUA. MATAGALPA: Matagalpa я 57'N, Nic 85°55'-56' O, M. pissed 3652 (MEXU, MO); Río Yasica, 20 km al E de Matagalpa, D. Neill 1983 SOUSA —LEGUMINOSAS 729 Fruto. La infru- (HNMN); 1.5 km М de Matagalpa, ca. 12%57'N, 85*55'O, Stevens 22527 (MEXU, MO). CHONTALES: orilla de Cuapa, 12?16'N, 85?23'O, Stevens 23155 (MEXU, MO) Lonchocarpus pilosus está emparentado tanto con L. phlebophyllus como con L. eriocarinalis, todas ellas poseen nervaduras laterales que se 730 resuelven en el margen foliolar, pero difiere de ellas por su distintiva pelosidad pilosa, enormes estípulas, y fruto de gran tamaño, además de vivir en hábitats húmedos de montaña y no en selvas caducifolias. El nombre específico se re- fiere a la pelosidad de sus partes vegetativas. Lonchocarpus verrucosus M. Sousa, sp. nov. TIPO: Mexico, Chiapas: Vega del Río Naranjo, en- tre Santa in y El Real, E de Ocosingo, Mpio. Ocosingo, 8 marzo de 1951, F. Mi- nds 7135. (holotipo, MEXU; isotipo, US). Figure 4. Arbores altae, sempervirentes; ramunculi lenticellas numerosas gerentes quae accrescentes coalitaeque de- mum producentes pagina verrucosa. Folia stipulata, stipulis deciduis, erectis, anguste ovatis vel subulatis, longus; ne, О ролет inter dum paru n to; se mitibus reniformibus, ш Cotyledones epigeae; eophylla а unifoliolat "Shi-inte", “Gusano colorado" (Chiapas, Mé- xico). Arboles hasta 20 m de alto, perennifolios; la corteza interior sin fluido resinoso al corte; ramas laxamente canescente velutinas, pronto glabras; las lenticelas de elipti linear oblongas, agran- dandose y algunas veces fusionandose con la edad dando una superficie verrucosa. Hojas estipula- de largo; raquis foliar glabro, 10—13.5 cm de lar- go; hojas 9-11-folioladas; foliolos cartaceos, epunteados, eliptico oblongos, algunas veces ovados, 6-11 cm de largo, 3.5—6.5 cm de ancho, base ligeramente cuneada, ápice de obtuso a re- tuso, ligeramente bicoloros, verdosos y glabros en el haz y blanquecinos y moderadamente ca- nescente sericeos en el envés, nervaduras late- rales 9-12. Inflorescencias pedunculadas, 11-15 cm de largo; floración coetánea; pedünculos flo- rales 2-3 mm de largo; pedicelos 3-4 mm d largo; bracteolas subopuestas a alternas, de la oO ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 mitad al tercio superior del pedicelo, filiformes, 1.1-1.2 mm de largo. Flores 6-7 mm de largo; cáliz ciatiforme, epunteado, 2-2.5 mm largo, 4 mm de ancho, esparcidamente a moderadamen- te canescente seríceo, casi trunco, solo presentes los tres dientes carinales, el medio 0.7 mm de largo; corola roja, epunteada, todos los pétalos canescente sericeos; estandarte erecto, su lámina cóncava, transversalmente ancho elíptica, 7.5 mm de ancho, moderadamente a densamente pelosa en las venaciones de la superficie externa; anteras glabras; ovario seríceo, 6-ovulado. Legumbre in- dehiscente, coriácea, de elíptica a oblonga, ate- nuada en la base, redondeada y rostrada en el ápice, lateralmente compresa, | o 2 semillas por fruto, 8.5-11 cm de largo, 1.8-2.1 cm de ancho, esparcidamente velutina a glabra, sutura vexillar en ocasiones alada, ala 1 mm de ancho, más comünmente ambas suturas angostamente aqui- lladas; semillas 13 mm de largo, 7.5 mm de an- cho, 2.5 mm de grueso, moreno-rojizas. Plántulas con cotiledones epígeos; eófilos opuestos, uni- foliolados. Distribución. De Veracruz y Chiapas en Mé- xico hasta Nicaragua, en selvas altas perennifo- lias con Terminalia, Manilkara y Swietenia, fre- cuente como ripario, en selvas altas y bosques de galeria con Taxodium, en suelos calizos, del nivel del mar a 1,200 m. Florece en marzo a mayo, fructifica a partir de mayo y aün tiene frutos en enero. EXICO. VERACRUZ: a 2 km al N de Uxpanapa, Mpio. J. Chavelas P. 3019, G. Alanis & MICH, US); Emiliano Zapata, nández P. 323 (МЕХО); entre Sand Rita y EIR de Ocosingo, F. oo 7137 (MEXU); entre Monte Libano al Cen E de Ocosingo, Miranda 7179 (MEXU ); orilla del Río Santo Domingo, a + altura del poblado de Santo Domingo, Mpio. Ocosingo, M. Sou- sa 12402, G. Davidse, A. Chater & E. бо (ВМ, DS, MEXU, М GUATEMALA. HUEHUETENANGO: 1-2 km ONO de San Antonio Huista, en el camino a Santa а Huista, Н. H. Iltis, К. Lind & A. Barrios 138 (F, W NICARAGUA. ZELAYA: 3.6 km SE uis San Isidro, Rio Rama, Río Escondido, m G. R. Proctor 27265, G. C. Jones & L. Face Sus relaciones son claras con Lonchocarpus teolas más chicas y pedünculos delgados. El 1986] SOUSA—LEGUMINOSAS 731 о 1 2 Зет £ c d Tom FIGURE 4. e verrucosus.—a. Rama con inflorescencia. —b. Estandartes.— c. Ala.— d. Quilla. — Fruto, mostrando el márgen vexilar ligeramente alado. — f. Plántula. Las hojas y flores "ums tomadas de Mirada 7179, el fruto de Ds 3019 y la plántula de Sousa 10402. 732 nombre especifico se refiere al aspecto de las ra- mas dado por las lenticelas prominentes. ROBINIEAE Coursetia apantensis M. Sousa, sp. nov. TIPO: Nicaragua, Dept. Matagalpa: Cerro Apante, 4 mayo de 1980, P. P. Moreno 175 (holo- tipo, MO). Figure 5. ices vel arbusculae, foliis imparipinnatis, (7—)9— 15- direi stipulae persist ps pen latae vel paene aciculares; rhachis tereticaulis, non alata, canaliculata, velutina, sericea vel subglabra; foliola stipellis persis- tentibus, acicularibus; pulvinuli 1-1.5 mm longi, dense tomentosi vel subglabri, laminae ellipticae, (1-)2.5-5.5 cm longae, (0.5-)1.5-2.5 cm latae, dorsaliter sparse ventraliterque dense vel sparse sericeae, margine re- volubili, nervis lateralibus 6-11, ventraliter inconspi- cuis. Inflorescentiae 4-9 cm longae; pedunculi pedi- cellique eglandulares, pedicellis gracilibus, 0.5-1.8 cm longis; flos 1.4-1.8 cm longus; calyx pilosus, tubo 3-4 mm longo, lobulis acutitriangularibus, 3-6 mm longis; corolla lutea; petala ae vexillum “e us mibus in base laminae dis en nti- -seri- ceum n vel subglabrum. Semina lenticularia, ‘pusilla, cas- tane “Ebano” (Esteli, Nicaragua). Arbustos a arboles pequenos hasta de 6 m de alto; la corteza se defolia en las ramas delgadas; ramas sericeas, pronto glabras, con iiti or- biculares a elipticas, cremas. jas imparipin- nadas estipuladas; estípulas ES л: alez- nadas a casi aciculares, 3-4 mm de largo; peciolo profundamente canaliculado, (0.5-)1.5-2.5 cm de largo; raquis foliar tereticaule, no alado, ca- naliculado como el peciolo, velutino o seríceo hasta glabrescente, (1.5—)3—5.5(—8.5) cm de largo; hojas (7-)9-15-folioladas; foliolos estipelados, estipelas persistentes, aciculares, 1-2 mm de lar- subc ticas, (1-)2.5-5.5 cm de largo, (0.5-)1.5-2.5 cm de ancho, ápice agudo a obtuso, mucronado a emarginado, brillantes a opacas y esparcidamen- te canescente seríceas en el haz y opacas y den- samente a esparcidamente canescente velutinas o seríceas en el envés, vena primaria realzada en el envés, laterales 8-11, inconspicuas. Inflores- cencias racemosas, laxas, 3.5-9 cm de largo; pe- dünculos y pedicelos eglandulares; pedicelos del- gados, 0.5-1.8 cm de largo; bráctea floral triangular aleznada, 3 mm de largo. Flores 1.4—1.8 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 cm de largo; cáliz ligeramente zigomorfo, piloso, su tubo 4 mm de largo, lóbulos triangular agudos, 3-6 mm de largo; corola amarilla; pétalos subi- guales en longitud; estandarte con dos callosi- dades pulviniformes bien marcadas en la base de la lámina; pétalos de la quilla falcados, 8 mm de ancho, ápices acuminados; gineceo seríceo a glabrescente. Legumbre linear, torulosa, 5-16.5 cm de a eia mm de ancho, coriácea, inicialm anescente serícea posteriormente e semillas lenticulares, 5 mm de diá- metro, 2 mm de grueso, color castaño. Distribución. Enel NO de Nicaragua, de Es- telí a Boaco, en selvas sub a ifolias, bosques de galería y sabanas, asi como en áreas pertur- badas; en altitudes entre 400 y 1,000 m. Florece de abril a junio, habiendo un reporte en agosto, fructifica de fines de mayo a julio, persistiendo frutos hasta noviembre. NICARAGUA. ESTELÍ: Isiqui, A. Laguna 396 (MEXU, ; La Gavilana, 13°03'N, 86?19'O, Р. P. Moreno 21881 (MEXU, MO); El Salto, La Estanzuela, 13?02'N, 86°22'0, M. Sousa et al. 12986 (MEXU, MO). JINO- TEGA: à 7 km a la entrada del camino viejo a Jinotega “El Eden”. 12°58'М, 85%58'0, S. Vega & W. Robleto 104 (MEXU, МО). MATAGALPA: Cerro Apante, Р. P. Moreno 132 (MEXU, MO); Dario, El Ojo de Agua, a 2 km de la Carretera Panamericana, 1 2°46'N, 86?00'O, Moreno 9320 (MEXU, MO); Darío, a 9 km de la Ca- rretera Panamericana, sobre la carr. a Terrabona, 12*45'N, 86°01'O, Moreno 9328 (MEXU, MO); Darío, Valle el Jícaro, carr. a Terrabona, a 6 km de la Carretera Panamericana, 12?45'N, 86?03'O, Moreno 9259 (MEXU, MO); 1. 1 km al so. ge ин наи x € Grande de Matagalpa, 12°38'М, 86?01'O, W. D. prion 9820 (MEXU. MO). a lo largo del camino entre Waswali Abajo y Waswali Arriba, 12%55'-56'N, 85?57'O, Stevens 20311 (MEXU, MO). BOACO: San José de Los Remates, “Га Majada”, 12°37'N, 85%51'0, Moreno 16295, Stevens 22279 (MEXU, MO). Esta especie pertenece al grupo Madrensis Rydberg (1224), con ACA араны, е$- das у pétalos de la quilla acuminados, se айтма del las otras de su grupo: por ser plantas eglandulares; hojas con ápices redondeados, márgenes revolutos; pulvinulos cortos y la pelosidad más densa tanto en hojas, raquis, pulvinulos y frutos. Especie cuya variación más extrema se encuentra en el Cerro Apante, del cual su epiteto deriva. Coursetia elliptica M. Sousa & Rudd, sp. nov. TIPO: Costa Rica. Guanacaste: A 16 km al NE de Liberia y a 4 al SO de San Jorge, 27 enero de 1983, M. Sousa 12732, L. D. Go- 1986] SOUSA—LEGUMINOSAS 733 d e f 7 A г FIGURE 5. Coursetia apantensis.—a. Rama vegetativa.—b. Estipelas у pulvinulos. —c. Rama con inflores- cencia.—d. Cáliz.—e. Estandarte.—f. Ala.—g. Quilla.—h. Androceo diadelfo.—i. Gineceo. Tomado de Moreno 175. 734 mez, G. Davidse, C. Humphries, R. Hamp- shire, N. Garwood & M. Gibby (holotipo, MEXU, isotipos, BM, CR, MO). Figure 6. Arbusculae, foliis imparipinnatis, еу Е tentes, subulatae vel paene acicular non alata, d foliola, = laribus; pulvi a- )2.2-5(-6 5) cm longae, atae, ventra liter glabrae, dor- saliter canescenti- -sericeae, nervis lateralibus 8-12, dor saliter inconspicuis. Inflorescentiae 2-6 cm longae; pe- 0.5-1.5 cm longis; flos 1.7-2.3 cm longus; “calyx ca- nescenti-sericeus, tubo 3-4 mm longo, lobulis deltoi- brescenti; petala subaequalia; vexillum duobus callo- sitatibus linguiformibus i in base laminae bene distinctis instructum; petala carinae falcata, 5-10 mm lata, in apicibus acuminata. Legumen lineare, и 4- 16 cm longum, 5-7 mm latum, coriaceum, glabrum. Semina lenticularia, pusilla, castanea. Arboles pequenos, hasta 7 m de alto; la corteza se defolia en las ramas delgadas; ramas espar- cidamente seríceas, pronto glabras, con nume- rosas lenticelas orbiculares, cremas. Hojas im- paripinnadas; estípulas persistentes, aleznadas a casi aciculares, 2-4.5 mm de largo; pecíolo pro- fundamente canaliculado, (0.5—)1.5—2.5(-3.5) cm de largo; raquis foliar tereticaule, no alado, ca- naliculado como el peciolo, esparcidamente se- ríceo a glabrescente, (2—)3.5—6(—8.5) cm de largo; hojas (S5—)7—9-folioladas; foliolos estipelados, es- tipelas persistentes, aciculares, 2-3.5 mm de lar- go; pulvinulos (1-)2-3 mm de largo, casi glabros; láminas membranosas a cartáceas, (1.5-)2.2- 5(-6.5) cm de largo, (0.5-)1.2-2.5(-3.2) cm de ancho, con los márgenes planos, elipticas, apice de agudo a obtuso, mucronado a emarginado, brillantes a opacas y glabras en el haz y opacas y canescente seríceas en el envés, vena primaria realzada en el envés, laterales, 8-12, inconspi- cuas. Inflorescencias racemosas, axilares, laxas, más cortas que las hojas, 2-6 cm de largo; pe- dúnculos y pedicelos eglandulares; pedicelos del- gados, 0.5-1.5 cm de largo; bractea floral trian- gular aleznada, 1.5-3 mm de largo. Flores 1.7— 2.3 cm de largo; cáliz ligeramente zigomorfo, se- ríceo, su tubo 3-4 mm de largo, lóbulos deltoides a triangular agudos, 3-8 mm de largo; corola amarillo pálida a blanquecina, el estandarte ro- jizo en su cara externa, en su cara interna con una guía de nectario verdosa en su base; pétalos subiguales en longitud; estandarte con dos callo- sidades lingüiformes bien marcadas, arriba de la base de la lámina; pétalos de la quilla falcados, ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 5-10 mm de ancho, ápices acuminados; gineceo sericeo. Legumbre linear, torulosa, 4-16 cm de largo, 5-7 mm de ancho, coriacea, glabra; se- millas lenticulares, 3.5-4. m de diámetro, 2 mm de grueso, color castaño. Distribución. En las vertientes del Pacifico de Nicaragua y Costa Rica, en selvas medianas ca- ducifolias a subcaducifolias y encinares tropi- cales; del nivel del mar hasta 580 m. Florece de enero a marzo, habiendo un reporte en junio; fructifica de enero a julio. CARAGUA. MANAGUA: S of Managua, J. M. & M. E Greenman 5736 (MO); 0, km N- NO de la Carretera srt O, Stevens 6176 (MEXU, MO); Managua, Lom de Villa Fontana, 12?06'N, 86?16'O Stevens 21450 (MEXU, MO). GRANADA: Isla Zapatera, Ensenada Los Chiqueros, entre Santa María y ladera O de Laguna de Zapatera, 11?45'N, 85?51'O, C. Sandino 1843 D MO); idem, entre Ensenada Los Chiqueros y Sandino 1887 em MO). RIVAS: Puerto San Jorge, M. & S. Cal- derón 92 (M COSTA Nore GUANACASTE: Questa, Parque Nacional de Santa Rosa, R. Liesner 4552 (CR, MO); Llano Ji- cacal, Parque Nacional de Santa Rosa, M. Sousa et al. 12687 (BM, CR, MEXU, MO); a 3 km al E de Cua- jinicuil y 11 km del Parque Murciélago, Sousa et al. 12719 (BM, CR, MEXU, MO); Playa El Hachal, en la ue de oce Elena, Sousa et al. 12728 (BM, CR, О); 5 km al S de La Cruz, Г. O. & Г.Р. sa 24521 (F, MEXU). Coursetia elliptica también pertenece al grupo Madrensis Rydberg y es semejante a C. apan- tensis, pero tiene hojas con menor nümero de foliolos y márgenes planos; pulvínulos largos; caliz sericeo con pelos adpresos y en general la pelosidad tanto de hojas y frutos es más rala; su floración боште E periodos diferentes, asi como y geogr ráfica El nombre de la especie es por la forma eliptica de sus foliolos. Coursetia paucifoliolata M. Sousa, sp. nov. TIPO: icaragua. Dept. Esteli: En el camino Con- dega a Yalí, 16.9 km NE de la Carretera 1 ya3.5 km SE del Valle Santa Rosa (13?23'N, 86?17'O), 19 noviembre de 1979, W. D. Ste- vens 15813 (holotipo, MEXU: isotipo, MO). Figure 7. Frutices vel arbusculae, foliis imparipinnatis, 3—5- e foliolatis; stipulae persistente glabrescentes; laminae ellipticae vel anguste ellipticae, lanceolatae, apice acutae vel caudatae, (1.5—)3—6(-7.5) 1986] SOUSA—LEGUMINOSAS 735 AE EM NP. FIGURE 6. Coursetia elliptica. —a. Rama florífera. —b. Caliz.—c. Estandarte.—d. Ala.— e. Quilla.—f. Andro- ceo diadelfo.—g. Gineceo.—h. Infrutescencia. Las hojas y flores fueron tomadas de Stevens 9820 y los frutos de Stevens 20311. 736 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 с / 7 h А IGURE 7. Coursetia па —a. Rama florifera.—b. Арісе del raquis, mostrando estipelas у dd nulos. —c. Cáliz bs Estandarte. — e. Base de la lámina del estandarte, mostrando las callosidades. — f. Quilla.—h. An dro 0. —i. Ginec ceo. — j. Dicens Las hojas, inflorescencias y flores de Stevens 15813, la infrutescencia de SU 16055. 1986] cm longae, (0.5-)1.5-2.8(-3.5) cm latae, ventraliter ni- e centiae breviores vel longitudini foliorum aequale nl 7.5 cm lon dicellis gracilibus, 0.5-2.3 cm longis; fos 1.9-2. ра ст longus; calyx sparsim sericeus, tubo ongo, lo- bulis deltoideis, 2.4 mm m longis; corolla сш. vel i i ios vexillum in superficie externa purpu- talis subaequalibus; vexillum duobus callo- sitatibus ШЕ Н) in basi laminae bene distinctis ruc longum, 5-7 m mina lenticularia, pusilla. castanea vel nigricantia. Arbustos o árboles pequeños; la corteza se de- folia en las ramas delgadas, ramas esparcida- mente sericeas, pronto glabras, con lenticelas oupioulares, cremas, Hojas imparipinnadas, sciculides. 3-71 mm de largo; peciolo profunda- mente canaliculado, 1—2.5 cm de largo; raquis foliar tereticaule, no alado, canaliculado como el pecíolo, seríceo a glabrescente, 1-2 cm de largo, hojas 3-5-folioladas; folíolos estipelados, esti- pelas persistentes, puras 2.5-3 mm de largo; pulvinulos 3-4 m e largo, elabrescentes: lá- eas acus sa , lan- ceoladas, (1.5—)3-6(-7.5) cm de largo, (0. 51. 5- 2.8(-3.5) cm de ancho, con los márgenes planos, ápice de agudo a caudado, mucronado, brillantes y glabras en el haz y opacas y glabrescentes en el envés, vena principal y secundarias ligera- mente realzadas en el envés, laterales 4—7. In- florescencias racemosas, laxas, de más cortas a de igual longitud que las hojas, 2-7.5 cm de largo; pedünculos y pedicelos eglandulares; pedicelos delgados, 0.5-2.3 cm de largo, bráctea floral tri- angular aleznada, 2 mm de largo. Flores 1.9-2.2 cm de largo; cáliz ligeramente zigomorfo, espar- cidamente seríceo, su tubo 5 mm de largo, ló- bulos deltoides, 2-4 mm de largo; corola ama- rillo pálida a blanquecina, el estandarte púrpura en su cara externa, en su cara interna con una guía de nectario oscura verdosa en su base; pé- SOUSA — LEGUMINOSAS 737 talos subiguales en longitud; estandarte con dos callosidades lingüiformes, bien marcadas, en la base de la lámina; pétalos de la quilla falcados, 7 mm de ancho, ápices acuminados; gineceo se- ríceo. Fruto linear toruloso, 4.5-6.5 cm de largo, 5-7 mm de ancho, coriáceo, glabro, largo a corto atenuado hacia la base; semillas lenticulares, 4— 5 mm de diámetro, 2 mm de grueso, de castaño a casi negras. Distribución. Sólo del NO de Nicaragua, en dos departamentos, en altitudes entre 540 y 1, m. Florece y fructifica en noviembre y diciembre, habiendo un reporte en mayo y persistiendo los frutos hasta febrero. NICARAGUA. MADRIZ: El Rodeo, 4 km S de Somoto, 13°26'N, 86°34'О, P. P. Moreno 20082 (MEXU, MO); 17321-a (MO). EsrELÍ: El Chayote, 13?16'N, 86200, Moreno 22799 (MEXU, MO). Coursetia paucifoliolata es otra especie del grupo Madrensis Rydberg, se distingue por sus pocos folíolos por hoja, el ápice de acuminado a а caudado de sus folíolos, poca Е en sus ombre especifico hace énfasis en los pocos dete por hoja. LITERATURA CITATA HERMANN, Е. J. 1948. Studies in Lonchocarpus and latadi II: Mi 1] Middle America Lonchocarpi. J. Wash. Acad. Sci. 38: 11-14. 65. 10. Lonchocarpus H.B.K. In Flora of Pa ap (Family 83. oe Ann. Missou- t. Gard. 52: 39-4 Bun. L. R. & L. J. de OVEDA A. 1975. Arboles de Costa Rica 1. Centro Científico Tropical, San José, Costa Rica RYDBERG, P. A. 1924. Coursetia DC. N. Amer. Flora 24: 229-233. THREE NEW SPECIES OF SOLANUM SECTION GEMINATA (G. DON) WALP. (SOLANACEAE) FROM PANAMA AND WESTERN COLOMBIA! SANDRA KNAPP? ABSTRACT Three new species of Solanum are described: S. eut from eastern Panama, S. unifoliatum from the low elevation wet forests of the departmen of Chocó in Colombia, and S. dolosum from higher elevations in the department of Valle de me in Colombia. Relationships of eni species are discussed. Solanum section Geminata (G. Don) Walp. is one of the largest sections of Solanum, with 82 species, most of these Neotropical. They are gen- erally small trees and shrubs, often growing in primary forest understory, an unusual habit in Solanum. Characters useful in recognizing the section are 1) difoliate sympodial units with gem- inate leaf clusters or unifoliate sympodial units, 2) leaf-opposed inflorescences, 3) plants glabrous or with simple uniseriate trichomes, 4) small white or greenish-white flowers, and 5) hard, green fruits at maturity (see Knapp, 1985, 1986 for discussion of these characters). While preparing a monograph of the group, I encountered many new species, both in the field and in the herbar- ium. Three of these are described here so that the names can be used in floristic works. Solanum darienense S. Knapp, sp. nov. TYPE: Panama, Darién, Cana near Río Setigandi, 540-580 m, 18 Apr. 1980, Gentry et al. 28541 (holotype, MO, location of isotypes unknown). Figure 1 Frutex; caules juniores sparse pubescentes, laete ala- apicem dem semina fusca ovoidea reniformia, tes- ta foveolat Shrubs with foetid foliage, 1—1.5 m tall; young stems sparsely hispidulous with erect uniseriate trichomes ca. 0.1 mm long, these often only on one side of the stem; young leaves glabrous; stems winged from the decurrent leaf bases; bark of older stems reddish-golden and shiny. Leaves elliptic to ovate, not geminate except on non- reproductive nodes, widest at or just below the middle, glabrous on both surfaces, occasionally minutely puberulent along the veins beneath, 1 1— 14 cm long, 3-4.5 cm wide, with 7-8 pairs of primary veins, these not prominently raised above, prominent and yellowish beneath, the apex acute to acuminate, the base truncate; petioles winged from the decurrent leaf bases, ca. 1 mm long. Inflorescences opposite the leaves, simple, thread-like, 0.5-1.5 cm long, 4—5-flowered, mi- nutely puberulent with erect uniseriate trichomes like those ofthe stems; pedicel scars evenly spaced 1-2 mm apart, slightly raised; pedicels at anthe- sis filiform, 0.6-1 cm long, tapering from the calyx to a slender base ca. 0.25 mm in diameter, sparsely puberulent with uniseriate trichomes; buds globose when young, hispidulous with uni- seriate trichomes like those of the rest of the inflorescence, the corolla soon exserted from the calyx tube making the buds elliptic to obovoid; calyx tube broadly conical, ca. 0.5 mm long, the lobes deltoid, 0.25-0.5 mm long, the margins paler, the lobes and tube minutely hispidulous with uniseriate trichomes ca. 0.1 mm long; co- rolla white, 5-7 mm in diameter, lobed nearly to the base, the lobes reflexed at anthesis, the tips and margins of the lobes minutely papillose; an- thers ca. 1.5 mm long, 1 mm wide, poricidal at ' I thank James Mallet, W. С. D'Arcy, and Michael D. Whalen for advice and encouragement; Bente Starke King for the illustrations; and the curators of the following herbaria for the loan of specim U research was funded by U.S. National Science Foundation grant BSR facilities, BH, COL, Е, MO, NY, US. This 8302773 to Michael D. Whalen and Sandra Kna ser Educational Foundation Fellowship to Sandra ens or the use of = by an American Association of University Women ? L. H. Bailey Hortorium, 467 Mann Library, Cornell University, Ithaca, New York 14853. ANN. MISSOURI Bor. GARD. 73: 738-744. 1986. 1986] 0 Solanum darienense (from Gentry et al. FIGURE 1. the tips, the pores becoming slit-like upon drying; free portion of the filaments 0.25-0.3 mm d the filament tube ca. 0.1 mm long; ovary brous; style straight, 3-3.5 mm long; stigma a slight broadening at the top ofthe style, minutely papillose. Berries globose, green at maturity, ca. 1.5 cm in diameter; fruiting pedicels deflexed, woody, 1.8-1.9 cm long, expanded at the apex, 0.5-0.75 mm in diameter at the base; seeds dark brown in dry material, ovoid-reniform, 3-3.5 mm long, 2-2.5 mm wide, the surfaces minutely pit- ted. Chromosome number: not known. Distribution. Inthe of eastern Panama, the only Se eae are M HER the vicin- ity of the gold mine at Cana, from 500 to 600 m elevation. Figure 2. E Solanum darienense is related to S. confine Dunal in DC., a species of the Andean foothills in eastern Peru, and to S. pertenue Morton KNAPP— SOLANUM SECT. GEMINATA 739 28541), scale bar equals 1 cm. Standley of montane Costa Rica and western nama. Solanum darienense is distinct from _ Species in its reddish-golden bark, similar, perhaps due to their primary forest hab- itat. D'Arcy (1973) recognized 5. darienense as anew entity from Panama but, due to the paucity of material available at the time, did not describe it. The species may also occur in adjacent Co- lombia or in other parts of the range of low mountains on the Panama-Colombia border. Additional specimens examined. PANAMA. DARIEN. Vicinity of airstrip at Cana gold mine, 480 m, 29 Jul. 1976, Croat 37963 (MO); vicinity of Cana, 1,750 ft, 23 Jun. 1959, Stern et al. 477, 661 (MO, US). Solanum unifoliatum S. Knapp, sp. nov. TYPE: Colombia, Chocó, Municipio de Chocó, 740 © 5 darienense № 5. unifoliatum © S.dolosum as, FIGURE 2. Carretera Quibdó-Tutenendo 15 km de Quibdó, 45 m, 6 Sep. 1976, Forero & Ja- ramillo 2544 (holotype, MO, isotype, COL, not seen). Figure 3. rutex; caules glabri laete alati, cortice viridi-fusco; ка unifoliati; ; folia elliptica vel anguste elliptica utrin- uminato, basi atten posit ices, cicatri dicelli arcte et aequaliter dispositis; pedicelli sub anthesi deflexi fili- formes; rregulari minuta pallide viridis, lobis sub globosa viridis, pedicello frugifero erecto vel deflexo Shrubs or subshrubs, 1-3 m tall; stems gla- brous, lightly winged between the nodes with the decurrent leaf bases; bark light greenish brown, sparsely lenticellate, in age becoming paler and ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 COLOMBIA Distribution of S. darienense, S. unifoliatum, and S. dolosum, scale bar equals 200 km. exfoliating. Sympodial units unifoliate. Leaves elliptic to narrowly elliptic, not geminate, widest at the middle, glabrous on both surfaces, 18.8- 27.5 cm long, 4.8-7.3 cm wide, with 5-6 pairs of primary veins, only the midrib raised above, prominent below, the apex long acuminate, the cm long, -flowere tally; pedicel scars closely and evenly packed, but not overlapping; pedicels at anthesis 5-6 mm long, deflexed, filiform, less than 0.25 mm in diameter at the base, abruptly widening to the calyx tube; buds globose, translucent, minutely 1986] KNAPP—SOLANUM SECT. GEMINATA FIGURE 3. Solanum unifoliatum (from Forero & Jaramillo 2808 & Gentry & Fallen 17585), scale bar equals cm. papillose; calyx tube ca. 0.5 mm long, cup-shaped, the lobes rounded and irregular, ca. 0.5 mm long, minutely papillose; corolla minute, pale green, ca. 5 mm in diameter, lobed three-quarters of the way to the base, the lobes reflexed at anthesis, the tips and margins of the lobes minutely papil- lose: anthers ca. 1.5 mm long, 0.5 mm wide, poricidal at the tips, the pores becoming slit-like upon drying; free portion of the filaments less han 0.1 mm long, the filament tube less than 0.1 mm long; ovary glabrous; style straight, ca. 2.5 mm long, in short styled flowers ca. 0.5 mm long; stigma clavate, dark papillose at the tip. Berries globose, slightly umbonate when im- mature, green at maturity, 0.8-1 cm in diameter; fruiting pedicels erect or somewhat deflexed, 742 woody, 1.8-2 cm long, 0.5-1 mm in diameter at the base; immature seeds flattened-reniform, ca. 2 mm long, 2 mm wide, the surfaces minutely pitted. Chromosome number: not known. Distribution. Found only in the pluvial forest in the department of Chocó in Colombia, near sea level. All known collections are from the up- per Rio Atrato basin. Apparently a plant of both primary and secondary forest. Figure 2. Solanum unifoliatum is very closely related to S. triplinervium Morton, also of the Chocó flo- ristic province but known only from Isla Gor- gona off the department of Narino. The two species share unifoliate nodes, unusual in the sec- tion, crowded inflorescences with closely packed pedicel scars, and large glabrous leaves. Solanum unifoliatum differs from S. triplinervium in its longer inflorescences and leaf venation. Solanum triplinervium, as the name suggests, has three main veins from the leaf base, and the strongly parallel tertiary venation is very like that of a member of the Melastomaceae (Morton, 1944). Solanum unifoliatum has leaf venation like other members of the Solanaceae. Only two flowers are present on the type spec- imen, and one of these is apparently short-styled. Andromonoecy has not been demonstrated in section Geminata, but this is one of the many species with short and long-styled flowers. The large numbers of fruit set on n each inflorescence of S. at the species is not andromonoecious (see Whalen & Costich, 1986), but further work is clearly needed. Additional specimens examined. COLOMBIA. CHOCO, Quibdo, Guayabal, Rio Hugon, ca. 80 m, 12 Sep. 1976, Forero & Jaramillo 2808 (MO); 7 km W of Tutenendo о Yuto, ca. 50 т, 18 Jan 24410 (MO, NY) . 1979, Gentry & Renteria A. Solanum dolosum Morton ex S. Knapp, sp. nov. TYPE: Colombia, Valle de Cauca, Cordillera Occidental, La Cumbre, 1,600-1,800 m, 14- 19 May 1922, Killip 5705 (holotype, US, isotype, NY). The sheet at US bears an an- notation label in Morton's handwriting dat- ed Nov. 1935 stating “Solanum dolosum Morton TYPE”, but a description was never provided for this species. This sheet is there- fore designated as the holotype. Figure 4. ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Frutex; caules juniores pubescentes valde alati; и unifoliati; folia lanceolata vel linearia supra glabra sub alba, lobis longe acuminatis sub anthesi reflexis; Расса et semina ig- notae. Shrubs or climbing shrubs, of unknown height; oung stems and leaves hirsute with unseriate golden trichomes ca. 0.1 mm long; the ultimate branchlets slender; bark of the older stems par- tially glabrate, golden green; stems strongly winged from the decurrent leaf bases, and also from de novo wings arising ca. 1 mm below the inflorescence. Sympodial units unifoliate. Leaves lanceolate to linear, not geminate, widest just below the middle, 5.5-11.2 cm long, 1.4-1.7 cm wide, with 6-10 pairs of primary veins, these indistinct above except for the raised midrib, prominent, yellowish, and puberulent with uni- seriate trichomes 0.05-0.1 mm long, the apex long acuminate, the extreme tip blunt and round- ed, the base cuneate, decurrent on the petiole and stem; the leaf margins erose, ciliate near the apex; petioles winged, 1-5 mm long. Inflores- cences opposite the leaves or occasionally inter- nodal, filiform, simple, 0.5-1.2 cm long, 2- 6-flowered, densely to sparsely hirsute with uniseriate golden trichomes like those of the young leaves and stems; pedicel scars irregularly spaced 0.5-2 mm apart, beginning 2-3 mm from the base of the inflorescence; pedicels at anthesis 4-6 mm long, deflexed, tapering from the calyx tube to a slender base ca. 0.5 mm in diameter, sparsely pubescent with uniseriate trichomes; buds globose, the calyx lobes swollen and knob- like; calyx tube ca. 1 mm long, sparsely pubes- cent, the lobes broadly deltoid, 0.5-1 mm long, sparsely pubescent with the same golden unise- riate trichomes as the rest of the inflorescence, minutely papillose at the tips; corolla white, 8— 9 mm in diameter, lobes three-quarters of the way to the base, the lobes long acuminate, re- flexed at anthesis, minutely papillose on the tips and margins; anthers 1.5-2 mm long, the ter- minal ca. 0.2 mm thickened and paler, 1-1.5 mm wide, poricidal at the tip, the pores becoming slit-like upon drying; free portion ofthe filaments 0.5-1 mm long, the filament tube ca. 1 mm long; ovary glabrous; style straight, ca. 4 mm long; stigma not distinguishable from the rest of the style, minutely papillose on the extreme tip. Ber- 1986] KNAPP— SOLANUM SECT. GEMINATA 743 FIGURE 4. Solanum dolosum (from Killip 5705, 11589), scale bar equals 1 cm. ries globose, glabrous, only immature ones seen. de Cauca in Colombia, from 1,200 to 2,000 m Chromosome number: not known. in elevation. Figure 2. Distribution. On the western slopes of the Solanum dolosum is most closely related to an Cordillera Occidental in the department of Valle undescribed species from high elevation eastern 744 Bolivia. It shares with that species leaves that are ciliate at the apex, filiform inflorescences, and zig-zag stems. The inflorescences of S. dolosum are much shorter, ca. 1 cm as opposed to 3-5 cm, and the leaves are acute at the base. Solanum dolosum is also quite similar to S. longevirgatum Bitter, also found on the western slope of the Cordillera Occidental, but from higher eleva- tions in the department of Cauca. Solanum do- losum is distinct from S. longevirgatum in its lanceolate to linear leaves that are ciliate at the apex, shorter pubescence, and much smaller (8- 9 mm in diameter versus 1-2 cm in diameter) flowers. The two species share unifoliate nodes, fleshy calyx lobes in bud, prominent yellow leaf venation, and erose leaf margins. Additional specimens examin COLOMBIA. VALLE ned. DE CAUCA. Cordillera Occidental, W slope, Rio Digua, right bank between Queremal & La Elsa, 1,160-1,200 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 m, 27, 29 Mar. 1947, Cuatrecasas 23897 (US); La Cumbre, Cordillera Occidental, 1,600-2,100 m, 25-27 Sep. 1922, ‘Killip 11589 (NY, US). LITERATURE CITED D’Arcy, W. G. . Solanaceae. Jn В. E. Woodson & R. W and Central America. Ann. Missouri Bot. Garden : 558-569. 1986. A Revision id Solanum Section Gem- . Ph.D. Thesis. Cornell Some South American species of Solanum. Contr. U.S. Natl. Herb. 29: 41-72. LE D. E. CosricH. 1986. Andromo- noecy in Solanum. In W. G. D'Arcy (editor), So- lanaceae: Systematics and Biology. Columbia Uni- versity Press, New York. A SURVEY OF SOLANUM PRICKLES AND MARSUPIAL HERBIVORY IN AUSTRALIA! D. E. SYMON? ABSTRACT survey of the prickliness of Solanum in Australia is made and the distribution of the most prickly survey appe of the diets of vertebrate herbivores in Australia is presented. It is that the development of prickles is not a response to the dirae environment but to browsing by marsupials belonging primarily to the group known as wallabie Itis commonly asserted (Kerner, 1897; Fritsch & Salisbury, 1953; Davis & Heywood, 1967) that prickles are an adaptation to arid environments or to heavy grazing. If the second assertion is correct, one could expect to find some spacial relationship between the presence of herbivores and the distribution of prickly species. A survey ofa large genus displaying varied prickliness over a substantial area may be a way of showing any relationship and is attempted here. The avail- ability of a recent revision of Solanum in Aus- tralia in which data on the species and their dis- tribution are available (Symon, 1981) enables such a comparison to be made. The genus So/a- num is not unique in its varied prickliness and occurrence over large parts of Australia; other large genera that might also be surveyed are Aca- cia or Hakea but both of these lack comprehen- sive revisions. Familiarity with Solanum in Australia has lead me to doubt any special relationship between prickles and aridity and a glance at Figure 1 does not indicate any special concentration of prick- liness in the most arid areas nor is this supported in Table 2, which shows the distribution of prick- ly species in relation to moisture index. This as- pect will not be followed further here. The survey is presented in three sections. First, an estimate of prickliness was determined for all species of Solanum in Australia and the distri- bution of the most prickly species was mapped. Second, an extensive survey of the reported diets of herbivores was made in an attempt to discover what animals were significant herbivores and ex- actly what they ate. An appalling lack of knowl- edge of diets was revealed but all principal her- bivores are discussed in detail. Third, a brief survey is made of exotic Solanum species on some islands where vertebrate herbivores are ab- ent. Two sections of the genus native to Australia are completely unarmed, sect. Solanum (Black nightshades) and sect. Archaesolanum (Kanga- roo apples). The species are small or large shrubs growing mostly well within the range of herbi- vores. An alternative strategy of plant protection must be operating and in this regard it is notable that the level of alkaloids in these two sections is considerably higher than in the sections with prickly species (Bradley et al., 1978). Ф MATERIALS AND METHODS A measure of prickliness for each native Sola- num species was obtained by counting the num- ber of prickles on 4 cm lengths of stem on her- barium specimens. At least 10 collections of each species were examined (or all that were available in rare species). The lengths of about 50 prickles were measured for each species. Old weathered stems were excluded and where possible a dis- tinction was made between those specimens in juvenile phases of growth and mature specimens. the sample, or details of prickle distribution on the plant. RESULTS A total of 83 species was examined (Table 1). Two species were so inadequately represented that no worthwhile estimate was made; S. ca- ! [ am particularly grateful to P. B. Copley and A. C. Robinson for allowing me access to unpublished data on the diet and distribution of Petrogale xanthopus, to the latter for advice and assistance on the distribution of many species of poss and to Dr. G. D. Sanson for "helping to sort out the browsers and grazers. This f the paper was part of the Second at the Missouri Botanical Garden on 3—6 Augus DIVIVEY 2. ? Botanic Garden, North Terrace, Adelaide, ite Australia, 5000. ANN. MISSOURI Bor. GARD. 73: 745-754. 1986. 746 Distribution of the most prickly species FIGURE 1. of Solanum. taphractum A. M. Cunn. ex Benth. from the north west coast of Western Australia and S. spora- dotrichum F. Muell. from the Queensland rain- forest, and these two will not be discussed fur- ther. Six species were unarmed in the specimens available. No prickles were found on S. dense- vestitum F. Muell., S. nemophilum F. Muell., or S. tetrandum R. Br., and these three are believed to be wholly unarmed. The few specimens of S. dunalianum Gaudich that were available were unarmed but it is known from specimens from New Guinea that this species has the potential to produce prickles and a range of juvenile plants and more collections are needed to get any es- timate of prickliness. This species is restricted in its distribution in Australia to the far north of Cape York. Solanum viridifolium Dunal is a small tree found in the rainforests of Queensland and is unarmed in the many herbarium sheets available. Young plants collected by B. Hyl juvenile phase and the taxon certainly has the potential to produce prickles although it is com- pletely unarmed when mature. Solanum tumu- licola Symon is an herbaceous perennial found about Daly Waters in the Northern Territory. It is unarmed in most collections, but a few plants do have some small weak prickles towards the base and, like S. viridifolium, the species has the capacity to produce prickles. All other species can be described as prickly to a greater or lesser extent. The lower level of prickliness is repre- sented by S. esuriale Lindley, 10 collections from each of Western Australia, Northern Territory, and Queensland had no prickles whereas collec- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 tions from New South Wales and South Australia averaged about one prickle per 4 cm length. At the other extreme are 5. semiarmatum Е. Muell. from eastern Australia with a mean of 70 prickles per 4 cm length of stem and S. ashbyae Symon from Western Australia with 43 prickles. Several species are conspicuously prickly on the main stem in the juvenile phases of growth and have fewer prickles on mature and distal twigs. Few herbarium specimens represent such species adequately and more may show this char- acteristic than have been recorded. Conspicuous examples are: S. inaequilaterum Domin, 37 be- low, six above; 5. macoorai Bailey, 26 below, five above; S. asymmetriphyllum Specht, six be- low, 0.3 above; S. clarkiae Symon, 39 below, 21 above. The first two are found in the rainforest of eastern Australia and the second two at the foot ofthe escarpments in northern Arnhemland. All but SS. clarkiae will form small treelets, and if the prickles do deter herbivores it is an at- tractive guess that the later, lightly armed leaves and twigs are out of reach. In almost all species the stems are more strong- ly armed than the leaves. No count of prickles per unit leaf area has been made, but S. /acu- narium from the Darling-Murray river system and with an inland population near Lyndhurst in South Australia has leaves that are more prick- ly than the stems. Two possible reasons for this difference between stems and leaves may be sug- gested. Leaves may be more expendable than stems because a leaf once lost may be replaced by anew one but the loss ofa stem is more serious damage. Secondly, to be effective prickles are best held rigidly, and flaccid leaves may not be fully effective bases for prickles. In most cases prickles on leaves are concentrated on the strong- er main veins and lower portions of the leaf. If leaves are bitten as a mouthful, relatively few prickles may be effective. Little is known of what grazer to be able to nibble such leaves with rel- ative ease. Although it is true that some degree of prickliness would deter vertebrate herbivores, many species are so lightly armed that the effec- tiveness of their prickles can be seriously ques- tioned. Contemporary accounts of the effectiveness of prickles seem to be very sparse. Janzen (1981), in a review of legume defences against herbi- vores, only mentioned prickles in discussing the 1986] SYMON—SOLANUM PRICKLES TABLE 1. Prickliness of Australian Solanum species. Mean number per 4 cm length of stem and mean length TABLE l. Continued. 747 (mm). Length No. (mm) = ant gilesii 6.3 4.8 i hamulosum 6.7 3.6 adenophorum 6.3 3.9 hesperium 4.3 1.6 ashbyae 43.3 5,2 heteropodium 32.8 4.2 asymmetriphyllum young 6 4.4 hoplopetalum 24.5 5.2 old 0.3 3.3 horridum 40.2 4.9 beaugleholei 24.7 5.0 ystri. 17.1 4.3 ownii 4.0 6.0 inaequilaterum young 37.4 5.9 campanulatum 16.4 4.4 old 6.5 4.8 carduiforme 22.3 4.3 karsensis 1.6 7.8 centrale 2.0 3.0 lachnophyllum 38.2 4.5 chenopodinum 1.5 6.3 lacunarium inland 2.1 2.1 chippendalei 14.5 4.1 main river 3.5 2.3 inereum 10.0 7.6 lasiocarpum (ferox) 15.1 3.0 clarkiae young 39.5 4.8 lasiophyllum (Giles) 24.1 4.9 old 21.0 4.0 W.A. 16.7 3.2 cleistogamum W.A. 18.0 4.9 leopoldensis 16.3 4.3 N.T. 16.0 3.1 lucani W.A. 18.8 3.8 coactiliferum W.A. 3.7 3.8 N.T. 23.7 4.5 N.T. 2.2 4.1 macoorai young 26.1 6.6 N.S.W. 2.4 3.6 | 5.3 4.1 east S.A. 4.3 4.4 melanospermum young 16.0 — west S.A. 2.1 4.0 old 4.0 — cookii 16.6 5.6 multiglochidiatum 27.0 3.75 corifolium 1.4 5.0 nemophilum 0 cunninghamii 8.6 2.8 nummularium 3.7 6.7 chii 0.7 4.6 oedipus 40.5 7.0 densevestitum 0 0 oldfieldii 3.3 3.5 ianthophorum 13.7 4.3 oligacanthum 4.0 5.6 dimorphispinum 3.0 4.6 orbiculatum ssp. orbic 2.0 6.4 ioicum King Leopold 8.5 3.1 ssp. macro 1.9 7.3 ibb & Hann R 4.0 3.8 papaverifolium 14.1 3.4 Tanami Desert 18.4 2.3 parvifolium 4.7 4.7 lor 4.8 5.0 petraeum 23.4 3.2 diversiflorum 4.4 6.0 petrophilum (Flinders) 12.5 6.4 dunalianum 0 0 phlomoides 19.1 4.3 eardleyae 33.9 4.9 plicatile 3.4 2.8 eburneu 15.9 4.1 prinophyllum 9.1 7.8 echinatum 9.5 3.5 puguinculiferum 2.7 9.2 elachophyllum 4.0 10.5 pungetium 3.5 4.8 elegans 2.6 5.9 quadriloculatum 18.0 3.9 ellipticum typical 11.8 4.4 seitheae 21.5 3.7 NT prickly 29.3 6.8 semiarmatum 70.1 3.1 SW Eyre Pen. 7.8 4.6 sporadotrichum (1) 26.0 4.4 eremophilum 8.3 4.2 stelligerum 1.5 5.3 esuriale W.A. 0 0 sturtianum W.A. 1.3 3.3 N.T 0 0 S.A 0.8 3.3 Q. trace terraneum 3.3 2.4 N.S.W. 1.1 2.4 tetrandrum 0 S.A. 0.7 2.5 tetrathecum 1.2 7.6 ferocissimum N.T. 4.1 4.5 ununggae 1.2 2.6 , 4.0 6.1 tumulicola 0 S.W. 5.2 5.7 vansittarensis 0.7 5.8 furfuraceum 1.4 7.6 viridifolium 0 gabrielae 3.7 4.0 yirrkalensis 2.8 3.5 [Vor. 73 ANNALS OF THE MISSOURI BOTANICAL GARDEN 748 * ж * * * * * * * * * x * * * * * * * * * * * * í * . * ж ж ж ж ж c0» * * * ж ж ж ж Fr0-cO0 * * * * * * * * * * * 90t0 * * * 8'0-9'0 * = * * * * * * * * * * * * * * * * * * * * * * * * * * * 8« v9 — 9t tt Ct Ot 8E Ot pE CTE OE 8C OC vc cc Oc 81 91 ФГ CI OI 8-L 95 pb 371 0 UOSBIG IWUN sappoug JO J9QUINN id 31n1SIOJA ‘((udy—AON) иоѕеәѕ 1ouruins XIPUI aunjsroui әц 01 uorje[a1 ur sarods UNUDJOS jo uonnquisq. c 318v] 1986] ant-thorn Acacia. There is no mention of con- ventional prickles, thorns, pungent leaves, or vertebrate herbivores. Main (1981), in discussing plant responses to herbivory, mentioned nutri- tional value and toxicity but not tomentum or prickles. Tomlinson (1962), in discussing the prickles of palms, did consider that they are a response to herbivory and gave some interesting information on their origin and distribution. A SURVEY OF THE PUBLISHED DATA ON THE DIETS OF HERBIVORES IN AUSTRALIA INSECTS No adequate survey of insect herbivores on Solanum is available. Many leaves bear evidence of insect activity. I have observed that flea beetles (Fam. Chrysomelideae), looper caterpillars (Fam. Geometrideae), grasshoppers (Fam. Acridideae), and leafminers (Fam. Tineideae) are all relatively common on species of Solanum. No doubt there are many more. The smaller insects, e.g., flea beetles, can move readily between most prickles, and severe leaf damage on some very prickly species (by human standards), e.g., S. prino- phyllum Dunal, is evidence of this. In some cases dense prickles could conceivably keep large in- sects off the plant surface, but this must be quite rare and prickly species like S. oedipus Symon with well armed leaves and stems are extensively attacked by grasshoppers (pers. obs.). In looking through herbarium specimens I have concluded that apparent insect damage is more closely re- lated to the length and density of the tomentum than to prickliness. REPTILES Although a few Australian reptiles (the larger Skinks) are herbivorous, their diet seems re- stricted to flowers and fruits. I have found none that regularly eat leaves of Solanum and they can be disregarded as significant herbivores of these species in Australia. BIRDS The two large flightless birds Emu and Cas- sowary are both herbivorous. The Cassowary is known to eat fruit (Stocker & Irvine, 1983), es- pecially fallen fruit, and is not known to graze to any extent. Its distribution is strictly limited to rainforest areas of northern Queensland and it can scarcely be considered a significant factor in the more extensive browsing of plants. SYMON- SOLANUM PRICKLES 749 By contrast, the Emu was extremely wide- spread and is omnivorous. Fruits, flowers, a wide range of vegetable matter, and insects make up its diet. It certainly eats Solanum fruits, even those with prickly calyces (Symon, 1979). Davies (1978) gave details of Emu diets in Western Aus- tralia. Within a very wide range of tolerance they take advantage of locally abundant fruit, flowers, herbage, and insects. Shoots and herbage formed an important part from January to October and grasses and Compositae shoots were ranked among the most abundant. The leaves of S. ash- byae were eaten in April and August. Davies did not list Solanum fruits in his tables or Appen- dices despite their importance in eastern Aus- tralia as cited by Noble (1975). Davies pointed out that Emus prefer succulent food items to dry ones, and such a diet does not include dried herb- age or grasses or the mature leaves of shrubs. Three prickly plants provided significant amounts of food, i.e., Acacia tetragonophylla, A. victoriae, and Scaevola spinescens. In all cases the Emu ate the fruits and may be a dispersal agent rather than predator. No native mammalian herbivores occurred in New Zealand and the only large vertebrates were the now extinct Moa. Greenwood and Atkinson (1977) suggested that the long-term effect of Moa browsing in New Zealand was the development of a large number of shrubs with an intricate divaricate growth habit, but scarcely any prickly shrubs. This thesis is challenged by McGlone and Webb (1981). The action of an Emu would be of plucking, pulling and breaking; the chewing or nibbling of an ungulate is scarcely possible. I have found no comment on the possible effect of Emus on the morphology of the Australian flora. A few intricate divaricate shrubs occur in Australia, but nowhere in the proportion that they occur in New Zealand. Their number could include S. nummularium S. Moore from West- ern Australia, of which the Emu would eat the unlikely that the Emu is of sign browser of fresh foliage, although their role in seed dispersal is undoubted. MAMMALS In Australia mammalian herbivores would have been marsupials and to a lesser extent na- tive rodents. No ungulates were originally native to Australia. The marsupials occur in great array ranging from kangaroos 2 m tall to the smallest wallabies, bandicoots, and possums, some no 750 larger than a rat, and these must be considered the major herbivores of the Australian flora. Some fossil species were larger than any living species and were considered to be browsers (Sanson, 1978). Because some of these became extinct in relatively recent times, it is likely their influence may still be reflected in the present vegetation. Family Thylacomyidae, bandicoot. Watts (1969) in a paper on the distribution and habits of the rabbit-eared bandicoot (Macrotis lagotis) showed that plant material composed most of the diet and listed Solanum seed and Solanum roots as items of diet. Solanum seed remains in the feces formed 23% of the whole in one sample and in many cases were intact. Although once more widely spread, this bandicoot is now ex- tremely restricted in number and area (an en- dangered species) and seems unlikely to have been an effective browser on Solanum. Family Phalangeridae, possums. Van Dyck (1979) reported the destruction of wild tobacco tree (S. mauritianum) by mountain possums (Trichosurus caninus) in south east Queensland. This alien Solanum is a small tree, quite un- armed, and the animals ate the leaves and in particular the bark, often stripping the stems to ground level and causing their death. Van Dyck also gave other unpublished reports of the cop- регу brushtail (T. vulpecula) eating this species. Proctor-Gray (pers. comm.) reported that this species also eats the leaves and unripe fruits of S. seaforthianum, an unarmed, alien, climbing species. The native t species of Solanum are rarely robust enough to sustain the weight of the possums, but their densely prickly main stems could reduce predation and protect the bark. Pos- sums do not now occur over large areas of Aus- tralia, they are largely arboreal, and few Solanum in Australia are robust enough to sustain them. Freeland and Winter (1975) in a study of the brushtail possum (7. volpecula) reported that this species spent up to 2396 of its feeding time on the ground and that the animals “probably” ate the leaves of S. nigrum (?S. americanum), an unarmed species. Possums eat unarmed intro- duced Solanum but have not been reported eat- ing prickly native species. Family Petauridae, possums and gliders. data located. Family Macropodideae, kangaroos and wal- labies. Sanson (1978), on the basis of masti- cation characteristics, divided the Macropodidae into a browsing grade and a grazing grade. The first was considered ancestral and included those No ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 genera whose diet was substantially of soft, un- abrasive, low fibre plant, whereas the latter was a derived grade specialised to eat the more ab- rasive, siliceous, high fibre grasses. The browsers , Dendrolagus spp. (tree kangaroo), La- gostrophus (hare wallaby), the New Guinea genera Dorcopsis, Dorcopsulus, and the extinct genera Protemnodon, Hadronomas, Dorcopsoides, and Sthenurus. An intermediate grade includes trogale (rock wallaby) and the fossil Troposodon. The grazing grade included Onychogalea (nail- tail wallaby), Lagorchestes (hare wallaby), many, but not all of Macropus/Megaleia (kangaroo and euro), and the fossil Peradorcas, Fissuridon, and Procoptodon. In his (1978) paper, Sanson did not give specific names and the species included in his genus Prionotemnus are not stated. For this study and from Sanson (1980), I have considered Macropus bernardus, M. fulginosus, M. gigan- teus, M. parma, M. parryi, M. rufogriseus, and M. rufus to be grazers and Macropus agilis, M. dorsalis, M. eugenii and M. robustus to be in- termediate; M. irma : 8и known апа М. greyi is now extin Sanson believed Ке m browsers did not con- sume large quantities of grasses, that extant browsers tend to show relict patterns of distri- bution, and that the oldest fossil members of the Macropodideae were all browsers. In contrast, the advanced grazers are distributed in recently developed arid and savannah grassland regions. Increasing aridity, the extinction of denser wet forests in the central regions of Australia, and an increase in grassland and shrub steppe at the end of the Miocene provided the impetus for change. Bolton and Latz (1978) make no mention of Solanum in the brief list of plants eaten by the western hare-wallaby (Lagorchestes hirsutus). The diet given, though admittedly inadequate, sug- gests a grazer rather than a browser. Christensen (1980), in a study of the woylie (Bettongia penicillata), showed that fungi com- posed a significant proportion of its diet through- out the year. Some starchy and cellular material was present but no mention was made of leafy herbage. No mention was made of Solanum in the diet or in the environment (likely to be rare in that area) and the woylie does not appear to be a significant herbivore of leafy material. Stocker (1982) reported that the red-legged pademelon (Thylogale stigmatica) is active in stripping and eating the bark from the lower 50 1986] cm of the alien Solanum mauritianum to the extent that the tops may be killed, though the base may coppice. This species is quite unarmed. George (1982) stated that “the tree kangaroo (Dendrolagus) spends much of its time on the ground and presumably feeds predominantly on ground herbage." No precise details of its diet appear to be available. Storr (1964) in a study of the quokka (Setonix brachyurus) on Rottnest Island, Western Aus- tralia recorded Solanum simile as a minor item of diet in January to March. The maximum per- centage ofthe diet recorded comprising Solanum was 1296. This Solanum 1s quite unarmed. From the list of plants eaten it appears that the quokka is a browser rather than a grazer. Elsewhere Storr (1962) stated “the leaves, young stems and bark off older stems are only eaten in areas where little or no other herbage is available. Ripe fruits are eaten in summer and autumn. Ordinarily the plant is not important as food, it is more valuable as shelter when growing in dense thickets in burnt- out country." No prickly species of Solanum is listed for the island. Solanum simile belongs to the section Archaesolanum, all species of which are unarmed. The section has significantly higher levels of solasodine alkaloids than do the stellate- haired prickly species. Christensen (1980) reported on the stomach contents of the tammar (Macropus eugenii). In a thesis by Kelsall (1965) (not seen) it was stated that tammars browse predominantly on scrub species, particularly Acacia but are also attracted to grassy areas. An analysis by Christensen of the stomach contents of four animals disclosed 95, 90, 50% grassy material in three and 95% di- cotyledonous material in the fourth. In contrast to Bettongia, very little fungal matter was pres- t. Maynes (1977) studied aspects of the biology of the parma wallaby, Macropus parma. This species is native to wet sclerophyll with a thick shrubby understory in association with grassy areas. Although observed to graze in grassy patches, no precise details of their diet were giv- en Eight studies of the diet of the large red and grey kangaroos (Macropus rufus and M. robus- tus) were located (Chippendale, 1962; Kirkpat- rick, 1965; Griffith & Barker, 1966; Chippendale, 1968; Storr, 1968; Bailey et al., 1971; Griffiths et al., 1974; Ellis et al., 1977). These studies confirm that red kangaroos are essentially grass feeders and that browse is a mi- SYMON — SOLANUM PRICKLES 751 nor component of their diet and neither can be considered as a major herbivore of Solanum al- though species of the latter occur wherever the larger kangaroos occur. The most prickly species of Solanum are not a feature of = plains where the larger grazing kangaroos occu Ealey and Main (1967) па aay ei the ecology ofthe euro (Macropus robustus) in north-western Australia recorded a fruit of Solanum lasio- phyllum in the mouth of one euro, but the genus was not otherwise recorded in their diet. How- ever, Dawson and Ellis (1979) when comparing the diet of the euro and Petrogale xanthopus showed that though the euro showed a higher percentage of grass at all sampling times the com- ponent of plants with stellate trichomes was equally high in July 1974 and February 1978 and low in September 1976. It must therefore be con- sidered as a herbivore of Solanum. Dawson and Ellis (1979) compared the diet of Petrogale xanthopus (yellow-footed rock-walla- by) and sympatric herbivores in western New South Wales. The wallaby is now restricted in distribution in New South Wales and South Aus- tralia and occurs in rocky ranges. The plants with a tomentum of мее hairs were classed to- gether and included S (S. petrophilum), Amaranthaceae, and Malvaceae. After three sampling periods the percentage of particles (plants with stellate trichomes) in the feces of the four animal species was wallaby 6-15%, euro 1-13%, goat 5-23%, rabbit 4-7%. The percent- age of samples with stellate hairs was greatest in February 1978, “а notable component.’ Copley and Robinson (1982) in their study of the diet of Petrogale xanthopus Gray (yellow- footed rock-wallaby) discussed six major dietary components of which one was plants with stellate trichomes (members of the families Solanaceae, Malvaceae, and Amaranthaceae), which could be recognised in the feces because of their char- acteristic stellate hairs. While this category con- stituted only about 596 of the diet in winter, it increased to 18-2496 in summer (when species of these three plant families would be in active growth), and when fruits of Solanum were eaten as well evidenced by the seeds of Solanum pe- trophilum that occur in their feces. All the di- etary components occurred in their highest pro- portion in close association with the rock outcrop. Although Copley and Robinson listed 5. petro- philum and S. sturtianum as the species in- volved, both were common at the site. Solanum petrophilum was eaten throughout the year, in- 752 FIGURE 2. Distribution of the known browsing marsupials. cluding flowers and fruit in spring and summer, and S. sturtianum was eaten in the summer. Hornsby (pers. comm.) has shown me a vid- eotape of a female yellow-footed rock wallaby in the Flinders Ranges eating the stems of Solanum ellipticum. The conditions were very dry at the time and the plants desiccated. The hands were used to break up the plant, after which the animal held pieces ofthe stem in its mouth and appeared to brush them vigorously with its hands to re- move or break the prickles off before eating the remainder. The dried leaves did not appear to e eaten. Hornsby also reported then, that at other times of the year the animals selected and ate the flowers of Solanum. The bushes were carefully searched and only the flowers were eat- en. Eremophila flowers were also selectively eat- en. In experiments in October to November 1981, the animals were offered wilted and fresh ma- terial of 5. ellipticum, S. petrophilum, and 5. sturtianum. The flowers and fruits of S. ellipti- cum were eaten. Mature material of S. sturtia- num was ignored but flowers were eaten. A joey showed distaste on sampling a leaf. The same joey later browsed some of the 5. ellipticum. It is apparent from these studies that both the yel- low-footed rock wallaby and the euro consume Solanum spp. and that particularly in summer Solanum may compose a measurable portion of their diet. Short (1980) studied the ecology of the brush- tailed rock wallaby (Petrogale penicillata) at Kangaroo Valley (near Nowra) and Goulbourn River (near Muswellbrook) in New South Wales. Here grasses were a dominant element in the diet (33-50%), forbs (28-35%) and shrubs (12-30%) ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 together about equalled the grass. No Solanum was listed for the area or recorded in the diet. The intermediate position of Petrogale as both a grazer and browser was confirme In a limited study of the swamp wallaby: Wal- labia bicolor, Edwards and Ealey (1975) con- cluded that the species was a browser and that although the animals had access to grass and pasture these were scarcely eaten. The animals preferred coarse browse provided by shrubs and bushes. The swamp wallaby occurs in eastern Australia from Cape Yorke to Victoria and though Solanum was not mentioned in its diet or its environment, species occur in its area of occur- rence. There is evidence that the browsing wal- labies do eat prickly Solanum. For distribution of browsing marsupials see Figure 2. Family Phascolarctidae, koala. The koala (Phascolarctos cinereus) eats Eucalyptus only and is very specialised in its choice of species. It need not be considered in this context as a browsing herbivore. Family Vombatidae, wombats. The few species are grazers, not browsing herbivores and feed principally on monocotyledons (Wells, pers. mm Q o Family Tarsipedidae, honey possums. Al- though herbivorous, this species has a specialised diet of pollen, nectar, and fruit and is not a browsing herbivore. Rodents. Three studies of the diets of Aus- tralian rodents (Watts, 1970, 1977; Watts & Braithwaite, 1978) record Solanum seed as a mi- nor element of the diet of Notomys alexis at Yuendumu (Northern Territory) and there is no other mention of Solanum P. M. Hyland and B. Gray (pers. comm., 1981) stated that rats are known to eat the bark of S. mauritianum and S. torvum. Both these are alien weedy species that occur in wet tropical areas in eastern Australia. The first is unarmed, the second is prickly. Though possibly significant locally or when in plague numbers there is no evidence to suggest rodents are important as grazers. THE RESPONSE OF SOLANUM IN THE ABSENCE OF HERBIVORES Hawaii has always been an oceanic island and originally had a very limited vertebrate fauna. The nucleus of the whole flora arrived by long distance dispersal and has since speciated richly. There are about six native species of Solanum (St. John, 1973); of these, five are unarmed and 1986] only one, S. haleakalaense, is well armed. This last species may have been a late arrival that has not yet lost its prickles or the prickles may deter snails, which are abundant in Hawaii. I have not seen it growing in the field. Carlquist (1970) dis- cussed the prickles on some of the lobelioids in Hawaii. The family Campanulaceae subfamily Lobelioideae is not noted for the development of prickles. The plants are often acrid, bitter, and unpalatable, and many contain indole alkaloids. In Hawaii many species are densely prickly in the young stages and may even have prickly in- florescences and flowers. Carlquist considered this a defense against snails because no other herbi- vores are well developed in Hawaii. It is possible, although I have no evidence to support it, that the Australian species of Solanum with ex- tremely dense stem prickles that grow in the high rainfall sites, such as S. macoorai, S. oedipus, and S. semiarmatum, may also be protected against snails. Fiji, without mammalian herbivores, has six species of Solanum, probably native, all of which are unarmed, including S. repandum Forster, whose close relatives in South East Asia and South America are prickly (Whalen et al., 1981). New Caledonia also lacked mammalian herbivores. Heine (1976) listed 10 native species, of which seven are unarmed and three have traces of prickles. Included in these three is S. vacci- nioides, which varies from being unarmed to having prickles present, as might be expected if the original immigrant was prickly and the se- lection pressure for prickles was relaxed. Tomlinson (1962) discussed the pungency of palms that may develop prickles from leaves, stem emergences, or the roots. He suggested that their role is defence of the single large apical bud and that there are now few large animals capable of tearing open the heart of the palm. Without discussing it further he pointed out that many unarmed palms occur on the isolated islands of the Pacific, where there are few or small verte- brate animals. the thesis that prickles are Ре a a response to herbivory and have little relationship to the physical environment. THE EFFECT OF EXTINCT MARSUPIALS The extinct marsupials, some of which were very large, must have had some effect on the vegetation and in some cases may have acted as seed dispersal agents. Some of these animals sur- vived to the late Pleistocene and certainly over- SYMON—SOLANUM PRICKLES 753 lapped the arrival of aboriginal man in Australia. Sanson (1978) considered most of the extinct macropodids to be browsers. Many of the others erbivores, e.g., diprotodontids, phasco- larctids, palorchestids, and vombatids, and in- cluded both browsers and grazers (Archer, 1981). Janzen (1982) claimed to detect a suite of Central American plants whose fruits were probably dis- persed by the recently extinct large herbivores of that area. A parallel in Australia is not imme- diately obvious, but the influence of the browsers on some ofthe remaining vegetation, particularly in those areas now occupied by the grazing kan- garoos, might be worth searching for. CONCLUSION The evidence assembled suggests that prickles on Solanum are not an adaptation to aridity. The scattered and incomplete records of the diet of Australian vertebrate herbivores suggests that the relic group of marsupials called wallabies and prickly Solanum do in general coincide. The dis- tribution of wallabies may have contracted in geologically recent times due to increasing aridity but prickly Solanum still remain in some areas where wallabies are now extinct. LITERATURE CITED ALLAN, H. H. 1 ra of ae Zealand. Gov. Printer Wellington, New Zea ARCHER, M. 1981. A review of de origins and ra- diations of eie. чш mammals. 1 п А. Кеаз! (еа- itor), Ecolo Mono- graphic Seige: :1437- .1488. Dr. W. Junk, The Bases: ex , P. N. MATENZ & R. BARKER. 1971. The T kangaroo Megaleia rufa (Desmarest) in north-western New South Wales 11. Food. CSIRO BOLTON, B. L. & P. K. LATz. The western hare- wallaby Lagorchestes ко, (Gould) (Macro- podidae) in the Tanami Desert. Austral. Wild. Res 5: se ae BRADLEY, V., D. J. CoLLINS, P. G. CRABBE, F. W. T M. C. IRVINE, J. M. Sid & D. sa SYMON. A survey of Australian Solanu plants for potentially useful sources E laa Austral. J. Bot CARLQUIST, S. 1970. Hawaii: А Natural History. Nat. Hist. Press, New York. ‚М. Я Batanmcal examination of e ttl dida n tents. Austral. J. Sci. 25: m a The plants grazed by red kangaroos Megaleia тй (Desmarest) in Central Australia. c. New South Wales 93: 98-11 80. The biology of Bettongia Proc. Linn. и. Р. E. ad ei (G 1837) Macropus eugenii (Desmar st 18. 1 INi in relation to fire. Forests Dept. W. Aust d 754 CoPLEY, P. G. 1982. Studies of the yellow-footed rock wallaby, Petrogale xanthopus. 1. Distribution in South Australia. peer Wild. Res. 10: 47-61 ROBINSON. 1982. Studies of the yel- low-footed rock wallaby, Petrogale xanthopus Gra (Marsupialia, Macropodidae) II. Diet. Austral. : 63-76. 1978. The food of emus. Aust. J. Ecol. 3: 411-422. Davis, P. Н. & V. Н. HEywoop. 1967. Principles of Angiosperm Taxonomy. Oliver & Boyd, Edin- b urgh. Dawson, T. J. & B. A. ELLis. 1979. Comparison of the diets of yellow-footed rock-wallabies and sym- patric herbivores in Western New South Wales. Austral. Wild. Res. 6: 245-254. us a Н. М. & А. В. MAIN. 1967. oe na o, Macropus robustus dus CSIRO Wildl. Res. 12: 53-65. EDWARDS, G. P. & E. H. M. EALEv. 1975. Asp of the ecology of the D wallab y Wallabia! color (Marsupialia: Macropodidae). Australian Mam 3 1 malogy 1 7- Е1115, B. A., Е. M SS J. DAWSON & C. J. F HAR 97 easonal changes in diet pref- erences of free-ranging red kangaroos, euros & sheep in Dd New South Wales. Austral. Wild. Res. 4: 127-14 FREELAND, W. J. J. W. WINTER. 1975. Thee lutionary consequences of eating: Trichosurus umi сабы Deer ape and the genus Eucalyptus. J. Che 439-4 FRITSCH] T | SALISBURY. 1953. Plant Form and Function. Bell GREENWOOD, R. . E. ATKINSON. 1977. Evo- lution of divaricating plants in New Zealand in о to Moa browsing. Proc. New Zealand Ecol. c. 24: 21- , M. & R. BARKER. 1966. The plants eaten by sheep and kangaroos grazing together in a pad- dock 1 Ў erg Po ern Queensland. CSIRO Wildl. Res 45- — — & L MCLEAN. 1974. Further obser- grazing together in a -a in south- western Queensland. ge 1. Wild. Res. 1: 27-43. НЕШЕ, Н. 1976 ore de la Nouvelle Caledonie et dependances. pil Nat. d'Hist. Nat. Paris 7: 119- 212. JANZEN, D. H. 1981. The defenses of legumes against herbivores. Jn R. M. Polhill & P. H. Raven (ed- i , Advances in Legume System- atics 2: 951-977 1982. Neotropical anachronisms: the fruits the Gomphotheres ate. Science 215: 19-27. KERNER, A. 1895. The Natural History of Plants. Oliver translation. Blackie & Son, London. KIRKPATRICK, T. H Food preferences of the grey kangaroo (Macropus major Shaw). Queens- land J. Agric. Sci. 22: 89-93. Main, A. R. 1981. Plants as animal food. Jn J. S. Pat Comb (editors), The Biology of Australian Plants. University of Western Australia Press MAYNES, G. M. 1977. Distribution and aspects of the biology of Parma Wallaby Macropus parma ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 in New South Wales. Austral. Wild. Res. 4: 109- 125 MCGLONE, М. S. & C.J. WEBB. 1981. Selective forces influencing the evolution of ne plants. New Zealand J. Ecol. 4 NOBLE, J.C. 1975. Difftvenee ir in size of Emus on two contrasting diets on the riverine plain of New South 1 һе Ети 75: 35—37. The evolution and significance acropodidae. Australian ales. SANSON, G. D. of mastication in n the M Mammalogy 2: © morphology and occlusion of the molariform cheek teeth i some Macropodinae ). Austral. J. Zool. 28: F 1-365. . 1979. Hair types as taxonomic characters Solanum. Pp. 307-319 in J. G. Hawkes, R. N cae & A. D. Skelding (editors), The Biology and Taxonomy of the Solanaceae. Academic Press, ondon. SHORT, J.C. 1980. Ecology of the Brush Tailed Rock Wallaby (Petrogale penicillata Griffith, oo & Pidgeon). M.Sc. Thesis. Univ. of Sydne . 1973. List of flowering plants i m Hawaii. 0-301. А 1983. Ѕееа dispersal T Cassowaries (Casuarius casuarius L.) in the rth Q forests. Biotropica 15: 170- 76. Storr, G. M. 1962. Annotated flora of Rottnest Is- 2 Western Australia. West. Austral. Nat. 8: 109— s 964. Diet of the Quokka Setonix brachyurus (Quoy & Gaimard), on Rottnest Island, Western Australia. Austral. J. Biol. Sci. 17: 469-481. 68. Diet of kangaroos Megaleia rufa and Macropus robustus and merino sheep near Port Hedland, Western Australia. J. & Proc. Roy. Soc. Western Australia 51: 25-32. Symon, D. Fruit diversity and dispersal in Solanum in Australia. J. Adelaide Bot. Gard. 1: 321-331 A revision of the genus Solanum in Australia. 7. Adelaide Bot. Gard. 4: 1-367. TOMLINSON, P. B. 1962. Essays on the morphology of palms. VII. A digression about spines. Principes 6: 44-5 VAN Dyck, s. Destruction of wild tobacco trees (Solanum алиби by mountain possums (Trichosurus caninus). Mem. Queensland Mus. 19: 367-371. 1969. sein oe and habits of the Roy. Soc. South WATTS, C. Н. S. rabbit bandicoot. Trans. Australia 93: 135-141. 0. е foods eaten by some Australian ustral. Naturalist 44: 71-74. e foods eaten by some Australian rodents ide Austral. Wild. Res. 4: 151- 157. desert о S. R. W. BRAITHWAITE. 1978. The diet of Rattus lutreolus and ne other dec in Southern Victoria. ig al. Wild. Res. 5: 47— WHALEN, M. D., D. E. о B. HEISER. 1981. Тахопоту of Solanum section Lasiocarpa. Gentes Herb. 12: 41-129. ZAPOTECA: A NEW GENUS OF NEOTROPICAL МІМОЅОІРЕАЕ! Héctor M. HERNANDEZ? ABSTRACT The existence of ie ed variation patterns in polyad characteristics, seedling morphology, chro- tures within x pers species of Calliandra Benth. , the first one including the species of has revealed the presence b. two w ell defined ta Laetevirentes. New combinations are presented for C. caracasana, C. formosa, C. lambertiana, C. media, C. mollis, C. portoricensis, and C. tetragona. RESUMEN La existencia de patrones ics de variación en las características de las enn morfología de Í como en aproximadamentre 25 taxa distribuidos desde el suroeste de lo hasta el norte de Argentina, con un claro centro de diversificación en el sur de Mé sen también nuevas combinaciones para C. caracasana, C. formosa, C. lambertiana, C. media, C. mollis, C. ст у С. tetragona. This paper reports part of a biosystematic study of the genus Calliandra (Leguminosae: Mimos- oideae, tribe Ingeae). As currently circumscribed, this genus includes a relatively heterogeneous as- semblage of about 150 species native predominantly in tropical and sub-tropical re- gions in the Americas. In addition, a few species are found in continental Africa (Thulin et al., 1981), Madagascar, and India (Paul, 1979). The genus was originally described by G. Bentham in 1840, who subsequently (1844), and more for- mally in his monograph of the Mimosoideae (1875), subdivided it into five s series Ще on leafandi cro- phyllae, Laetevirentes, Pedicellatae, Nidaa. and Racemosae. Since then, many new species have been described, and the genus has been included ! I would like to thank E. Forero, Ph. Guinet, D. Neill, for CarCiull grateful to P. H. Raven for his s providing helpful criticisms, J. Dwy and M. Veith for assistance in the SEM Lab at Washington University, Biology Department. I a upport and encouragement throughout the study. The financial support from in a number of floras; nevertheless, Bentham’s original concept of the genus remains unchanged. Palynological studies have contributed sub- stantially to the understanding of the infrageneric relationships in Calliandra (Guinet, 1965, 1969, 1981: Guinet & Barth, 1969; Sorsa, 1969; Niez- goda et al., 1983). The New World species may be classified into two basic pollen groups, accord- ing to their contrasting polyad features. The first group (Type IB in Guinet, 1965) includes all the species of ser. Laetevirentes, and two species placed by Bentham into ser. Macrophyllae. Poly- ads in this group are 16-grained, radially sym- metrical, acalymmate, with granular ektexine structure. These polyad characteristics are found in all the other genera of the tribe Ingeae; how- ever, species in this portion of Calliandra have L. I. Nevling, I. Nielsen, R. M. Polhill, and M. Sousa er for checking the Latin description, am cabo i The Garden Club of America, World Wildlife Fund —U.S. (Tropical Botany Fellowship), the Mi da otia ? Missouri Botanical Garden, O NO a e E - = Ф DP [e] 5 =й nN o2 — nN е po] =| о N O 5 e > Qa б. =} O Ф : erbario Nacional, Instituto de Biologiá, Universidad Nacional Autónoma de México, Apartado Postal 70- 367, сат 04510 .F. México ANN. MISSOURI Bor. GARD. 73: 755—763. 1986. 756 FIGURES 1, 2. (cult. at MO from: Mexic z, Johns ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Polyad types o a = e C о — 1. Unacetolysed 16-grained polyad of Z. tetragona 582).—2. Unacetolysed 8-grained polyad of C. houstoniana. The sticky appendage on nih Mis cell iS ME to the SEM stub (cult. at MO from: Mexico, Chiapas, Johnson 363-78). Scales = 50 u very distinctive circular thickenings in the cen- tral cells generally on one side of the polyad (Fig. 1), which are an excellent diagnostic character. The remaining species of the genus, in contrast, have specialized 8-grained, bisymmetric, calym- mate polyads, with columellar/granular exine structure. In addition, the 8-grained polyads have a mucilaginous structure on the “basal” cell (on the narrow part of the polyad), which is associ- ated with transfer by pollinators (Fig. 2). The nature of polyad cohesion (calymmate), the char- acteristics of the ektexine, and the presence of the sticky appendage in the “basal” cell of the 8-grained polyads in Calliandra, form a com- bination of features found nowhere else in the Mimosoideae. It is important to mention that no intermediate forms have been found d the two pollen groups (Ph. Guinet, pers. com The outstanding differences between Pn 16- and 8-celled polyads have provided evidence to suggest that the genus should be subdivided (Guinet, 1969). Moreover, Niezgoda et al. (1981) have suggested a separate generic or even tribal status for the species of Calliandra with 8-celled polyads, in light of their great distinctiveness in the whole subfamily. On the basis of this evidence, several studies were initiated in order to test the general validity of this relationship within Calliandra. These studies, the results of which will be outlined be- ow, include surveys on seedling morphology, reproductive biology, and chromosome numbers of species throughout the genus. Although these investigations are still in progress, they show the same variation pattern found in the palynolog- ical work. The dichotomy resulting from the analysis of these characters acquires a special — elastically dehiscent pods. All other evidence in- dicates that ser. Laetevirentes and the rest of Cal- liandra species do not constitute a monophyletic assemblage and suggests that these taxonomic units must be placed into separate genera. The genus name Calliandra Benth. has been conserved over Anneslia R. A. Salisbury (Int. ode Bot. Nomencl. p. 267, no. 3444. 1961), with C. houstonii (L’Hér.) Benth., nom. illeg. [Gleditsia inermis L., C. inermis (L.) Druce] as the type species. Bunting (Taxon 16: 469-472. 1967) has proposed amending the conservation 1986] of Calliandra with C. grandiflora (L'Hér.) Benth. as the type species, because of the difficulty of typifying C. inermis; however, this proposal was rejected (Taxon 17: 466. 1968). Bentham (1875) considered C. houstonii as belonging to his ser. Racemosae. Similarly, the synonyms of Callian- dra, Clelia ornata Casar. [= Calliandra harrissii (Lindley) Benth.], and Codonandra purpurea Karsten. [= Calliandra codonandra (Karsten.) Benth.] were included in sers. Macrophyllae and Nitidae respectively. Consequently, as no pub- lished generic name seems to be available for any member of ser. Laetevirentes, the following new name is proposed. Zapoteca H. Hernández, gen. nov. (Legumino- sae-Mimosoideae, trib. Ingeae). TYPE: Z. tet- ragona (Willd.) H. Hernández, comb. nov. Acacia tetragona Willd. Species Plantarum 4(2): 1069. 1806. Venezuela, Caracas, in ri- pis arenosis fluvii Guairito, Bredemayer 17 [lectotype: B (Herbar Willdenow no. 19147) here designated; phototypes MO!, MEXU!]. Calliandra Benth. ser. Laetevirentes Benth. Hook. J. Bot. 3: 97. 1844. Frutices ramosi, erecti vel scandentes, inermes, ram- is teretibus vel А Stipulae ѕаере conspicuae, foliaceae, persis end HE bipinnata, foliolis saepe membranaceis iy coriaceis. Inflorescentia capitata, pedu cilia, piece vel in pseudopanicula, ma vel homogama, bracteata. Calyx cupulatus, dentatus, glaber vel pubescens. Co- rolla infundi formis vel лы lis r volutis. Stamina numerosa (ca. 30—60); filamentis longe exsertis, 19-4 m angis, slbi roseis | vel purpureis; staminum tubo inclus osis. Pollinia 16-cellularia, du и cellulis cen- illatis. Ovarium | breviter stipitatum, 10- mate apicali, cupulato. Legumen membranaceum ad oriaceum, siccum, lineare, rectum, plano-compres- sum, margine incrassato, valvis ab apice ad basin elas- tice dehi il S ina ind ta, exarillata. x = 13. Shrubs ramose, erect or scandent, unarmed, glabrous or pubescent, with the young branches terete or 4-angled. Phyllotaxy distichous. Sti- pules usually conspicuous, leafy, usually persis- tent in mature leaves. Leaves bipinnate; petiole rarely with nectariferous glands; leaflets oppo- site, sessile, usually membranous or rarely co- pseudopaniculate, pedunculate, capitate, densely flowered, homomorphic, heterogamic or homo- gamic, bracteate. Flowers sessile, actinomorph- HERNÁNDEZ- ZAPOTECA 757 ic, 5- rarely 6-merous; calyx cup-shaped, dentate, glabrous or pubescent; corolla campanulate or infundibuliform, membranous, with the petals revolute, valvate in bud; nectariferous disk pres- ent in all flowers; stamens numerous (ca. 30-60), 19-43 mm long, rple inal tube inserted; anthers еси. Ае. eglandular, every anther containing eight polyads; polyads grained, discoid, heteromorphic, with eccentric lens-shaped thickenings on central cells; ovary 1, shortly stipitate, with ca. 10-15 ovules; style filiform, ca. 15-56 mm long in fertile pistils; stig- ma apical, cup-shaped. Legume dry, membran- aceous to coriaceous, linear, straight, plano-com- pressed, with the margins thickened, usually with constrictions in interseminal areas, valves de- hiscing elastically from the apex to the base. Seeds in one series, hard, ovate to rhomboid, non-ar- illate, non- is usually with irregular or regular 90 percent pleurogram. Seedlings phaneroepi- geal, with the cotyledons ephemeral, foliaceous, sessile, elliptic to elliptic-ovate; the first and sec- ond eophylls opposite, the third and fourth al- ternate; leaflets usually membranous. x — 13. Z. caracasana (Jacq.) H. Hernández, comb. nov. Mimosa caracasana Jacq., Collectanea 4: 142. 1791. Fig. 3: 20, t.632. 1793. Z. formosa (Kunth) H. Hernández, comb. nov. Acacia formosa Kunth, Mim. p. 102, t. 32. Jacq., Icon. pl. rar. 1822. Z. lambertiana (G. Don) H. Hernández, comb. nov. Acacia lambertiana G. Don, Edwards Bot. Reg. 9: t. 721. 1823. TYPE: México, lo- cality unknown; cultivated by G. Don, from material sent to him by Mr. Lambert from Boyton, where plants of this species had been raised from seeds collected in México (lec- totype, K!, here designated). Z. media (M. Martens & Galeotti) H. Hernandez, 3362 pro parte (lectotype, BR!, here desig- nated). Z. mollis (Standley) H. Hernandez, comb. nov. Calliandra mollis Standley, Contr. U.S. Natl. Herb. 17: 431. 1914. TYPE: Costa Rica, Puntarenas, near Nicoya, Tonduz 13536 [holotype, US! (sheet no. 578114); isotypes, BM!, F!, G!, GH!, K!, NY!, US!]. 758 Z. portoricensis (Jacq.) H. Hernandez, comb. n Mimosa ва Јаса., li. р. 143. 1791. Fig. Jacq., Icon. pl. rar. 3: 20, 1. 633. 1793 This genus is named to pay homage to the Zapotec people, who since 500 B.C. (Flannery & Marcus, 1983) to the present have inhabited the varied lands of Oaxaca, Mexico, the center of diversity of this group of plants. Zapoteca is a rather homogeneous assemblage morphologically, consisting of about 25 taxo- nomically difficult taxa, distributed from north- ern Mexico and southwestern United States to northern Argentina, ranging from sea level to ca. 2,150 m. The greatest concentration of species is in southern Mexico, especially in the state of Oaxaca. Natural populations occur primarily in disturbed habitats derived from a variety of vegetation types, from semi-arid scrub to rela- tively wet forests; however, the greatest diver- sification has occurred in tropical dry forests. Several species have very wide ranges (e.g., Z. formosa, Z. portoricensis, Z. tetragona) and evi- dently have outstanding ability to colonize new habitats. Other species are narrow endemics (e.g., Z. mollis). The totality of published specific epithets (45) included within Calliandra, ser. Laetevirentes is clearly comprised in Zapoteca. As mentioned above, 25 taxa are currently recognized in the genus, and new combinations are proposed here for seven of them. The remaining ones will be treated in a future revision, due to the fact that their taxonomic status— whether species or sub- species—is still unclear. Because the species of Zapoteca are quite uniform morphologically, the genus is easy to identify. There are, however, some Calliandra species, not members of ser. Laetevirentes, that must also be taken into con- sideration because they have some features typ- ical of Zapoteca. These are C. amazonica Benth. and C. aculeata Spruce ex Benth., both placed by Bentham (1875) in his ser. Macrophyllae, who, however, had previously noted similarity be- tween the structure of the inflorescence of C. amazonica and that of Z. portoricensis (Benth- am, 1844). Subsequently, Guinet (1965) found that the polyad characteristics of these two species are essentially the same as those of Bentham’s ser. Laetevirentes. The small number and larger size of the leaflets in C. amazonica and C. acu- leata, and the presence of stipular spines in C. aculeata, are the main characters differentiating ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 them from the species of Zapoteca. In the light of present evidence, it seems clear that these two species do not fit within the limits of Calliandra as proposed in this paper; however, their inclu- sion within Zapoteca or another genus needs fur- ther study. DESCRIPTION OF SOME CRITICAL CHARACTERS IN ZAPOTECA AND CALLIANDRA This section provides the supporting evidence that reflects the heterogeneous nature of Callian- dra as structured by wh Penman. The characters analyz edared гитсе accounts will be provided in a subsequent paper. Seedling morphology. A total of 95 seedling samples have been studied, representing nine species of Zapoteca and 31 of Calliandra. In describing the seedlings I have selected four main characters: |) degree of persistence of cotyledons, 2) morphology of cotyledons, 3) type of succes- O these, cotyledon features revealed the most strik- ing contrast between the Zapoteca seedlings and those of Calliandra. All the species of Zapoteca have seedlings with elliptic or elliptic-ovate, ses- sile, foliaceous, and ephemeral cotyledons (Figs. 3, 4). In contrast, Calliandra seedlings have sag- ittate, petiolate, fleshy, and relatively persistent cotyledons as common characters (Figs. 5, 6). While the type of succession of the eophylls and the leaflet characteristics were remarkably con- stant in the Zapoteca seedlings, these characters display a high degree of variation in Calliandra. This variation allows a further grouping into five subtypes, which correspond to a certain extent with Bentham's series. In sum, the morpholog- ical characteristics in the seedlings of Zapoteca, which are notably homogeneous, differ from those of Calliandra primarily in the basic features of the cotyledons. The absence of intermediate seedling forms between Zapoteca and Calliandra reflects the lack of close relationships between the two genera. Reproductive biology. The species of Zapo- teca differ from those of Calliandra in a number of aspects of their reproductive behavior. In Za- poteca the individual flowers are always arranged in compact, homomorphic, spherical inflores- cences (Fig. 7). In Calliandra, in contrast, there are more a patterns of inflorescence organ- ization. The basic type seems to be the axillary, obconiform жыла (Fig. 8) from which other types have been derived. The racemose 1986] HERNÁNDEZ- ZAPOTECA 759 FIGURES 3-6. 2 types of Zapoteca and Calliandra. —3. Z. sp. nov. (cult. at MO from: Mexico, Oaxaca, IGURE Torres eim . Z. tetragona (cult. at M (cult. at MO nud о. Torres 4 192). Scales = 10m inflorescences of the species of ser. Racemosae (Fig. 9), a group basically endemic to Mexico and Central America, represent the type that has de- parted most from the ancestral type of inflores- cence. An interesting reproductive feature found in most species of Calliandra ser. Nitidae, Mac- rophyllae, and in some of the Pedicellatae is the presence of heteromorphic inflorescences (Fig. 10). In general, these have one to several central flowers whose morphology is modified in rela- tion to the peripheral ones. The central flowers are the only ones bearing nectaries; these are larg- er and have the staminal tube well exserted from the corolla apex. This floral specialization, which is also found in other genera of the tribe Ingeae co provide nectar rewards, the visitors forage ac- tively on them, seeking them out and therefore increasing the possibilities of pollinating a higher number of hermaphroditic flowers in the inflo- rescence in the course of a single visit. A detailed analysis of sex expression carried MO from: Mexico, Я David, Johnson 929-79). —6. С. Е (cult. at MO from: Mexico, Oaxaca, Veracruz, H. Hernández 167).—5. C. calothyrsus out on flower material collected in natural pop- ulations has shown that andromonoecy, express- ed morphologically by gynoecium abortion, is widespread in both genera. In contrast to the opinion of Nevling and Elias (1970), this study has shown that variation in the presence of an- dromonoecy is not consistent with the variation patterns found in other characters (e.g., polyads, seedlings, chromosome numbers) Another reproductive feature characteristic of Zapoteca species is the stigma type. All the species in this genus have delicate, cup-shaped stigmas, with a very narrow area of receptivity that can hold just a single 16-celled polyad (Fig. 11). As far as it has been possible to observe, this stigma type is found in all the genera of the tribe Ingeae, except Calliandra (see for example, Koptur, 1983; Hoc, 1981), and in 17 genera ofthe Mimosoideae (Lewis, pers. comm. via Polhill; see also Simpson et al., 1977; Nevling & Niezgoda, 1978; Kenrick & Knox, 1981; Lewis & Elias, 1981; Hopkins, 1984). In contrast, the species of Calliandra have very distinctive expanded, fungiform, discoid, or capitate stigmas with a much wider area of poly- ad receptivity (Fig. 12). The stigma types in Cal- 760 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 7-12. Reproductive characters of Zapoteca = Calliandra. —7. Spherical, homomorphic inflores- cence of Z. tetragona (cult. at MO from: Mexico, Veracruz, Johnson 1582).—8. Obconiform, heteromorphic inflorescence of C. rubescens (Mexico, Veracruz, H. Hernánde ez 156).—9. Racemose reci rt of C. calo- thyrsus (Mexico, Chiapas, H. na 519).—10. Heteromorphic inflorescence of C. rubescens (Mexico, Veracruz, H. Hernández 156).— Cup-shaped stigma of Z. caracasana ue a i tained d in the stigmatic cavity (cult. at MO foie gris 2494-82).—12. Fungiform stigma of C. houstoniana. The arrows indicate polyads attached to the style (cult. at MO from: Mexico, Chiapas, pope 363-78). Scales = 40 mm in Figures 7-10; 50 um in Figure 11; and 1 mm in Figure 12. liandra are clearly a derived condition with im- ioral reproductive characteristics in Zapoteca portant reproductive consequences; however, ап and Calliandra is associated with particular analysis of their adaptive significance is out of classes of pollinators. Field and greenhouse ob- the scope of this paper. servations, in conjunction with some literature The specific set of morphological and behav- reports, have allowed us to characterize these 1986] HERNÁNDEZ- ZAPOTECA 761 TABLE 1. Chromosome numbers in Zapoteca and Calliandra. The asterisk (*) indicates first report for the species. Chromo- some Number Species n 2n Reference Zapoteca tetragona* 26 México, Veracruz, Mun. Catemaco, a | km de Sontecomapan por camino a Coxcoapan, H. Hernández 167 (MO, MEXU 26 . México, Veracruz, Mun. Córdoba, sobre la autopista Fortín-Córdoba, H. Hernández 827 (MO, MEXU) 26 Costa Rica, Cartago, Finca Perla, entre Reserva de Vida Silvestre Ta- pantí y Purisil, H. Hernández 675 (MO, MEXU) Z. portoricensis* 26 México, Veracruz, 6.8 km N de Xalapa por camino a Naolinco, H. Hernández 154 (MO, MEXU) Z. formosa* 26 México, Oaxaca, Dist. Tehuantepec, 11.1 km SW del entronque carr. Tehuantepec-Oaxaca, por camino a Buenos Aires, R. Torres 4262 (MO, MEXU 26 México, Yucatán, 20 km S of Progreso, D. Johnson 483-78 (MO) Z. media* 26 México, Hidalgo, Mun. Metztitlán, Paso del León, Barranca de Vena- dos, A. Delgado 53 (MEXU) Z. sp. nov. (1)* 26 México, Oaxaca, Dist. Mixe, 6 km М de Mitla por camino a Totonte- pec, R. Torres 3947 (MO, MEXU) Z. sp. nov. (2)* 26 México, Oaxaca, Dist. Tehuantepec, El Manguito, SW de El Limón, R. Torres 4167 (MO, MEXU) Ser. Nitidae Calliandra surina- 16 — Atchinson (1949)'; Goldblatt (1981b) nsis C. surinamensis 8 Bir et al. (1980) C. magdalenae 16 Goldblatt (1981b) C. haematocephala 16 Atchinson (1951) C. tweedyi 16 Bir and Kumari (1978, 1979) C. pittieri 32 Shibata (1962) Ser. Macrophyllae C. tergemina* 16 México, Jalisco, Mun. la Huerta, Estación de Biología Chamela, E. Lott 1686 (MO, MEXU) Ser. Racemosae C. houstoniana 8 Bir et al. (1980) C. calothyrsus 22 Goldblatt (1981b)” C. rusbyi* 22 México, Guerrero, Mun. Coyuca, Torre de Microndas, 5 km NW de El Zapote, por carr. Atoyac-Cayuca, H. Hernández 865 (MO, C. sp. nov.* 22 México, Oaxaca, Dist. Putla. 11.5 km NE de Putla por carr. a Tlaxia- co, H. Hernández 465 (MO, MEXU) ! Referred to as C. haematocephala. ? Referred to as C. confusa. genera in terms of their pollination mechanisms. The pollination of species of Zapoteca appears f to be exclusively nocturnal, and a number o settling moth species (families Noctuidae, Pyral- idae, and Geometridae, etc.) consistently have been observed foraging on inflorescences of Z. portoricensis, Z. formosa, Z. media, and Z. tet- ragona (pers. obs.). Calliandra, however, is more diverse in its pollination systems. Several species of the genus have been reported as hawkmoth- pollinated (Cruden et al., 1976; Haber & Frankie, 1984; pers. obs.), whereas others are humming- bird pollinated (Snow & Snow, 1972; Arroyo, 1981; pers. obs.). Interestingly, as was noted by 762 ANNALS OF THE MISSOURI BOTANICAL GARDEN (VoL. 73 TABLE 2. Comparative characters of Zapoteca and Calliandra. Calliandra Zapoteca Leaflets thin membranaceous, rarely coriaceous Inflorescence spherical heads; homomorphic Polyads 16-grained, discoid Stigmas cup-shaped, with narrow area of polyad receptivity Legume membranaceous to coriaceous Seedlings cotyledons ovate, sessile, foliaceous, phemeral Chromosome x= 13 numbers chartaceous to coriaceous obconiform, racemose, rarely spherical; homo- morphic or heteromorphic 8-grained, bisymmetric, with viscid appendage in “basal” grain expanded; fungiform, discoid, or capitate more rigid; coriaceous, ligneous, rarely mem- branaceous cotyledons saggitate, petiolate, fleshy, persistent п = 8, 11 Arroyo (1981) for other genera of the Mimoso- ideae, the limits between sphingophily and or- nithophily are not well established in a number of species of Calliandra [e.g., C. magdalenae oe ex DC.) Benth., С. A (L'Hér.) , ndley]. In these species the filaments a are spi or a combination of white and red, and the flowers remain in an- thesis during the night and the first hours of the day, allowing a succession of sphingid moth and hummingbird visits. The relative effectiveness of each pollinator class will be analysed elsewhere. A similar situation is probably found in C. cal- othyrsus Meissn. (= C. confusa Sprague & Riley, and C. similis Sprague & Riley), а ех have been reported to be visited by hon ers (Coerebidae; Skutch, 1981), bats (Holdridge & Poveda, 1975), and sphingid moths (pers. obs.). Visitors of high energetic requirements often seek such flowers with their copious nectar product if the nectar is relatively accessible. Cytogenetics. Although Calliandra sensu Bentham is a large genus, very little information on its chromosome numbers has been published (Table 1). The counts I have obtained so far in the hitherto cytologically unknown Zapoteca are п = 13 (Table 1), which is consistent with the basic chromosome numbers reported for all the genera of the Ingeae (х = 13; Goldblatt, 198 1a). Calliandra sensu stricto, in contrast, represents Furthermore; С. se инна Tu un unum and other members of ser. Racemosae are known to be = 11, and C. pittieri is polyploid, n= 16 (Shibata, 1962). Table 2 shows a summary of the basic characters of Zapoteca and Calliandra. GENERAL REMARKS The segregation of the species of Calliandra ser. Laetevirentes into the genus Zapoteca as pro- posed in this paper constitutes the first step to- ward a comprehensive study of Calliandra (Fo- rero, 1984; Hernandez, 1984; Romeo, 1984). The Calliandra species from Madagascar and India must be studied critically in the same way as the Zapoteca species because these are palynologi- etic. We must pursue an integral in- frageneric classification of this genus in order to improve Bentham’s taxonomic framework. Sev- eral taxa, primarily South American, do not fit into this classification, and evidence indicates that sers. Macrophyllae and Pedicellatae are arti- ficial. The presence of two species of Calliandra in restricted areas of eastern Africa provides an in- g polyad characteristics of these species are basi- cally the same as those of the neotropical Cal- liandra (Thulin et al., 1981) with the exception that the cells are free from one another (acalym- 1986] mate). The existence of this feature is a further indication of the primitive nature of these species. LITERATURE CITED ARROYO, M. T. K. 1981. Breeding systems and pol- lination biology in Leguminosae. Jn R. M. Polhill . H. Raven (editors), Advances in Legume Sys- tematics. Proceedings of the International Legume Conference. Royal Botanic Gardens, Kew ATCHISON, E. 1949. Studies in the Leguminosae. IV. Chromosome numbers and geographical rela- tionships of ccc Leguminosas, J. Elisha Mitchell Sci. Soc. 65: 2. 19 Studies in 1 шй сайа, VI. Chr mosome numbers among tropical woody ate Amer. J. Bot. 38: 538-547. BENTHAM, G. 1840. Contributions towards a flora of South America. — enumeration of plants collected by Mr. о in British Guiana. Hook. J. Bot. 2: 127-146. 1844. o tes in Mimoseae, Sine a Synopsis of ‘the species. London J. Bot. 3: 82-112. 1875. Revision of the ide шне . 1978. In IOPB a ' number reports a Taxon 27: 53-61. & 1 Cytological evolution of leguminous flora ш the Punjab | Plain. In S.S. B (editor), Recent E yani Publishers, New Deli, India. 1980. In IOPB chromo- some number reports 69. Taxon zi 703-730. CRUDEN, R. W., S. KINSMAN, KHOUSE II & Y.B. LINHART. 1976. Pollination, Scu dic and the distribution of the moth-flowered plants. Bio- tropica 8: 204-210. FLANNERY, K. V. & J. MARCUS (editors). 1983. The Cloud People. Divergent Evolution of the Zapotec and Mixtec Civilizations. Academic Press, New York. FORERO, E. 1984. Revision of Calliandra: a multi- ИУ approach. Bull. Groupe Int. Etude 2: 14-15. mosome ndi in pues II. Ann . Missouri p Gard. 68: 551- GUINET, PH. 1 tude des caracteres ж pollen dans le genre Calliandra (Mimosaceae). Pollen & Spores 7: 157- 1969. Les TN Etude de palynologie fondamentale, correlations, evolution. Trav. Sect. Sc. Techn. Inst. Fr. Pondichery 9: 1-293. 981. priced the characters zd id pollen grains. Jn В. M. Polhill & P. H. en (editors), Advances in e Systematics. a ceedings of the qiii ти Conference. Royal Botanic Gardens, Kew & O. M. BARTH. ы L'exine S Callian- dra (Mimosaceae) observee en microscopie pho- tonique et en microscopie loa ada Pollen 4 Spores 9: 211-227. HERNÁNDEZ- ZAPOTECA 763 HABER, W. A. & G. FRANKIE. 1985. Characteristics and organization of a tropical hawkmoth com- munity. Biotropica (in press). HERNÁNDEZ, H. M. Contribution to the sys- tematics of Calliandra, with particular reference to its infrageneric relationships. Bull. Groupe Int. -18. Hoc, P. S. . El género Pithecellobium en la Ar- gentina. Darwiniana 23:523-558. HOLDRIDGE, L. & L. PovEbA. 1975. Arboles de Costa Rica, Volume 1. Centro Cientifico Tropical, San José, Costa Rica. . 1984. Floral biology and pollination ecology of the neotropical species of Parkia. J. Ecol. 72: 1-23. KENRICK, J. & В. B. Knox. 1981. Structure and his- tochemistry of the stigma and the style of some Australian species of Acacia. Austral. J. Bot. 29: —745. Кортов, S. 1983. Flowering phenology and floral biology of Inga (Fabaceae: Mimosoideae). Syst. Bot. 8: 354-368. Lewis, G. P. & T. S. Euras. 1981. Mimoseae. Jn В. M. Polhill & P. H. Raven (editors), Advances in Legume Systematics. Proceedings of the Interna- tional Legume Conference. Royal Botanic Gar- NEVLING, L. I. & T. Euras. 1970. Calliandra, pollinia, and s dee implications. Amer. J. Bot., Suppl. 59: 753 tr.). & CH. ] NIEZGODA. 1978. On the genus е (Leguminosae-Mimosoideae). Adan- sonia, ser. 2, 18: 345—363. Nen CH. J., S. M. FEUER & L. I. NEVLING. 1983. Pollen ultrastructure of the tribe Ingeae о oideae: Leguminosae). Amer. J. Bot. 70: 650- 667. PAuL, S. R. 1979. The genus Calliandra (Mimosa- ceae) in the Indian subcontinent. Feddes Repert. -164. RAVEN, P. Н. & D. Н. AxELROD. 1974. ib CIR biogeography and past continental move Ann. Missouri Bot. Gard. 61: —673. ROME 1984. Preliminary ch taxo i - vestigations of Colombian леа ma in n . Bull. Group nonprotein amino ac Int. Etude Mimos. 12: 1 шге. К. . colombianas silvestres у cultiv Tokyo Nogyo Daigaku 8:4 SIMPSON, B. B., J. L. NEFF & A. В. we 1977. Prosopis flowers as a resource. /n B. B. Simpson (editor), Prosopis: Its Biology in Two Desert Scrub Ecosystems. US/IBP Synthesis Series 4. Dowden, Hutchinson & Ross, Inc., Stroudsburg, Pennsyl- Estudios. citológicos de plantas к Jour. Agric. vania. SKUTCH, A. F. 1981. New studies in tropical Amer- ican birds. Nuttall Ornithological Club. 19. Snow, B. K. & D. W. SNow Feeding niches of hummingbirds in a Trinidad Valley. J. Anim. Ecol. 41: 471-485. SonsA, P. 1969. Pollen morphological studies in the Mimosaceae. Ann. Bot. Fenn. 6: 1- THULIN, M., PH. GUINET & A. HUNDE. 1981. Callian- dra (Leguminosae) in continental Africa. Nord. J. Bot. 1: 27-34. DISTRIBUTION OF NONPROTEIN IMINO AND SULPHUR AMINO ACIDS IN ZAPOTECA! JOHN T. ROMEO? ABSTRACT Canari Рр a 11 г con taining сопа amino acids and nonprotein imino acids, and f the patterns d with those of Calliandra species, S-(8-carboxyethyl)-cysteine, the major free seed amino acid in all Calliandra species examined, an i: S- (8-carboxyisopropyl)-cysteine, a lesser constituent, were not detected in Zapoteca species. these same comp iandra species. The chemica interpreted as supportive evidence for the distinctiveness к these two closely allied groups of pee Calliandra (Mimosoideae) is a large hetero- geneous group of tropical-subtropical woody shrubs and trees found mainly in the Americas, extending from the southern U.S. to Argentina. About 200 species have been described. Zapo- teca, a new genus (Hernandez, 1986) consists of includes all plants paced previously under Ben- tham's series t of Calliandra (1875). Species of both genera are characterized by large amounts of rare unusual imino acids that accumulate in both seeds and leaves. Nine com- pounds, derivatives of pipecolic acid, previously have been isolated from various Calliandra species (Marlier et al., 1972, 1979; Bleecker & Romeo, 1981, 1983; Romeo et al., 1983). They include four BE E pipecolic acids trans-4-, trans-5-, cis-4-, and cis-5-hydroxypi- pecolic acids); four а | pipecolic acids (trans-cis-, trans-trans-, cis-trans-, and cis-cis-4,5- oo acids); and trans-4-acetyl- nopipecolic acid. The last two compounds are incon reed panel Calliandra. Populational sampling of a number of species has established that there is essentially no intraspecific variabil- ity in the rare compounds (Romeo, 1984). Com- plementary laboratory experiments have shown that the compounds are stable under various conditions of imposed water and UV stress (Bleecker & Romeo, in prep.; Balis & Romeo, unpubl. data). The rarity of the compounds cou- pled to their apparent ecological stability indi- cates that they are suitable for use in taxonomic studies. Analysis of Calliandra leaf material of over 100 species has produced several different chemical patterns (Romeo, ever, is yet to be determined. Calliandra is cur- rently undergoing systematic revision, and the chemical findings will be interpreted in conjunc- tion with results emerging from other fronts (see below). The seeds of Calliandra are characterized ad- ditionally by large amounts of nonprotein sul- phur-containing amino acids. S-(8-carboxy- ethyl)-cysteine (S-CEC) is the major compound. It is rapidly metabolized upon germination and is absent from mature foliage, but continues to appear in new emerging leaves up to 70 days after germination (Swain & Romeo, 1986). Less- er amounts of other related S-containing amino acids are also present in seeds. S-(8-carboxyiso- propyl)-cysteine (S-CIC), djenkolic acid, and acetyldjenkolic а ii "un frequently along with fth aara (Krauss & Reinbothe, 1970; Romeo & Swain, unpubl.). The presence of these two classes of unusual amino acids in Calliandra has led to a large-scale study of the distribution of these compounds in leaves and seeds. Calliandra, which has received little significant taxonomic treatment since the monograph by Bentham (1875), is now the focus ! This material is based upon work supported by the National Science Foundation under grant no. BSR 8400277. I thank Lee Swain, who pla yed a major role in the collection E vu data, for his assistance. ? Department of Biology, ш of South Florida, Tampa, Florida ANN. MISSOURI Bor. GARD. 73: 764-767. 1986. 1986] of much research (Forero, 1984). In addition to species), and Hernan can Mexican species), developmental, palynological, chromosomal, and chemical studies are under- way. The work described herein is part of this multidisciplinary project. With the establish- ment of a new genus Zapoteca (Hernandez, 1986) to include several species previously considered under Calliandra, it isan appropriate time to put forth some of the data that are emerging from the chemical studies. The purpose of this paper is to present chemical data on Zapoteca species, and to discuss the significance of these findings particularly in relation to Calliandra. MATERIALS AND METHODS Collection of Material. voucher data in Tables | a Extraction and chromatographic meth- ods. One hundred mg of material were ground and extracted by the method described in Romeo et al. (1983). A MeOH-CHCI,;-H;O (12:5: 1) ex- tract was separated into an upper aqueous and lower СНС. layer by addition of H,O and СНС... The aqueous phase was removed, evaporated to dryness and redissolved in 25% EtOH. This was used for paper chromatography and high voltage paper electrophoresis. Samples for quantitative analysis on the amino acid analyzer (Dionex D-300) were prepared by evaporating aliquots of the above solution and redissolving them in 0.2 M Nat buffer (Dionex Femto buffer 1A, pH 2.0). Chromatographic solvents used were 1) BuOH- НОАс-Н.О (12:3:5), 2) 80% PhOH-H,O (w/v) in the presence of NH, vapor, 3) BuOH-HCO;H- H,O (15:3:2). High voltage electrophoresis was performed on paper in buffer (pH 1 Identification of compounds. Compounds were identified from К, values and ionic mobil- ities (Seneviratne & Fowden, 1968; Romeo et al., 1 , comparison with authentic standards, and color reactions with the location reagents ninhydrin, isatin, and iodoplatinate. Species are listed with nd 2 RESULTS There is a striking dichotomy in sulphur chem- istry between Calliandra as circumscribed here, which includes members of Bentham’s series Macrophyllae, Nitidae, and Racemosae (seed material of the small series Pedicellatae is not yet available for study), and the new genus Za- ROMEOC- ZAPOTECA IMINO AND AMINO ACIDS TABLE 1. Distributi f acids in seeds of species of Zapoteca compared with oe S-CEC = S-(8-carboxyethyl)- x gps d = S- rer -cysteine; 5-2 = d amino acid. п 1 No. Compounds Bentham Species* S- S- Series Examined CEC СІС 8-2 Laetevirentes Zapoteca) 8 = — ++ Macrophyllae 8 dp + T Nitidae 23 +++ + T Racemosae 8 +++ + T +++ = >10 mg/g dry wt; ++ = 1-10 mg/g dry wt; + = 0.1- y у wt; Т = <0.1 mg/g dry wt; — = not detec Е E are listed by name, collection A and n file. ec Hernández, J = Johnson, L = Lott, Martinez, MFR = Forero-Romeo, MS = Sousa = ‚Ё = Robbertse, $ = Shilom, SM = Smith, T= Torres. Zapoteca Z. caracasana J-2494, MO; Z. ee T- sie ; H- ; zie pin J-2246-80, J-1532-80, H-800, H-827, MO; . nov. (1), T-4167, MO; Z. sp. nov. (2) T- :3947, Саша ndra Macrophyllae: C. angustifolia 15051, COL; C. car- bonaria 33509, COL, SM-6471 peo C. glaberrima 55261, 117125, F-9325, COL; C. mexicana 6064, USF; C. rekoi MS- 12545, MO; c. seemannii i 1860709, F; C. C . aff. tergemina L-1684, Nitidae: C. caeciliae 176561 11, F; C. wp на E 607, COL; C. eriophylla H- 191, MO; C. glom ulata 15864, COL; C. haematocephala 2817, USF; € humilis D-1201, MO; C. magdalenae N-23070, H-674, MO; C. matisiana 25785, COL, C. medellinensis F-9415, COL; C. pittieri 66977, COL; C. pubiflora MFR-603, COL; C. purdiei F-9955, COL; C. purpurea F-9920, 05360, COL; C. redacta R-1168, MO; C. re- ticulata 1437, TEX; C. rigida 52044, COL; C. rubes- cens DS-11139, H-762, H-763, J-12273 MO, 1633009 F; C. schultzei F-9419, F-9861, F-9888, F-9889, F-9890, F-9897 COL; C. selloi MFR-538 COL; C. tenuiflora 133081, COL; C. tolimensis 14913, COL; C. turbinata 128406, COL; C. sp. nov. (1), H-867, MO А 7, р. 54, MO: С. juzepezukii Н- 400, МО; С. parviflora 1841137, F; C. rusbyi H-865, M 766 TABLE 2. Distribution of imino acids in leaves and ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 seeds of Zapoteca species. TT (trans-trans), CC (cis-cis), TC (trans-cis), CT (cis-trans)-dihydroxypipecolic acid; C4 (cis-4), T4 (trans-4), C5 (cis-5), T5 (trans-5)-hydroxy- pipecolic acid; PIP— pipecolic acid. Species Col. No. TT CC TC CT C4 T4 C5 T5 PIP Z. caracasana J-2494 ++ ++ + FI Z. formosa T-4262 ++ + ++ q T-4063 ++ + +} de Z. lambertiana S-3405 ++ + + ++ +++ Z. media D-53 ++ + + Z. mollis G-4164 ++ ++ + + Z. portoricensis H-154 ++ + ++ + Н-465 ++ + ++ + Z. tetragona J-2246-80 ++ ++ ++ J-1532-80 ++ ++ ++ H-800 ++ + ++ ++ + H-827 ++ + ++ ++ + Z. sp. nov. 1 T-4167 T + +++ Z. sp. nov. 2 T-3947 ++ + ++ +++ = >1 mg/g dry wt; ++ = 0.1-1 mg/g dry wt; + = «0.1 mg/g dry wt. DISCUSSION poteca, which includes those members of Cal- liandra belonging to Bentham’s series Laetevi- rentes. All species of Calliandra that have been examined contain S-CEC as the major free seed amino acid (Table 1). This compound accounts for as much as 2.5% of the total seed weight. S- CEC has not been detected at all in members of Zapoteca (Table 1). The distribution of a second compound, S-CIC, which is found in much lower concentrations, follows the same pattern. Con- versely, another nonprotein amino acid (S-2), as yet unidentified because of insufficient material but believed to be a sulphur compound, is the major free amino acid of Zapoteca. This com- pound is lacking or present only in trace amounts in Calliandra (Table 1). The distributional data for the imino acids of Zapoteca are shown in Table 2. The presence of large amounts of trans-trans-4,5-dihydroxy- pipecolic acid together with the monohydroxy- pipecolic acids trans- 4- and cis-5- is ther common attern. Th absolute configuration best the 4- and 6e -саг- bons as the monohydroxy isomers with which it most frequently co-occurs. This Zapoteca pat- tern is one of several different combinations of mono and dihydroxy compounds that are seen in Calliandra species (Romeo, 1984 and unpubl. data). Of the other three dihydroxypipecolic acid isomers that may characterize some sections of Calliandra, only the cis-cis isomer was detected in significant concentrations in one species of Zapoteca. Nonprotein amino acids are useful as chemical systematic markers in legumes (see Bell, 1971, and Harborne & Turner, 1984 for reviews). Some taxa have been characterized by the accumula- tion of a single unusual amino acid. For example phenylalanine (Bell & Janzen, 1971), and 5-hy- droxy-L-tryptophan is found only in species of Griffonia (Bell et al., 1976). More often, however, it is an association of nonprotein amino acids that characterizes a taxon. Biochemical subge- neric groupings have been established for Lath- yrus and Vicia (Bell, 1971), and Acacia (Sene- viratne & Fowden, 1968; Evans et al., 1977). Recently, a reassessment of four major genera (Derris, Tephrosia, Lonchocarpus, and Millettia) of the Tephrosieae (Fabaceae) was suggested by studies of the distribution of basic nonprotein amino acids in this group (Evans et al., 1985). The significance of the findings reported herein is difficult to evaluate for several reasons: for sulphur compounds (8 of 18 taxa of Zapoteca, 39 of the ca. 200 species of Calliandra). Ben- tham's four major series are well represented in this group, however, and many species are rep- resented by more than one sample (Table 1); 2) some ch ways of the secondary compounds are not yet known; 3) relationships within the large genus Calliandra are still unclear, and a suitable mor- 1986] phological framework upon which to base chem- ical interpretation does not exist. Nonetheless, tentative judgements can be made. The sulphur amino acid patterns are distinctive enough that, on the basis of these data alone, one can assign species to one or the other group with a reason- able amount of certainty. When considered to- gether with the a data (Guinet, 1965, 1 li i N 1983) and the mh oed. cytological, eco- logical, and developmental findings (Hernández, 1986), these chemical data can be viewed as sup- port for the segregation of the Zapoteca species from Calliandra. The imino acid pattern of Zapoteca, in con- trast, while generally consistent within the genus, is not a particularly distinctive one vis à vis Cal- liandra. Imino acid patterns observed in Cal- liandra consist primarily of the absence or pres- ence of one or more dihydroxypipecolic acid isomers in association with the monohydroxy- lated compounds that are the probable precur- sors of those isomers. All four possible dihydroxy compounds have discontinuous distributions within Calliandra. The usual trans-trans, trans- 4, cis-5 pattern of Zapoteca is common in Cal- liandra and is found in several Latin American species. This sharing of rare imino compounds emphasizes the close phyletic relationship of Za- poteca to Calliandra just as the differing sulphur amino acids emphasize their distinctness. Since the distribution of these compounds is not known from other related genera (Inga, Lysiloma, Pithe- cellobium), it is not yet possible to speculate about the apomorphic or plesiomorphic nature of the chemical characters. As the metabolic relation- ships of the various compounds become eluci- dated, and as interspecific relationships within Calliandra itself become better understood with systematic revision, we can expect new phylo- genetic insight into the relationship of Zapoteca to Calliandra. LITERATURE CITED BELL, E. A. 197 discard biochemistry of non- protein amino acids. de 79-206 in J. B. Har- borne, D. Boulter & B deis urner (editors), Che- hue ii ofthe Leguminosae. Academic Press, w Yor & D. H. JANZEN. 1971. Medical and ecolog- ical considerations of L-DOPA and 5-HTP in seeds. Nature 229: е WS & I M. QUERESHI. 5- hydro dig с рһа келеи del in Griffonia. и 15: 823. , 1976. and ROMEO— ZAPOTECA IMINO AND AMINO ACIDS 767 Берра С. 1875. Revision of the suborder Mi- oseae. w^ Linn. Soc. London 30: 335-664. sure A. Romeo. 1981. 2,4-Trans- 4,5- sie 5- dihydroxypipecolic acid and cis-5- hydroxypipecolic acid from leaves of Calliandra angustifolia and sap of C. confusa. Phytochemistry 20: 1845-1846 & . 1983. 2,4-Cis-4,5-cis-4,5-dihy- droxypipecolic acid—a naturally occurring imino acid from Calliandra йге. Bl dome: 22: dp Evans, C. S., M. Y. QURESHI & E. A. BELL. 1977. Free amino acids in the seeds of Pie species. Phytochemistry 16: 565-570. Evans, S. V., L. E. FeLLows & E. A. BELL. 1985. Distribution and systematic significance of basic nonprotein amino acids and amines in the Te- phrosieae. Biochem. Syst. Ecol. 13: 271-302. FORERO, E. 84. Revision of arg T Ta disciplinary approach. Bull. IGSM 12: GuINET, P. 1965. Etude des characters du e dans le genre Calliandra (Mimosaceae). Pollen et Spores 7: 157-1 Le Mimosacées. Etude de palynologie fundamentale, corrélations évolution. Trav. Sect. Tech. Inst. Fr. Pondichery 9: 1-293. B pS B. L. TURNER. 1984. Plant Che- rk. I m g ÁN Neot ropical Mimosoideae. Ann. Missouri Bot. Gard. 73: 755- ,G. J. & H. REINBOTHE. 1970. Die Amino- säuren der Gattung Albizzia. Biochem. Physiol. Pflanzen Bd. 161: 243-265. MARLIER, M., G. A. DARDENNE & J. CASIMIR. 1972. 4,5- Dihydroxy- L-pipécolique a partir de Callian- dra haematocéphala. Phytochemistry 11: 2597- 599. —— —, & ————. 1979. 2S,4R-Carboxy-2- acétylamino- pipéridine dans les feuilles de Cal- liandra haematocéphala. Phytochemistry 18: 479- NEVLING, L. I. & T. ELIAS. 1970. и к d ic impli Supp J. Bot., § P 59: 753 NIEZGODA, C. J., S. M. FEUER & L. I. NEVLING. 1983. Pollen ultrastructure of the tribe Ingeae (Mimo- soideae). Amer. J. Bot. 70: 650—667. iu. J. T. 1984. Preliminary chemotaxonomic in- vestigations of Colombia Calliandra species based on nonprotein imino acids. Bull. IGSM 12: 19- 23. Я n & А. B. ре 1983. ay pa hydroxypipecolic acid and 2,4-cis-4,5-trans-4,5- dihydroxypipecolic acid from Calliandra. Phyto. chemistry 22: 1615-1617 . ROMEO. 1986. Persistence of the te in seed amino acid S-(6-carboxyethyl)- cysteine in young leaves of Calliandra rubescens: ecological implications. J. Chem. Ecol. 12: 2089- SENEVIRATNE, А. S. & Г. FOwDEN. 1968. The amino acids of the genus Acacia. Phytochemistry 7: 1039- 45. A COMPARATIVE STUDY OF THE EMBRYOLOGY OF LUDWIGIA (ONAGRACEAE): CHARACTERISTICS, VARIATION, AND RELATIONSHIPS! HIROSHI TOBE? AND PETER Н. RAVEN? ABSTRACT Based ono our ti f 1 q м ї 1 ls чы кү un to 40 characters, t 1 {һе genus L udwigia (Onagrac a cece line separate from all other Onagraceae, so it occupie of relationships within the family and between Onagraceae and атра families of M ake position in in considerations he typical distinctive 4- е Oenothera- -type embryo sac was present in all m , studied displaying no s includi distinctive feature ubicuitous occurrence of starch embryological features that we examined in Ludwigia, seed coat an maturity. This character cannot therefore be used to demonstrate credits io between ‘these families, as suggested by earlier embryologists. Two derived embry by Onagraceae and Lythraceae, however, and might indicate ор grains in the nucellus; and (2) tr еи na ns. cal features that are shared between them are (1) e st ger tp in y be mo considering о within the genus. For example, the specialized ве structure of sect. Dantia is found only in sections are closely related. Despite the fact that it clearly belongs in the order Myrtales, Onagraceae is very distinctive within that order (Cronquist, 1981; Dahlgren & Thorne, 1984; Johnson & Briggs, 1984). One of the most distinctive aspects of the family is its embryology, and in particular its unique 4-nu- cleate Oenothera type embryo sac (Seshavata- ram, 1970; Raven, 1979; Tobe & Raven, 1983). Within Onagraceae, the genus Ludwigia, con- sisting of about 80 species found mainly in the tropics and in temperate North America, is the only genus of tribe Jussiaeeae (Raven, 1979). Ludwigia has been studied intensively from var- ious points of view, partly because it has been considered to be a primitive group, closest to a prototype of Onagraceae (see Parmentier, 1897; Melchior, 1964; Takhtajan, 1980). Earlier stud- ies of Ludwigia have been concerned with bio- systematics (Duke, 1955; Schmidt, 1967; Ra- mamoorthy & Zardini, 1987; Peng, 1982); chromosomal observations (Gregory & Klein, 1960; Kurabayashi et al., 1962; Raven & Tai, 1979): pollen morphology (Skvarla et al., 1975, 1976, 1978); reproductive morphology and anat- omy (Eyde, 1977, 1978, 1981); and leaf anatomy (Keating, 1982). These works as a whole have ! Grants to Raven from the Ww acknowledged. We are also grateful to din C.H W. Dolan for the collection of the ? Biologica ratory, O; Labo 3 Missouri Botanical Garden, P.O. Box 299, St. ANN. MISSOURI BOT. GARD. 73: 768-787. 1986. n members of sect. Microcarpium, thus supporting the hypothesis that the two indicated both the intersectional diversity of Ludwigia and its distinctiveness as a genus. Eyde (1981) provided a comprehensive review of the interrelationships of the sections of Lud- wigia. He suggested, based on his analysis of its characteristics, that Ludwigia is the sister group of all other Onagraceae, a concept that is well supported by evidence drawn from many differ- ent lines of investigation. The present study is concerned with the em- bryological characters of the genus; i.e., the de- velopment of anthers, ovules, embryos, seeds, and gametophytes. All of these characters have represent the range of variation in the genus, have been examined in detail. This study is in- tended to illuminate the pattern of variation in embryological characteristics within the genera, and to provide a sound basis for comparisons with other members of the family Onagraceae and the order Myrtales. REVIEW OF EARLIER EMBRYOLOGICAL STUDIES Anthers and microspores. Seshavataram (1967, 1970) described these features in Lud- U.S. National Science Foundation, most recently BSR-8214879, are gratefully och, Ching-I Peng, Barbar nb Briggs, Rebecca R. Sharitz, and Yoshida Colles: Kyoto University, Kyoto 606, ee Louis, Missouri 63166. 1986] wigia octovalvis (Jacq.) Raven. The anther wall is six-layered and consists of the epidermis, fi- brous endothecium, three middle layers, and the tapetum; the tapetum is glandular, and its cells are two-nucleate; cytokinesis in the microspore mother cells is simultaneous. No other details were described. Ovule and megagametophyte forma- tion. Tackholm (1915) reported the presence of the 4-nucleate Oenothera type embryo sac in Ludwigia octovalvis (Jacq.) Raven (= “Jussieua cfr. villosa Lam.,” “J. cf. suffruticosa DC.””). His report came anly a few years after Geerts (1908) first detected this unique type in Oenothera gla- zioviana Mart. (= O. lamarckiana de Vries). Subsequent papers dealt with L. peploides (HBK. ) Raven (= “Jussieua repens”; Ishikawa, 1918); L adscendens (L.) Hara (= “Jussieua repens"; Ma- heshwari & Gupta, 1934; Khan, 1942); L. epi- lobioides Maxim. (= “L. prostrata"; Ishikawa, 1918), L. perennis L. (= “L. parviflora”, Ma- heshwari & Gupta, 1934); and again with L. oc- tovalvis (Seshavataram, 1967, 1970). Taken to- gether, these papers indicate that in Ludwigia the ovule in anatropous, bitegmic, and crassinucel- late; the micropyle is formed by both integu- ments; the archesporium is one-celled (in most reports) or multi-celled (Khan, 1942); an arche- sporial cell cuts off the primary parietal cell; the tetrads of megaspores are linear; the micropylar megaspore in the tetrad is functional and devel- ops into a 4-nucleate Oenothera type embryo sac; and the chalazal megaspore may also develop to a certain degree, resulting in an additional ru- dimentary embryo sac in an ovule. Fertilization and endosperm. Porogamous fertilization and Nuclear endosperm formation were reported in Ludwigia octovalvis (= “‘Jus- sieua cfr. villosa"; Tackholm, 1915; Seshavata- ram, 1967, 1970) and L. adscendens (= “Jussieua repens"; Khan, 1942). Tackholm (1915) stated that the free endosperm nuclei became cellular only at the micropylar region, whereas Sesha- vataram (1970) observed that the endosperm be- came absolutely cellular. In this connection, Jo- hansen (1931: 23) implied that Ludwigia, which e considered a primitive genus, had a dense and distinctly cellular endosperm as a heritage from its ancestor, whereas Epilobium (= “ Zauschner- ia"), an advanced genus, had a coenocytic en- dosperm, or no endosperm at all at maturity. Hypostase. Tackholm (1915) first simply de- scribed that a hypostase was differentiated in Ludwigia octovalvis (= “Jussieua cfr. villosa"). TOBE & RAVEN—LUDWIGIA EMBRYOLOGY 769 A later paper that was concerned with L. pe- ploides (= “Jussieua repens”) and L. epilobioides (= “L. prostrata"; Ishikawa, 1918) reported that a hypostase was absent. Later, Johansen (1928a, 1928b) declared that there was no hypostase in Ludwigia, based on the observations of a Lud- wigia that he called L. mullertii. He contrasted its absence in Ludwigia with its presence in such genera as Oenothera, Gaura, and Clarkia. Jo- hansen explained the function of the hypostase (and the epistase) as follows: “They serve to sta- bilize the water balance of the resting seed over the long period of dormancy during the hot dry season." Maheshwari and Gupta (1934) did not observe the hypostase in L. perennis L. (= “L parviflora”) and L. adscendens (L.) Hara (= “Jus- sieua repens") either. In contrast, Khan (1942) reported that in L. adscendens (= “Jussieua re- hypostase was formed subsequently. Seshava- taram (1967) confirmed the presence of a hy- postase in the older stages of development of the ovule of L. octovalvis. Starch grains in the nucellus. Ishikawa (1918) prostrata") and few in L. adscendens, in contrast with their abundance in Oenothera, Gaura, and Circaea. Because of his belief that Ludwigia lacked a hypostase, Ishikawa inferred that its presence was correlated with a lack of starch grains in the nucellus. Embryogeny. Only the Onagrad type of em- bryogeny has been recorded in Ludwigia. Re- orts include those of Souéges (1935) for L. pa- lustris (L.) Elliott, Khan (1942) for L. adscendens; and Seshavataram (1967, 1970) for L. octovalvis. Seed coat. Corner (1976) first gave general descriptions of the seed coat histology of Lud- i dotesta composed of crystal cells that are lignified and stellately lobed with thickened inner and ra- dial walls. The tegmen is also two-layered with the fibrous exotegmen. Eyde (1978) provided general descriptions of the seed coat histology of species of seven sections of Ludwigia, including L. peruviana. This review ofthe available literature indicates clearly that, although quite a few reports on the embryology of Ludwigia have been published, 770 TABLE 1. ANNALS OF THE MISSOURI BOTANICAL GARDEN Vouchers of the Ludwigia species used. Species Vouchers Sect. Myrtocarpus L. peruviana (L.) Hara Sect. Macrocarpon L. lagunae (Mo- rong) Hara L. bonariensis (Michx.) Hara Sect. Ludwigia L. maritima Harper L. virgata Michx Sect. Microcarpium L. о El- L. id Walter subsp. glandulosa Sect. Dantia L. repens Fors- er L. arcuata Wal- Sect. Seminuda L. leptocarpa (Nutt.) Hara Sect. Oligospermum L. peploides (Kunth) Ra- ven Australia. Mascot District, Sydney, B. G. Briggs 7143, 7226 (NSW) Paraguay. Asuncion, Luque, 7. Ramamoorthy 1019 (MO). Paraguay. Caaguazu, 7. P. Ra- mamoorthy 1097 (MO). Argentina. Buenos Aires, 7. P. Ramamoorthy 1005 i odd Argentina. Tucumán Ramamoorthy 1015 ae USA. South Carolina, Colleton Co., C. Peng 3920 (MO). USA. Alabama, Mobile Co., H. Tobe s.n. (MO). USA. Georgia, McIntosh Co., C. Peng 4139 (MO). USA. Louisiana, Cameron Parish, C. Peng 4367 (MO). USA. Florida, Lee Co., C. Peng 4290 (MO). USA. Florida, Hillsborough Co., C. Peng 4320 (MO). Brazil. Bentos, Santa Catarina, T. P. Ramamoorthy 1134 (MO). USA. Georgia, Barnwell Co., R. W. Dolan 1 (MO). USA. Missouri, St. Louis Co., H. Tobe s.n. (MO). there still remain several gaps in the available information. Also, the discrepancies in literature regarding the ovular archesporium, the endo- sperm, and the hypostase remain to be resolved. MATERIALS AND METHODS The 11 species examined in this study are list- ed in Table 1, together with voucher informa- TABLE 2. A summary of embryological data on anthers and microspores. Thickness of Anther Mature Microspore Delimitation of Endo- thecium Tetrads! Pollen Microspores Tapetum opment Epidermis Wall Taxa Sect. Myrtocarpus 2-celled usually tetrahedral/ persistent fibrous glandular 2-nucleate simultaneous Basic 5-6 layers L. peruviana very rarely isobila- teral [Vor. 73 TOBE & RAVEN—LUDWIGIA EMBRYOLOGY 771 1986] (2861) рїшцә$ моцој AouoanbaJj 10} suoissa1dxq 'spe1191 pojoajas 06 uo poseg | P2][99-Z — — әз®әүәпи-г Je[npueys snolqy 109151519 ISP SI9ÁB] 9—6 sapiojdad "T wnia4adsosi]o “1998 9j?ssno -эр А|]элелиетрэЧ P2][99-Z -21J9] sABMe 4үіеәи snoouej[nuiis 9jeo[onu-c Te[npue[3 snoJqg 1091515194 21529 SJÁL] 6 01020112] "T DPNUIWAY “1998 P2][99-Z — snoouej[nuiis әзеәүәпи-г Te[npue[3 snoJqy juojsis1od 21529 SIAR] 9-6 DIDNIAD "] ILS -snoop Á[[euorseooo P2][99-Z /тетрәцедә1 A[[ensn зпоэцепии$ 9jeopnu-z Iejnpue[g — snolqy juajsisJod ISP SIDAP] © suada4 `7 DIJUD( “1993S ojessnoop K[[£UOIS?220 ÁJIA P2][99-7 /erpoyess) 4епѕп snooue}[Nus 9jeopnu-z Je[npue[8 — snoiqy 109181819 21529 SIOART С psojnpunjs «dsqns ю$о[прир]8 "T NLS -snoap Á[[LUOISBIIO РэПээ-с /erpoyens} Aj[ensn snooauej[nuiis 9jeo»ponu-z Ie[npue[. поду 109151519 21529 SIOAPT Ç DID]O9IUD] "T wnid4p20421JA. “1928 рәцәә-г — snoouej[nuiis ojeoponu-cz Iejnpue[g. — snoIqy 309151519 21529 SIJÁB] 9-S DIDSAIA "] P2][199-7 — — — — — — — — ошілрш v DISIMPNT 159$ P2][99-Z — snoouej[nuiis 3JBIMNU-Z Je[npue[8 snoiqy 09151519 21529 SIAR] 9-6 Si$u214DuOq "T [&1o1e[IqOST Á]9Ie1/91essnoop [BUOISBIDO Алэл РэПээ-с [payen Аепѕп — 3JPINU-Z Je[npuve[3 snoJquy 1091515194 21529 SIDA] © әрип80| “T UOdADIOLIDJY “1998 иәоа 15репәр soJodsoJorJA e) umnjede] штпоәҷ sruuopidq јчәшао IHEM ехе 9Jn]eJA 910dsoJotJA[ jo uonejrurmqoq deL -ория тәләф зЈәцјиу ЈО ul JO зам пем Ssouyory L JoquinN respon yy Iogyuy 1129 JO ЭЧАТ, 'penuguo) “7 ATV IL FA tion. The samples of flower buds and fruits in various stages of development were fixed with FAA (5 parts stock formalin; 5 parts glacial acetic acid; 90 parts 70% ethanol). Subsequently, they were dehydrated through a t-butyl alcohol series, and then embedded in Paraplast with melting point 57-58°С for microtome sectioning. Serial sections cut 5-7 um in thickness were stained mounted in Histocla Observations of microtome sections were made with a Zeiss Standard microscope equipped with a phase-contrast condenser. The photomicro- graphs were mostly taken using an Iris dia- phragm for bright-field observations. In some, annular stops were used for the phase-contrast observations in order to show the location of starch grains in the nucellus (e.g., Figs. 8—12, 15, 16) RESULTS Embryological data concerning all species studied are summarized and compared in Tables —6. A discussion of these characteristics is pre- sented in the following pages. N ANTHERS AND MICROSPORES (TABLE 2) The thickness of anther wall varies from five to six cell layers (Fig. 1). There is no conspicuous difference in thickness from species to species, nor from section to section. The anther wall is composed of an epidermis, an endothecium, two or three middle layers, and a tapetum (Fig. 1). Following the definitions of types by Davis (1966: 8-11), the wall formation of all the species stud- ied conforms to the Basic type. In it, the middle layers have a common histogenetic origin with both the endothecium and the tapetum (Fig. 1). As the anther develops, the middle layers are completely crushed (Fig. 2). The epidermis is ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 persistent and remains uncrushed until the time of anther dehiscence; in some species such as Ludwigia leptocarpa (Fig. 3) and L. peploides, the epidermal cells are more or less flattened, but in most other species, they are enlarged as much as the endothecial cells (Fig. 4). The endothecium always develops fibrous thickenings, as shown by Eyde (1977). The Tn is glandular and its cells become two-nucleat Meiosis in the microspore a cells is ac- companied by simultaneous cytokinesis (see ar- row in Fig. 2), as recorded in L. octovalvis (Se- shavataram, 1967, 1970). The arrangement of microspores in a tetrad is usually tetrahedral, occasionally decussate, and rarely isobilateral throughout the genus; there is no conspicuous difference in the frequencies of the different types of arrangement from species to species. The pol- len grains are two-celled when shed (Fig. 5). OVULES (TABLE 3) The characteristics of the ovule are remarkably consistent throughout the genus: the ovule is anatro р pao: pa bitegmic. The inner integument 1 t yered. The out- er ion is two-layered in zii species but is three-layered in its basal portion in Ludwigia peploides (sect. Oligospermum). The distinctive mode of initiation of the integuments in L. pe- ploides is discussed in a separate paper (Tobe & Raven, 1985). The micropyle is always formed by both integuments. MEGAGAMETOPHYTE FORMATION AND NUCELLUS (TABLE 4 The archesporium is uniformly one-celled (Fig. 6), contrary to a report by Khan (1942), who wrote that it looked multicellular in Ludwigia adscendens. Very rarely, more than one mega- spore mother cell (archesporial cell) is found in RES 1-5. Anthe — ers and microspores of Ludwigia.—1. L. leptocarpa. Transverse section (TS) of a du FiG xm The wall is S-layered, Werde sac of аклан ср, {нн кие, о two porns layers (ml), and tape m.—2. L. bon (t) (mc = microspore mother cell). Sca middle layers and 2-nucleate glandular tapetal i т epidermis x bean (et). A cell at telophase of meiosis II. Sca 3. L. lep of a flattened persistent ene (sp) Hey a s fibrous e ps Le (et). Scale = 100 um.— S of an older anther epic collapsed arrow indicates isced anther. Шы "wall i is com carpa. TS of a deh pos 4. L. linearis e 420, MO). TS of a dehisced anther. The wall is Den eins of well-developed persistent epidermis (ep) a fibrous endothecium (et). Scale = 100 um.— 5. L. lept the normal tetrad formation found ocarpa. Two-celled pollen grain artificially separated mda at dehiscence; stained vegetative nucleus and a smaller generative one. Scale — with 196 aceto-carmine. Arrows indicate a large 50 um TOBE & RAVEN—LUDWIGIA EMBRYOLOGY 713 1986] ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 774 juosoud IO pue п POJIÁBIA EZ POJSÁBI-Z c 9jej[»onuisse15 snodojeue sapiojdad "T шишаэ4$о08 О “1998 juosoud IO pue IT PaJSÁBI-Z POJ9ÁBI-Z c 9je[[oonuisse1o snodoujeue 011020112] `7 ррпитшәѕ 1298 quasoid IO pue п POJSÁBI-Z POJIÁBI-Z c 9je[oonuisseJo snodomeue DIDNIAD "] quosoid IO pue п PosaRI-Z Po19AP]-7 с IJP[[SINUISSPII snodomeue suadad "T DIJUD “1939S jussoid IO pue п P9J9ÁBI-Z P9J9ÁBI-Z C 91e[[oonursseJo snodoijeue рѕојприо8 «dsqns psojnpunjs `T juosoid IO DU? п рэлэА®1-с pouoKe[-c C 9je[[oonursse1o snodojeue DIDJOIIUY] "] wnid4p20421Jy. 1998 1uosaud IO pue IT рәләАр]-г рәләАр[-г с 9jej»onuisse1o snodojeue DIDSAIA "T — — = — T — — puio 7] D181Mpn'T “1998 quasoid IO pue п POJSÁBI-Z P9J9ÁB]-Z c 91e[[oonuisse1o snodomeue SISU214DUOQ "T juosoud IO pU? п P9JIÁBI-Z P9JI9ÁBI-Z ré 9je[[oonursseJo snodojeue ovun3o] `1] и0@41020420 И! “1998 1цә$әла IO pue п P9J9ÁBI-Z рәләАр]-г C 9je[[oonursseJo зподоцеце рирлапаэа 7] SNAIDIOUAPY “1998 SNIJOONN әчз o[AdouorjA IO JO ssouyory р Ir Jo SSIUADIY | е SNJNN JO 2unjeN 2dÁ] IMAO exe UI SURIN jo uon e?uno -ngoju[ Ҷ21е3$ jo Joquinyy "1uoum$ojur 19jno “о зиәшпЗә1ш JOUUT “п :suonerAa1qqy ‘зэтло uo еер [eorSojoA1quio Jo Aewwns y e aay L TOBE & RAVEN— LUDWIGIA EMBRYOLOGY 775 1986] [epiosdiq[o t D421[]0u2() Je[AdoJorui Ieouly] pauoKe[ 9-6 I sapiojdad "T wuni42adso8i]o “1998 [eououds t 7421/10u2() Ie[AdoJorui тәй рәләАр| p-€ 1 241020142] 7 юрпиїшә; “1938 тертозаэ t 7421/10u2() Ie[ÁdoJjorui жәй рәләЛАр| S—p 1 DIONIAD "] [ерто$атә t 742410u2() Ie[Ádouorui Іеәш] рәләҝе S—p I suada4 "T DIJUD( 129g [epiosdijo Ӯ 242410u2() Ie[AdoJorui Je2ul[ рэтэАе] 9-p I psojnpunjs «dsqns psompurvja `T терто$аэ t D421/10u2() Je¡AdoJoru Jeoul[ рэлэАе] S—p I 0]0]0әәир] `T wnid4p20421JA. ‘429$ ртоло Ӯ D42110u2() Ie[&doJorui еә рәләЛАр] с-ф I вова "] = = m = = — — puio 7] DIBIMPNT “1998 pioAo t 0421/10u2() Je[AdoJorur Je2ul[ рэлэАе| с-ф I 51$иәмриод 7] PIoAo Ӯ D421[10u9() Je[Ádouorui Jeoul[ рәгәќе 9-p I әрип80] “I моалрәоләэр[}ү “1998 тертозаэ Ӯ 0421/10u2() Te[AdoJorur Ieoui[ рэлэА®| c-p I puviansad 7] snda4p20)A4]A. ‘129$ oes OAIQUIJ aes uoneulo, DES o1odse33JA peno[ onssr] [219224 DICO) exe] әтер јо adeys оќЈашя oAIquiq јо AdÁ] [euorjoun y o1odse33]JA Jo SSIUADIY L jetiods әлеу JO uonisoq jo иәш -эцәгу ш эм -23UBLIY jo J3QUINN JO эашам "uoneuLioj o31udojoure3e3oui uo ejep [eorgo[oÁ1quio jo ќлешшпѕ y ‘p 3TEVL 776 a given nucellus. This condition was not char- acteristic of any species examined. Each arche- sporial cell divides periclinally into two: the up- per primary parietal cell and the lower primary sporogenous cell. The primary parietal cell fur- ther divides periclinally (Fig. 7). Its daughter cells repeatedly divide periclinally and anticlinally, which results in a three- to six-layered parietal tissue being formed above the young embryo sac. The primary sporogenous cell directly develops into a megaspore mother cell (Figs. 7, 8) that later dm meiosis, forming megaspores (Figs 11). enis division in the meiosis of the megaspore mother cell results in the production of a dyad usually composed of a larger micro- pylar and a smaller chalazal cell (Figs. 9, 10). This contrasts sharply with the situation in most angiosperms, in which a dyad consisting of a smaller micropylar cell and a larger chalazal cell is common. The subsequent homotypic division usually occurs in both micropylar and chalazal cells, resulting in the formation ofa linear tetrad of megaspores. Occasionally, it occurs only in the micropylar cell, in which case a linear triad - megaspores is formed. In either a tetrad or a iad, the micropylar megaspore is always the nae cell, and the functional one (Fig. 11). This megaspore enlarges, while the others degenerate (Fig. 12). Occasionally, we have observed an en- larging chalazal megaspore opposite the enlarged micropylar megaspore, a condition that was re- ported repeatedly in earlier works (e.g., Ishikawa, 1918). We have never observed twin mature em- bryo sacs derived from both the micropylar and the chalazal megaspore, however, nor have we ever seen a single mature embryo sac derived from the chalazal megaspore alone The functional megaspore undergoes two suc- cessive nuclear divisions, forming a megaga- metophyte. As has often been indicated in the earlier works, the nucleus of the functional mega- spore moves toward the micropylar side before the first nuclear division; the two nuclei resulting from its first meiotic division remain on the mi- cropylar side (see Fig. 13), with one slightly above the other. In the second nuclear division, the upper nucleus divides horizontally, whereas the lower one divides vertically (Fig. 13). It appears that the two nuclei derived from the upper nu- cleus develop into the two synergids, whereas those derived from the lower nucleus develop into the egg and the polar nucleus, as mentioned ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 by Khan (1942). Thus, megagametophyte for- mation in Ludwigia conforms exactly to the Oe- nothera-type: the organized embryo sac contains the egg, two synergids and one polar nucleus, and lacks any antipodal cells (Figs. 15, 16). Very rare- ly an embryo sac with more than four nuclei is formed when other megaspores, which normally would have degenerated, are released from the nucellus into the embryo sac (Fig. 14). Such an aberrant embryo sac clearly does not indi y similarity to any other type. In Ludwigia, the mature embryo sacs are di- verse in shape, varying from spherical or ovoid to ellipsoidal. There seem to be no significant differences in the shape of the mature embryo sac from one species to another, however. During megasporogenesis and megagameto- genesis, no particular specialization occurs in the nucellar tissue. Apical epidermal cells of a nu- cellus do not divide periclinally. This character is significant because the more or less related members of the Myrtales (e.g., Combretaceae) show remarkable periclinal divisions of the nu- cellar apical epidermal cells to form a nucellar cap (see Venkateswarlu & Rao, 1972). STARCH GRAINS (TABLE 3) The accumulation of starch grains in the nu- cellar tissues of all species of Ludwigia is con- spicuous, contrary to the observation of Ishikawa (1918). In some species—e.g., L. virgata—it is less conspicuous than in others. In the details of this process of starch accumulation, Ludwigia is exactly similar to Epilobium (Rodkiewicz & Bed- nara, 1974). Starch grains first appear on the up- per, or micropylar side, of the megaspore mother cell (Fig. 8). During meiosis I, they are observed both on the micropylar and on the chalazal side, more abundantly on the former (Fig. 9). In the dyad stage, the starch grains of the upper cell are localized on the micropylar side, whereas those of the lower cell are restricted to the chalazal side (Fig. 10). In the tetrad, the functional megaspore, which is micropylar, has starch most abundantly on the micropylar side, while they disappear for the most part from the chalazal megaspore (Fig. 11). The two intervening megaspores have no starch grains t the functional megaspore stage numerous more and more starch grains are accumulated in 1986] TOBE & RAVEN—LUDWIGIA EMBRYOLOGY 777 FIGURES 6-11. Archesporium and megasporogenesis in Ludwigia arcuata. сит sections of young ovules from the archesporial cell stage to the megaspore tetrad stage. Arrows in 8-11 point out the location of plastids synthesizing starch grains (st). Scales = 10 um.-6. Archesporial cell stage. Note that the archesporium (arc) is one-celled.—7. Primary sporogenous cell stage; га parietal cell (p), primary sporogenous cell (5). — . Megaspore mother cell stage; megaspore mother cell (mc). — 9. Telophase of meiosis I.— 10. Megaspore dyad stage.— 11. Megaspore tetrad stage. Note that the micropylar megaspore (c) is larger than the lower three (c), and that starch grains are localized only in the micropylar megaspore 778 ANNALS OF THE MISSOURI BOTANICAL GARDEN (VoL. 73 4 № Je Од FIGURES 12-17. ея and fertilization in Ludwigia. Longitudinal sections of older and fer- tilized ovules.— 12. L. arcuata. Functional megaspore stage; functional megaspore Maid remnant of к iona Scales = 10 um.— 17. L. ri tip ns. Fertilized ovule stage showing a remnant of a pollen tube (pol) penetrating into the micropyle. Scale = 20 u 1986] TOBE & RAVEN—LUDWIGIA EMBRYOLOGY 779 TABLE 5. A summary of embryological data on fertilization, endosperm, embryo, and seed. Type of Endo- Endo- Cellular sperm in Type of Path of sperm ndo- Mature mbry- Taxa Pollen Tube Formation sperm Seed ogeny Hypostase Sect. Myrtocarpus L. peruviana porogamous Nuclear formed absent — present Sect. Macrocarpon L. lagunae porogamous Nuclear formed absent — sent L. bonariensis porogamous Nuclear formed absent Onagrad present Sect. Ludwigia L. maritima porogamous Nuclear formed absent Onagrad present L. virgata porogamous Nuclear — — Sect. Microcarpium L. се porogamous Nuclear formed absent — present L. glandulos subsp. а porogamous Nuclear formed absent Onagrad present Sect. Dantia L. repens porogamous Nuclear formed absent — present L. arcuata poro ous Nuclear formed absent — present Sect. Seminuda L. leptocarpa porogamous Nuclear formed absent — present Sect. Oligospermum L. peploides porogamous Nuclear formed absent Solanad present both the embryo sac and ш nucellar | tissue (Figs. 15, 16). Ultimately, star d abun- dantly even within the cells of the proembryo. FERTILIZATION, ENDOSPERM, AND EMBRYO (TABLE 5) Fertilization in Ludwigia is porogamous. We often observed a remnant of a pollen tube pen- etrating into the micropyle (Fig. 17). Endosperm formation is of the Nuclear type (Fig. 18). During the early, free-nuclear stage, the endosperm nuclei form separate groups on the micropylar and the chalazal sides of the embryo sac, rather than being scattered peripherally (Fig. 18). At this stage, the endosperm nuclei on the chalazal side characteristically have dense cy- toplasm and form a very obvious group (Fig. 18). Wall formation in the free endosperm nuclei al- ways begins in the micropylar group, ultimately reaching the chalazal one. Cellular endosperm probably never fills the embryo sac, because the endosperm is apparently absorbed by the grow- ing embryo more quickly than it is formed. Ma- ture seeds in all Onagraceae completely lack en- dosperm (Fig. 23). n a few species, we observed the details of embryogeny. Ludwigia bonariensis, L. ma- ritima, and L. glandulosa all had the same On- agrad type embryogeny as reported in different species by all earlier workers (Souéges, 1935; Khan, 1942; Seshavataram, 1967, 1970). In L. Onagrad and Solanad types, the basal cell of the two-celled proembryo plays only a minor or no part in the subsequent development of the em- bryo (Figs. 21, 22). The Solanad type differs from the Onagrad type in that the apical cell of the two-celled proembryo divides by a transverse wall (Fig. 20), instead of by a longitudinal wall as in the Onagrad type (Fig. 19). The suspensor is very short, mostly three or four cells long (Figs. 21, 22). The embryo in the mature seed is straight and has two equally developed cotyledons (Fig. 23). 780 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 18-23. Endosperm formation and embryogeny in Ludwigia. — 18. L. maritima. one wenn section (LS) of a young seed showing free endosperm nuclei (fe) and hypostase (hyp). Scale = 50 um.— 19. L. maritima. 1986] HYPOSTASE (TABLE 5) In all of the species that we studied, we were able to confirm that the differentiation of the hypostase “consists of a well defined but irreg- ularly outlined group of thick-walled cells at the chalazal end of the ovule" (Johansen, 1928a; see also Fig. 18). The hypostase was not observed, however, until the two- to four-celled proem- bryonal stage, at the earliest. In this respect, our results agree with those of Khan (1942) and Se- shavataram (1968), who mentioned that the hy- postase is not differentiated in younger ovules. Starch grains appear in early megasporogen- esis, whereas the hypostase cannot be distin- guished until the early proembryonal stages. Consequently, the hypostase cannot, contrary to the views of Ishikawa (1918), play a role in the accumulation of starch grains. SEED COAT HISTOLOGY (TABLE 6) Our comparisons are based on mature seeds that no longer contain endosperm, and on those portions of the seed coat that are formed of both inner and outer integuments; descriptions are based on longitudinal sections of mature seeds. In all of the species that we examined, the seed coat had basically the same histological struc- ture, consisting of a two-layered tegmen and a two-layered testa (see Figs. 25-31; Corner, 1976; Eyde, 1978) As regards the tegmen, all the species had sim- ilar cells in the exotegmen and the endotegmen, respectively. The cells of the exotegmen are char- acteristically elongate and tracheidal, with spiral wall thickenings. In contrast, the cells of the en- dotegmen are also elongate, but more or less col- lapsed. Ontogenetically, the cells of the endoteg- men become elongate and tanniferous even prior to fertilization (Fig. 24). They may play a role in the early development of the seeds. In contrast, there was considerable diversity in the structure of the testa, particularly in its TOBE & RAVEN— LUDWIGIA EMBRYOLOGY 781 thickness, within Ludwigia. The cells of the exo- testa showed differences in cell substance as well as in size. The exotesta of L. bonariensis con- tained abundant tannins from the early stages of development onward (Figs. 24, 26). In most species, the cells of the exotesta were thin-walled, but those of L. peploides were exceptionally lig- nified and thick-walled (Fig. 31). The endotesta was found to be much more specialized than the exotesta. It was consistently lignified and com- osed of crystal cells in every species, and varied considerably in thickness from species to species, and from section to section. The endotestae of L. peruviana (sect. M. yrtocarpus; Fig. 25), L. spermum, Fig. 31) had approximately the same thickness throughout an individual seed. The en- dotestae of L. repens (Fig. 29) and L. arcuata (sect. Dantia) and those of L. glandulosa (sect. Microcarpium, Fig. 28) differed markedly in that they showed a difference in thickness from part to part: the middle part was thickest (i.e., about 38-44 um thick), and the portions close to both ends were thinnest (i.e., about 8-17 um thick). DISCUSSION Characteristics of Ludwigia. This study has presented a considerable amount of new embry- ological data on Ludwigia and also critically re- viewed the existing literature on the genus. In the light of this information, the embryological characteristics of Ludwigia may be summarize as follows: Anther wall 5- to 6-layered; wall formation conforming to the Basic type; anther epidermis persistent; endothecium fibrous; two to three middle layers ephemeral; tapetum glandular, and its cell 2-nucleate. Cytokinesis in microspore mother cells simultaneous; microspores in a tet- а nes section of a three-celled proembryo; apical cell of the two-celled proembryo and its derivatives C was divided by a transverse wall, and not by a longitudinal o —21 and 2 b). Note that the apical cell was divided by a transverse wall. 2. L. arcuata. Longitudinal sections of о older globular proembryos. Derivatives of Ed basal cell (cb) at the two-celled proembryo contribut ort suspensor. Endosperm (e) becomes cellular. Scales = 10 џи te only - a minor part of the embryo and mostly form 3. L. repens. LS of a mature seed showing —2 a р unte dictyledonous zu (em), which occupies the ET space of the embryo sac. Note that endosperm is absent. Scale = 100 um [VoL. 73 ouou snoJa3ji[[Te1s A15 [eproyoen ouou € 9T'p 90I+L (Oc Sapiojdad "T wniu42dso31]() “1998 эцоц snoJ3jr[[e1s 15 јертәцэвд 3U0U UIZ0'61 L VEZ EZ C 011020)02] 7 DPNUIWAS 129g z [071—8] ^ әиои ЅПОЈӘЈ RISI [eproyoen ouou 6 9I-LTI CCF-0'8€ C suada “7 < опира “1998 4 [691-8 v1] 5 9uou snoJ3jI[[e1s A15 [eprayoen ouou UIC-8 eI rrtt-08€ c psojnpunjs ‘dsqns 7, ю$о]прир]8 "] £ əuou SNOIIJI[eISAIO jeprayorn ouou 1741-88 $'8-5'6 c DIDJOIIUD] "] 8 штп1йлр20421ү ‘159$ = әиои SNOIIjIT[VISAIO [eproyoen ouou 9'01-5'8 L'17-6°91 С ршпырш 7l 9 DISIMPNT 1298 = ouou snoi3jr[[eis A15 [eproyoen əuou €'8-£'9 821-1751 га 518иә11риод "T m T и01020422 W 1296 5 әиои snoJa3jr[[e1s A15 [eproyoen ouou 9°0I-S'8 LT7I-9 Ol C Duvianaod 7] 4 snd4v20144]y “1998 Ё €]$2310X' 8]5310pug иәш8210х9 иәш32] (wn) (wn) (SISAP] uoxe L < -opuq B1S910xq е15ә]0рия ПӘ? jo Joquinu) B1S9] uonezi[eroodg jo adÁT SIZÁBT 1807) JO ssouyory L `зәАе] әці JO иопіоа ѕәишці BY} JO SSIUADIY) ә]еотрш $1әўәвлд ш є1$әзорпә JO SSIUADIY] 10} sa1n3r. p "urojeue 1209 poos uo елер [eorgo[oA1quia jo Атешшп$ у “9 118V L 782 1986] rahedral, decussate or и tetrad; pollen grains 2-celled when Ovule anatropous, pes and biteg- mic; inner integument 2-layered and outer in- tegument also 2- Hem rarely 3-layered (Lud- wigia peploides), micropyle formed by bot integuments. Archesporium one-celled; archesporial cell cutting off a primary parietal cell and forming a parietal tissue three to six cell layers thick; mei- osis in megaspore mother cells resulting usually in a linear tetrad of megaspores, occasionally in a linear triad; micropylar megaspore functional, developing into a 4-nucleate Oenothera type embryo sac, comprising an egg, two synergids, and one polar nucleus; apical nucellar epidermal cells not dividing periclinally; prominent accu- mulation of starch grains in the nucellus com- mon. Fertilization porogamous; endosperm forma- tion of the Nuclear type; free-nuclear endosperm becoming cellular later; seed exalbuminous; em- bryogeny conforming mostly to the Onagrad type, rarely to the Solanad type (Ludwigia peploides); suspensor short; embryo straight and with two equally developed cotyledons; hypostase differ- entiated in proembryonal stages. Seed coat composed of 2-layered tegmen and 2-layered testa; exotegmen tracheidal; endotesta crystalliferous, varying in thickness. Relationships of sections Microcarpium and Dantia. In this discussion, we shall consider only those characteristics that vary within the genus as an index of relationships. As seen in Tables 2-6, however, there are fe characteristics among the embryolosical features of Ludwigia. Furthermore, some of the charac- teristics that do vary within the genus (e.g., thick- ness of the anther wall, thickness of the nucellar parietal tissue, thickness of the seed coat, shape of mature embryo sac, and type of embryogeny) do not fall into distinct classes clearly enough to distinguish species or sections from one another. The most diversified and presumably significant character is the histology of the seed coat, which differs from section to section in the degree of specialization of its constituent layers—particu- larly that of the endotesta. The seed coats of many sections (i.e., Ludwig- ia, Macrocarpon, Myrtocarpus, Oligospermum, and Seminuda) exhibit a condition that we re- gard as generalized; in it, all of the constituent layers are of uniform thickness over the entire seed. We consider structure of this kind to be — е TOBE & RAVEN—LUDWIGIA EMBRYOLOGY 783 basic because of its wide distribution in Lud- wigia. In contrast, sect. Dantia exhibits a more specialized structure, in which the endotesta is much thicker at the median part of the seed than at the two ends; the thickest part of the endotesta in species of sect. Dantia is as much as 38-44 um thick (see Table 6). We examined two of the five species of sect. Dantia, and Eyde (1978) ob- tained the same results for a third. In sect. Mi- crocarpium, one of the two species examined, L. lanceolata, had a seed coat of uniform thickness, whereas another, L. glandulosa, had a structure similar to that of sect. Dantia. We have checked this character in the other species of sect. Mi- crocarpium, and found that some of them (i.e., L. alata, L. linearis, L. pilosa, L. polycarpa, L. ravenii, and L. sphaerocarpa) have a more or less thin uniform endotesta about 4-16 um thick, whereas the others (i.e., L. curtisii, L. linifolia, L. microcarpa, L. simpsonii, and L. stricta) have a thicker endotesta similar to that of sect. Dantia. Earlier students, basing their conclusions on other features, pointed out that sect. Microcar- pium is diverse (see Raven & Tai, 1979). Species differ in seed surface pattern (Eyde, 1978; Peng, 1982), floral vasculature (Eyde, 1981), mode of capsule dehiscence (Peng & Tobe, 1987), his- tological structure of the capsule wall (Peng & Tobe, 1987), and whether the pollen is shed singly or in tetrads (Raven, 1963). Overall, sects. Dantia and Microcarpium appear to be closely related (see Raven, 1963; Eyde, 1978, 1981), and hybrids between species of these two sections are easily obtained and fairly frequent in nature (Schmidt, 1967; Peng, 1982). The evidently de- rived seed coat structure found in six of the 13 species of sect. Microcarpium and at least three of the five species of sect. Dantia also suggests that these groups are closely related; the kind of specialized seed coat found in these groups is unknown elsewhere in the genus. Eyde (1978: 663) implied that the seed coat structure of these two sections differed, stating that "seeds of sec- tions Dantia and Microcarpium are histologi- cally similar to those of section Myrtocarpus, but in Ludwigia palustris (i.e., sect. Dantia) the cells of the crystalliferous layer can differ greatly in size from one part of the seed to another, those in the middle of the seed being much larger than those ofthe ends." We have found, however, that six of the 13 species of sect. Microcarpium are identical to sect. Dantia in seed coat structure. Different species of sect. Microcarpium, although they are clearly related to one another, have both ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 24-31. _Seed coat structure of са Sei 1. bonariensis. Longitudinal section of a young о 100 um. — 25-31. Longitudinal sections of mature 1986] the generalized and the derived types of seed coat structure, a relationship that makes it almost certain that the specialization of the seed coat structure occurred within this group. Thus by the occurrence in sect. Dantia of two synapomor- phies—the specialized seed coat and its entirely opposite leaves—that section is clearly derived in the context of Ludwigia as a whole Relationships with other Onagraceae and Myrtales. In contrast with the traditional view, which holds that Ludwigia is the primitive genus closest to the prototype of Onagraceae and there- fore the best link with Lythraceae and other Myr- tales (see Melchior, 1964; Takhtajan, 1980), re- cent 1 t it inc reasingly clear shal Ludwigia is, in fact, the sister group of all other Onagraceae (Eyde, 1981: 404; see also Ra- ven & Tai, 1979; Eyde, TU 1978). It therefore y line, but would by no means be expected to have a monopoly on primitive features within the family. Nonetheless, a comparison of Ludwigia with other Onagraceae affords us one important way to evaluate the characteristics of their common ancestor. For example, that Ludwigia has the typical distinctive 4-nucleate Oenothera type embryo sac like all other Onagraceae clearly in- dicates that this feature evolved in the ancestor of the family or its derivative before the evolu- tionary line leading to Ludwigia diverged from that leading to the rest of the family Earlier embryologists (Tischler, 1917; Mau- ritzon, 1934; Joshi & Venkateswarlu, 1937) re- а кай suggested that the Oenothera type em- bryo sac might have been derived from the ih Cape e embryo sac of Lythraceae, which like Oenothera lacked antipodal cells at maturity because they were Sphemersl we now know, however, that acteristic of most Myrtales and do not, a support a hypothesis of a direct relationship be- tween Onagraceae and Lythraceae (Tobe & Ra- ven, 1983); moreover, the lack of antipodal cells in mature embryo sacs has a completely different basis in Onagraceae, where they are never formed, from that in other Myrtales, where they are formed and then lost. The comparison is a false one, based on superficial convergence. epresents 9 distinct TOBE & RAVEN—LUDWIGIA EMBRYOLOGY 785 The Oenothera type embryo sac is quite dif- ferent from the normal-type embryo sacs in the following two respects: (1) the micropylar mega- spore in a tetrad, instead of the chalazal one in the case of the normal type, always functions to develop into an embryo sac; (2) the nucleus of the functional megaspore always divides twice, instead of three times, which results in a 4-nu- cleate embryo sac. With respect to a reversed polarity of the functional megaspore, Rodkie- wicz and Bednara (1974) have postulated that the uneven distribution of starch grains and dic- tyosomes in megaspores as well as of callose on megaspore walls may prevent the pisse i of the chalazal megaspore into an embryo sa Starch grains accumulate in the dn megaspore but not to as great an extent as in the chalazal one; dictyosomes occur at a greater den- sity in the chalazal megaspore than in the mi- cropylar one; and callose is laid down on all megaspore walls except for the upper wall of the micropylar megaspore. The kind of embryo sac that is characteristic of Onagraceae, which has only four nuclei, might be regarded as a result of neoteny: functioning at an earlier stage of development than is char- acteristic of most plants. The several substantial differences between the Oenothera type of em- bryo sac and that characteristic of most angio- sperms suggest that the former may have origi- nated as a result of the accumulation of several different mutations. It is probably because of the complex nature of these differences that the Oe- nothera type embryo sac is unknown except in Onagraceae. Judged from its ubiquity in the fam- ily, this unique type of embryo sac must have been present in the common ancestor of the two fundamentally different lines leading, respec- tively, to Ludwigia and to the rest of the family. Onagraceae resemble Lythraceae in a number of features, including wood Vei oig Pe 1975), leaf qiu i: (Hickey, pe „in Dahlgren & Thorne, 1984), seed s ше (Согпег, 1976), aah petal venation pattern (Chrtek, 1969). Our results confirm that these two families share a specialized tracheidal exo- tegmen, which, however, is also found in two unrelated families of Myrtales; i.e., Combreta- <— seed coats, each of which is composed of a two-layered aa and a two-layered tegmen; exotesta (exts); endotesta . L. (ents); exotegmen (extg); endotegmen (entg). Scales = 20 џи . peruviana. bonariensis. —27. maritima.—28. L. glandulosa. — 29. L. repens. — 30. L. ла xd. L. роде. 786 ceae and Trapaceae (Corner, 1976). In Myrtales generally, starch grains accumulate in the nucel- lus only in Onagraceae and Lythraceae. Ishikawa Hubert (1896) in Cuphea (Lythraceae). We here report their ubiquitous occurrence in Ludwigia as well. We are at present studying the embryology of all other 16 genera of Onagraceae as well as of some Lythraceae. Further detailed comparisons will be presented in subsequent papers. Embryology and relationships within Ludwig- ia. Many past studies of embryology have been too limited in scope to warrant the general con- clusions that have been drawn from them. Most have been concerned only with the development of the male and female gametophyte and that of the embryo, and the number of actual observa- tions made have often been too few to warrant the generalizations that have been based on them. Our present study has been based on a sufficient number of species to represent the range of vari- ation of the genus and has included, insofar as possible, all of its embryological features. In the light of this information, it is possible to offer some suggestions about joe E within Ludwigia, based on our observatio Most embryological features in eae did not vary throughout the genus. However, differ- ences in two important features have been noted within the genus, and these are the type of em- bryogeny and the structure of the endotesta. Along with earlier students of Ludwigia embryology, we found that most species we examined had the Onagrad type embryogeny. The only exception was L. peploides (sect. Oligospermum), whic had the Solanad type embryogeny. This finding agrees with other evidence in suggesting an iso- lated position for sect. Oligospermum within the genus, although the earlier report of Onagrad type emb oge ny in L. adscendens (= “Jussieua repens"; Khan, 1942), a species that is very closely related to L. peploides, needs to be investigated. Embryogeny often varies considerably at a fam- ily level (see Davis, 1966: 25-26), and a sufficient number of species need to be investigated before conclusions are drawn about this feature. As we mentioned above, the species of Lud- wigia sect. Dantia and some species of sect. Mi- crocarpium have a more specialized endotestal structure than the rest of sect. Microcarpium an the other species of the genus. Based on the oc- currence of this clearly apomorphic feature, it ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 seems certain not only that the groups are di- rectly related, but that sect. Dantia, with its op- posite leaves—another clearly apomorphic fea- ture, unique for Ludwigia—was derived from an ancestor that would, if it were known, be placed in sect. Microcarpium. Although diverse, sect. Microcarpium seems clearly to consist of ele- ments that are directly related to one another. LITERATURE CITED CARLQUIST, S. 1975. Wood anatomy of Onagraceae, with notes on alternative modes of көрү acme е in dicotyledon woods. Апп. Misso Bot. Gard. 62: 386-424. CHRTEK, J. 1969. Die Kronblattnervatur in der Fam- ilie Lythraceae. Preslia 41: 323- Corner, E. J. H. 1976. The Seeds of Dicotyledons, ridge. sification of Г Flowering Plants. Columbia Univ. ess, New York. DAHLGREN, R. . F. THORNE. 1984. The order М, угїа1е$: circumscription, variation, and relation- 33-699. Systematic Pub of the ngiosperms. John Wiley & Sons, New York. DUKE, J. A. 1955. Distribution and speciation of the genus Ludwigia in North Carolina. J. Elisha Mitchell Sci. Soc. 71: 255-269. EYDE, В.Н. 1977. Reproductive structures and evo- lution in Ludwigia (Onagraceae). I. Androecium, placentation, merism. Ann. Missouri Bot. Gard 5 . 1978. Reproductive structures and evolution in Ludwigia (Onagraceae). II. Fruit and seed. Ann. Missouri Bot. Gard. 65: 656-675. 1981. Reproductive structures and evolution in Ludwigia (Onagraceae). III. Vasculature, nec- taries, conclusions. Ann. Missouri Bot. Gard. 68 470-503. GEERTS, J. M. 1908. Beitráge zur Kenntnis der cy- tologischen Entwicklung von Oenothera MUN iana. Ber. Deutsch. Bot. Ges. 26: 608-6 GREGORY, D. P. & W. M. KLEIN. 1960. и of meiotic chromosomes of six genera in the On- 1. Recherches sur le зас embryon- naire des plantes grasses. Ann. Sci. Nat. Bot. 2: 37- ISHIKAWA, M 18. Studies on the embryo sac and fertilization in Oenothera. Ann. Bot. (London) 32: 7 -317. JOHANSEN, D. A. 1928a. The hypostase: its presence and function in the ovule of the Onagraceae. Proc. Natl. bo ad. U.S. 10-713. 8b. The hypostase and seed sterility in the a Madroño ud 5-167. . 1931. rphology of the On- nus characterized by irregular embryology New York Acad. Sci. 33: 1-26. JOHNSON, L. А. S. & В. G. BRIGGS. 4. Myrtales and Myrtaceae—a phylogenetic analysis. Ann Missouri Bot. Gard. 71: 700-756. 1986] Josui, A. C. & J. VENKATESWARLU. 1936. Embry logical studies in the Lythraceae. II. Proc. Indian Acad. Sci. 3: 377-400. KEATING, В. C. 1982. The evolution and systematics of Onagraceae: leaf anatomy. Ann. Missouri Bot. Gard. 69: 770-803. KHAN, R. 1942. A contribution to the embryology of Jussiaea repens Linn. J. Indian Bot. Soc. 21: 267-282. KURABAYASHI, M., H. Lewis & P. H. RAVEN. 1962. A а study of mitosis in the Onagraceae. Amer. J. Bot. 49: 1003-1026. MEHESHWARI, P. & B. L. Gupta. 1934. The devel- opment of the female gametophyte of Ludwigia Ropero and Jussiaea repens. Current Sci. 3: 107- 108. MAURITZON, J. 1934. Zur Embryologie einiger Lyth- raceen. Acta Horti Góthob. 9: 1-21. MELCHIOR, H. 1964. Myrtiflorae. Pp. 345-366 in H. Melchoir (editor), A. tory r’s Syllabus der Pflan- zenfamilien. 12. Aufl. Bd. 2. Gebriider Borntrager, Berlin-Nilglasse. PARMENTIER, P. 1897. Recherches anatomiques et taxonomiques sur les Nee ray et les Halo- ragacées n. Sci. | PENG, C.-I. 1982. А о, ‘Stady of а sect. Microcarpium (Опаргасеае). а $ег- tation. Washington iia St. Lou obe. 1987. Capsule wall 76 in relation to capsular dehiscence in Ludwigia sect. Microcarpium (Onagraceae). Amer. J. Bot. (i press). RAMAMOORTHY, T. P. & E. M. ZARDINI. 1987. The о and evolution of Ludwigia sect. Myr- u lato (Onagraceae). Monogr. Syst. S The Old World species of Lud- wigia (including Jussiaea), with a vir of the genus dw ceae). Reinwardtia 6: 327-427. ——. 1979. А survey of os ке in бше New Zealand J. Bot. 93. & W. Tar. 1979. к. of chromo- somes in Ludwigia ewm Ann. Missouri Bot. Gard. 66: 862-8 KIEWICZ, B. & J. n ak 1974. Distribution of organelles and starch grains during megaspo- rogenesis in Epilobium. Pp. 89-95 in H. F. Lin- skens (editor), Fertilization in Higher Plants. North- Holland Publishing Company, Amsterdam SCHMID, R. 1982. Descriptors used to indicate abun- ax © Я TOBE & RAVEN—LUDWIGIA EMBRYOLOGY 787 dance and frequency in ecology and systematics. 94 . 1967. A —— study of Lud- wigia sect. Dantia (Onagraceae). Ph.D. Disserta- tion. Stanford University, Stan “fo rd. SESHAVATARAM, V. A contribution to the em- bryology of Ludwigia octovalvis (Jacq.) Raven subsp. sessiliflora (Mich.) Raven. Proc. 54th In- dian Sci. Congr. (Hyderabad) Pt. III Abstr. p. 328- . 1970. Onagraceae. In B. В. Seshachar (editor), Proceedings of the Symposium on Comparison Embryology of Angiosperms. Bull. Indian Natl. Sci. Acad. 41: 220- b , W. F. со & М. 8. Ап i de Baa tudy of viscin threads in Onagraceae pollen. Pollen & Spores 20: 143. — & J. PRAGLOWSKI. The evo- lution of pollen tetrads in Onagraceae. Amer. J. 2: 6-35. , & ——. 1976. Ultrastructural sur- vey of Onagraceae pollen. Pp. 447-479 in І. К. Ferguson & J. Muller (editors), The Evolutionary Significance of the Exine. Academic Press, Lon- don SOUEGES, R. 1935. Embryogénie des Oenothéracées. e l'em- TACKHOLM, G. 1915 menentwicklung einiger Onagraceen. Svensk Bot. Tidskr. 9: 294-361. TAKHTAJAN, A. L. 1980. Outline of the classification of flowering plants POSUER Bot. Rev (Lancaster) 46: 225- TISCHLER, G. 17. e ae Entwicklung und phy- logenetische Bedeutung des Embryosacks von Ly- thrum salicaria. Ber. Deutsch. Bot. Ges. 35: 233- 246. Tose, H. & P. H. RAVEN. 1983. An embryological analysis of ae its definition and UN cter- istics. Ann. Missouri Bot. Gard. 70: 71- 85. The owes ый еуо- lution of Е A “os ceae. Ann. Mis- souri Bot. Gard. 72: 4 d VENKATESWARLU, J. & P. S. es Rao. 1972. Embry- ological geri in some Combretaceae. Bot. Not. 125: 161- KOEHNERIA, A NEW GENUS OF LYTHRACEAE FROM MADAGASCAR! SHIRLEY A. GRAHAM,” HIROSHI TOBE,? AND PIETER BAAS‘ ABSTRACT The new monotypic genus Koehneria is described, based on Pemphis madagascariensis, one of two species of the lythraceous genus Pemphis. It is distinguished from Pemphis acidula and other members of the Lythraceae by a combination of characters including: glandular trichomes, strongly reflexed calyx lobes, doubled episepalous stamens, a conspicuous ovary stipe, elongated inner epidermal cells of the ovary wall, septifragal capsule dehiscence, 3-pseudocolpate pollen, and wood with septate fibers, scanty parenchyma, and erect ray cells. Morphological, palynological, and anatomical comparison of Koehneria is made to the Old World genera Pemphis, Lagerstroemia, Woodfordia, and the New World genera Adenaria and Pehria. The last three genera share with Koehneria a three-character synapo- morphy absent from Pemphis and Lagerstroemia and all other genera of the family. A common ancestral origin is postulated for Koo ria Woodfordia, Adenaria, and Pehria, but subsequent ex- ether wi th forms, limits understanding | of more PCIE inter- -generic relationships. Among the 15 monotypic or ditypic genera of the family Lythraceae is the Old World genus Pemphis, whose two species have long been re- garded as highly disparate (Koehne, 1901). Pem- phis acidula Forster, the type of the genus, is a well-known, widely-distributed species of strand and dune habitats occurring on the east African coast and eastward on islands throughout the Indian and western Pacific Oceans. Pemphis madagascariensis (Baker) Koehne, in contrast, is an endemic of semiarid savannahs of the south- ern half of Madagascar. Recently, new collec- tions of P. madagascariensis became available for study, allowing the question of the species' generic status and its affinities to other members of the family to be reassessed from a more ex- tensive data base. Pemphis madagascariensis was originally de- scribed as a Lagerstroemia from south-central Madagascar in 1881 (Baker, 1881, 1882). At that time, only flowering material, without fruits, was although it was noted that authentic material had not been seen by the author (Koehne, 1883). In 1896, a second species of Pemphis, P. punctata Drake, was described from the same general geo- graphical area as L. madagascariensis and was distinguished from P. acidula by the presence of punctae on the leaves, buds, and ovaries and by the long-exserted stamens and style. Koehne, in monographing the family Lythraceae, remarked that he had yet to see a single flower of Lager- stroemia madagascariensis (18 years after his treatment of Lagerstroemia) and noted in the same work that P. punctata was quite different from P. acidula although they shared the same inflorescence type (Koehne, 1901). In completing the monograph of the Lythraceae, Koehne finally was able to study material of this taxon from s transferred Lagerstroemia madagascariensis to emphis, citing P. punctata in synonymy (Koehne, 1903). The generic position of Pemphis madagascar- iensis has depended exclusively on inflorescence and floral morphological features observed at the turn of the century. This study presents detailed floral morphological and anatomical features, leaf ' Supported in part by grants from the National Science Foundation to Shirley A. Graham and Peter nd P. C. Ho Raven. We are grateful to W. G. D'Arcy, L. J. Dorr, a dry materials for this study; to B. van Heuven for sectioned leaf and wood materials Dorr for valuable comments and information in reviewing the manuscript. We also thank to P. Raven and L. A. Lourteig for providing specimens and extremely helpful information on types at ipd a leaf trichomes and pollen, and the curators of BM, K, реа Н. ch for their help in providing preserved and for P, A. Graham for SEM L, MO, P, and TAN for loan of herbarium т клеш of Biological Sciences, Kent State University, Kent, Ohio 44242. 3 Depart ment of Biology, Faculty of Science, Chiba University , 1-33 Yayoi-cho, Chiba 260, Japan. * Rijksherbarium, P.O. Box 9514, 2300 RA Leiden, The Netherlands, ANN. MISSOURI Bor. GARD. 73: 788-809. 1986. 1986] and wood anatomical characteristics, and pollen morphology for P. madagascariensis, and pro- vides comparison to P. acidula and other puta- tive allies. The unique circumscription of P. madagascariensis described here supports its treatment as a new genus of the Lythraceae. MATERIALS AND METHODS Observations on floral and seed morphology and floral anatomy of Pemphis madagascariensis were based on FAA-preserved material (D'Arcy & Rakotozafy 15317, MO; Dorr et al. 3923, 3933, MO), supplemented by dry materials from her- barium specimens (TAN and MO). Anatomical her vances of flowers were made using micro- d following a standard par- affin: method; scanning electron-microscopic ob- servations were made according to standard techniques using a JSM-25S (JEOL) microscope. he following additional fixed materials were compared: Adenaria floribunda H.B.K., Knapp & Schmalzel 5208 (MO); Pehria compacta (Rus- by) Sprague, Berry 4006, 4028 (VEN); Pemphis acidula Forster, Raulerson s.n. in 1982 (МО); Woodfordia fruticosa Kurz, Dice & Musial s.n. in 1983 (MO), Nasir s.n. in 1983 (MO), and Bird s.n. in 1983 (MO) For leaf anatomical descriptions of Pemphis madagascariensis, material of R. Decary 8940 (L) and L. Bernardi 11181 (L) was studied. For leaf-anatomical comparisons, data scattered in the literature and summarized by Solereder (1899, 1908), Metcalfe and Chalk (1950) and Napp- Zinn (1973, 1974) were complemented with orig- inal observations for Adenaria, Pehria, Lager- stroemia, and Woodfordia in order to contribute to an assessment of the affinities of this new ge- nus. Sections of other genera represented in the Rijksherbarium slide collection (Ginoria, Law- sonia, Lafoensia, and Physocalymma) were also included in the comparison. For wood anatomy, twigs of P. madagascariensis 3 and 4.5 mm in diameter were studied from the collection D'Arcy & Rakotozafy 15317 (MO) and ca. 12 mm di- ameter from Dorr et al. 3923, 3933 (MO). The paper by Baas and Zweypfenning (1979) served as a source of comparative data, complemented by sections of twig material of Pemphis and La- gerstroemia. ight microscope and scanning electron mi- croscope (SEM) comparisons of pollen of Pem- phis madagascariensis were made with pollen of all genera of the Lythraceae utilizing the pollen GRAHAM ET AL.—KOEHNERIA 789 collections of A. Graham, Kent State University. A Cambridge Stereoscan SEM 100 was em- ployed for pollen and leaf trichome photos. FLORAL MORPHOLOGY Flowers are homomorphic, actinomorphic, perfect, and basically 6-merous (Fig. 1). The flo- ral tube is campanulate, and the length from the base to the tip of the calyx lobes is 4.5-8.5 mm. The calyx lobes are six in number, deltoid, and strongly reflexed (Figs. 1, 4). A conspicuous ap- pendage (the epicalyx) is present at the sinus be- tween adjacent lobes and its length varies from 0.0 to 2.0 mm. There are six petals opposite the appendages. These are widely obovate with a short claw, 5.5-10.0 mm long and 4.0-7 mm wide (Fig. 2). The venation ofthe petals is simple, with a thick midvein present in the center, and four or five secondary veins. There are no anas- tomoses between the secondary veins (Fig. 2). The androecial position is perigynous. The sta- mens are commonly 18 in number (occasionally fewer or more) of which 12 (10-14) are opposite the sepals, and 6 opposite the petals. Episepalous stamens are inserted at nearly two-fifths the height of the floral tube, and epipetalous stamens at a slightly higher level (Fig. 3). The episepalous sta- mens are generally paired; in Figure 3 ten of the 11 episepalous stamens are paired, the single re- maining one is indicated by an arrow. The paired stamens seem to have been derived by early split- ting of the stamen meristem (chorisis), an oc- currence common in several Lythraceous genera such as Lagerstroemia, Crenea, Lawsonia, Ne- saed, dee and Ginoria (Tobe et al., in prep.). Rar produced by split- ting ui a filament 1-2 mm above the point of insertion to form two stamens. Filaments of the episepalous stamens are 10-11 mm long, some- what longer than those of the epipetalous sta- mens, which are ni mm long. Exsertion of epi- sepalous and ep equals the length of the floral tube (Fig. 4). The ovary is superior and globose, with a con- spicuous stipe, 1.0-1.2 mm long (Figs. 4, 7). Pres- ence of a comparable ovary stipe is seen else- where in the family only in Adenaria and Pehria. The style is narrow, slightly exceeding the sta- mens, and 7-15 mm long. It tapers to a punc- tiform stigma. The capsule is globose, and the mode of dehiscence is probably septifragal mar- ginicidal. Although sometimes incomplete, de- hiscence occurs from the top of the capsule by 790 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 FIGU Koehneria madagascariensis. —1. Flower.—2. Petal о ike 8% KOH.—3. Transverse section (TS) o ower. Note that, while TSs of filaments of episepalous stam appear as paired small circles, TSs of Neca Bd of Dum stamens do not appear yet at this level. An arrow indicates a non-paired, single episepalous stamen . Lateral view of dissected flower. Abbreviations: ca, calyx lobe; pe, petal; sta, stamen; stl, style; p, а ir to a petal; s, vascular bundle to a stamen; ost, ovary stipe; ov, ovary. Scale equals 5 mm for Figure 1; | mm for Figure 2; 2 mm for Figures 3 and 4 1986] two or three longitudinal lines along the ovary septa to approximately half the length of the cap- sule. The seeds are more or less conical with the chalazal part broadest (Fig. 9). They are small, 1.2-1.3 mm long and 0.9-1.0 mm wide. The seed coat surface is almost smooth, without special- izations such as the inverted epidermal hairs found in a number of lythraceous genera. FLORAL ANATOMY In anatomical section 16 tissue may be seen forming a narrow ring at the base of the floral tube surrounding the ovary (Fig. 7). Cells of the tissue are densely cytoplasmic, making the tissue easily recognizable in thin section because of its high stainability. The number of locules (or constituent carpels) of the ovary is often three (Fig. 3), not always two as has been described (Koehne, 1903). The ovary wall is thin and four to five cells thick with inner epidermal cells of the wall showing re- markable radial enlargement (Fig. 8). TRICHOMES Two types of trichomes occur in the species, one a simple unicellular falcate hair, the other multicellular, globose, and glandular. The uni- cellular hairs form a very dense cover on the abaxial leaf surface (Fig. 12) and a sparser one on the adaxial side. Glands are also more abun- dant and larger on the abaxial than on the adaxial surface. Both types are abundant on leaves, young branches, pedicels, and the outer surface of the floral tube (Figs. 5, 6, 12). Both types also occur on the upper half of the ovary, although the glan- dular hairs are most abundant (Fig. 6; see also Fig. 8). Unicellular trichomes are 90-180 um long; glandular trichomes are 70-130 um in di- ameter. There is no size difference between glan- dular trichomes on the flower surface and those on the ovary surface. LEAF ANATOMY In surface view the leaves have an indumen- tum of unicellular hairs and globular, multicel- lular glands (dark dots as seen with a hand lens) (Figs. 12, 13). The unicellular hairs are proxi- mally broad and slender and thick-walled dis- tally (Fig. 14). The glands on the abaxial surface are 100-130 um in diameter, on the adaxial sur- face 70-90 um. They are sessile, with a very short multiseriate base, and composed of one (locally GRAHAM ET AL.— KOEHNERIA 791 two) layers of cells enclosing a large cavity with more or less granular contents (Fig. 15; the struc- ture of mature glands suggests a lysigenous or schizolysigenous development of these cavities). Epidermal cells have straight to curved anticlinal walls. Stomata are disais to the abaxial sur- face, anomocytic yclocytic, and the guard cell pairs are (1 pipes um wide, (22-)24(-26) um long. In transverse section the lamina is dorsiven- tral, ca. 170-200 um thick; the cuticle is 1 um thick or less. Epidermal cells are almost all with convex periclinal walls and mucilaginous, al- though sometimes with periclinal division walls and then only the inner daughter cell mucilagi- nous. The stomata are slightly raised but well protected by dense hair cover; inner and outer cuticular ledges are fairly well developed. A hy- podermis is absent. The mesophyll is composed of one layer of tall palisade cells and spongy tis- sue. The midrib is provided with a single, bi- collateral vascular bundle, linked through a col- lenchymatous bundle sheath extension to the upper epidermis and by a parenchymatous to collenchymatous ground tissue to the lower epi- dermis, without supporting sclerenchyma. Mi- nor veins are mostly embedded in the mesophyll; vascular bundles of the major lateral veins are vertically transcurrent through parenchymatous to collenchymatous bundle sheath extensions. Crystals are present as druses in the vicinity of vascular bundles. WOOD ANATOMY Growth rings are faint to distinct. Vessels are diffuse or wood is semi-ring-porous in some of the growth increments. Vessels number ca. 150- 200 per sq. mm, ca. 25-4596 are solitary, the remainder in radial ое of 2-4(-7) ог more rarely in small clusters, round to oval, the tan- gential diam. (20-)40(-60) um, the radial diam. up to 70 um, the walls 3-4 um thick. Vessel member length is (180-)360(-420) um. Perfo- rations are simple in oblique to nearly horizontal end walls. Inter-vessel pits are crowded, alter- nate, vestured, round to polygonal, 3-5 um, with slit-like apertures. Vessel-ray pits are similar but half-bordered. The fibers are (370-)580(-770) um long, with walls medium to very thick, some- times weakly gelatinous, with simple to minutely bordered pits confined to the radial walls, sep- tate. Parenchyma was not seen. Rays are mainly uniseriate, partly with low biseriate portions, al- 792 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 1986] GRAHAM ET AL.—KOEHNERIA 793 FIGURES 10-13. Scanning electron micrographs of lower leaf surfaces of "€ acidula [Cramer 4935 (MO)] and Koehneria madagascariensis [D'Arcy € Rakotozafy 15317 (MO)].— Seriaceous lower leaf surface of P. acidula. — 11. Enlargement of unicellular trichomes of P. acidula. — 12. T ubi lowe t leaf surface of K. malana —13. Enlargement of multicellular, globose, glandular tric ве ome, surrounded by unicellular, vet oe eae trichomes of K. madagascariensis. Scale equals 200 um for Figures 10 and 12; 50 um for Figures lla most entirely composed of upright cells, central tals were not observed. Many fibers with dark- cells sometimes square to weakly procumbent staining contents (probably living fibers) are (this ray type would probably develop into het- present (Figs. 16, erogeneous type I in a more mature stem). Crys- Additional stem aa are: Phloem de- <— FIGURES 5-9. Koehneria madagascariensis.—5. Scanning electron шүн of trichomes оп the outer surface of the floral tube. — 6. Scanning electron epic e trichome sober MS of the upper half of the ovary.— 7. Longitudinal section (LS) of flower. — 8. LS of ovary wall.—9. Sc canning electron micrograph of nearly mature seed [see specimens examined, Réserves Naturelles 2639 (TAN). Abbreviations ne, nectariferous tissue; ep, inner epidermal cell. Scale equals 100 um for Figures 5, 6, 8, and 9; 1 mm for Fig 794 Camera lucida drawings of leaf FIGURES 14, 15. trichomes of Koehneria кое, — 14. Uni- cellular trichomes; three t richomes n the left from lower epidermis and one on the right on upper epi- dermis.— 15. Multicellular, glandular trichome from lower leaf surface. Scale equals 20 u void of sclerenchymatous elements; secondary phloem with narrow bands of crystalliferous cells containing druses; interxylary phloem well-de- veloped; pith with nests of thick-walled sclereids POLLEN MORPHOLOGY The pollen is prolate-spheroidal and tricol- porate with three faint pseudocolpi (Figs. 18-20, 24). The colpi are meridionally elongated, equa- torially arranged, equidistant, straight, 10-16 um long, extending to within 3 um of the pole and tapering to an acute apex; costae colpi are nar- r over the pore; pseudocolpi are shorter than the colpi (7-9 um) and represented by thinned me- socolpal regions with increased granulation of the exine (Fig. 19). The poles are sala s d at the midpoint of the colpus, 2.5-3 um lam. The grain size is 22-26 um long (P) x 1620 um wide (E) at the equator. RELATIONSHIP TO PEMPHIS ACIDULA The great difference between the two species currently assigned to Pemphis is apparent in comparative summaries of their reproductive and anatomical characters (Tables 1, 2). The most striking floral features of P. madagascariensis are the distinctive glandular trichomes and conspic- ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 uous ovary stipe, both of which are lacking in P. acidula. In addition, the combination of septif- ragal marginicidal capsule dehiscence, doubled condition of episepalous stamens and bilevel sta- men insertion, strongly reflexed calyx lobes, re- stricted stigma development, ae basal pro- duction of elongated inner epidermal c ee the ovary wall clearly set P. madagascariensis apart from P. acidula. In contrast, P. acidula has circumscissile capsule dehiscence, single stamens inserted at one level, erect calyx lobes, bilobed stigma, nectarif- erous tissue lining the floral tube to the level E stamen insertion in the upper one-third of the ovary that harden in association with its circumscissile dehiscence mode. The morphology of the flower in P. acid- ula is related to its distylic condition (Lewis, 1975; Gill & Kyauka, 1977). The flowers of P. mad- agascariensis are homomorphic. The seeds are similar but much smaller in P. madagascariensis than in P. acidula. The difference in vestiture between the two taxa is another indicator of their disparity. Both are well adapted to xerophytic habitats, but the trichome complements are entirely different. In Pemphis madagascariensis a mixture of sessile, globose, glandular hairs and unicellular, falcate, non-glandular hairs occurs (Figs. 12, 13). In P. acidula the dense seriaceous indument is com- posed entirely of long, straight, thick-walled, uni- cellular hairs ringed by a basal collar of the thick- ened epidermal cell wall (Figs. 10, 11). Pollen differences between the two species are as great as between any two genera in the family. The pollen of Pemphis madagascariensis is tri- colporate with three faint pseudocolpi, one to each mesocolpal area, and the exine is faintly scabrate (Figs. 18-20, 24). The pollen of P. acid- ula is twice as large and is consistently tetracol- porate (ca. 1% tricolporate) with eight faint pseu- docolpi, two to each mesocolpal area. The exine is characteristically psilate to faintly scabrate (Figs. 21-23). From the summary of important anatomical attributes (Table 2), it is clear here again that Pemphis madagascariensis ane P. guida show dosi any n- atomical characterization of P. acidula, de centric xylem parenchyma, and by other ana- tomical features (Baas & Zweypfenning, 1979). Only a few of these can be interpreted as spe- 1986] GRAHAM ET AL.—KOEHNERIA FIGURES 16, did 8. зу ра кош of yc ipl madagascariensis. — 16. Transverse section of stem 5.—17. showing cork, cortex, phloem and secondary xylem adial ee TAS section through the sec- 540. 3 ondary xylem eund septate fibers "eft and narrow he ©) elements (right), cializations related to its xerophytic habit (thick isobilateral gg with stomata on both sides; cf. Kienholz, 1926). Vegetative anatomy un- ambiguously адаш separation of the two taxa. Features shared by Pemphis madagascariensis and P. acidula, such as actinomorphic flowers and diplostemonous stamens with long episep- alous filaments, are common to most genera of the Lythraceae. The generalized nature of these characters does not negate their inclusion in this comparison but does invalidate their usefulness as indicators of phylogenetic relationship. All evidence suggests there is no direct relationship between P. madagascariensis and P. acidula. Support for removal of the former from the genus Pemphis is well founded on all grounds inves- tigated. The new generic name, Koehneria, is henceforth used in this discussion in place of Pemphis madagascariensis. RELATIONSHIP TO LAGERSTROEMIA The original disposition of Koehneria as a species of Lagerstroemia appears to have been made through ignorance of the definitive char- acters of Lagerstroemia or incomplete knowl- edge of the fruiting condition of Koehneria. La- gerstroemia uniquely has unilaterally winged seeds with a pyramidal seed body and revolute cotyledons. In addition, its inflorescences are multi-flowered racemes or panicles, the ovary is sessile, and the capsule is loculicidally dehiscent. In Koehneria the seeds lack wings and have straight cotyledons, the flowers are mostly soli- tary and axillary, the ovary is stipitate, and the capsule is septifragally dehiscent. Metcalfe and Chalk (1950) recorded glandular trichomes for Lagerstroemia, but this is apparently based on L. madagascariensis. Furtado and Srisuko (1969) 796 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 1986] loosely applied the terms **dotted" and “‘gland- otted" to the leaves of a number of Lagerstroe- mia species, but glands were not found on her- barium specimens at Leiden of some of those species. Both Koehne (1903) and Furtado and Srisuko (1969) indicated that L. ovalifolia Teysm. & Binnend. has leaves that are black-punctulate beneath, and thus they might bear glands com- parable to those of Koehneria. Examination of leaves of this species from four countries, how- ever, revealed no glandular trichomes but in- stead numerous large subepidermal secretory cavities. These are especially abundant on the lower leaf surface and macroscopically appear as dark, not black, dp specimens are dried. Glandular tricl he Koehneria type prob- ably are absent from E Lagerstroemia. Similarities in wood anatomical characteris- tics of Lagerstroemia and Koehneria (Table 2) are plesiomorphic in nature, whereas the nu- merous wood anatomical features in which they differ are clearly apomorphic and indicate that entities. Pollen features of cate only their distant common ancestral origin; they suggest no close relationship between the taxa. Lagerstroemia is the only genus of Lythra- ceae with a center of speciation in southeastern Asia. No native species are found either in Africa or Madagascar. The species closest to Madagas- car occur 4,000 miles to the northeast in Indo- nesia. From all evidence, there is no close phy- logenetic affinity between the two genera and Koehne was unquestionably correct in removing Koehneria from Lagerstroemia. RELATIONSHIPS TO OTHER LYTHRACEAE Koehneria is unique in the Lythraceae in pos- sessing the following suite of apomorphies: sol- GRAHAM ЕТ AL.— KOEHNERIA 797 itary, axillary, 6-merous flowers with strongly reflexed lobes; 18 stamens, of which 12 appear as pairs on the six dicm a conspicuous elongate ovary stipe; elongate inner epidermal cells of the ovary; a septifragally dehiscent capsule; and glo- bose, glandular trichomes on vegetative and flo- ral parts. Among all other members ofthe family, the genus bears closest resemblance to the New World genera Adenaria and Pehria, and the Afro- Asian genus Woodfordia, by the appearance in all of the unusual globose glandular trichomes, an ovary stipe, and enlarged inner epidermal cells of the ovary wall (Table 3). These synapomor- phies are unknown elsewhere in the Lythraceae and point to close common origin of the genera. Two other lythraceous genera, the zygomorphic sa a stipe but lack other specialized characters that would ally Koehneria with this group. The stipe is regarded as having evolved independently in these taxa. In Woodfordia the ovary stipe is much less developed than in the other genera. In some specimens it appears to be absent; the nectarif- erous tissue then is fused directly to the base of e ovary. In other specimens, a very short stipe, free from the surrounding nectary, is discernible. Some variation exists in the glandular tri- chomes as well. In Woodfordia and Adenaria a canalicular apical extension may develop on some glandular trichomes. This form was first ob- served in Woodfordia (Shome et al., 1981) and now is also noted in Adenaria. The necked form is restricted in Woodfordia to the outer surface of the floral tube and pedicel (Fig. 27); the stems and leaves bear the globose form (Figs. 25, 26). In Adenaria necked glandular trichomes, iden- tical in appearance to those of Woodfordia, are seen sparsely mixed with globose forms on the ovary only on specimens from southeastern = = — FIGURES 18-24. Pollen of Koehneria madagascariensis and Pemphis acidula. —18. Scanning electron micro- torial view graph of K. madagascarien by arrow). доме uals 5 18-20 from n Figures 18- of a colpus t micrograph of P. O um in Figures 23 and 24. Figures Arcy & Rakotozafy 15317 (MO); Figures 21 and 22 from Kibiwa 1204 (MO); Figure 23 from von Mueller 3 ^ Figure 24 from Lam & Meeuse 5430 (K). ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 798 snoJoumnu sno[e -j2dido “snoJaumu o[8urs sno[ejod o[8urs sno[e1od Burs sno[e1od Buts sno[e1od 218015 sno -[E19dido *po[q -nop sno[edosido JO э[841$ sno[edosido -1da 2p sno[edosido -Id2 2р sno[edosido -Ido 29 sno[edosida -Ido 29 snoyedasida 10111504 uswies 00c-SI 71 (8) (0D 8 (0D 8 cl (0738191) Iequinu uurs (2501 aya *apdund-2so1 pa pol aed 10) зуцм oy o[dund-5soJ JO[O9 [8194 шш шш 50-70 шш 9°1—Є`[ шш c[-Q'I шш (y9-g'6 шш /-—0 Oc-t x шшщ z-s x шш 0-01 x шш $ -сф х ШШ Z'p-0'p х шош (8-6, x шш Q[-<¢°¢ 9215 [Bled 318] 318] 318] IJPLAOQO-IJBAO -O39UB] Á[MOJIBU -Оә2ие| Á[MOLIBU | -OoDU?[QO-9]PAOQO ondi 318A0QO Á[PLOJq odeys үезәд +9 9 (f) (С) ($) # 9 9 Joquinu [е1э4 шш (4) 2-00 шш £'07c0 шш g'0-c0 шш (00 шш $0 "eo шш c-0'0 q13uo[ xÁ[eoidg 19919 рәхәрәз о} 12219 12219 JO poxoga1 Күрәм 12919 12213 poxoyos Á[3uo.x]s 90111504 23qo' шш 11-6 шш (07-)61—0'6 шш 0°6-0'5 шш $6 шшщ 1-0% шш ¢°8-S'P y13u97 Je[nq әзеүпирашеә -N] 0] шоре ULIOJIYIBÁO эзетиеаше 9je[nuedureo oje[nueduirgo odeus $поләш-(8—)9(—р) $поләш-9(—р) $поләш-(с 10) y SNOJOWI-(¢ JO) y $поләш-9 $поләш-9 ÁUSOIIA orydiow oryd -08ÁZ Күҳеәм ornqduourounoe -JOUWOBÁZ ÁJALIM *ompduourournoe orgdourounoe orqduourouroe orqduouroure АцэшиА$ эщаюшопош orqduourououir эщаошоџош эщазошщ эщаошір эщаошоџош SULIOJ [21014 АЗотоцатош [e10L4 sno9) sno9) $1025 -eqioy A[ureurud -BLIOD + Jy} Kap ‘snosoeqioy AIp 'snoooequoq АЧ$ЭН “отці -BLIO9 F ‘JYI SABI] 3500] 2500] шло} *pa1oMog-A^3] “PIJIMOY-MIJ -Шэдига *osuap әзе] ‘эшАэ pououelq ‘эшАэ рәцәивд ‘эшАэ рәцәир1д -norued Jo ISOUIIDPI -qonui *joeduioo -qonui *j2eduioo -uQonui ‘Pedwo Árej[rxe *&re1t[os Kre][rxe “4121105 ээцэ$эло Ци] (‘ds Г) BIsy ^3 Y тео 2уіоеа -sesepeyy 9 (ds BIQUIO] eunuasiy ogioed 01 12558 uon "A “15у (MA 29 a D Vuy 'A??'3 -00 01 S&puoH `м оз ОЗ AS -EPEN Y POL y 71 JeosesepeW -NQUISIP orude130or) t$ С I I I I sanads jo J3QUINN DIVUUIOAIS1ISO] eDIpsofpoo m Е DIIDUIPY siyduodg DILIUYION SI9]2EIEQ-) ‘(€061) эццэочу uo pue SIOYINE JYI jo SUOMBAJISQO uo poseq ве ‘5эЩе JANBINA рие :р11и204542807 'stjdui2d чим риәицәоу jo uosueduio;) ' a18v[ GRAHAM ET AL.—KOEHNERIA 799 1986] рэутрноз $ sno рәЗлеүиә S[[99 рэ8леиэ $[[ә2 рэ8летиэ 5$[[ә2 -Iaytuue} Алело рэ8тетиэ $Пээ иоце21[ето é тешлэртАэ лэцит [euLlapids лэцит тешлэр!Аэ зәиш јо #1 Jaddn jo $[[ә2 јешләртаә лэцит -24$ ем АтелО ssou é $135 ^79 5192 £79 $195 6-8 $195 05-2 $1925 -p 41 Пем АлелО ѕиәш oqni елор SUJUIB]S 01 -B]S 0} Аему[ец 5097035 0} oqni егор juosqe jo aseq ‘зиэзэла əseq uioJj *juosoud əseq WO 3uosoJd 9seq woy 3uosa1d jo aseq “uasald SILIBIDIN Ѕүезәа ‘Атело Алело sred jo деч 1əddn jo деч 1əddn sued [ел quasqe эапз елор Jano ‘aqn} телор Jano “3qn3 ¡e10y Jano juasqe 29 23qn1[e10y Jano -ор uo чоцеоо] yoou цим yoou Чим juosqe osoqo[8 ‘әѕ040[8 2509018 2504018 *2soqo[8 juasqe osoqo[3 odeys Ie[npue[£) $us ‘9524 yeom “еүпә2ц yeam “empon Хо Чим 9383 yea] SnOLIBA 3183183 “Je]n[reorun -mw 29 1e[nj[oorun -mw w= 1e[nj[oorun -uo[o '1e[nj[oorun ojeo[ej je[n[[oorun ио Je[npuer3-uoN ѕәшоцоц], ayeride) uiognound ÁK[o1eq-unognound uLiojound рэаопа Apounsrp 91e1ideo uiiojround adeys etu3ns suouiejs Зитрэээхэ “D[QBLIBA шш p[-€] шш 8-/ шшщ p-] шш 6-1 шш ¢[-Z 413931 9]A1g 10959Р О] ‘1104$ juosqe Алэл *3uosoud yuasoid quasoid 109592 1495244 э4т1$ ÁIBAO 9€ С [^ С t £ JO cC Jequinu 3]NIOT Suo[qo—osoqo[8 WIOFJISNy F 9soqo[8 3J2UIQINI 9jeurqunj 2504918 odeys АтелО [епбә пе Jo 198и0] sno[edosido 12840] sno[edosido [enbo + [re enbo + е Joduo[ sno[edosido 1o3uo[ sno[edosido $41843] uoure]s Атао suəw -e]s snorejadida JO *sno[e1odido 7 sno suaulejs SNOTe -[еӣәѕтаә ш 1ussoud juosqe juosqe juosqe juosqe -dosida ш juosoud SISO UJUIEIS 2142041542871 «DIPAOfPOO м 014Чә4 DIIDUIPY siyduwod DIdQUYIOY $19]29EIEq.) 'penuguo) ‘| a18v[ ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 800 "snus3 Аир SIY} ш 25021174] DIpsofpoo 4 UO A[ureurud poseq теа . 05 —91 = ис ‘зпоцел 91 = uc c£ = UZ TE = uc TE = uc UMOUuxun ‘sou sWIOSOWIOIYD Я win 9p q ur" 91 Я ur 81 q ur oc 3 un ps Я ur 07-p1 ST xdUmsce-ceg -ZI x d W” 2-91 -91I xdungz2Z -91 x d W” Zzz-81 -S£ x d ur cc-op x d шт 97-61 (3 x 4) 2215 aeq IJPINIIDA a1e1qeos э1е14е5$ 9]eJqeos -e5s Ápurej-oje[rsd э1е14е5$ Аршеј oInjd[nos әшхд sndjoo snd[oo -OSOUI/Z “ите; 9 JO 0 0 0 0 ѕпаүо2оѕәшут ‘g -озәш/ “шер c Idjooopnasg t t t t 14 t Jaquinu 10109 АЗојоца:ош uo[[od 9jn[oAo21 эчеа oue[d oue[d oue[d oue[d 40111504 uopo[410)) juasqe juoso1d 109592 109592 juasqe juosqe ѕпец рәәѕ juasaud juosqe jussqe juosqe juasqe jussqe ZUIM pə шш шш ¢'0-+'0 uu ф'0-5`0 шш 9'07e'0 шш 6'[-6] шш (y [—-6'0 $-£ x uiu c[-/ x шш 80-10 x шш 6'0-/70 x шш [' [760 x unu cec-ote x WU Ст 9215 poss rep -Dornooj “pour “ү -snouea 30111115 АПетхере 3un [ертш ээиээ [eprornoo] JO JU9ISIYIPUI -11ds 'reprorpnoo[ JUIISIYIPUI I[ISSIDSUIMIJIO -1eur-[e3e1jr)dos -$Чэр oa[nsde;) 7114204154287] JDIP40Jp00M 014Чә DIIDUIPY siyduod ризицаоу $1ә}ютецг) 'penuguo) “| TIAVI 1986] GRAHAM ET AL.—KOEHNERIA TABLE 2. Anatomical character summary of Koehneria and putative allies. Koeh- Pem- Ade- Wood- Lager- neria phis naria Pehria fordia stroemia* Leaf Anatomy Unicellular hairs broad-based + = + + + zo Glo sent + — + + + =. ы Stomata on both leaf surfaces — + + —, + — - Epidermal mucilage cells ++ ++ ++ —, + =,+ + Lamina isobilateral - T = = = pum Major veins vertically transcurrent + > + + + + Wood Anatomy Fibers septate + — + + + + Parenchyma scanty or absent + = + + + —‹ Rays mainly of erect cells + = + + + — a Leaf anatomical data on Lagerstroemia are scanty e the large number of speci ^ Metcalfe and Chalk (1950) recorded glands for Lagerstroem Furtado and Srisuko (1969) described a number of Lagerstroemia species with (gland-) dotted leaves; е records could not be confirmed (see also morphism, a feature absent fro ++ = character very pla or abundant; + = character present; + infrequent. Mexico [R. Torres 1369, Breedlove 25581, 38133, (KE-Graham)]. None have been seen on collec- tions from Central or South America. Trichomes vary in the length of the neck produced. In Wood- fordia they can be seen in all stages of develop- ment from the common globose form to short- or long-necked forms. The thick orange content, possibly resin, extends part way into the neck but is not extruded at the apex as in typical glan- dular trichomes. There is no evidence on which to determine whether the globose or necked form is the derived state. However, if the globose gland is derived from a typical multicellular gland through loss of the apical extension, the loss is nearly complete, being retained on limited parts of the flower only. Multicellular glandular hairs are uncommon in the family (common only in Cuphea and Pleurophora), have a broad multi- cellular base rather than a narrow, nearly stalked one; and produce a clear, sticky resin unlike the thick orange contents of the globose glands. This suggests globose glands are not homologous to other multicellular glandular trichomes in the Lythraceae but may be a new trichome type de- veloped in the ancestor of this group of genera. Anatomical attributes of Koehneria are closely comparable to those of Pehria, Woodfordia, and Adenaria (Table 2). Pehria differs by possessing specialized chambered crystalliferous fibers (Baas & Zweypfenning, 1979). This difference, how- group of Lagerstroemia species has scanty parenchyma but is characterized by conspicuous fiber di- m Koehneria. The other species of PONE have abundant parenchyma. — character weakly pronounced or ever, does not preclude close phylogenetic affin- ity and, on the whole, anatomy supports the as- sociation of the genera circumscribed by the trichome and ovary characteristics. The pollen of Koehneria differs from that of Adenaria, Pehria, and Woodfordia by the pres- ence of faint pseudocolpi. Koehneria pollen, however, does not display any specialized fea- tures that point to a specific generic relationship. Several genera or portions of genera show the same general shape, triporate condition, and ex- y some species of Lagpnstroenia. The pollen data do not rule against a h Adenaria, Pehria, and Woodfordia but к no direct evidence for any more specific relationship. Although they share a number of specialized morphological and anatomical features, genera of this alliance differ to a substantial degree from each other (Table 3). Adenaria is a monotypic, small-shrub genus of New World seasonally ev- ergreen and semievergreen forest zones. Its leaves are herbaceous with glands and uni- and multi- cellular weak hairs, which are most abundant along the veins on the abaxial side of the leaf (Fig. 29). The four-merous flowers with erect ca- lyx lobes appear in dense umbelliform axillary clusters and are weakly tristylic. There are three 802 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 Table 3. Comparison of selected morphological features of Koehneria, Adenaria, Pehria, and Woodfordia. Derived characteristics shared between two or more genera boxed; characteristics unique to a genus bold-boxed. Character Koehneria Adenaria Pehria Woodfordia Inflorescence Leaf texture Floral forms Floral symmetry Merosity Shape Calyx lobes Petal shape Petal color Stamen number Stamen lengths +coriaceous monomorphic actinomorphic 6-merous ? campanulate [strongly reflexed| broadly obovate rose-purple 2 lengths Paired episepalous stamens Stigma Ovary stipe Globose glands Nectary vary inner epidermal cells Capsule dehiscence Seed hairs Pollen pseudocolpi punctiform compact, much branched cyme, umbelliform clusters herbaceous actinomorphic compact, much branched cyme, open clusters herbaceous monomorphic compact, much branched cyme, open clusters | *conaceous] monomorphic actino- to weakly zygomorphic weakly zygomorphic | 4-merous 4-merous | campanulate erect 6-merous | cyathiform cyathiform to tubular weakly reflexed to erect erect obovate - narrowly narrowly oblanceolate lanceolate lanceolate - linear white to rose [red red | 8 12 [1 length 1 length | 2 lengths absent absent absent | bilobed strongly} punctiform punctiform [present, long present, long present, long | m mem - present. мү short abse ito present present basa ell- basa' one-half e to present [basal to stamens] present | basal, well- splitting variously separated from distanc separated from stam со stamens [enlarged enlarged enlarged enlarged | septifragal- indehiscent loculicidal indehiscent or marginicidal absent absent absent | present, spiral] 0 0 a. Polarity of floral merosity at the family level in Lythraceae is not clear. Both the 4- and 6-merous condition о е been derived merous, but zi Ф mn rom an ancestral 5-merous a vary trom 5- to 6-me commonly 6-merous. Another view regards 6-mery as primitive in the family (Tobe, unpubl. data stamen lengths, but within a single flower there is only one stamen whorl and the filaments are more or less equal in length (Nevling, 1958). The breeding system of the genus has not been fully explored but limited pollen viability data suggest the genus might be evolving from heterostyly to dioecy in a way similar to that observed in certain heterostylous Rubiaceae (Bawa, 1980). In two collections of short-styled flowers, 4496 viable grains were produced (based on counts of 50 1986] grains stained 1 in colon SABE: lactol phenol). In n pollen was 6096 viable. In contrast, flowers from four long-styled collections formed no pollen, the microsporangia in each instance containing amorphous, dark- stained contents (Graham, unpubl. data). Other characteristics of Adenaria not found in Koeh- neria are the unusual (for Lythraceae) deeply bi- lobed stigma, more extensively developed nec- tariferous tissue, and indurate, indehiscent capsule. Pehria is similar to Adenaria in habit and dis- tribution. It is also a monotypic shrub genus of Central and northern South America found at elevations from 90 to 1,300 m in a variety of habitats ranging from dry deciduous woods to pine-oak and tropical cloud forests. Its leaves, like those of Adenaria, are herbaceous, without tomentose indument, and, in addition to the glo- bose glands, bear mostly multicellular, uniseri- ate, weak trichomes mixed with sparser short unicellular ones (Fig. 28). The flowers are 4-mer- ous, as in Adenaria, but are monomorphic with a more elongate floral tube, deep red petals, a punctiform stigma, extensive nectary develop- ment extending from the base of the ovary to the stamen insertion, and thin-walled, loculicidally dehiscent capsules. Ad ia and Pehria are more similar morphologically than either is to Koeh- neria; anatomically Koehneria and Adenaria are most similar. he genus Woodfordia comprises two species of shrubs of semixeric habitats. Woodfordia uni- flora is found in Nigeria, Cameroon, eastern dan, Ethiopia, and Uganda (Keay, 1954). Wood: fordia fruticosa grows with Koehneria in Madagascar (possibly as an early introduction because of its alleged aphrodisiacal properties?) and further ranges from Pakistan through north- ern India and southern China to Sumatra, Java, and Timor. In total leaf vestiture W. fruticosa and Koehneria are virtually identical in having к sunken glandular trichomes interspersed n a dense эе of unicellular falcate hairs (cf. ‚ 26). The leaf texture is also attributed to a close phylogenetic relationship, although it is also possible this is the result of evolutionary convergence as a response to the similar the sea- sonally dry habitats in which these two genera occur. In most other derived characteristics, beyond the previously mentioned shared trichome and GRAHAM ET AL.—KOEHNERIA 803 ovary characters, Koehneria and Woodfordia are dissimilar. The flowers of Woodfordia are six- merous as in Koehneria but have elongate tubes that appear weakly zygomorphic due to the ar- rangement of stamens and style on the lower side of the flower and to the frequent splitting of the persistent floral tube and enclosed capsule along only the adaxial side of the flower at maturation. The same zygomorphic tendencies are also seen in Pehria. Its floral tube and petals are bright red and the ovary stipe is scarcely developed to ab- sent, rather than conspicuous. A further apo- rphy of Woodfordia, not present in the other genera of this group, is the spiral, inverted epi- dermal hairs of the seed coat. These unusual hairs occur in only three other genera of the family, in the zygomorphic genera Cuphea and Pleuro- phora, and in the actinomorphic Lafoensia. Most likely, they have been independently derived in Lafoensia. They are absent from the seed coat of Koehneria, Adenaria, and Pehria. In summary, Koehneria’s nearest relatives are the genera Adenaria, Pehria, and Woodfordia, an interpretation based on a three-character syn- apomorphy unique to the group. Beyond sharing this character set, the genera have differentiated extensively and more specific relationships can- not be еш from present information. Koehne among the four, displays the most derived е (Table 3), but differentiation has occurred in different directions and at different rates for individual characters in each of the gen- lishing phylogeny difficult. Koehneria the inflorescence has been reduced to a solitary axillary flower, while at the same time the primitive campanulate flower shape and bas- al nectary have been retained. In Woodfordia a less advanced inflorescence type, i.e., numerous flowers in a basically cymose inflorescence, is combined with a more advanced elongate floral tube, in which the nectary has remained basal. The present distribution of Koehneria and al- lied genera in two hemispheres apparently re- flects an early and extensive dispersal from a common African center of origin. The earliest incontrovertible fossil records of the Lythraceae are from the lower Eocene London C India and suggest an Old perate origin for the family (Graham & Graham, 1971). The diversity oftaxa present in the Eocene indicates a history for the family extending at least back into the Paleocene. The fossil record in combination with distribution patterns of ex- 804 ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 \ uM a "n )- « M ^ AE i » te” N gor hn v j - 4 um н d as STARS, VS TNA 2 FIGURES 25-29. Scanning electron micrographs of trichomes of Woodfordia fruticosa [USDA-Miami PI 19882 (KE-Graham)], Pehria compacta [Ortiz 878 (KE-Graham)], and Adenaria floribunda [Torres 1369 (KE- 1986] tant lythraceous genera suggests east Africa as a plausible place of origin for the family. The dis- tribution of the Koehneria alliance is consistent with this model. Migrations during the early Ter- tiary, to dei Adenaria and Pehria or their ancestors in the American flora, would have been possible geri. and climatically, either westward to Sout a or via a north At- lantic land route (Raven & Axelrod, 1974; Smith et al., 1981). A history of overland dispersal of plant forms related to Koehneria finds some sup- port from fossil evidence. Fossil seeds from the Miocene of Denmark have been related to mod- ern Koehneria seeds (Friis, 1985). Identical seeds are recorded from the older Oligocene floras of England (Chandler, 1957) and the northeastern United States (Tiffney, 1981; flora presumed to be Oligocene) with affinities tentatively attrib- uted to Lythraceae by the authors. A separate, later migration from Africa is necessary to ex- plain the present distribution of Woodfordia fru- ticosa in Pakistan, northern India, southern China and islands southward. Koehneria may have evolved from ancestral, less xerophytic, forms in Africa and spread to Madagascar where it sur- vives today, or it may have evolved in situ. Un- derstanding of differentiation in this alliance and in the family generally would be greatly enhanced by discovery of fossil remains from the critical Paleocene epoch in Africa. In the absence of a substantiating fossil history, the present mor- phological and distributional patterns of Koeh- neria and relatives are regarded as the result of да абран md divergence and extensive dis- persa ctions. These processes have obscured or erased evidence of the intervening history of the genera to the extent that determination of phylogenetic affinities of Koehneria may be made to the level of the four- genera alliance, but not to a specific modern ge- nus. SYSTEMATIC TREATMENT In view of the significant differences in mor- phological, anatomical, and palynological fea- tures, Pemphis madagascariensis is here as- signed to the monotypic new genus Koehneria. GRAHAM ET AL.—KOEHNERIA 805 The generic name honors B. A. Emil Koehne (12 Feb 1848-12 Oct 1918), Berlin, whose finely de- tailed and self-illustrated monograph of the Ly- today the singl of the family. Koehneria S. A. Graham, H. Tobe & P. Baas, gen. nov. TYPE: Koehneria madagascariensis (Baker) S. A. Graham et al. = Lagerstroemia madagascariensis Baker. = Pemphis mad- agascariensis (Baker) Koehne. Fam. Lythraceae. Frutex, alt. 0.5-2(-4) m, undique cinereo-tomentellus. Folia petiolis subnullis, ‚ oblonga, cuneato-compressa. Ab glanduliferis, staminibus 18, ovario stipitato Shrubs with ash-gray bark and short, terete branches bearing white indument; nodes slightly buttressed; buds ovoid to triangular, com- pressed. Leaves opposite, subcoriaceous, sessile or nearly so, oblanceolate to linear-oblong, both sides with black punctulate glandular trichomes, these denser on the abaxial side and interspersed in a tomentum of white, falcate, non-glandular hairs. Flowers solitary, or rarely 2, axillary, ped- icellate, campanulate, glandular punctulate and pubescent, 6-merous, homomorphic; lobes strongly reflexed at anthesis, alternating with 6 short appendages — Petals 6, bright rose- purple, longer than the ral tube. Stamens (16-)18(-20), the pie ee ай ones paired, the epipetalous ones single and inserted higher in the tube. Pollen prolate-spheroidal, tricolporate with circular pore and 3 faint pseudocolpi, the exine faintly scabrate. Ovary globose, distinctly stipi- tate, 2-3-locular, glandular-punctate and lightly pubescent at the apex; style long exserted; stigma punctiform. Capsule dehiscence septifragal mar- ginicidal; seeds deltoid- to cuneate-compressed; cotyledons flat. A monotypic genus with the following species: — Graham)]. — surrounded by unicellular, non- -glandular trichomes on lower leaf surface o glandular trichome with nec glandular trichomes and stomata prominent.— —25. Tomentose lower leaf surface of W. fruticosa. — 26. M aperi globose, glandular trichome W. fruticosa. — 27. Multicellular, . Lower leaf чИ of Р. сотрасіа with globose we trichomes and elongate non-glandular trichomes prominent. Scale equals 100 um in Figures 25, 28, and 29; 50 um in Figure 26; 25 um in Figure 27. 806 FIGURE 30. Koehneria madagascariensis. —a. Flowering branch, х 1.— b. Stamen insertion on opened partial floral tube and gynoecium, x 3.— c. Flower with reflexed lobes, x 3. From Koehne, 1903. Koehneria madagascariensis (Baker) S. A. Gra- ham, H. Tobe & P. Baas, comb. nov. (Fig. 30). Lagerstroemia madagascariensis Bak- er, J. Linn. Soc. (Bot.) 18: 270. 1881; J. Bot. (n. s.) 11: 112. 1882. Pemphis madagascar- 6. Heft 17. p. 187. 1903. PE: Madagascar “Ibara” (Bara) country, communicated May 1880, Langley Kitching s.n. (holotype, K!). — punctata Drake seit — Bull. Soc. Linn. s 2: 1222. 1896. L TY species name is written on the in the hand of Drake, while the other specimen bears the name in the hand of a French botanist, ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 P. Danguy (pers. comm., A. Lourteig, 1984). Since both syntypes are otherwise equally representa- tive of the species, the specimen annotated by the author of the species is selected as lectotype. Shrubs 0.5—2(—4) m tall, trunk diam. to 4 cm, with smooth ash-gray to gray-brown bark ex- corticating in very narrow to thread-like whitish strands, the inner bark red-brown; stems much branched, the branches terete, short, 4-10(-20) cm long, covered in fine, white or red-brown in- dument; nodes opposite, buttressed by the raised bases of the leaf scars, especially on the youngest branchlets; internodes telescoped toward the branch extremities with the season's leaves then tending to cluster at the ends of the branches; leaf scars semi-circular, projecting outward and tapering downward from the stem, bearing a row of red-purple, erect, thick, stipular processes adaxially between the scar and the stem, the pro- cesses falling away with age; buds ovoid to tri- angular in outline, compressed parallel to the stem surface. Leaves deciduous, opposite, entire, thick, rigid, subcoriaceous, sessile or nearly so, the petioles to 2 mm long; blades oblanceolate to oblong, rarely narrowly linear, 15-70 mm long, 5-20 mm wide, the bases attenuate, tapering to a brief petiole, the apices rounded to acute, oc- casionally thickened to a minute apiculate tip; upper blade surfaces deep green, conspicuously black glandular punctate when dry, the glandular hairs orange when fresh, interspersed with mi- nute, fine, whitish soft hairs, the venation not prominent; lower blade surfaces white tomen- tose, thickly invested with minute, fine, sofi pointed hairs interspersed with black Е аіп rarely 2, axillary, mostly appearing at the distal- most nodes of the branchlets; pedicels 2-12 mm long, erect to spreading, stout when short, fili- orm when long, bearing 2 linear, caducous brac- teoles 0.5-1.5 mm long at or near the base. Floral tubes (5-)6-merous, homomorphic, obovoid- turbinate in bud with 6 processes (appendages sensu Koehne) projecting outward from the bud at the sinuses ofthe calyx lobes; floral tubes cam- panulate at anthesis, 2.5-5 mm long excluding lobes, 2.5-5 mm wide at the mouth, coriaceous, whitish pubescent, with 6 strongly reflexed del- toid calyx lobes, the lobes 2-2.5 mm long, 1.5- e subtending tube; ap- pendages present between the lobes to varying 1986] degrees as a result of greater or lesser fusion of the margins of adjacent lobes, mostly ca. 0.5-1 mm long; external surface of the flower and ped- icel glandular-punctate and finely pubescent, sometimes suffused rose in color; internal surface glabrous, deep rose-purple. Petals 6, attached within the tube at the base of the calyx sinuses, bright rose-purple, obovate to broadly oblong, venation simple, the midvein thick, bearing 4 or 5 secondary veins, without anastomoses between the veinlets, 5.5-10 mm long, 4-7 mm wide, including a claw 0.25-1.5 mm long. roa (16-)18(-20), occasionally increased by splitting of a filament near its base to form a double sta- men, or reduced by absence of 1 of an episepal- ous pair, 6 single stamens epipetalous, 12 paired stamens episepalous, the epipetalous ones higher in the tube than the episepalous ones at ca. three- fifths vs. two-fifths the length of the tube, often slightly shorter than the episepalous stamens; fil- aments slender, well exserted, deep rose-purple, ca. 10 mm long, frequently persistent in the fruit- ing floral tube; anthers bright yellow, bispo- rangiate, dorsifixed mediany y with a slender con- nective, strongly re caducous Pollen prolate-spheroidal, tricolporate, with three faint pseudocolpi; colpi meridionally elongated, equatorially arranged, equidistant, straight, 10- 16 um long, extending to within 3 um of the pole, tapering to an acute apex, the costae colpi nar- row, ca. 1 um wide; exine faintly scabrate, tectate, with an ektexine bridge frequently present over the pore; pseudocolpi shorter than the colpi, 7- 9 um long, represented by a thinner granulated exine; pore circular, at midpoint of the colpus, 2.5-3 um in diam.; grain size 19-26 um long (P) x 14-20 um wide at the equator (E). Ovary 2- or 3-locular, globose, thick-walled at anthesis, ар red-purple with mixed white pubescent апа ack glandular-punctate apex, stipitate, the stipe ca. 1 mm long, 0.5-0.7 mm wide at anthesis, surrounded at the base by a narrow ring of nec- tariferous tissue; placentation axile, the placenta swollen, fleshy, nearly globose; septa incomplete at the apex of the ovary; ovules numerous in each locule, attached over the entire placental surface, anatropous, erect; style red-purple, well exserted, 7-15 mm long, distinctly longer than the sta- mens, narrowing to the apex; stigma punctate, no wider than or only slightly wider than the narrowed apex of the style. Capsule exceeding the persistent floral tube by 2-3 mm at maturity, dry-walled, red-brown, septifragal marginicidal- ly dehiscent from the apex for ca. one-half the GRAHAM ЕТ AL.—KOEHNERIA RE 31. piens m of Koehneria madagas- cariensis in Madaga length. Seeds numerous, deltoid- to cuneate- compressed, the chalazal end broadest, 1.2-1.3 mm long, 0.9-1.0 wide, the cotyledons flat, not revolute. Common names. Pisopiso and Kipisopiso, meaning *'cat-cat," used by the Bara tribe, who find the flowers of the plants pretty, just as they find cats pretty (Rakotozafy, pers. comm Dorr); Hazobotsy, meaning “white used by the Bara tribe (Dorr, pers. comm.). Phenology. Flowering peaks approximately from late June to September, beginning prior to appearance of the new season’s leaves. However, some flowering occurs virtually all year, depend- ing on local conditions, the flowers then borne with current season’s leaves. Capsule maturation beginning in August; seeds freely dispersed, with empty capsules persistent until the next flowering period. Distribution. (Fig. 31.) Savannahs of semi- arid western and southern Tuléar Province and 808 bases of the central-south mountains in Fiana- rantsoa Province; in open grasslands and dry thickets on red calcareous soils, at altitudes of 100 to 800 m (Perrier de la Bathie, 1954). These areas are characterized by a distinct dry season (six months or more) that is most prolonged in the southwest. Rainfall is irregular, between 500 and 1,000 mm per year, but less than 500 mm in the southwest. Mean temperature of the cold- est month is 20°C (Koechlin et al., 1974). Spec- imens studied are cited below by phytogeograph- ic province following Humbert (1955) and then alphabetically by collector. к examined. Est: Dist. Farafangana, An- mbohobe-Ivohibe, Rakotovao s.n. (Réserves Na- SU 6102) (TAN); Antahambohobe-Ivohibe, with- out о (Service des Eaux et Foréts 1945) (ТАМ). Centre: Mahabo, J. Bosser 9749 (TAN); 75 km sud de э. J. Bosser 13943 (TAN); Tsivory, Catat 4358 (P); Common in the valleys to east of Ihosy, W. Deans Cowan s.n. (BM); 9 km N of Ihosy, W. D’Arcy & A. Rakotozafy 15317 (MO); Domaine du centre moyen, au sud нЕ, В. Descoings 977 ENT Envi- ( dés L J. Dorr 3923 (MO); R. N 3 о В. М. а . М. 7, L. O) "Est d Ihosy, H. Humbert 4897 (BM): a, prés poste forestier d'Imonty (Am- У М. Keraudren ena ae Graham, P); Ihosy, montée vers | M. Peltier 2764 (TAN); ё Euphorbia forest, A. Richard P. Morat 2525 (TAN); Plateau de Bemaraha, without col- lector, Herb. Jard. Bot. Tananarive 6154 (MO); Am- pandrandava pas Betioky, without collector, Herb. Jard. Bot. Tananarive 5111 (TAN). Sud: Ouest d’Ejeda, J. Bosser 229 (MO); Imanombo, Bosser 3877 (МО); Ifo- taka, J. Bosser 4029 (MO); Environs de Tuléar, F. Chauvet 300 (KE-Graham, P); Androy, R. Decary 2607 (US); Antanimora, R. Decary 2935, 9198 (TAN); Be- hara, without collector (Réserves Naturelles 2689) (K, TAN); Tranomoro, R. Decary 9013 (TAN); Environs Tuléar, Dequaire 27541 (KE-Graham, P); Sud, without precise locality, B. Descoings 542 (TAN); Bas Man- oky, B. Descoings 778 (TAN); Vallée Moyenne du Mandrare prés d'Anadabolava, H. Humbert 12437 (KE- еш Р, TAN, US); Rt. Nationale по. 7 at 63 km of Tuléar, D. се 1923 lee pa A. Rak- MALA н ага, Amboa Tsilizy s.n. éserves Naturelles 2 (TAN), eed Antanimora et , AN); Ampanihy, route Ejeda, without collector, Herb. Jard. Bot. Tananarive 4217 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 (TAN). Without precise locality or collector, (Herbier de la station agricole de l'Alaotra 4615, 24086) (MO). The genus is showy in bloom, with the large bright rose petals extending well beyond the shorter floral tube, and the flowers scarcely ob- scured by the elongate leaves, which are not fully developed at first flower appearance. The plant is widely distributed on the drier side of the is- land and is still quite common in the south-cen- tral area and the southwest. At the time of its original discovery, it was reportedly very com- mon in the type area, where it grew with the equally abundant Woodfordia fruticosa (Baker, 1882). The tribal position of Koehneria is not speci- fied because tribal limits in the family were erect- ed on an erroneous distinction between complete septal development (Nesaeeae) and incomplete septal development (Lythreae). Tobe (in prep.) has found that, in fact, all genera are character- ized by incomplete septal development. Studies in wood anatomy and palynology support the interpretation that there is no major natural di- vision in the family corresponding to the present tribes (Baas & Zweypfenning, 1979; Lee, 1979). EXCLUDED NAMES 1 т = ы] ~ Rev. Gen. PI. 1: 250. 1891, nomen AA et superfi. ES madagascariensis (Baker) H. Perrier, in Hum- bert, H., ed., Flore de Madagascar et des Comores. Fam. 147, Lythracées. P. 22. 1954, comb. superfl. LITERATURE CITED Baas, P. & R. C. V. J. ZWEYPFENNING. 1979. Wood anatomy of the Lythraceae. Acta Bot. Neerl. 28: . 1881. Notes viri sp m of flowering pp made by L. Kitching, Esq., scar 79. Lagerstroemia al J. Linn. Soc. Bot y^ 270. 1882. А d "ri flora of Central Madagascar. 5 Bot 12. Bawa, K. S. 80. P of РТ x flowering plants. Ann. Rev. Ecol. Syst. 11: 15- CHANDLER, М.Е. Ј. 1957. The Oligocene bn of the + Tracey Lake Basin E Bull. Brit. Mus. (Nat. Hist.), Geol. 3: 7-123. Еви$, E. M. 1985. Angiosperm puc and seeds from the middle Miocene of Jutland (Denmark). Kon- gel. Danske Vidensk. Selsk. Biol. Skr. 24: 1-165. FURTADO, C. X. & M. SRISUKO. 69. A revision of Lagerstroemia L. ее Gardens’ Bull. (Singapore) 24: 185-334. GILL, Г. S. & P. S. KYAUKA. 1977. Heterostyly in Pemphis acidula Forst. AM in Tanzania. Adansonia 17: 139-14 1986] GRAHAM, A. & S. A. GRAHAM. 1971. The geologic history of the — Brittonia 23: 335-346. Les жо теме di eh im e. Ann. Biol. (Paris) 31: 439- 448, ma AY ы ; qii Pp. 163-166 in J. Hutchin n & J. M. Dalziel (editors), Flora of West Tr al Africa, 2nd edition. Crown Agents, Lon- 1926. An ecological anatomical study . Proc. Amer KIENHOLZ, R. h vegetation in the Philippines RAT. 1974. Flore et Végétation de Madagascar. J. Cramer, Va- duz KOEHNE, E. 1883. Lythraceae monographice descri- buntur. XX. Lagerstroemia L. (ampl.). Bot. Jahrb. Syst. 4: 12-35. . Lythraceae novae. Bot. Jahrb. Syst. 29: 1903. Lythraceae. Pp. 1-326 in A. Engler (editor), Das Pflanzenreich, IV. 216. Heft 17. Wil- helm Engelmann, Leipzig. LEE, S. 1979. Studies on the pollen e in the Lythraceae. Korean J. Bot. 22: 115-133. Lewis, D. 1975. Heteromorphic о dia tem under disruptive selection. Proc. R London, Ser. B., Biol. Sci. 188: 247-256. GRAHAM ET AL.— KOEHNERIA 809 METCALFE, C. R. & L. CHALK. 1950. ин of the Dicotyledons. Clarendon Press, Ox 1973 & 1974. An 2 "des Blattes. pe NEVLING, L. I 58. ao Lythraceae nn. Missouri Bot. Gard -115. PERRIER DE LA sp r 1954. Fam. 147. a- cées. Pp. 1-26 i . Humbe pt et des perte Typog. Firmin- Di. dot. RAVEN, н. жр. I. AXELROD. 1974. Angiosperm biogeography and past continental movements. Ann. Missouri Bot. Gard. 61: tu 673. SHOME, U., S. MEHROTRA & H. P. SHARMA. 1981. Pharmacognostic studies on the flower of Wood- fordia fruticosa Kurz. Proc. Indian Acad. Sci. 90: SMITH, A. G., А. М. HURLEY & J. C. BRIDEN. 1981. Phanerozoic Paleocontinental World Maps. Cam- e Univ. Press, Cambridge E. H. 1899 & 1908. Systematische Anato- mie der Dicotyledonen & Erganzungsband. Enke, TIFFNEY, B. H. 1981. Fruits and seeds of the Brandon lignite. VI. Microdiptera (Lythraceae). J. Arnold Arbor. 62: 487-516 WOOD ANATOMY OF LYTHRACEAE—ADDITIONAL GENERA (CAPURONIA, GALPINIA, HAITIA, ORIAS, AND PLEUROPHORA)'? PIETER BAAS? ABSTRACT e wood anatomy of the lythraceous genera Capuronia (Madagascar), d ad Pi Haitia (Hispaniola, Dominican Republic), Orias (= Lagerstroemia excelsa from Chin Е | | in their wood structure. The a to the last two g parallel de Tetrataxis in its wood anatomy, У ш also c si morphological features and may be due t Orias supports its modern ccount characters from reproduc parallel alton iced the very different ecologies of the two taxa. . Pleurophora may be more closely allied to Woo The likelihood lfordia of parallel develo pieni of chambered crystalliferous fibers and vascular tracheids in the evolution of the Lythraceae is discussed, and an earlier phylogenetic classification of the family based on wood anatomy is modified to a purely phenetic scheme, not necessarily indicating mutual affinities. In an earlier study a broad survey of the wood anatomy of the Lythraceae was presented (Baas & Zweypfenning, 1979). Together with studies on the genera A/zatea (Baas, 1979), Punica (Bridgwater & Baas, 1978), and Koehneria (Gra- ham et al., 1986), this left only the woody genera Capuronia, Galpinia, Haitia, Orias (now La- gerstroemia), and Pleurophora as taxa whose wood anatomy was unknown. Thanks to the help of Peter Raven (Missouri Botanical Garden) and several other botanists, vouchered wood samples have recently become available for study. This paper thus completes the wood anatomical sur- vey of the family at the generic level. There is still scope for substantial extension of wood an- atomical studies below the genus level, because many genera show an interesting wood anatom- ical range that can aid in arriving at more natural classifications. Four genera without truly woody representatives, viz. Didiplis, Hionanthera, Pep- lis, and Rotala are of course beyond the scope of this stud Methods employed are similar to those of the earlier studies and the descriptive style more or less follows the same format, with the exception that for material from thin stems full quantita- tive characters are presented here. It should be stressed that these values can be taken only as a very rough indication of what the values in thick- er stems might be. As is generally known, vessel diameter, vessel member length, fiber length, and ray width increase from the pith outward in ju- venile wood, while vessel frequency decreases. Moreover, the rays in juvenile wood usually have a much higher proportion of strongly upright cells than mature wood. All wood samples and slides studied are kept at the Rijksherbarium (L,). When known, the location of the herbarium vouchers is indicated under material studied using the abbreviations from Index Herbariorum. WOOD ANATOMICAL DESCRIPTIONS AND TAXONOMIC NOTES CAPURONIA LOURTEIG (FIGS. 1—3) Material studied: C. madagascariensis Lour- teig: Madagascar, Dorr et al. 4136 (twig of 6 mm g (1979), Bridgwater and Baas (1978), and Graham ! The other genera were d ibed in Baas and 7 г et al. (1986 ? Тат greatly indebted to P. Ra (Mi i Bot material for this study. Thanks are ‚ also due to T. A. Zanoni (Jardin Bota Graham (Kent State University, Ohio) a arriving at the conclusions presented her | Garden, St. Lou is) fo timul obtaining nico Nacional: Dominican Republic), orrespondence with S. A nd H. Tobe (Chiba University, Japan) have been most helpful for ere. 3 Rijksherbarium, P.O. Box 9514, 2300 RA Leiden, The Netherlands. ANN. MISSOURI Bor. GARD. 73: 810-819. 1986. 1986] diam. with many indistinct growth increments, MO), Léandri 2669 (twig of 2.5 mm diam. with 7 growth increments, L), Bernardi 11425 (twig of 3.5 mm diam. with 5 growth increments, L). Shrub in dry tropical forests, West Madagas- car. Growth rings faint to distinct. Vessels diffuse or wood weakly semi-ring-porous. Vessels ca. 200-300 per square mm, ca. 30% solitary, the remainder in radial (rarely oblique to tangential) multiples of 2-4(-6), sometimes including very narrow vessel elements, round to weakly angular, tangential diam. (20-)30-35(-45) um, radial diam. up to 50 um, walls 2-6 um thick. Perfo- rations simple in oblique end walls. Vessel mem- ber length (130-)210-240(-380) um. Intervessel pits alternate, pag round to weakly polyg- ona m ssel-ray pits similar but half- бойдегед. Vessels al with gummy con- tents. Spiral thickenings absent. Fibers thin- to medium thick-walled, or thick-walled and gelat- inous, (220-)340-390(-560) um long, with sim- ple to minutely bordered pits mainly confined to the radial walls, septate, partly chambered crys- talliferous, mostly with cytoplasmic contents, y scan typ | in 3- 4- celled strands. Rays ca. 16 per mm, exclusively uniseriate, (1-)3-6(-20) cells high, composed of upright to weakly procumbent cells (juvenilistic to Kribs’ heterogeneous type I). Crystals solitary prismatic, large and one per chamber i in crystal- ber of minute, irregularly shaped crystals. Silica bodies absent. Taxonomic note. When Lourteig (1960) de- scribed the genus Capuronia she considered it close to the genus Nesaea of Koehne’s subtribe Nesaeinae (1892). Other genera in this subtribe with woody species are Adenaria, Crenea, De- codon, Ginoria, Heimia, Pehria, and Tetrataxis. Haitia would also belong there (see below). In its wood anatomy Capuronia resembles Galpi- nia, Ginoria, and Pehria rather than Nesaea. The former genera as well as Punica (a putative Ly- thraceae or Punicaceae, cf. Bridgwater & Baas, 1978) share chambered crystalliferous fibers, a character that was considered to be strongly in- dicative of mutual affinity (Baas & Zweypfen- ning, 1979). Capuronia and Galpinia share the presence of foliar nectaries but differ substan- tially in floral and pollen morphology (Graham, pers. comm.). Both Graham and Tobe (pers. comm.) consider Ginoria, Pehria, and Punica at BAAS—LYTHRACEAE 811 most distantly related to Capuronia. This might indicate that chambered crystalliferous fibers have evolved independently in at least two dif- ferent groups of genera of the Lythraceae. A wood anatomical comparison of Capuronia and Galpinia yields as the only relevant differ- ence the exclusively uniseriate rays of the former and the uni- and biseriate rays in the latter. This is a very minor difference, especially because bi- seriate rays are quite rare in Galpinia and because the possibility cannot be excluded that Capuro- nia is capable of forming some biseriate rays in more mature wood. GALPINIA N.E. BR. (FIGS. 4—6) Material studied: G. transvaalica N. E. Br., South Africa, Balsinhas 3636 (stem of 3 cm diam., PRE). Shrubs or small trees in rather open mountain vegetation. Growth rings faint to distinct. Wood diffuse- to weakly semi-ring-porous. Vessels ca. 140 per square mm, ca. 2096 solitary, remainder in radial multiples of 2-5(-10), occasionally in clusters in- cluding very narrow vessels, weakly angular, tan- gential diam. (20-)40(-60) um, radial diam. up to 65 um, walls 2-4 um thick. Perforations sim- ple in oblique to nearly horizontal end walls. Vessel member length (300-)390(-500) um. In- tervessel pits alternate, vestured, round to po- lygonal, 5-7 um. Vessel-ray pits similar but half- bordered. Some vessels with gummy contents. Spiral thickenings absent. Fibers thin- to medi- um thick-walled or thick-walled and gelatinous, (390-)580(-720) um long, with simple pits main- ly confined to the radial walls, septate, partly chambered crystalliferous; non-crystalliferous fi- bers rich in starch grains. Parenchyma extremely scanty paratracheal, in 2-celled strands, inter- grading with wide, septate paratracheal fibers. Rays ca. 17 per mm, mainly uniseriate, partly with a low biseriate central part, (1-)2-8(-20) cells high, heterocellular, composed of upright and square to procumbent cells (Kribs' hetero- geneous types I-II). Crystals solitary prismatic or irregularly shaped, one to several per chamber in crystalliferous fibers. Silica bodies absent. Taxonomic notes. Galpinia was doubtfully placed in an alliance with Diplusodon and Pem- phis (Diplusodontinae in Koehne’s classifica- tion, 1892) by Brown (1894) when he described this monotypic genus from South Africa. Koehne 812 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 t LL Г ' * кез 4 TY ДЫ) ^, " MIRA .* ХЕЛ АА ААА га. і Г d — s а 4 < 7 . L^ a d w^ PA JAA в То (уч. кор lr ws . ; .....:39 +7 ә ә. Y · se eens age Mex ES 1-6.— 1-3. Capuronia madagascariensis. — 1. TS. — 2. TLS, note septate fibers. — 3. RLS, note cham- bered crystalliferous fibers. — 4—6. Galpinia transvaalica. —4. TS.—5. TLS, note septate fibers. —6. Maceration, chambered crystalliferous fiber. 1986] (1903) followed this suggestion but also com- mented on the resemblance in habit to Lawsonia and Lagerstroemia. The main wood anat Galpinia and Diplusodon lies in the nonseptate, noncrystalliferous fibers of Diplusodon. Pemphis is wood anatomically quite isolated within the family on account of its well developed vasicen- tric parenchyma (Baas & Zweypfenning, 1979) and differs in many other wood anatomical char- acters from Galpinia. In contrast, there are very close wood anatomical similarities between Gal- pinia and Capuronia, Ginoria, Pehria, and Puni- ca. On the basis of morphological characters (Graham, pers. comm.) only a close affinity with Capuronia seems to be supported. namical diffe het DUU UU HAITIA URBAN (FIGS. 7, 8) Material studied: H. pulchra Urban & Ekman, Haiti, Zanoni et al. 28433 (stems of 2, 3, and 4 mm diam., MO). Shrubs of dry limestone terraces. Growth rings faint to absent. Vessels diffuse, ca. 100 per square mm, 15% solitary, remainder in radial (to oblique) multiples of 2-4(-5), rounded to angular, tangential diam. (20-340 (-50) um, radial diam. up to 60 um, walls 1.5-3 um thick. Perforations simple in oblique end walls. Vessel member length (300-)390(-500) um. Intervessel E alternate, vestured, round to polygonal, 4—7 um. Vessel-ray pits similar but half-bordered. Kem thickenings and vessel con- tents absent. Fibers thin- to medium thick-walled or thick-walled and gelatinous, (390—)580(—7 20) um long, with simple pits mainly confined to the radial walls, septate. Parenchyma very scanty paratracheal, in 2-3-celled strands. Rays ca. 13 рег mm, 1-2(-3)-seriate, (1—)2-12(-24) cells high, heterocellular with strongly upright marginal cells and (in case of taller rays only) square to pro- cumbent central cells (Kribs' heterogeneous type I). Crystals and silica bodies absent. Taxonomic note. Haitia was described by Urban (1919) as a relative of the genus Ginoria. Graham (pers. comm.) supports this affinity, but wood anatomically Haitia is closer to Tetrataxis and Lafoensia (cf. Baas & Zweypfenning, 1979). logenetic links between Tetrataxis and Ginoria so that in an indirect way wood anatomy is not BAAS—LYTHRACEAE 813 entirely in conflict with a presumed affinity of Haitia with Ginoria. Graham and Lorence (1978) concluded, however, that the relationships of Tetrataxis to other members of the Lythraceae remains unknown, thus invalidating this argu- ment. The main difference between Ginoria and Hai- tia concerns the chambered crystalliferous fibers in the former, which are absent from Zaitia. However, in the restricted material of Ginoria, we also encountered one specimen (of G. amer- icana) that was devoid of crystalliferous fibers (Baas & Zweypfenning, 1979). Studies of more materials of Ginoria and Haitia might thus re- veal a wood anatomical overlap between the two genera, since they are similar in most other wood anatomical features. ORIAS DODE (FIGS. 9, 10) Material studied: Lagerstroemia excelsa (Dode) Chun ex S. Lee & L. Lau (= Orias excelsa Dode), China, Sichuan, Liu et al. 5220 (stem of ca. 2 cm diam., Trees, 2,000 m. MO). in rainforests between 1,200 and Growth rings distinct. Wood semi-ring-po- rous. Vessels ca. 28 per square mm, 35% solitary, remainder in radial multiples of 2-3 or in small clusters, round to oval, tangential diameter (30-)75(-120) um, radial diam. ир to 160 um, walls 2-5 um thick. Perforations simple in hor- izontal to oblique end walls. Vessel member length (200-)300(-360) um. Intervessel pits al- ternate, vestured, round to polygonal, 6-8 um. Vessel-ray pits and vessel-parenchyma pits al- ternate to reticulate, with reduced borders to simple, infrequently unilaterally compound or elongate. Thick-walled tyloses present in some walls, thin- to medium thick-walled, (280-)520 (7740) um long; and ‘parenchyma-like’ fibers, all thin-walled, 200-400 um long, and also with simple pits mainly in the radial walls, in poorly defined broad tangential, discontinuous bands, including some true parenchyma strands. Both types of fibers septate and partly chambered crys- talliferous. Parenchyma scanty paratracheal and scattered between the ‘parenchyma-like’ fibers, in 2-5-celled strands. Rays ca. 16 per mm, al- most exclusively uniseriate, rarely with a low biseriate portion, (1—)2-8(-20) cells high, weakly — = чә r amm е этә eee ee » oe, “Ф — a 9 A "a we Oe w- — - => —— e eee — __ ANNALS OF THE MISSOURI BOTANICAL GARDEN o. © — — ЕЕЕ - ~ чю ч + A oe = = — ¿gan ro a= - Oo O ТА. a -o — == T: => а» ар A” CEI TT = = = 9 -> - @e о-о о ® - - и =: - = е» оф - - =, 0». o EA” em Осо о > О О >> | — „e нь = a — — Ge == de = тҮ ч - о Т aem оо „ = - ө-т ә . > > o - ——— HÀ > э ањ - mm Ф > > T S — FIGURES 7-10.—7, 8. Наша pulchra. —7. TS.—8. TLS, note septate fibers and heterocellular rays. —9, 10. Lagerstroemia excelsa. —9. TS, note faintly expressed fiber dimorphism.— 10. TLS and thin-walled ‘parenchyma-like’ fibers on the right (including some chambe e norma A р fibers on the left red crystalliferous fibers). 1986] heterocellular with one row of square to weakly upright marginal cells or completely composed of procumbent cells (Kribs’ heterogeneous type III to homogeneous). Crystals solitary prismatic, large and one per chamber or accompanied by smaller crystals of irregular shape, in crystalli- ferous fibers. Silica bodies absent. Taxonomic note. Lagerstroemia excelsa was originally described as Orias excelsa by Dode ( who noted that it was close to Lager- stroemia and Lawsonia (subtribe Lagerstroemi- inae). In the flora of China, Orias was reduce to Lagerstroemia by Lee and Lau (1983) and L. yangii was recognised as a synonym. Dode (1909) and Furtado and Srisuko (1969) stressed affini- ties of this species with L. subcostata. Wood anatomically L. excelsa fits very well into the group of Lagerstroemia species with fi- ber dimorphism. This group also includes L. sub- costata. The wood anatomical differences be- tween L. excelsa and L. subcostata are very slight: in the limited material of the latter species stud- ied only its diffuse-porosity and more promi- nently half-bordered vessel-ray pits contrast with the situation in L. excelsa (cf. Baas & Zweypfen- ning, 1979). Other Lagerstroemia species with fiber dimorphism are L. calyculata, L. floribun- da, L. indica, L. loudonii, and L. tomentosa and probably a number of other species that have hitherto never been described wood anatomi- cally. PLEUROPHORA DON Material studied: P. patagonica Spegazzini, Argentina, Grondona 23859 (stem of ca. 8 mm diam. with 14 annual rings, MO); P. saccocarpa Koehne, Argentina, Schinini & M. Martinez Crovetto 22640 (stems of 3 and 5 mm diam., swollen stem base of 8 mm diam., root of 1.5 mm diam., MO) Fruticose or herbaceous perennials or annuals of dry places. In view of the substantial wood anatomical differences between the two species studied, sep- arate descriptions are given for each. P. patagonia (Figs. 11-13) Growth rings distinct. Wood semi-ring-po- rous. Vessels ca. 250—500 per square mm, mainly in multiples (radial, oblique or tangential; the extent of the multiples is difficult to estimate due to intergradation of narrow vessels and vascular BAAS—LYTHRACEAE 815 tracheids in the latewood), vessels rarely solitary, 2-3 um thick. Perforations simple in oblique end walls. Vessels of two types: ‘normal’ ones, and very narrow ones intergrading with vascular tra- cheids in the latewood. Length of ‘normal’ vessel members bb дш, um; of the very nar- row vessel members and vascular tracheids (170-)215(-300) um. cree pits alternate, vestured, polygonal to rounded, 3-7 um. Vessel- ray pits similar but half-bordered. All vessels and vascular tracheids with closely spaced spiral thickenings. Some vessels with dark-staining gummy contents. Vascular tracheids (see above) forming the ground tissue in the latewood, thick- walled. Fibers thin- to medium thick-walled, forming ground tissue in the early and inter- mediate wood only, (210-)285(-360) um long, with simple to minutely bordered pits mainly confined to the radial walls, nonseptate, mostly with cytoplasmic contents. Parenchyma virtual- ly absent; only one 2-celled paratracheal strand noted. Rays ca. 12 per mm, exclusively unise- riate and very low, 1-8 (mostly only 1-2) cells high, entirely composed of upright cells. Crystals and silica bodies not observed. P. saccocarpa (Figs. 14-16) The various parts of stems and root do not differ substantially from each other; the descrip- tion is based on the stem of 5 mm diameter. Growth rings faint to distinct. Wood diffuse- to semi-ring-porous. Vessels ca. 400 per square mm, mainly in radial multiples of 2-6(—10), rare- ly solitary or in short oblique to tangential mul- tiples, rounded to slightly angular, tangential diam. (10-)20(-40) um, radial diam. up to 45 um, walls 1-4 um thick. Perforations simple in oblique to nearly horizontal end walls. Vessels of two types: ‘normal’ ones, and very narrow ones intergrading with vascular tracheids asso- ciated with vessel multiples and in zonate bands coinciding with growth ring boundaries. Length of‘normal’ vessel members (150-)235(-350) um, of the very narrow vessel members and vascular tracheids (180-)260(-360) um. Intervessel pits alternate, vestured, round to polygonal, ca. 6 ит Vessel-ray pits similar but half-bordered. Spiral thickenings absent. Some vessels with dark- staining contents. Vascular tracheids (see above) forming ground tissue of narrow tangential bands associated with growth ring boundaries (but not 816 ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURES 11-16. Pleurophora.—11-13. P. patagonica. —11. TS, note latewood zones of thick-walled tra- cheids.— 12. TLS through latewood. Note low rays and tracheids with spiral wall thickenings. — 13. Maceration, part of vessel element with spiral thickenings.— 14—16. P. saccocarpa. — 14. TS. Arrows indicate indistinct zones with tracheids. — 15. TLS, note relatively tall rays. — 16. Maceration, vessel element with smooth wall. 1986] always identifiable as latewood), thick-walled. Fibers thin-walled or thick-walled and gelati- nous, (240-)350(-480) um long, with simple to minutely bordered pits mainly confined to the radial walls, mostly nonseptate, very rarely sep- tate, often with cytoplasmic contents and starch grains. Parenchyma absent. Rays ca. 24 per mm, exclusively uniseriate, (1-)2-10(-25) cells high, largely composed of upright cells, but some with square to procumbent cells in part of the rays. Crystals and silica bodies not observed. Taxonomic note. Pleurophora saccocarpa differs from P. patagonica in its lack of spiral vessel- and tracheid wall thickenings and its much greater ray height. Pleurophora patagonica is really very unusual in exhibiting such very low rays, so far found else in the Lythraceae. In other wood anatomical features the two species are quite similar. Their most striking feature is in the occurrence of bands of (vascular) tra- cheids. Koehne (1892) incorporated Pleurophora in the Lythrinae, a subtribe in which some genera include woody species: Ammannia, Cuphea, Ly- thrum, and Woodfordia. Tobe (pers. comm.) considers Pleurophora close to Cuphea but the latter genus lacks tracheids in its wood. Wood anatomically P. patagonia is strikingly similar to Heimia, through the shared vascular tracheids in the latewood and the prominent spi- ral thickenings in both vasacular tracheids and vessel all other wood features (cf. Baas & Zweypfenning, 1979; a ma- ture wood sample of Heimia myrtifolia was also studied for the present comparison). Pleurophora saccocarpa is also quite similar to Heimia, but lacks spiral thickenings and has lower rays. The only other genus in Lythraceae with nu- merous narrow vessels intergrading with vas- Msi tracheids at growth ring boundaries is ordia, a genus that in other wood anatom- м dun is also similar to Pleurophora. Thus an affinity of Pleurophora to one of the members of the subtribe Lythrinae can be supported. The similarities between the wood of Heimia and Pleurophora can be contrasted by numerous significant differences in floral and pollen mor- phology (Graham, pers. comm.; Koehne, 1903). Because of these differences a close mutual affin- ity seems unlikely and this makes it probable that the wood anatomical similarities are due to parallel evolution. This parallelism cannot be ac- counted for by similar selective pressure induced BAAS—LYTHRACEAE 817 by environmental factors on the hydrosystem: Heimia is a genus of shrubs occurring along river margins and in wet ditches; Pleurophora is a ge- nus of herbaceous and fruticose annuals and pe- rennials of dry places. DISCUSSION In an earlier paper a tentative phylogenetic classification of the Lythraceae was proposed based exclusively on wood anatomical diversity patterns (Baas & Zweypfenning, 1979). It would be quite easy to fit the genera described in this paper into this classification. However, assump- tions on the monophyletic nature of chambered crystalliferous fibers and fiber-dimorphism, as well as classical transformation series for types of ray tissue according to Kribs (1935), played important roles in constructing this phylogenetic scheme. In the taxonomic notes for Galpinia and Capuronia it had to be concluded that on the totality of evidence from reproductive and veg- etative morphology and wood anatomy, close affinities of these genera with Pehria and Ginoria are unlikely, which implies that their shared chambered crystalliferous fibers must have orig- inated independently. The same conclusion probably applies to vascular tracheids in Pleu- rophora, which apparently arose independently from those in Heimia. Ray specialisation as al- ready emphasised in the earlier paper (Baas & Zweypfenning, 1979) and later advocated in a more general context (Baas, 1982), is probably highly subject to parallelism and reversibility and cannot be used as an unambiguous criterion for determining advancement levels or mutual affin- ity. This is especially the case in the Lythraceae in which the wide range of habit categories (from annual herbs, via fruticose perennials to large trees) makes the developmental phase of the out- er wood of a plant of small stature incomparable to that of a large tree. The distribution of septate and nonseptate fibers in Cuphea and Pleuro- phora (which both may or may not have septate fibers) also suggests parallel development or re- versibility of these character states For the reasons outlined above, I refrain from using the data on Capuronia, Galpinia, Haitia, and Pleurophora for the elaboration of a phylo- genetic classification of the Lythraceae. Instead the most important data on these genera are sum- marised together with those on other Lythraceae in a phenetic scheme (Fig. 17). Even as a phenetic scheme, Figure 17 cannot give an ideal picture ANNALS OF THE MISSOURI BOTANICAL GARDEN a Ginoria + (9) Galpinia + Ф Ammania — Pehria + Ф Nesaea — (s) Lafoensia + Decodon + Tetrataxis + Lythrum + [CS ~ | Crenea + Pleurophora (+) (s) vt Diplusodon — [VoL. 73 Heimia + (s) v Cuphea * / — Adenaria + Koehneria + Capuronia + Ф Haitia + Woodfordia +, vt onsi[ruaAn( SÁB Y — p Scanty paratracheal to absent — 818 Nn 3 © 5 Lagerstroemia A + Ф Lagerstroemia B + 4 O E O ~ T oo M с 2 A = mE [^4] ^ |5 = E Physocalymma * (9) = | & м 0 8 MEE: ы la | | MEM 8|. з a Pemphis — Lawsonia + $ A Б (parenchyma o 50 vasicentric) о |2 з |2 Ф wn ы Ф A = — AMEN A | > n - 3 © o > |e s |8& = |g g o I Aliform to banded Rather scanty; |. Fibre dimorphism Parenc "Tes distribution FIGURE 17. on the right has juvenilistic rays composed mainly o characters indicated in the dia and fibre dimorphism Grouping of Lythraceae genera on wood anatomical characters. Ray tissue characters (according to Kribs, 1935, 1968) on the vertical axis*; parenchyma characters o n the horizontal axis. The column fiber rs nonseptate: s: spiral thickenings Ка оп the vessel walls; vt: vascular tracheids common; ( ): feature only present in part of the species ог speci ы Within individual genera or even species or specimens two sequential ray ie types may intergrade. In this diagram only the most frequently occurring type has been taken into acco of the wood anatomical similarities and differ- nificance. Differences between two sequential ray types may be taxonomically very insignificant (cf. Baas & Zweypfenning, 1979) and yet they cause Capuronia to be in a different cluster of genera in Figure 17 from that of Galpinia, Gi- noria, and Pehria, with which it has more in common than with Haitia, Woodfordia, and Di- plusodon. Similarly, Woodfordia, although sim- ilar to Heimia and Pleurophora, figures in a dif- ferent column because of relatively minor differences in ray composition. Therefore Figure 7 should be interpreted with caution and be seen as only one possible way to present the wood anatomical variation in the Lythraceae in an or- derly manner. There is no reason to abandon our earlier views that within Lythraceae the combination of het- erogeneous type I (sensu Kribs) rays, septate fi- bers, and scanty paratracheal parenchyma pre- sents the plesiomorphic condition for the family. Specialisation in rays towards homogeneity in — 1986] the tree genera on the one hand and towards juvenilism in small shrubs and fruticose peren- nials on the other hand, acquisition of vascular tracheids, spiral vessel wall thickenings, crystal- liferous fibers, fiber dimorphism, and banded or vasicentric parenchyma, as well as loss of fiber septation in some taxa, all probably represent apomorphic character states. The likelihood that these apomorphic states evolved independently in some only remotely related Lythraceae be- came apparent in discussions with specialists of the family who are engaged in taxonomic and morphological studies of the family (S. A. Gra- ham and Hiroshi Tobe). The problem at this stage is that we have no ways to identify in which combination of taxa the apomorphic state is monophyletic and in which it is not. It seems to me that further progress in interpreting the wood anatomical diversity in terms of character po- larity and phylogenetic significance can be achieved only by extending our wood anatomical data base with many more species, and by active cooperation with taxonomists. The extended generic survey of wood anatomy of the Lythraceae does not alter the wood anat- omist’s view of the position of the family within the order Myrtales (Van Vliet & Baas, 1984). LITERATURE CITED Baas, P. 1979. The anatomy of A/zatea Ruiz & Pav. (Myrtales). Acta Bot. Neerl. 28: 156-158. 1982. Systematic, phylogenetic, and ecolog- Уу Wood Anatomy. Nijhoff/Junk, Dordrecht/Bos- ton/London. BAAS—LYTHRACEAE 819 & R. C. V. J. ZWEYPFENNING. 1979. Wood anatomy of the Lythraceae. Acta Bot. Neerl. 28: 7- BRIDGWATER, S. D. & P. Baas. 1978. Wood anatomy of the Punicaceae. IAWA Bull. 1978/1: 3—6. 1894. Galpinia. Bull. Misc. Inf. Kew FURTADO, C 69. A revision of Lagerstroemia L. a Gard. Bull. Sin- gapore 24 GRAHAM, S. A. Eo “LORE NCE. 1978. The redis- covery of Tetrataxis Hooker fil. (Lythraceae). Bot. J. Linn. Soc. 76: 71 с. GRAHAM, S. A., Н. Tose, P. BAAS & Р. Н. RAVEN. 1986 [1987]. a a new genus of Lythra- ceae from Madagascar. Ann. Missouri Bot. Gard. 73: 788-809. KOEHNE, A. 1886. Die ra Verbreitung der Lythraceen. Bot. Jahrb 1-61. Lythraceae. | a ler 8 Prantl's natürlichen Pflanzenfamilien, Edition 1, 3(7): 1 16. Die 03. rhe e In ^. Engler, Das Pflan- zenreich 17 (iv.2 Kriss, D. A. 1935. m lines of structural spe- cialization in the wood rays of dicotyledons. Bot. Gaz. 96: 547-557 1 Commercial Foreign Woods of the American Market. Dover Publications, New Yor LEE, S.-K. & L.-F. LAU. pep Lagerstroemia excelsa (Dode) Chun ex S. Lee et L. Lau. + Flora Rei- publicae Popularis Sinicae 52(2): 1 LOURTEIG, A. 1960. Une Lythracée Sine Capu- j / 7 j nov. sp. nov. de Mad agascar. Comptes Rendus Hebdom. Acad. Sci. Paris 251: 1033-1034 . 1919. Sertum antillanum 9. аа Repert. : 132-1 . 1984 [1985]. M i anatomy and classification of the Myrtales. Missouri Bot. Gard. 71: 783-800. NEW NEOTROPICAL SPECIES OF MELIOSMA (SABIACEAE)! ALWYN H. GENTRY? ABSTRACT Gentry, from Nicaragua; peyton solomonii A. Gentry from Bolivia are ve A number of undescribed species of Meliosma have been encountered in the process of prepar- ing the familial treatments of Sabiaceae for the Floras of Nicaragua and Peru and in the course of general identification. Meliosma is a largely warm-temperate to montane-tropical genus with at least 15 Asian and ca. 40 neotropical species. A characteristic but rather nondescript element of many Latin American cloud forests, it is very poorly collected and as a result is poorly known taxonomically. Indeed this paper presents the first report of the genus for Peru. Each of the new species proposed here is represented by several collections, making possible an assessment of patterns of intraspecific variability. In addition to these six species, there are many unidentified Meliosma collections, especially from Peru, that seem to fit none of the described taxa and may belong to as many as ten or more additional new species. Alternatively these additional collec- tions might represent many fewer relatively poly- collections will provide clearer understanding of specific delimitations. Meliosma hartshornii A. Gentry, sp. nov. TYPE: Costa Rica. Heredia: road to Volcán Barba, 2 km N of Sacramento, 2,750 m. alt, 29 Dec. 1974, G. Hartshorn 1608 (holotype, MO). Arbor usque 15 m alta. Folia oblanceolata vel an- guste elliptica, usque 11 cm longa et 4 cm lata, domatiis axillaribus pubescentibus. Inflorescentia sparsim ad- presso-puberula, pedicellis 1-2 mm longis, florum se- palis 1.5 mm longis, petalis ca. 2 mm longis. Fructus globosus, 0.8-1.0 cm diametro. Tree to 50 cm dbh and 15 m tall; branchlets somewhat angled, glabrous or with a few minute Meliosma hartshornii A. Gentry, as Costa Rica; M. corymbosa A. Gentry and M. nanarum A. i A. Aei and M. vasquezii A. Gentry from Peru; and M. ibe and inconspicuous reddish trichomes, the bark finely longitudinally ridged, with scattered raised lenticels. Leaves alternate to irregularly clus- tered, oblanceolate to narrowly elliptic, 2-11 cm long, 1-4 cm wide, acute to apiculate at apex, cuneate and often somewhat marginally inrolled at base, coriaceous, the 2° and 3° venation sub- prominulous above and below, the surface minutely and densely punctate above and lepi- dote-punctate below, glabrous above and below except for conspicuous tufts of simple trichomes in the axils of the lateral nerves below, drying grayish-olive above, olive or brownish-olive with n main veins below; petiole 0.2-1 cm long. Inflorescence pyramidal-paniculate with a well- developed central axis, 8-17 cm long, usually opposed to a subterminal leaf or short branch, appearing more or less terminal, glandular-pap- illose and also sparsely appressed-puberulous, the thick pedicels 1-2 mm long. Sepals 5, ovate, ca. 5 mm long, glandular-papillose, the margin strongly ciliate; petals mostly caducous and miss- ing on type, ca. 2 mm long; stamens with slender 2 mm long filament, the bottom half fused with the bottom half of the narrow inner petal, the anthers subglobose, ca. 0.6 mm long, subtended by the thickened connective; ovary ovoid, ca. 1 mm long, densely puberulous, the style linear, 1 mm long. Fruit red when fresh, globose, 0.8-1.0 cm in diameter, not stipitate, sparsely puberu- lous with flexuous trichomes when young, essen- tially glabrate, 1 -seeded, 0.8-1.0 cm in diameter. pparently endemic to the cloud forests of Volcán Barba between 2,700 and 2,800 m d £e Additional f examined. COSTA Rica. EREDIA: road to Volcán Barba, 1.5 km N of Sacra- mento, 2,700 m; tree 50 cm dbh, 15 m tall, fruit red, 27 Jul. 1975, Hartshorn 1759 (MO). ' The field work that resulted in discovery of several of the new species described here was supported by grants from the National Science Foundation (DEB-75-20325; DEB-8006253) and USAID (DAN-5542-G-SS-1086- ? Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. ANN. MISSOURI Bor. GARD. 73: 820-824. 1986. 1986] This species is distinctive in the genus in its small leaves and smallish fruits. It is closely re- lated to M. irazuensis Standl., known only from the type from Volcán Irazú. The main difference is in the larger, pedicellate flowers of M. harts- hornii as opposed to the tiny (<1 mm long) ses- sile flowers of M. irazuensis. The inflorescence of M. irazuensis differs in being narrower, having simple lateral branches with more crowded ses- sile flowers, and, especially in the very different, much denser pubescence of erect reddish tri- chomes. Vegetatively that species apparently can be distinguished from M. hartshornii by its nar- rower oblanceolate leaves and pubescent young branches. Another small-fruited, usually small- leaved relative of M. hartshornii is M. idiopoda Blake, common at somewhat lower elevations in Costa Rica and Chiriqui. The material of M. hartshornii was originally identified as a variant of M. idiopoda, but its much thicker inflores- cence branches and pedicels, larger flowers, thicker more coriaceous leaves, and consistently larger fruits distinguish it from that species. Meliosma corymbosa A. Gentry, sp. nov. TYPE: Nicaragua. Matagalpa: Cordillera Dari- enense near Aranjuez, 15 km N of Mata- galpa, 1,400 m alt., lower montane moist forest; tree 20 m, 12 Aug. 1977, Neill 2342 (holotype, MO; isotypes, HNMN, to be dis- tributed) Arbor usque 20 m alta. Folia oblanceolata vel an- guste elliptica, 5-13 cm longa, 1.4-4 cm lata, glabra. Inflorescentia terminalis, corymboso-paniculata, pu- я туе А т гис tus asymmetrice subglobosus, 1.6-1.8 cm longus, 1.5- 1.6 cm latus Tree 20 m tall, the branchlets somewhat an- gled, glabrous, the bark very finely longitudinally ridged. Leaves irregularly arranged, in part op- posite or subopposite, always in part clearly al- ternate, oblanceolate to very narrowly elliptic, 5-13 cm long, 1.4-4 cm wide, acute at apex, narrowly cuneate at base, coriaceous, entire, drying dark above, olive brown below, the lateral veins plane and inconspicuous above, promin- ulous below, completely glabrous except for a der, indistinctly demarcated from the tapering leaf base, ca. 0.5-2 cm long, glabrous except for a very few minute and inconspicuous scales. In- florescence terminal, corymbose-paniculate, flat- topped and greatly exceeding the uppermost GENTRY — MELIOSMA 821 leaves, dense, puberulous with tiny suberect, red- dish trichomes, ca. 10 cm long and 15 cm ac- cross, the flowers ultimately on short pedicels, mostly ca. 1-2 mm long. Sepals 5, ovate, minute, less than 1 mm long, glandular-lepidote, the mar- gin minutely ciliate; outer petals broadly ovate, ca. 2 mm long but the rounded apex inrolled, ca. 1.5 mm wide, the inner ones narrow, ca. 2 mm long; fertile stamens 2, the anther thecae thick and suborbicular, ca. 0.5 mm long, subtended by the broad connective; staminodes 3, ca. 1 mm long; ovary ovoid, ca. 1 mm long, merging with the short style. Fruit asymmetrically subglobose, 1.6-1.8 cm long and 1.5-1.6 cm wide, short- stipitate, minutely glandular-papillose, other- wise glabrous, 1-seeded. Apparently endemic to the montane forests of north central Nicaragua in Matagalpa and Jino- tega Provinces. Additional а examined, NICARAGUA. MATAGALPA: Macizos de Pefias Blancas, SE side, N of da. San Martin, 13°14- 15'N, 85°39'W, 950-1 000 m alt., border with Jinotega, 24 Nov. 1981, W. Stevens & R. Riviere 20916 (MO). JINOTEGA: road from Hwy. 3 to Fundadora, 13%2-4'N, 85%54-55'W, 1,200-1,400 m, large tree in cafetal, 9 Dec. 1983, Stevens 22542 (MO). This is a remarkably distinctive species on ac- count of its large corymbose terminal panicle. No other Neotropical species of the genus (and no Asian species represented at MO) has an even remotely similar flat-topped inflorescence; all other species have pyramidal (or variously re- duced) inf aspect flowerin collections of M. corymbosa somewhat resemble some species of Viburnum much more than any Meliosma. The first collection of this species was filed at MO for some time as a familial “indet.” and at first I suspected that it might not belong to Sabiaceae at all. However, the flowers and fruits are undoubtedly the highly distinctive ones of Meliosma. In general g Meliosma nanarum A. Gentry, sp. nov. TYPE: Nicaragua. Zelaya: Cerro El Hormiguero, west range, ca. 13?44'N, 85?00'W, 1,100- 1,183 m, dense virgin elfin forest, tree ca. 7 m, fruit pendent, green, 15 Apr. 1979, J. Pipoly 5169 (holotype, MO; isotypes, HNMN, to be distributed). r 6-15 m alta. Folia oblanceolata vel anguste snipes 3-16 cm longa, 1.2-4.5 cm lata, plu glabra. dep imd ramiflora, pyramidato-panicu- lata, eem adpresso-puberula. F obpyrifor- mis, 1.7-2 cm шр Y 3-1.9 cm latus. 822 Tree 6-15 m tall, the branchlets rather thick and crooked, irregularly angled, glabrous or with a few appressed trichomes at extreme tip, with a few scattered large raised round lenticels. Leaves alternate to clustered, oblanceolate to narrowly elliptic, 3-16 cm long, 1.2-4.5 cm wide, acute to short-acuminate at apex, cuneate at base, cori- aceous, entire, the venation plane above, intri- cately prominulous below, essentially glabrous, sometimes with a very few appressed trichomes scattered along midvein below, the surface densely and minutely papillose-glandular, drying dark gray above, brownish below; petiole slen- der, 1-2 cm long. Inflorescence (in fruit) from below the leaves, rather small, ca. 9-15 cm long, pyramidal-paniculate, with thick well-developed central axis, sparsely minutely appressed-pub- erulous. Flowers not seen. Fruits obpyriform, 1.7— 2 cm long, 1.3-1.9 cm wide, with the poorly demarcated thick basal stipe having a conspic- uous asymmetric squarish corner at base, gla- brous except for some scales, drying rather light rown. Apparently locally endemic in the cloud for- ests of the isolated Cerros La Pimienta and Hor- miguero of Zelaya Province, Nicaragua, where it is reported to be locally very common Additional collections examined. NICARAGUA. ZELAYA: Cerro La Pimienta, N slope facing La Gar- rapata, 13?45'N, 84°59’W, 900-1, $e m, cloud Ps and lower elfin forest; tree 6 m, frt. green, common lower elfin forest, 16 Mar. 1980, J. Pipoly 6049 (MO): Cerro La Pimienta, eastern range, 13?45'N, 84?59'W, 900-1,160 m, tree 7 m, lower elfin forest, fruit green, majority Pese locally common, 17 Apr. 1969, J. Pipoly 5258 (MO); Cerro El Hormiguero, 800-1,000 m, 13?44'30"N, 84*59'30"W, arbol 15 m, frutos verdes, 14 Apr. 1979, A. Grijalva 318 (MO). This is a rather nondescript species character- ized by small leaves similar in size to those of M. idiopoda Blake or M. matudae Lundell but more coriaceous. However, the fruits of M. nan- arum are much larger than those of other small- leaved species and in size approach those of M. glabrata Urb. or M. occidentalis Cuatr., which differ not only in their larger more membrana- ceous leaves but apparently also in a thicker fleshier fruit. Meliosma peytonii A. Gentry, sp. nov. TYPE: Peru. uzco: Urubamba, Machu Picchu 0.5 km N of junction of Sayacmarca and Aobamba Rivers, 2,410 m, humid low montane sub- tropical forest on a ridgeline, 13 Oct. 1982, Peyton & Peyton 1483 (MO). ANNALS OF THE MISSOURI BOTANICAL GARDEN [Vor. 73 чн 5-10 т alta. $e oblanceolata, plerumque 13-29 cm longa, 4-9.5 c ta, domatiis axillaribus ie А a е sparsim ри berula. Fructus asymmetrice subglobosus, 1.5-1.8 c longus, 1.3-1.6 cm latus Spindly tree 5-10 m tall, the branchlets some- what angled, glabrous or glabrescent except for a few minute scales, the surface grayish, very finely striate-ridged, with scattered raised lenti- cels. Leaves clearly alternate, oblanceolate, (5-) 13-29 cm long, (1.6-)4- m wide, rounded but often subapiculate at apex, cuneate at base, coriaceous, essentially entire to sharply but shal- owly serrate, drying olive brown to grayish above and below, the main veins more or less im- pressed above and raised below, the surface pap- illate above, scattered lepidote-glandular below, with conspicuous tufts of trichomes in axils of lateral nerves below, otherwise glabrous; petioles well demarcated, thickened at base, 1-3 cm long. Inflorescence (seen only in fruit) paniculate, ax- illary and terminal, glabrescently puberulous, the fruits subsessile on short thick woody pedicels. Flowers not seen. Fruits yellow-green turning red when ripe, asymmetrically subglobose, 1.5-1.8 cm long, 1.3-1.6 cm wide, slightly contracted into a very broad and poorly demarcated basa stipe, the surface glabrous, papillate-glandular. Apparently endemic to the Machu Picchu area of the middle Urubamba Valley, between 2,100 and 2,400 m. — р ан examined. PERU pus Picchu, 0.5 km > = o < oO . > 5 РА e = — e e O > o = eyton E Peston 1472 Rio M , 2.5 km from Machu Picchu, 2230 m Peyton E Buon 431 (MO). This tree is reported by B. Peyton (pers. comm.) to be one of the most commonly occurring trees in his study areas in the narrow belt of low, very humid montane forest between 2,100 and 2,300 m around Machu Picchu and in the nearby Lu- cumayo drainage. Field notes indicate that the species is often climbed or marked by spectacled bears and B. Peyton (pers. comm.) suspects that these bears eat its fruits. A second sheet of Peyton & Peyton 431, though superficially very similar to the fertile one, ap- parently consists of part of a compound leaf of a Cupania and is not included in the above de- scription This s is а 0 М. к Cuatrec. & Idro t that species has d a less strongly i dus sec- 1986] ondary veins, a black-ripening fruit, and less ta- pered leaf base. Meliosma solomonii A. Gentry, sp. nov. TYPE: olivia. La Paz: Provincia Nor Yungas, val- ley 85 кб Coroico, Sacramento, 10 km NE of Chuspipata, 2,450 m, dense ridge-top cloud forest, 67%48"W, 16?18'S. Tree 6 m, flowers white, 27 Jan. 1984, Gentry & Sol- отоп 447 10 (holotype, MO; isotypes, LPB, MO, to be distributed). Arbor usque 6 m alta. Folia lanceolata vel anguste elliptica, 5-19 cm longa, 1-3.9 cm lata, praeter squa- mas lepidotas glabra. Inflorescentia pyramidato-pa- niculata, sparsim pub la. fl palis 2 mm longis petalis 2 mm longis. Fructus ignotus. Tree to 6 m tall, the branchlets somewhat an- gled to subterete, puberulous with short stiff as- cending hairs when young, glabrescent and glan- dular-punctate when older, finely longitudinally ridged and with scattered raised lenticels. Leaves alternate, usually irregularly arranged and clus- tered, lanceolate to very narrowly elliptic, 5-19 cm long, 1-3.9 cm wide, acute to acuminate at apex, cuneate at base, coriaceous, mostly entire, sometimes in part remotely dentate at least to- ward apex, drying olive to brownish-olive above, brownish below, the secondary veins raised be- low, plane and inconspicuous above, completely glabrous below except for scattered lepidote glands, above glabrous except for a few scales and reddish trichomes on the deeply impressed midrib; petiole slender at apex, thicker at base, 0.7-1.8 cm long, grooved and puberulous adax- ially. Inflorescence terminal and axillary in up- permost leaves, pyramidal-paniculate with an angled well-developed central axis, sparsely pu- berulous with short subappressed trichomes, with triangular-subulate 1-3 mm long bracts sub- tending each branch, the individual flowers sub- tended by a triangular bracteole ca. 1 mm long, subsessile or with a 1 mm long pedicel. Sepals 5, broadly ovate, ca. 2 mm long, glabrous except for ciliate margin; petals broadly ovate, ca. 2 mm long; stamens with slender 1 mm long filaments, fused at base with the small narrow bifid inner petal, the anthers small, widely separated by the thickened connective, including connective ca. 0.5 mm broad; pistil 1 mm long, the ovary ovoid, glabrous, ca. 0.5 mm long, tapering to a linear style ca. 0.5 mm long. Fruit not known Known only from the steep slopes of the Co- roico valley where it is a common component of GENTRY — MELIOSMA 823 the cloud forest, especially on ridge-tops, aver- aging about 30 plants 2 2.5 cm dbh per hectare. Additional collections examined. BOLIVIA. LA PAZ: (both from type locality), 29 Jan. 1984, Gentry & Sol- omon 44775 (LPB, MO), 44786 (LPB, MO) Only the second species of Sabiaceae known from Bolivia, M. solomonii is very different from M. boliviensis Cuatrec., which has much larger reddish pubescent leaves. The new species is characterized especially by the unusually small, lanceolate to very narrowly elliptic, coriaceous leaves, which are similar in size and shape to (but more coriaceous than) those of M. sellowii Urban of coastal Brazil. The other small-leaved Andean species of Meliosma (M. tachirensis Steyerm. & A. Gentry (ined.), M. meridensis Las- ser) have denser inflorescences, as does mostly lowland M. herbertii Rolfe (Antilles and Vene- zuela). Colombian and Venezuelan M. meriden- sis (which should include M. uberrima Idrobo & Cuatrec.) is most similar in leaf texture to M. solomonii but its leaves have rounded or obtuse apices. Meliosma vasquezii A. Gentry, sp. nov., TYPE: Peru. Loreto: Provincia Maynas, Caseria Al- ianza, Rio Tamshiyacu, non-inundated for- est on lateritic soil, trail toward Río Maniti, a m E 72°58'W, 4*5'S, | Aug. 1980, Gentry, Vasquez, Jaramillo, Andrade & piss ms (holotype, AMAZ; isotypes, AAU, F, G, IBE, MO, NY, USM). Arbor 12-20 m alta. Folia obovata, 17- 35 a a onea 7.5-15 cm lata centia laxa paniculata, saltem 30 cm longa, p Fructus oblongo-obovoideus, 2-2.4 cm E 1.5-2 cm latus Tree 12-20 m tall, the branchlets somewhat angled to terete, finely puberulous. Leaves clearly alternate, obovate, acuminate, 17-35 cm long, 7.5-15 cm wide, the acumen to 2.5 cm long, the base cuneate, chartaceous, rather remotely sharply serrate, drying dark above, brownish be- low, the main veins subplane to slightly im- pressed above, prominently raised below, the ter- tiary venation plane above, prominulous below, glabrous above except on midvein and some- times on secondary veins, below puberulous along veins and sparsely over surface; petiole 3.5-6 cm long, rather densely puberulous with suberect reddish trichomes. Inflorescence (seen only in fruit) large, openly paniculate, to at least 30 cm ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 FIGURE 1. New Meliosma species. — A, B. B. Close-up of portion of ¡caos (line i is 5 mm) (Gentry & а. 44710). С. М. vasquezii A. Gentry (line is 2 ст) (Gentry et al. 29270). long, somewhat persistently puberulous, the fruits on thick woody pedicels. Flowers not seen. Fruits turning black, sessile, oblong-obovoid, 2-2.4 cm long, 1.5-2 cm wide, glabrous Widespread but rarely collected in lowland Amazonian Peru. Additional collection examined. PERU. LORETO: Varadero de Mazan from Rio Amazonas to Rio Napo, 22 Aug. 1972, Croat 19444 (MO). This is one of the largest-leaved and largest- fruited species of the genus. In addition to the large leaf size it is characterized by the very long petioles and rather thin leaf texture. It is one of the very few species of the genus that reach low- M. solomonii A. Gen — А. Flowering branch (line is 2 cm).— land Amazonia. Perhaps as many as three other similarly large-leaved but more glabrescent species of Meliosma have been collected in low- d Amazonian Peru and Ecuador. One ofthese, keeled fruit as well as in vegetative characters; another, which may be conspecific with Pana- manian and Costa Rican M. allenii Standley & L. Williams, has entire leaves and a tendency toward parallel tertiary venation. The Croat collection was originally deter- mined as Licania egleri (Chrysobalanaceae), and duplicates were distributed under that name. NOTES MORE ON THE TECHNIQUES FOR COLLECTING AQUATIC AND MARSH PLANTS In 1980 we collected aquatic plants in the vi- cinity of Vancouver, British Columbia. We were accompanied by rt R. Haynes and shared with him the joy of collecting aquatics from flo- ristically rich lakes. There were many unforget- table moments. When we read Haynes’ recent paper (Haynes, 1984), however, we were sur- prised to discover the difference between our col- lecting philosophy and techniques and those he described. We cannot refrain from making a few comments. . We do not agree with Haynes’ notion, that “if ilis specimen cannot be determined, it might as well be left in nature." No new taxa would ever be described if one took this notion literally. Although Haynes does not qualify his statement, his instructions are meant “for persons collecting in predominantly tropical areas" (R. R. Haynes, pers. comm.). While we agree that in floristically rich areas it is usually difficult to identify sterile specimens of aquatics, in the temperate zone and floristically poorer areas the situation is simpler. Collectors should not be discouraged, but rather encouraged to collect sterile specimens of those aquatic plants that they cannot find in bloom or fruit. Many aquatic plants do not flower often. Of the approximately thirty localities of Megalo- donta (Bidens) beckii known in British Colum- bia, there are only two where one finds the species in flower regularly. Although it may be difficult to identify sterile Megalodonta using identifica- British Columbia (where it is known to occur in about fifteen localities) but again, with experi- ence, one can easily identify sterile specimens. The identification of sterile specimens can be difficult in critical groups, but alternative deter- mination methods can be developed for sterile specimens of aquatics. The anatomical charac- ters of the stem were used to identify broad- leaved species of Potamogeton (Ogden, 1943), and minute microscopic characters, such as hairs inside bladders, for identification of Utricularia (Komiya, 1972). Different flavonoid patterns of morphologically similar taxa were reported for ANN. MISSOURI Bor. GARD. 73: 825-827. 1986. Potamogeton (Haynes & Williams, 1975), Elo- dea (Mues, 1983), Isoetes (Кой & Britton, 1982), etc. We used thin layer chromatography of fla- vonoids for routine identifications of Myrio- phyllum (Ceska, 1977; Ceska & Ceska, 1986) and Ceratophyllum (Ceska & Ceska, 1980). Since one must use destructive methods for identification, or may send duplicate specimens to specialists for their opinions, it is necessary to collect more specimens of sterile aquatic plants than of plants in flower. With more material there is a better chance for identification. Of course, before you collect sterile material you should look around the site carefully for plants having flowers or fruits. In some cases it is important to keep cultures of living aquatic plants. The formation of floating leaves of Batrachium can be induced in culture (Cook, 1966), members of Lemnaceae are rela- tively easy to keep and study in cultures (cf. Lan- dolt, 1980), and temporary cultures are often re- quired in order to get root tips for chromosome counts. On the one hand, one should be equipped with enough collecting supplies (“‘ziploc”’ plastic bags, vials, etc.), on the other, one should not collect more taxa than it is possible to accom- modate in the available aquaria or culture flasks. 2. A collecting pole is a must. Ours was con- structed from an aluminum pike-pole (also known as a boat-hook) which is generally used for han- dling logs in water sorting grounds or saw mill ponds. It is about 3 m long and 3 cm in diameter. The iron hook at the end of the pole was replaced by a small three-toothed garden weeding fork. The pole is sturdy, but not heavy. In fact, it is made to float on water when dropped, whereas an ordinary garden rake would sink to the bot- tom. It fits inside a van or on the roof rack of a station wagon. We use the pole every time we collect aquatic plants, whether we do so from shore or a small inflatable boat. C. D. K. Cook (pers. comm.) uses bunches of welding rods weld- ed together in threes at one end. For collecting Lemnaceae, W. P. Armstrong (pers. comm.) uses an aluminum pole with a coffee can attached to the end. . Success in collecting depends on whether it is undertaken from shore or a boat. In listing the 826 flora of individual water bodies, it is important to know the intensity of collecting activities. For this reason we always note whether or not a boat was used, and all adverse factors (rain, wind) that can lower visibility in water. 4. For pressing, most aquatic plants should be floated. Haynes described the technique in suf- ficient detail. However, always use a pan (a pho- tographic tray, 14 x 18 inches), and do not use newspaper for floating the plants on. Use acid free paper, either mounting paper or white paper of slightly lower weight. Most of the plants do stick to base paper, which can then be glued eas- ily on a regular mounting paper to make a per- manent herbarium specimen. Use only clean water for floating and gently rinse the specimens in another tray or bucket before floating them onto the sheet of paper. Avoid the use of chemicals (alcohol, glyc- erol, etc.) for the treatment of plants before press- ing. Plant pigments, which can provide useful information on specimens, are dissolved in or- ganic solvents and lost (cf. Coradin & Giannasi, 1980). Also, do not use drastic pest control mea- sures, such as dipping the specimen into a mer- cury-chloride solution, in the herbarium. We found that mercury-chloride treatment destroys flavonoids in dried specimens and makes them useless for chromatographic investigations. 6. Specimens of floating Lemnaceae can be prepared easily using the following technique. Wash the collected mass of plants gently and let them spread on water in a tray or small bowl. Take a sheet of dry writing paper of a suitable size (index card format is the best, two or four sheets fit onto a herbarium sheet) and lay it gently on the surface of the duckweed plants. Peel the sheet from the surface of the water and most of the floating plants (about 70 percent) will stick to the surface of the paper in a single layer. Put the sheet with the plants in between newspaper drying sheets and dry in the normal fashion. One can repeat this peeling of duckweed plants sev- eral times with new sheets of paper, but the yield diminishes with every repetition. Dried specimens of Lemnaceae can easily fall off base paper. In the herbaraium they have to be protected in cellophane envelopes or, as W. „Р. Armstrong (pers. comm.) alsa recom- f Wolffia in 70-90 percent ethanol, ance the shape and ANNALS OF THE MISSOURI BOTANICAL GARDEN [VoL. 73 size of fronds are critical characters. He does not recommend using formalin because the fronds become very fragile At this point it na be useful to draw attention to an rigidus technique for photographing Lemnaceae and other small floating plants. Witham m published excellent photographs ofthese plants by floating them in a tray of milk(!), which provided a neutral white background with no shadows or reflections. For their photographs Armstrong (1984) and Armstrong ana Thorne (1984) used a di t bstage lighting with a fluorescent illuminator to produce a white background and excellent details of di- agnostic characters (W. P. Armstrong, pers. omm.) Collecting aquatic plants can be a rewarding experience when the resulting specimens are more than a blob of dried plant material on a mounting sheet. The preparation of aquatic plant speci- mens requires more care than that of dry land This effort is repaid, however, because it facilitates the handling and study of the speci- mens. Each collector tends to develop his own collecting style and in this process practice is more important than a set of guidelines. We will be grateful, however, if our discussion paper helps anyone interested in aquatic plants to develop good о habits. W. P. Armstrong, C. D. K. Cook and R. R. Haynes for their comments on our manuscript. W. P. Armstrong provided us with useful tips on collecting, preserving, and pho- tographing duckweeds. [e] LITERATURE CITED ARMSTRONG, W. P. 1984. acie gi соз Spirodella punctata (С. Т. W. Meyer) Thomps (Lemnaceae). Madroño 31: 123-1 124. — & В.Е. THORNE. 1984. The genus Wolffia (Lemnaceae) in California. Madroño 31: 171- Ad CESKA, A. & O. СЕЗКА. 1980. Additions to the flor. of British Columbia. Canad. Field-Naturalist 94: CESKA, O. 1977. Studies in aquatic macrophytes, En XVII: phytochemical differentiation of M phyllum taxa collected in British Columbia. Water Investigation Branch, Ministry of Environment, Victoria, B.C 1986. Myriophyllum (Halora- gaceae) i in British Columbia: problems with iden- in ceae species. Aquatic Plant Man agement Society, Vicksburg 9 Cook, C. D. K. 66. A iai ЕРИ study of Ra- 1986] nunculus subgenus Batrachium (DC.) A. Gray. Mitt. Bot. Staatssaml. München 6: 47-237. CORADIN, L. & D. E. GIANNASI. 1980. The effects of chemical — on plant collections t to be used in chemotaxonomical surveys. Taxon 3- 40 Haynes, R. R. 1984. Techniques for collecting aquat- ic and marsh plants. Ann. Missouri Bot. Gard. 71 1 D. С. Wiliams. 1975. Evidence for the hybrid origin of Potamogeton ay emer (Pota- mogetonaceae). Michigan Bot KoMivA, $. 1972. Systematic studies on SN Lenti- bulariaceae. Nippon Dental College, Tokyo. KOTT, . M. BRITTON. 1982. Comparison of chromatographic spot patterns of some North American Isoetes species. Amer. Fern J. 72: 15- 18 NOTES 827 1980. Biosystematische Unter- ilie der Wasserlinsen (Lem- ff. Geobot. Inst. ETH, LANDOLT, E. (editor). suchungen in ie Fam . Veró : 1 -247. Species specific flavone glucuronides . Syst. 11: 261-265. The broad- leaved species of Po- tamogeton of North America h of Mexico. odora a ie 119-163, 171-214. жиш H. s of San Diego County. San Diego = Nat. Hist. San Diego. — Adolf Ceska, British Columbia Provincial Mu- seum, Victoria, B.C., V 1X4; and Oldriska Ceska, Department of Biology, Univer- sity of Victoria, Victoria, B.C., Canada V8W 2Y2 CHROMOSOME NUMBER IN SARCOLAENACEAE Sarcolaenaceae are a family of ten genera' and approximately 35 species, the living members of which are endemic to Madagascar (Cavaco, 1952; Capuron, 1963, 1970, 1973). The very distinc- tive tetrad pollen of the family is also recorded from the early Miocene of southern Africa (Coet- zee & Muller, 1984). We report here the first the fam- ily previously unknown cytologically (Raven, 1975; Goldblatt, 1981, 1984). Based on counts from root tips, obtained from germinating seed- lings, we have determined a chromosome num- ber of 2n = 22 in two species of Sarcolaena and one of Leptolaena. The systematic position of the Sarcolaenaceae is not certain but currently the family is believed to be a member of Malvales (Takhtajan, 1980; Dahlgren, 1983). Takhtajan has suggested that it is related to Tiliaceae and Dipterocarpaceae. The Malvales as circumscribed by Takhtajan (1980) share many features with Flacourtiaceae (Vio- lales) as well as with the primitive members o the Theales, especially Ochnaceae. Cronquist (1981) treated Sarcolaenaceae as transitional be- tween Theales and Malvales while Hutchinson (1973) placed Sarcolaenaceae in Ochnales, an or- der that he considered to be derived from Theales. Recent bark and wood anatomical studies of Sar- colaenaceae (den Outer & Vooren, 1980; den Outer & Schütz, 1981) do not support a close relationship with Ochnaceae, but do suggest that Sarcolaenaceae are placed best in Malvales sensu Takhtajan. Thus despite the difference of opin- ion concerning the systematic position, Sarco- laenaceae are consistently considered to belong near or to be derived from Theales. Chromosome number appears to have little to contribute to the understanding of the relation- ships of Sarcolaenaceae. It is, however, worth restating (after Raven, 1975) the broad cytolog- ical patterns in the orders Malvales and Theales. Relatively high base numbers characterize Ster- culiaceae, which may have x = 10, while reported numbers in Elaeocarpaceae suggest a base num- ber of x = 14. Numbers of n = 18 and n = 11 have been reported from two genera of Scyto- petalaceae and Bombacaceae frequently have n = 36 and 72, but also n = 43, 44, and 48, and possibly n = 14 and 28 in Durio, the latter two numbers requiring confirmation. Dipterocarpa- ceae, treated in Theales by Cronquist but more often placed in Malvales (Dahlgren, 1983), have x = 7, 6, 11, and 10 (Raven, 1975). Raven con- cluded that x — 7 is possibly basic for Malvales, but that x = 10 appears basic for Tiliaceae and Sterculiaceae, possibly an ancient reduction from n — 14. Sarcolaenaceae would fit conveniently into this hypothetical reduction series from an early polyploid base of x = 14 in Malvales. In Theales, high basic numbers of x — 18, 15, 11, and 10 characterize many genera of Theaceae, but Raven has suggested that x = 7 may be basic for the family and also for Ochnaceae. Thus Sar- colaenaceae would appear to accord as well cy- tologically with Theales as with Malvales. ucher information for the species counted is as follows: Sarcolaena oblongifolia Gérard 2n = 22. Mad- agascar. Prov. Antananarivo. Antananarivo. Cultivated, Parc de Tsimbazaza (original locality unknown), Dorr 2746 (MO,TAN). S. multiflora Thouars. 2n — 22. Madagascar. Prov. Tulear, Fort Dauphin, Forét de Mandena, Dorr et al. 3996 (MO,TAN). Leptolaena bojeriana (Baillon) Cavaco. 2n = 2. Madagascar. Prov. Fianarantsoa, Col de Ta- pia, entre Antsirabe et Ambositra, Dorr et al. 3844 (MO, TAN) Field work in Madagascar was conducted with the generous cooperation of the Musée des Col- lections Scientifiques (C.N.R.T.), Parc de Tsim- bazaza, Antananarivo, Madagascar. LITERATURE CITED CAPURON, R. 1963. Contributions à l'étude de la flore de Madagascar. XVI. Deux nouveaux Schizolaena Prius (1 е elevated сс Thouars subgenus Mediusella Cavaco to generic rank, but he did not validly e change in rank since e did not provide a full and direct Mid to the place of valid Место a the basionym (ICBN. Art. i 2). s Fees is validly published as follow Mediusella (Cavaco) Dorr comb. et stat nov. (Le dg sér. 2, 23: 135. Bull. Mens. Soc. Linn. Paris 1: 564. ANN. MISSOURI Bor. GARD. 73: 828-829. 1986. na subgenus Mediusella a Bull. Mus. Hist. Nat. 1951). Type: Mediusella nu (Baillon) Dorr comb. nov. (Leptolaena bernieri Baillon, 1886). 1986] Dupetit-T! (S 1 2.3: 1 392-400 70. Obse rvations sur les Sarcolaenacées. pn sér. 2, 1 1973. Un Pentachlaena en nouveau. Adansonia, sér. 2, 289-293. CAVACO, A. 126° Fa ua — Chlénacées (Chlaenaceae). Flore de Madagascar et des Co- ). Adansonia. sér Paris. . & J. MULLER. 1984. The аря graphic significance of some extinct Gondwan pollen types from the Tertiary of the southwestern Cape (South Africa). Ann. Missouri Bot. Gard. 7 1088-1099. CRONQUIST, A. 1981. An Integrated System of Clas- sification of Flowering Plants. Columbia Univ. Press, New York. DAHLGREN, R. 1983. General aspects of angiosperm evolution and macrosystematics. Nordic J. Bot. 3: 119-149. GOLDBLATT, P. 1981. Index to plant chromosome numbers 1975-1978. Monogr. Syst. Bot. Missouri Bot. Gar 1984. Index to plant chromosome numbers NOTES 829 1979-1981. Monogr. Syst. Bot. Missouri Bot. Gard. 8. HUTCHINSON, J. 1973. The Families of Flowering Plants, rag о со Press, Oxford. OUTER, R. м & P. В. ScHÜTz. 1981. Wood anatomy мы ы som e Sa bse aenaceae and Rhopalo- carpaceae and ie ir putas position. Meded. Landbouwhogeschool 81-8: | OUTER, К. W. DEN & А. P. VOOREN, 1980. Bark anatomy of some Sarcolaenaceae and Rhopalo- carpaceae and their systematic position. Meded. Landbouwhogeschool 80-6: 1-15. RAVEN, Р.Н. 75. The bases of angiosperm phy- logeny: cytology. Ann. Missouri Bot. Gard. 62: 724-764 1980. Outline of the a ophyta). Bot TAKHTAJAN, A. L, of flowering plants шаш (Lancaster) 46: 225—3 — Peter Goldblatt, B. A. Krukoff Curator of Af- rican Botany, and Laurence J. Dorr, Missouri Botanical Garden, P.O. Box 299, St. Louis, Mis- souri 63166-0299. LEPIDIUM SOLOMONII (CRUCIFERAE), A NEW SPECIES FROM BOLIVIA Lepidium solomonii Al-Shehbaz, which is named after its collector (James C. Solomon), is American lepidiums treated by Thellung ое Hitchcock (1945), апа Boelcke (1964, 1984). It is a dwarf, matted, caespitose plant with a thick caudex, the branches of which are covered with the persistent petioles from previous years, and each branch is a scape with tiny, simple, entire or 1-4-toothed, rosulate, glabrous. leaves. T presence of four claws are rare features in Lepidium L., which characteristically has two or six stamens and un- appendaged claws. Two other South American species, Lepidium ын (О. Kuntze) Thel. mea quitense urcz., have four st sedie but these are very different from L. so- lomonii in habit and in having elliptic fruits and pubescent or puberulent leafy stems. The new species superficially resembles three North erican species, L. nanum S. Watson, L. da- я + petal n ed styles (Hitchcock, 1936; Rollins, 1948; Reveal, ). It is, however, clearly unrelated to any of these. They differ from L. solomonii in having six stamens, ovate to elliptic fruits, unappend- aged claws, and leafy stems. Lepidium solomonii Al-Shehbaz, sp. nov. TYPE: olivia. Depto. az: Prov. Los Andes, 6.6 km NW of Batallas on the principal road along Lake Titicaca; 16°15’S, 68°33’W; elev. 3,850 m; rocky hillside with Stipa, Tetra- glochin, Caiophora, and Baccharis, 5 Feb. 1984, J. C. Solomon 11448 (holotype, MO!; isotypes, GH!, LPB). Figure 1. Herba perennis caespitosa, pulvinata, multicaulis; caudex ramosus, ramis crassis scaposis; folia и petiolata, lineari vel anguste oblanceolata, 6-17 т lon mm lata; E persistentes, basi valde complanati, 3.5-6 mm lon culata, retus mm longa; stylus eee eae 0.4-0.5 m ANN. MISSOURI Bor. GARD. 73: 830-831. 1986. bentes. Caespitose perennial herbs, forming cushions. Caudex thick, branched, densely covered with papery petiole bases from previous years; ulti- mate branches rosulate, scapose, 1-2 cm long. Leaves petiolate, glabrous, somewhat fleshy; blade linear to narrowly lanceolate or oblanceo- late, 6-17 mm long, 0. m wide, entire or 1-4-toothed, acute or subacute, attenuate at base; midnerve prominent in the proximal half, ob- scure in the distal one; petioles persistent, strong- ly flattened, 3.5-6 mm long, 1.5-1.8 mm wide at base. Inflorescence an ebracteate, few-flow- ered, subcorymbose raceme, 1-1.5 cm long; rachis abrous; fruiting pedicels 3-5.5 mm long, as- cending, straight, winged. Sepals ascending, gla- brous, ovate to broadly oblong, 1.5-1.8 mm long, 1.2-1.4 mm wide, broadly white margined, not saccate, equal at the base. Petals white, clawed, spatulate to broadly ovate, 2.2-3 mm long, 1- mm wide; claws 1-1.2 mm long, with a small adaxial appendage just below the blade. Nectar ands 4 .4 mm long, toothlike, opposite petals. Stamens 4, erect, white; filaments dilated at the base, 1.6-2 mm long, the lateral pair as long as or only slightly shorter than median pair; anthers ovate, ca. 0.6 mm long. Fruits glabrous, orbicular, 3.2-3.8 mm long, keeled, retuse and obscurely winged at the apex; sinus 0.1—0.2 mm deep; styles persistent, exserted, 0.4—0.5 mm long. Seeds ovate, minutely reticulate, wingless, com- pressed, ca. 1.7 x 1.2 mm, only slightly muci- laginous when wet; cotyledons incumbent. LITERATURE CITED BOELCKE, O. 1964. Notas di especies de Lepidium de la ооа иа па 13: 84. Notas sobre Gade Argentinas I. а еп a género Lepidium. Parodiana 3: 21-29. Нитснсоск, С. L. 1936. The genus -r ione in the United States. Madroño 3: 265-320 1945. The South American species of Lepid- ium. Lilloa 11: 75-134. Figs. REVEAL, J. L. 1967. Anew Say fora Utah Lepidium. Great Basin Nat. 27: 177- ROLuns, В. C. On s imu caespitose lepidiums of western North America. Madroño 9: 5. 1986] NOTES 831 % Y Жас ONO A 5 : И ate MLA AS TS K Ne “ee DS a 2m "m M AES | | ES : Т. Al-S. FIGURE 1. Lepidium descen = нар 11448, holotype).—a. Plant.—b. Flowering branch of cau- dex.—c. Leaf.—d. Variat n leav er (one sepal and two petals removed). —f. Sepal.—g. Petal. — аы. —1. Fruit and Poe пей. —). e еее я by the author. Perrito, А. 1906. Die Gattung Lepidium (L.) R. — [Лап А. Al-Shehbaz, The Arnold Arboretum Br. Eine monographische Studie. Neu Denkachr, of Harvard University, 22 Divinity Avenue, Cam- Allg. Schweiz. Naturf. Ges. 41(1): 1-3 bridge, Massachusetts 02138. VOLUME 73 ANNALS MISSOURI BOTANICAL GARDEN The ANNALS, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be ac- cepted. Authors should write the Editor for information concerning arrangements for publishing in the ANNALS. EDITORIAL COMMITTEE NANcy Morin, Editor Missouri Botanical Garden MARSHALL R. CROSBY Missouri Botanical Garden GERRIT DAVIDSE Missouri Botanical Garden JOHN D. Dwyer Missouri Botanical Garden & St. Louis University PETER GOLDBLATT Missouri Botanical Garden Colophon This volume of the ANNALS of the Missouri Botanical Garden has been set in APS Times Roman. The text is set in 9 point type while the figure legends and literature cited sections are set in 8 point type. The volume has been printed on 70# Centura Gloss, an acid-free paper designed to have a shelf-life of over 100 years. Centura Gloss is manufactured by the Consolidated Paper Company. Photographs used in the ANNALS are reproduced using 300 line screen halftones. The binding used in the production of the ANNALS is a proprietary method known as Permanent Binding. The ANNALS is printed and distributed by Allen Press, Inc. of Lawrence, Kansas 66044, U.S.A. © Missouri Botanical Garden 1986 ISSN 0026-6493 AGOSTINI, GETULIO. (See Sonia Yarsick, Nerida Xena de Enrech, Nelson Ramirez & Getulio Agostini) ................................... AL-SHEHBAZ, IHSAN A. Lepidium solomonii (Cruciferae), A New Species С ОПУ А ko ok A NSF EE ARONSON, J. A. (See A. Shmida & J. A. Aronson) ................... AXELROD, DANIELI. Cenozoic History of Some Western American Pines BAAS, PIETER. (See Shirley A. Graham, Hiroshi Tobe & Pieter Baas) BAAS, PIETER. Wood Anatomy of Lythraceae — Additional Genera (Ca- puronia, Galpinia, Haitia, Orias, and Pleurophora) ............... BAJAJ, RENU. (See Arthur Gibson, Kevin C. Spencer, Renu Bajaj & Jerry L. McLaughlin) ее ——————Ón BAKER, HERBERT G. Yuccas and Yucca Moths—A Historical Commen- BU A A ИР fore Восте, A. LINN. The Floral Morphology and Vascular Anatomy of the Hamamelidaceae: Subfamily Liquidambaroideae ................. CANNON, J. Е. М. (See M. J. Cannon & J. Е. M. Cannon) ............ CANNON, M. J. & J. F. M. CANNON. Studies in the Araliaceae of Nicara- gua, and a New Widespread Species of Oreopanax ............... CARR, GERALD D. & GORDON McPHERSON. Chromosome Numbers of New Caledonian Plants ...........sceeso soe Rh n CESKA, ADOLF & OLDRISKA CESKA. More on the Techniques for Collecting Aquatic and Marsh Plants .......... 5... eee ee een n CESKA, OLDRISKA. (See Adolf Ceska & Oldriska Ceska) .............. CRANE, PETER R. 8 RUTH A. STOCKEY. Morphology and Development of Pistillate Inflorescences in Extant and Fossil Cercidiphyllaceae CRONQUIST, ARTHUR. Commentary on the Status of the Hamamelidae CROSBY, MARSHALL R. Topics in North American Botany, A Symposium Commemorating George Englemann: The Thirty-First Annual Sys- tematics Symposium .......-.- инки сзсз D’Arcy, W. G. The Calyx in Lycianthes and Some Other Genera D’Arcy, WILLIAM С. & ARMAND К AFY. A New Species of Sola- num (Solanaceae) from Madagascar ............................. DILCHER, Davip L. & MICHAEL S. ZAVADA. Phylogeny of the Hamamel- idae: An Introduction ...............eeee RR In DILCHER, Davip L. (See Michael S. Zavada and David L. Dilcher) .... DILCHER, DAVID L. & DANIEL MACKLIN. Phylogeny of the Hamamelidae: Taxonomic Index .............................. ns Dorr, LAURENCE J. (See Peter Goldblatt & Laurence J. Dorr) ........ Down, J. М. (See P. G. Martin & J. M. Dowd) ..................... DRANSFIELD, JOHN. A Guide to Collecting Palms .................... ENDRESS, PETER K. Floral Structure, Systematics, and Phylogeny in Tro- chodendrales ................................................. ENGELMANN, GEORGE. Instructions for the Collection and Preservation of Botanical Specimens ........................................ GENTRY, ALWYN H. New Neotropical Species of Meliosma (Sabiaceae) GENTRY, ALWYN H. Notes on Peruvian Palms ...................... D A a wae A UM REIS GIBSON, ARTHUR, KEVIN C. SPENCER, RENU BAJAJ & JERRY L. Mc- LAUGHLIN. The Ever-Changing Landscape of Cactus Systematics . . GOLDBLATT, PETER & LAURENCE J. DoRR. Chromosome Number in Sar- colaenaceae .................................................. GOLDBLATT, PETER. Convergent Evolution of the ‘Homeria’ Flower Type in Six New Species of Moraea (Iridaceae-Irideae) in Southern Africa GOLDBLATT, PETER. Notes on the Systematics of Hesperantha (Iridaceae) in Tropical Africa ............................................. GOLDBLATT, PETER. Cytology and Systematics of the Moraea fugax Com- plex (Iridaceae) ............................................... GRAHAM, SHIRLEY A., HIROSHI TOBE & PIETER BAAS. Koehneria, a New Genus of Lythraceae from Madagascar .......................... GRAYUM, MICHAEL H. New Taxa of Caladium, Chlorospatha, and Xan- thosoma (Araceae: Colocasioideae) from Southern Central America and Northwestern Colombia ................................... HERNANDEZ, HECTOR M. Zapoteca: A New Genus of Neotropical Mi- mosoideae ................................................... HOLSINGER, KENT Е. & HARLAN LEwis. Description of a New Section and Subsection in Clarkia (Onagraceae) ............................. JONES, JAY H. Evolution of the Fagaceae: The Implications of Foliar Features ..................................................... KAUL, ROBERT B. Evolution and Reproductive Biology of Inflorescences in Lithocarpus, Castanopsis, Castanea, and Quercus (Fagaceae KNAPP, SANDRA. Three New Species of Solanum Section Geminata (G. Don) Walp. (Solanaceae) from Panama and Western Colombia . LANGE, CARLA. (See Michael T. Stieber & Carla Lange) .............. LESTER, RICHARD N. & PHiLIP A. ROBERTS. Serotaxonomy of Solanum, Capsicum, Dunalia, and Other Selected Solanaceae ............... LEVIN, GEOFFREY A. Systematic Foliar Morphology of Phyllanthoideae (Euphorbiaceae). I. Conspectus ................................. LEVIN, GEOFFREY A. Systematic Foliar Morphology of Phyllanthoideae (Euphorbiaceae). II. Phenetic Analysis ........................... Lewis, HARLAN. (See Kent E. Holsinger & Harlan Lewis) ............ 102 Lott, EMILY J. & JAMES S. MILLER. Bourreria rubra (Boraginaceae), a New Species from Coastal Jalisco, Mexico ....................... MACKLIN, DANIEL. (See David L. Dilcher & Daniel Macklin) ......... MARTIN, P. G. & J. M. Dowp. Phylogenetic Studies Using Protein Se- quences within the Order Myrtales .............................. MCLAUGHLIN, JERRY L. (See Arthur Gibson, Kevin C. Spencer, Renu Bajaj & Jerry L. McLaughlin) .................se esee MCPHERSON, GORDON. (See Gerald D. Carr & Gordon McPherson) ... MILLER, JAMES S. (See Emily J. Lott & James S. Miller) .............. Morin, NANCY R. (See Stanwyn G. Shetler & Nancy R. Morin) ...... PENG, CHING-I. A New Combination in Ludwigia Sect. Microcarpium (Onagraceae) оиа уе ее ds POHL, RICHARD №. A New Paspalum (Poaceae) from Mesoamerica ... RAKOTOZAFY, ARMAND. (See William G. D’Arcy & Armand Rakotozafy) RAMIREZ, NELSON. (See Sonia Yarsick, Nerida Xena de Enrech, Nelson Ramirez & Getulio Agostini) ...............................8... RAVEN, PETER H. (See Hiroshi Tobe & Peter H. Raven) ............. REDFEARN, PAUL L. & P.-C. Wu. Catalog of the Mosses of China ..... ROBERTS, PHiLIP A. (See Richard N. Lester & Philip A. Roberts) ...... ROMEO, JOHN T. Distribution of Nonprotein Inimo and Sulphur Amino Acids in Zapoteca a ARA RC Ee s ROMERO, EDGARDO J. Fossil Evidence Regarding the Evolution of Noth- ofagus Blume rare ROMERO, EDGARDO J. Paleogene Phytogeography and Climatology of South America cir RARA SHAW, ELIZABETH A. Changing Botany in North America: 1835-1860. The Role of George Engelmann ................................ SHETLER, STANWYN G. & NANCY R. Morin. Seed Morphology in North American Campanulaceae ..................................... SHMIDA, A. & J. A. ARONSON. Sudanian Elements in the Flora of Israel SMITH, ALAN R. (See Julian A. Steyermark & Alan R. Smith) ......... SMITH, LYMAN B. Revision of the Guayana Highland Bromeliaceae ... SousA S., MARIO. Adiciones a las Leguminosas de la Flora de Nicara- SPENCER, KEVIN C. (See Arthur Gibson, Kevin C. Spencer, Renu Bajaj & Jerry L. McLaughlin) o rs wed ENE OES ERE RAEN CRECEN STEYERMARK, JULIAN A. & ALAN R. SMITH. A Remarkable New Selagi- nella from Venezuela .......................................... STEYERMARK, JULIAN A. Holstianthus, a New Genus of Rubiaceae from the Guayana Highland ............ ccce esee hn 216 216 177 128 STIEBER, MICHAEL T. & CARLA LANGE. Augustus Fendler (1813-1883), Professional Plant Collector: Selected Correspondence with George Е И Louise hs eek Fs II STOCKEY, RUTH A. (See Peter R. Crane & Ruth A. Stockey) .......... Symon, D. E. A Survey of Solanum Prickles and Marsupial Herbivory in Australia .................................................. THOMAS, DUNCAN W. Notes on Deinbollia Species from Cameroon . THORNE, ROBERT F. A Historical Sketch of the Vegetation of the Mojave and Colorado Deserts of the American Southwest ................ TIFFNEY, BRUCE H. Fruit and Seed Dispersal and the Evolution of the Hamamelidae ................................................ Tope, HIROSHI. (See Shirley A. Graham, Hiroshi Tobe & Pieter Baas) .. TOBE, HIROSHI & PETER H. RAVEN. A Comparative Study of the Em- bryology of Ludwigia (Onagraceae): Characteristics, Variation, and Relationships ................. аа WAGNER, WARREN L. New Таха іп Oenothera (Опаргасеае) .......... Wu, P.-C. (See Paul L. Redfearn & P.-C. Wu) ...................... XENA DE ENRECH, NERIDA. (See Sonia Yarsick, Nerida Xena de Enrech, Nelson Ramirez & Getulio Agostini) ............................ YARSICK, SONIA, NERIDA ZENA DE ENRECH, NELSON RAMIREZ & GETULIO AGOSTINI. Notes on the Floral Biology of Couroupita guianensis Aubl. (Lecythidaceae) ............................................... ZAVADA, MICHAEL S. (See David L. Dilcher & Michael S. Zavada) ... ZAVADA, MICHAEL S. & DAvID L. DILCHER. Comparative Pollen Mor- phology and Its Relationship to Phylogeny of Pollen in the Hama- A sc VA DRE CV NN 219 177 Volume 73, No. 3, pp. 503-651 of the ANNALS OF THE MISSOURI BOTANICAL GARDEN, was published 1 оп 9 February 1987. Contents continued from front cover New Neotropical Species of Meliosma (Sabiaceae) Alwyn H. Gentry ..... 820 NOTES More on the Techniques for Collecting Aquatic and Marsh Plants Adolf Ceska and Oldriska Ceska 825 Chromosome Number in Sarcolaenaceae Peter Goldblatt and Lau- rence J. Dorr 828 Lepidium solomonii (Cruciferae), A New Species from Bolivia Ihsan A. Al-Shehbaz