Volume 17 Number 1 Volume 77, Number 1 Winter 1990 Annals of the Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, con- tributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be accepted. Authors should write the Editor for information concerning arrangements for publishing in the ANNALS. Instructions to Authors are printed in the back of the last issue of each volume. Editorial Cónimitics George K. Rogers _ Editor, Missouri Botanical Garden Amy Scheuler Editorial Assistant, Missouri Botanical Garden “Glenda Nau Magdalen Lampe _ Publications Staff Marshall Missouri Missouri R. Crosby Botanical Garden - Gerrit Davidse Botanical Garden Joti D. Dres Missouri Fimahical Cardo de Saint Louis aci Peter Goldblatt 5 Missouri Botanical Garden ie "Bale E. fond 6 Missouri Botanical Garden Henk van ЗБК Werff : Missouri Botanical Garden ч RON PANE a) cade contact ESR Eleven, P.O, Box 299, St. Louis, MO 63166. Sub- m __ scription price is $75 per volume U.S., $80 Canada — ; and Mexico, $100 all other countries. Four i issues volume. - E The ANNALS OF THE MISSOURI Вота pe (ISSN 0026-6493) is published quarterly by the Missouri Botanical Garden, 2345 Tower Grove Av enue, St. I 55 paid at St. > Post ASTE THE Misso s a РА ouis, MO 63110. Second class ees Louis, MO and additional mailing offi R: Send address changes to ANNALS © p URI BOTANICAL GARDEN, Departmen D. Box 299, Bt. SM MO es Volume 77 Number 1 1990 Annals of the Missouri Botanical (Garden NZ A CENTURY OF SCIENTIFIC PUBLICATIONS AT THE MISSOURI BOTANICAL GARDEN The Missouri Botanical Garden, founded in 1859, originated as the country estate of St. Louis busi- nessman Henry Shaw (1800-1889), who made the Garden and neighboring Tower Grove Park gifts to the city of St. Louis, Missouri. From the outset, the Garden was a center for horticultural display, education, and botanical research. The young in- stitution jumped quickly to prominence in botanical research through Mr. Shaw’s association with Dr. George Engelmann, an obstetrician and one of the leading botanists of the time. Dr. Engelmann took full advantage of the position of St. Louis as the Gateway to the American West by coordinating and participating in western botanical exploration. And to the east, he maintained correspondence with and passed specimens to Harvard botanist Asa ray. Botanical education followed closely on the heels of research. In 1885 Mr. Shaw established and endowed the Henry Shaw School of Botany at nearby Washington University. A rising star in the botanical world, Dr. William Trelease, previously at the University of Wisconsin, was selected to be Engelmann Professor of Botany in the new pro- gram. When Henry Shaw died in 1889, his will placed the Garden in the hands of a Board of Trustees and a Director. The Board called upon the Director to publish an Annual Report. The first report, in 1890, contained a biographical sketch of Mr. Shaw, details of his will and related documents, a report from the Henry Shaw School of Botany, accounts of activities at the Garden, and the First Annual Report by the Director (who by 1890 was William Trelease). The second Annual Report likewise contained the Director’s Report and accounts of institutional activities, and there appeared the Garden’s first scientific publication, “Revision of North American species of Epilobium,” by W. Trelease. (Onagra- ceae have made frequent subsequent appearances in our scientific publications.) The next year’s An- nual Report held two scientific papers, including the classic “Тһе yucca moth and yucca pollina- tion” by Charles Riley. Following issues regularly contained multiple scientific papers from the Henry Shaw School of Botany and from outside contrib- utors. The subject matter was broad: physiology, tax- onomy, phenology, pathology, economic botany, plant chemistry, soil science, limnology, and ad- ditional areas. Titles ranged from “An ecological cross section of the Mississippi River in the region of St. Louis” (1908) to “A consideration of the physiology and life history of a parasitic Botrytis on pepper and lettuce" (1912). By 1912, the year of the Twenty Third Annual Report, it was clear that two publications were necessary to cope with the Garden's need for sci- entific and nonscientific outlets. Since then the ANN. Missouni Bor. GARD. 77: 1-3. 1990. Annals of the Missouri Botanical Garden Bulletin of the Missouri Botanical Garden has een the vessel for accounts of Garden activities, popular articles, and annual reports from the Di- rector. The Annals of the Missouri Botanical Garden was founded at essentially the same time for scientific articles. Volume 1 of the Annals came out in 1914 and contained 20 highly diverse ar- ticles, including one dealing with air pollution by bacteria in human breath. Over the years, the focus of the scientific pub- lications narrowed considerably, most notably with abandonment of bacteriological, physiological, and pathological material. There appeared an intensi- fied focus on plant systematics and related subjects, including ecology at times. Emphasis has been trop- ical with special emphasis on Central America. In 1937 Robert Woodson, Jr., and Russell Sei- bert published “Contributions toward a flora of Panama. I. Collections in the provinces of Chiriqui, Coclé, and Panama,” the first glimmer of The Flora of Panama, which became a dominant theme and formal series in the Annals. The Flora was launched by Dr. Woodson and R. W. Schery in 1943 and was concluded in 1981 by W. С. D'Arcy—6,818 printed pages and 5,314 species later. (For a his- tory of the flora, a checklist, and an index by W. G. D'Arcy see Monographs in Systematic Botany from the Missouri Botanical Garden, Volume 17.) A second major long-term subproject of the Ап- nals began in 1969 with publication of the Gar- den's 16th Annual Systematics Symposium. A full history of annual symposium publications follows: Symposium Volume number (issue) Year Symposium 16 56(3) 1969 Tropical Island Biogeography 17 59(3) 1972 Hybridization, Evolution, and Systemat- ics [18, 19 not published] 20 61(3) 1974 Plant-Animal Coevolution 21 62(2) 1975 Biogeography 22 63(2) 1976 Evolution at the Population Level 23 64(2) 1977 Chemosystematics 24 65(2) 1978 Systematic Studies in Africa 25 66(4) 1979 Palynology and Systematics 26 68(1) 1981 Evolution and Systematics of the Gra- mineae 27 68(2) 1981 Reproductive Strategies in Plants and An- imals 28 69(3) 1982 Biological Studies in Central America 29 70(3) 1983 Biogeographical Relationships between Temperate Eastern Asia and Temper- ate Eastern North America 30 72(4) 1985 The Implications of Phylogenetic Anal- ysis for Comparative Biology 31 73(3) 1986 Topics in North American Botany, A Symposium Commemorating George Engelmann 32 74(2) 1987 The Biology of Epiphytes 33 75(1) 1988 Species Diversit 34 75(4) 1988[ 1989] Macromolecular Approaches to Phylog- e 35 77(1) 1990 Conserving Biological Diversity: Pros- pects for the 21st Century 59(2) 1972 Disjunctions in Plants 61(1) 1974 25 Years of Botany 61(2) 1974 Evolution of Systematic Characters in Ferns Volume 77, Number 1 1990 Rogers 3 Volume (issue) Year Symposium 62(3) 1975 The Bases of Angiosperm Phylogeny 64(4) 1977 Perspectives in Tropical Botany 71(1) 1984 The Evolution of Dioecy 71(3) 1984 The Order Myrtales 74(4) 1987 Contributions to a Symposium on the Evolution of the Modern Flora of the Northern Rocky Mountains 75(3) 1988 Reproductive Biology of Freshwater Aquatic Angiosperms 75(4) 1988 Rhizophoraceae—Anisophylleaceae 77(1) 1990 The Biological Diversity and Evolution of In 1983 a series of guides to plant collecting began. Published so far are guides to collecting bamboos (70-1), Pandanaceae (70-1), aquatic and marsh plants (71-1, 73-4), passionflowers (71-4), Araceae (72-2), palms (73-1), Lecythidaceae (74- 2), botanical specimens (73-3), and Cyclanthaceae (74-4). Several additional contributions are in prep- aration. By 1978 the need for a split in the scientific publications was satisfied by initiating Monographs in Systematic Botany from the Missouri Botan- ical Garden to accommodate works too large for the Annals. Monograph number 1 was A Provi- sional Checklist of Species for Flora North Amer- ica. The second monograph was E. R. Eisendrath’s Missouri Wildflowers of the St. Louis Area. The most recent addition to the series is number 31, Carlyle A. Luer's Icones Pleurothallidinarum. VI. Systematics of Pleurothallis—Subgenus Ancipitia Subgenus Scopula—and Trisetella. the Tarweeds As the first century of scientific publications ends, new challenges are emerging. How do we distribute our works to individuals when soaring publication costs force prices to levels only afford- able by large institutions? What is the optimal role for computers in the publication process? As the Neotropics, featured for so long in our publications, are deforested and transformed culturally, where do we aim our view, and in what tones? However we answer these questions, we remain confident that systematic botany and related fields are ex- panding, diversifying, integrating, and surging for- ward in exciting new directions. Here lies the real challenge and joy of stepping into our second cen- tury. —George К. Rogers, Editor THE REAL WORK OF Michael E. оше? SYSTEMATICS!” ABSTRACT Two recent revelations, that the number of species is much greater than previously thought, and that they are disappearing at a frightening rate, should impel ipa to question the implicit objectives of their discipline. It is impossible, using traditional methods, to describe ar nd classify most species in the lesser-known groups. It is suggested that systematists and other organism- -ecological biologists must collectively establish criteria for research priorities so that the “real work" of biology in the next few decades can be achieved. Research on many fronts is essential if we аге to maintain a significant fraction of the planet's Кок. ern binomial system of nomenclature. Living sys- tematists are the proud carriers of that tradition. Blessed is he who has found his work; let him ask no other blessings. —Thomas Carlyle, Past & Present, The project has accomplished a staggering amount Bk. III (1843) of work. In all, over 1.4 million species have been described and classified, using time-tested, if in- flexible, formulae. Until recently, it was understood that this project was about half finished. Now we know differently. It is apparent that the biological world is richer than any of us could have imagined just a few decades ago. Recent investigations in the tropics (Erwin, 1988) have increased our estimates of the number of fellow species on this planet from a few WHENCE THE CLASSIFICATION PROJECT Now? Two recent revelations should impel a deep anal- ysis of the premises on which systematics has op- erated. The first revelation is the discovery that 300 years or so of laborious taxonomic work has not brought civilization within 10% of the way to million to tens of millions, mostly tropical arthro- its goal of describing the biological world. This pods. What are the implications of these hordes of would be arresting even if it were not for the second undescribed species for systematics? Is this a mar- revelation, which is that the planet is on the verge velous challenge or a task of Olympian impossi- of an anthropogenic mass extinction that will an- bility? Either way, the great Linnaean project has nihilate much of its biotic diversity even before it received a shocking setback — one that must, soon- can be cataloged. er or later, trigger a searching appraisal by those These facts the unsuspected vastness of biotic scientists who see it as their mission to classify and diversity and its current vulnerability—are a ma- catalog all forms of life. jor challenge to the taxonomic enterprise, including Of course, the problems are greater for some its goal, its cultural role, and its methodologies. Оп taxa than for others. For vertebrates, the classi- the other hand, this is an opportunity for system- fication project is certainly more than half done, atists to take stock and ask what is the real mission, although ichthyologists are naming some 250 new the "real work" of systematics. species of fish each year, most of them small, fresh- Almost 250 years ago, a great project to classify water species. For a few popular groups— vascular all life was systematized by the Swedish naturalist plants, vertebrates except fishes, butterflies—the Linnaeus, who is considered the father of the mod- — project may be within 10% of its goal (Peter Raven, ' This and the four articles that follow it are the proc e of the 35th Annual Visi Sy mposium of t Missouri р al Garden, Conserving Biological Diversity: Prospects for the 2 151 he entury. The 5 үрөш) was held 7-8 Oct. 1988 at the Missouri Botanical Garden in St. Be uis, Missouri, U.S.A. The ШЫ а cs Symposium is үй эз) їй ы by a grant from the be onal Science Foundation. We а ас knowledge ean continued support for the 34th year of this 35-year series. 2 The title of the paper is d in Dan from Gary Snyder's book, The Real Work. To Joel Cracraft, Paul Ehrlich, Peter Rave en, ar n Warren H. Wagner I am grateful for much good advice, but, obviously, responsibility for 3 School of Natural enden University of Michigan, Ann Arbor, Michigan 48109, U.S.A. Current address: Environmental Studies, University of California, Santa Cruz, Santa Cruz, California 95064, U.S.A. ANN. Missouni Bor. GARD. 77: 4-12. 1990. Volume 77, Number 1 1990 Soulé 5 Real Work of Systematics pers. comm.), though formidable challenges re- main. Many hundreds of plant species are being described each year. However, the taxonomic task assumes mind-boggling proportions for such taxa as arthropods, polychaetes, and nematodes. Even before the scale of the recent revelations was apparent, Peter Raven (1977) wrote the fol- lowing: “It is too late in the history of the world to think that there is time to produce ordered classifications of all plants, animals, fungi, and mi- croorganisms, and then to employ these classifi- cations to seek new kinds of generalities while these organisms are still extant." Why is it too late? Let’s take a quick look at b the "extinction scenario," using plants as our ex- ample. Plants are numerous, their regional diver- sity is probably representative of other groups, and they are relatively well known taxonomically. About half of the 250,000 species of plants are associated with just 6-7% of the land surface—tropical for- ests. Most of these forests are likely to be destroyed or greatly disturbed during the next few decades. What kind of extinction scenario does this portend? The paucity of information on the geographic extent (range) of most tropical species currently prevents us from estimating accurately the con- sequences of so much habitat destruction. Most of the current estimates assume that the density of protected areas (nature reserves) is too sparse to capture many locally endemic species. If this prem- ise is wrong, and most tropical species are relatively widespread, extinction rates could be lower than most futurists predict, at least in the short run. A conservative estimate, I believe, would be that 25% of tropical plant species will be extinguished by the year 2020. In other words, we will see a loss of about 34,000 species of plants, 12.5% of the world’s flora, within the next few decades. A much larger fraction of insect species is likely to be lost, how- ever, assuming that they are restricted to smaller geographic ranges than plants. Returning now to the question of the great Lin- naean enterprise, do we have the time to finish the classification project of earth’s lesser-known taxa (e.g., arthropods, nematodes, mites) before most of them vanish? What would it take to accomplish this task? Recently E. O. Wilson (1988) assumed that there are, at most, about 1,500 systematists competent to deal with tropical taxa. Assuming an average output of new descriptions by this subset of taxonomists of about five per year (this being the historical average, assuming that the effort began a little over 200 years ago), or 7,500 total descriptions per year, and given there are 30 mil- lion more tropical species to collect, describe, and classify, it would take 4,000 years to do the job at the current level of effort. (The total number of animal species from all regions described each year is about 10,000 Next, let’s assume that it is possible to mobilize the necessary economic resources and the edu- cational and scientific talent to increase vastly the number of tropical arthropod taxonomists from 1,500 to a whopping 15,000 “barefoot taxono- mists”” (Soulé, 1989), thus increasing the output of descriptions of new tropical species from 7,500 to 75,000 per year, or 205 per day. Ignoring the potential information glut and the blizzard of reprint requests, the task would still require 400 years using current procedures. Recall that we have only two or three decades to get the job done, assuming current rates of habitat loss. So much for this Quixotic brute force option. The training of huge numbers of potentially un- employable systematists is futile, and it cannot be- gin to meet the extinction challenge. It is also futile to continue the task of describing biological diver- sity in the manner of the 18th entury, writing, in other words, the slightly premature obituaries of millions of bugs and worms and those of a few thousand randomly selected, undescribed species of plants and vertebrates. I am not saying that we shouldn't be putting more resources into training taxonomists. Quite to the contrary. For most taxa, there are obviously too few specialists and the need for them is growing rapidly. In addition, our institutions are under- funded; many museums, herbaria, zoos, and bo- tanic gardens have not been able to compete with fashionable “big science" projects, such as map- ping the human genome, strategic defense technologies, and huge particle accelera- tors. Finally, we are tantalizingly close to wrapping up several taxa (including plants, vertebrates, some marine phyla, and a few insect groups), and these groups must be completed as soon as possible, so that biogeographic and evolutionary analyses can be based on complete data sets. Nevertheless, the twin crises of “too many species and too much extinction” will not be solved by the mindless cloning of thousands of taxonomists. Priorities must be established (Raven, 1977; 1980; Soule & Kohm, 1989). For systematics and “business as “star wars” other organismic disciplines, it is not usual" in the closing years of the 20th Century. BARRIERS TO A REVOLUTION IN SYSTEMATICS In responding positively to these twin crises, we must look inward as well as outward, confronting Annals of the Missouri Botanical Garden AFRICAN BUFFALO NUMBER OF SUBSPECIES 0 Li Т T LI 1860 1880 1900 1920 1940 1960 1980 YEAR FIGURE 1. An example of excessive HE for Bu- balus caffer, the African buffalo. See te intrinsic and extrinsic problems. One intrinsic issue is our rhetoric. EXTINCTION HYPERBOLE? I am concerned that some of our own rhetoric may come back to haunt us. There is no doubt that humans are initiating a massive spasm of species extinction, and that the rates of extinction may be on the order of 100 species per day in a few decades. But in our panic and despair are we con- sciously or unconsciously obscuring the complexity of the catastrophe? Is there a conspiracy of silence on the issue of the identity of most of these species? To biologists, the fact that more than 95% of these threatened species are arthropods and nema- todes hardly makes the problem less serious, but to the man on the street, this bit of information would change matters considerably. Such little crit- ters don't arouse a lot of public concern— just the opposite. Most people don't consider arthropods and helminths to be animals. Most people, if they new this truth. about the upcoming extinction spasm, would probably say "good riddance." Among the more popular taxa, plants and ver- tebrates, the losses will not be in the millions nor will they come close to approaching a rate of ex- tinction of 100/day. The point that we need to make is that there aren't millions of kinds of birds and mammals. In fact, there are precious few. Relative to the vast number of beetle and mite species, there is hardly a handful of these verte- brate relative On the der hand, we need to explain that: (1) you don't have to have warm blood to be an animal, (2) these “creepy” species have the same right to existence as their green and charismatic cousins, and (3 ў worms is signaling the loss of habitat and ecosystem — the demise of vast numbers of bugs and services (Ehrlich & Ehrlich, 1981) on which all other species, including humans, depend. ANACHRONISMS AND CHALLENGES The closet of systematics has more than its share of skeletons. For example, many systematists have pointed out that taxonomy fell into disrepute in the early decades of this century because an extreme form of typology prevailed that led to the appli- cation of species and subspecies names to the most trivial intraspecific variants. (The field is still plagued by pockets of extremists.) As shown in Figure 1, for example, some people apparently got a little carried away in describing "new" African buffalo. was reached in about 1935 when Zukowsky on one occasion p e ferent, new racial names to the two horns single African buffalo Bubalus caffer (Ansell, 1971). We may smile at the excesses of the past, but these excesses are symptomatic of a perennial de- bate—the species problem. Should every geo- graphic (evolutionary) entity be elevated to species rank, as Cracraft (1983) suggested? Some con- servation issues would be instantly solved by such a solution, at least in theory. For example, if sub- species no longer existed, there might be less debate about whether they should be interbred in propa- gation projects. Of course, it is a delusion to think that this would really solve the problem, because genetic relatedness would still be a matter of de- gree. Fortunately, many zoos have already adopted genetic rather than strictly taxonomic criteria for such matters (Benirschke et al., 1980; Ryder, 1986; Ryder et al., 1988) Practices in taxonomy are changing, but are they changing fast enough (Ehrlich, 1964; NAS, 980)? Methodological stability in such a quasi- legalistic field is commen — able, but there must also be sufficient flexibility to accommodate changes in technology and other conditions. For example, the methods for storing museum specimens (most spec- imens are still preserved as dried organs or as bleached cadavers—in biochemical terms, degrad- ed protein and decomposed DNA) must take into account advances in technology as well as the unfortunate fact that many of today's specimens will become proxies for tomorrow’s extinct taxa. Of what value will these specimens be to a 22nd- Century biologist? Put another way, how much more valuable would they be if they were bio- chemically intact? This is not to say that specimens preserved in the traditional ways (drying, fumi- gating, denaturing) lack scientific value. DNA frag- ments can be recovered from specimens that have been in alcohol for decades (Páábe, 1985), but this Volume 77, Number 1 1990 Soulé 7 Real Work of Systematics bit of serendipity should not be used as an excuse to decelerate the rate of conversion in many mu- seums to more modern methods of preservation. These days, a collector without dry ice or liquid nitrogen is an anachronism. ““THERE ISN’T TIME FOR MORE RESEARCH” Extreme anxiety can also be a barrier to change. If the field of systematics is to live up to its potential in the campaign to preserve the diversity of life forms, then it cannot fall victim to slogans born of ignorance. To be specific, one often hears that we already know enough—that we have enough knowledge about the systematics, biogeography, natural history, ecology, and genetics of organisms and ecosystems to establish a rational and com- prehensive set of nature preserves that would pro- tect most species from the coming Armageddon. It is asserted that there isn't time to indulge in research— we must simply buy more land and lock it up. Is this true? Do we already know enough to protect biological diversity? We don’t. Research in conservation biology dur- ing the last 15 years has altered fundamentally the design criteria and management objectives for pro- tected areas. The rising curve of new, manage- ment-relevant discoveries shows no sign of asymp- totic approach. Why do I raise this issue here? It is not to point the finger at systematists, who are rarely the perpetrators. Rather it is to enlist the support of systematists for conservation biology sensu lato. Many kinds of knowledge are necessary for the successful design and management of protected areas and propagation projects. The first step in many situations is to describe, inventory, and map biotic diversity. This is because the exact locations of reserves are critical, particularly if the objective is to protect the maximum number of species. The entire conservation enterprise depends on system- atists and biogeographers for guidance about where to place reserves, particularly in regions where there exist “hot spots” of endemism and species diversity (Diamond, 1986; Gentry, 1986; Myers, 1988; Soule € Kohm, 1989). On the other hand, the maintenance of biodiver- sity in a fragmented landscape is a more complex matter. It depends on scientific progress along many fronts, among which systematics is just one. Even if a reserve is in the right place, it will gradually lose many of its species unless managers are at- tuned to the effects of climatic change, fire regimes, siltation and sediment load (in marine and aquatic systems), patch dynamics, sea level rise, pollution, edge effects, viability of keystone species, loss of mutualists, migratory life cycles, and current and potential human interactions with species and wild- lands. Ignorance of these and other phenomena will produce a bitter harvest of conservation failures and wasted resources. It goes without saying that most of the above kinds of research depend on a taxonomic foundation (Boom, 1988). Ecological research in the last decade has led to profound changes in the ways that conservation projects are designed and managed. For example, only recently have we realized the ubiquity of the deleterious effects of edges. The effective size of a reserve is often much smaller than its map size would indicate because many diversity diminishing agents penetrate great distances into reserves. Many of these edge effects are only beginning to be under- stood. For example, Appanah (1987) pointed out that meliponid bees nesting as far as 1 km from the edge of a reserve were returning to their nests with 100% pollen from plantations. Because these bees are important tree pollinators in the forests of south Asia, such behavior could lead to wide- spread reproductive failure and the gradual die-off of forest interior species. The future status of the Everglades National Park at the southernmost tip of Florida, one of the richest landscapes in North America, offers an example of how recent discoveries can and should modify our management methods and priorities. Larry Harris (pers. comm.) has pointed out that the gradient in this part of Florida is 1:25,000, and that, given the accelerating rates of sea level rise (Titus, 1986), there is virtually nothing we can do to prevent the disappearance of most of this park under Florida Bay and the Gulf of Mexico in 50 to 100 years. There are many endangered species in the Everglades, but existing recovery plans have ignored this inevitable source of habitat loss and its implications for population viability. Genetics provides many examples of how very recent studies have altered the management of conservation projects. Genetics was virtually ig- nored by managers until the late 1970s. A few prophets (e.g., Frankel, 1974; Seal, 1978) had earlier warned of the hazards of inbreeding and the loss of genetic variability, but there were hardly any data from rare or captive species, and there were certainly no concrete guidelines of use to managers. Only in the last 10 years has evidence of widespread inbreeding depression in captive groups been uncovered (Ralls et al., 1988, and references therein). It is only in this decade that guidelines and protocols for wild and captive stocks have been suggested (Franklin, 1980; Soulé, 1980; Annals of the Missouri Botanical Garden Frankel & Soulé, 1981; Schonewald-Cox et al., 1983; Templeton & Reed, 1984; Lande & Bar- rowclough, 1987). Notwithstanding that the orig- inal caveats accompanying these principles and guidelines have been largely ignored, it is impres- sive how rapidly genetics has been assimilated into the mainstream of captive breeding, recovery plan- ning for endangered species, and the management of small groups of large animals. The problem is that once assimilated, people tend to take such information for granted, and to forget how important conservation biology has been in shaping current management practices. During crises, the value of past research is often ignored. In doing this, we commit an even graver and more perilous sin: discounting the value of future re- search. Those who believe that we cannot simultaneous- ly secure land and do more research fail to ap- preciate that, metaphorically, the storage of a pre- cious book or painting in a secure vault does nothing to prevent its gradual, chemical deterioration. To put aside land without knowing how to manage it is folly. It should be noted parenthetically that agencies like the National Science Foundation (NSF) may be legally barred from purchasing lands in developing countries, but they are not barred from funding research that would help insure the pro- tection of biotic diversity on such lands. In our panic to secure the few remaining bits of wild nature, we should not forget that our un- derstanding of biological diversity, particularly in the tropics, 15 shockingly superficial. Therefore, we have no choice but to proceed urgently to study the basic mechanisms that fuel, threaten, and main- tain the biotic complexity of this planet. Given the rate of habitat destruction, much of this research must be accomplished within the next few years or decades at most. The maintenance of biotic diversity, in situ and ex situ, will depend largely on the quality and quantity of these studies. The next step is to ascertain the most critical research needs. Recently, a workshop sponsored by NSF was convened by the Society for Conser- vation Biology (SCB) at Hawk's Cay in Florida to frame a report on research priorities in conser- vation biology (Soulé & Kohm, 1989). Following are the most pressing and important initiatives and research needs agreed upon at the workshop. Sys- tematics is at the heart of the first priority. 4 crash program to carry out extensive surveys and mapping to identify areas that are critical for the protection of nature and genetic resources. Reiterating the recommendations of an earlier report (NAS, 1980), these critical. areas should be defined in terms, among others, of their high biotic diversity, high levels of endemism, or imminent destruction of critical or unusual habitats and/or biotas. These studies should emphasize taxonomic groups that are better known or those that would indicate parallel biogeographic patterns in groups less amenable to censusing. А byproduct of this research could be critical information on the rates of deforestation and other forms of habitat destruction. 2. [t is particularly important to understand how natural systems "work," especially in the tropics. Therefore, the group called for the im- mediate establishment of a small number (perhaps four to eight) of research sites in the tropics in order to carry out a coordinated program of com- parative research on populations, communities, and ecosystems in relatively pristine and secure situ- ations. The workshop participants agreed with the authors of the authoritative NAS report, Research Priorities in Tropical Biology (1980), in rec- ommending the establishment of several major eco- logical research sites in the humid tropics where in-depth, long-term, and globally coordinated stud- ies are supported. These focal sites would be especially valuable as sources of long-term, baseline information on global and ecological processes. The SCB/NSF workshop also recommended the active participa- tion of local students, professionals, and institutions in this program and other research projects in their developing countries. One reason that the group did not recommend a large number of such sites is because there are too few researchers with the necessary expertise. 3. Studies at all spatial scales to assess the kinds, mechanisms, magnitudes, and impacts of humans on ecological systems. Here are included the effects of habitat fragmentation, biotic mixing (introductions), and air, water, and marine pollu- tion. These studies should focus on the development and evaluation of alternative means of exploitation and land/water use, with the goal of improving human welfare while minimizing environmental de- terioration and the destruction of biological diver- sity. 4. Studies on the physiology, reproduction, behavior, ecological interactions (including dis- eases), and viability of individuals, populations, and species have been essential in the protection and management of reserves and other wild- lands. The group urged the enhanced support Volume 77, Number 1 1990 Soulé 9 Real Work of Systematics for research that focuses on these fundamental, species-level processes and relations, especially with regard to species of critical ecological or economic importance. 5. Education in conservation biology, wild- lands management, and related areas with the objectives of training basic scientists and natural resource managers, particularly in tropical, de- veloping countries. Much of this training should occur locally and regionally, and should benefit local institutions and strengthen the conservation and management infrastructures in developing na- tions. The above list omits direct mention of global phenomena that affect landscape arrangement and habitat quality. These phenomena are of para- mount importance for the protection of biodiversi- ty, and are being intensely studied by other groups of experts. To summarize, the maintenance of biotic diver- sity in protected areas and in ex situ facilities will depend on many disciplines. Success will require a level of tolerance and a degree of scientific plu- ralism that are uncharacteristic of organismic and environmental biologists, Systematists should take the lea h in this “new age of biological brother- WHAT ROLES FOR SYSTEMATISTS IN THE BIODIVERSITY CRISIS? Granting the central role of systematics in con- servation biology, what can be done to involve more systematists (and other organismic biologists and ecologists) in the campaign for the protection of biological diversity? First, let me restate the prob- em: there are far too many taxa to be named and classified in the short time remaining (Ehrlich, 1964). Systematists might consider the following suggestions, none of which are original with me. 1. For relatively unknown groups like tropical insects, concentrate on a few “representative” taxa, hoping that the phyletic, morphological, bio- geographic, and ecological patterns manifested by these taxa are typical of related ones. Which groups should be chosen? Who would choose them? How should interim conclusions be validated? Answers to these questions require an unprecedented level of cooperation, compromise, and organization. 2. Focus on ecologically keystone and indi- cator taxa and their mutualists. In other words, opt an ecological approach, letting nature’s structure help shape taxonomic priorities, in con- trast to the present haphazard selection process driven by economics (agriculture and entomology), personal whims (“I like sceloporine lizards”), and theoretical issues in evolutionary biology. One such approach is to concentrate on taxa containing keystone species. Within the last decade or so, conservation biologists have emphasized the critical ecological roles of keystone species. Defined operationally, a keystone species is one that, by its effective absence from a system, results in the virtual disappearance, directly or indirectly, of sev- eral other species, causing, i ther words, an extinction cascade. аны у is undefined, and further work on the utility of this concept is ob- viously necessary. In addition, the concept is in need of a great deal more empirical and theoretical analysis. Sev- eral avenues are being pursued (Mills & Soulé, in prep.). For example, one can divide keystone species into two categories, trophic keystones and struc- tural keystones. The absence of the former kind will lead to dramatic changes at one, two, or more trophic levels below. The absence of structural keystones leads to changes in habitat, which in turn causes significant shifts in abundance of other species that may or may not be trophically con- nected to the keystone. Table | lists some of the categories of keystone species and gives examples of the likely consequences of their effective loss (including human-induced rarity) from a commu- nity. Effective management of protected areas de- pends on an understanding of the interactions of keystone species. Terborgh (1986) found, for ex- ample, that figs and palm nuts are keystones in a Peruvian tropical forest, where they may be the only food resources during the annual period of food scarcity. Palm nuts escape from all but a small group of specialist species that are either large and with powerful jaws or else can gnaw the nuts. Thus, among the few customers are peccaries and capu- chins, comprising ca. 30% of the total biomass of fruit-eating animals. Fig trees, though they fruit irregularly, are often keystone producers; figs are heavily consumed by all larger primates, procyon- ids, marsupials, and many birds. Terborgh (1986) concluded that a group of only 12 plant species (out of 2,000) maintains almost all large frugivores for about three months of the year. If we are to succeed in the protection and man- agement of the remnants of biodiversity, there is an urgent need to refine and deepen our under- standing of keystone species, especially in the trop- ics where most of the planet’s biological riches exist Annals of th Missouri a Garden TABLE 1. Some kinds of keystone species and the effects of their effective removal from a system. Keystone category Effect of removal Examples Trophic/resource keystones Top predators Large increases in the abundances of prey species and smaller predators, and sub Felids, canids, fishes sequent extirpa- tions of some of the latter’s prey species Pollinators and other mutualists Ф за c rz E =: E Б ga © © = E B id O a > Providers of essential resources scarcity Structural keystones Species that maintain landscape features Herbivores that pre- vent succession Failure of Көнө and recruitment in certain Local extirpation of dependent animals, including fruit- and nectar-eating species during times о Disappearance of water holes and wallows, ponds, Return of cover and decrease in habitat diversity; Hymenoptera, Lepidoptera, sym- bi l fungi of plants, bleaching iotic algae, Trees, such as Ficus and trees that provide nesting and hi- bernating sites Tapirs, beavers, alligators Moose, elephants, rabbits disappearance of species dependent on early successional habitats and resources and are currently at great risk. Current research efforts in this regard are pitifully inadequate, and the low level of funding for basic research in trop- ical ecology in general is simply scandalous. 3. Focus systematic work on phylogenetic rel- icts or other evolutionary outliers. Evolutionary outliers contain precious information in their ge- notypes and phenotypes and should rank at the top for systematic attention. Obvious target taxa include oligotypic marine phyla, proboscidians and other remnants of the great mammal radiation, and the last representatives of genera and families that are about to disappear, such as Hibiscadelphus in the Hawaiian Islands (Gentry, 1986, and refer- ences therein). 4. Focus systematic work on local endemic taxa inhabiting vulnerable environments. Habitat fragmentation and regional and global climate changes will eliminate local endemics, including those that inhabit estuaries, reefs, boreal mountain tops in low latitudes, and tropical forests. Other kinds of biogeographic cul-de-sacs harbor some interesting examples. A case in point is the fauna that is restricted to the relatively cool waters of the north end of the Gulf of California, Mexico (Brownell, 1986; Perrin, 1988). As global warming develops, many of these taxa, including the por- poise Phocoena sinus, are likely to be squeezed out of existence, trapped between the land on the west, north, and east, and the warm waters to the south. 5. Develop new (interim) approaches for de- scribing and classifying, especially for groups like tropical arthropods for which traditional ap- proaches are too slow by orders of magnitude and for which binomials can wait (Soulé, 1989). Pm not speaking of DNA fingerprinting using restric- tion fragment length polymorphisms, nor the use of mitochondrial DNA, nor the use of electropho- resis for the detection of protein variants and the measurement of their frequencies, nor DNA hy- bridization studies, nor similar techniques. These are all too costly and time-consuming. I am refer- ring to automated methods for screening, cheaply and efficiently, very large numbers of specimens and assigning them to taxa of convenience. These methods might include three-dimensional tomog- raphy for computerized clustering based on mor- phology and automated chromatographic screening for clustering based on biochemical patterns in such tissues as cambium and hemolymph. Other suggestions have been made that would enhance data retrieval, comparability, and depth. For example, data entry formats should be devel- oped by international teams of systematists and conservation biologists (ecologists, geneticists, and biogeographers) to ensure that the data will be useful to those asking different sorts of questions. Unless such a team approach is implemented, and such formats and methods are standardized by in- ternational agreement, much of the current in- vestment in data banks for systematics will be wast- Education is also a high priority. Systematists are in a better position than most biologists to plant the seeds of a conservation ethic. They are often in a position to train local people and to engage leaders in discussions about the latter’s biotic patri- Volume 77, Number 1 1990 Soulé 11 Real Work of Systematics топу. Аз teachers, systematists can profess the love of nature to students, the younger the better. On OUR REAL Work The last of the preceding proposals brings me to my final point—it is OK for systematists to speak of love and beauty. It is even OK for systematists to express emotions in public. Real adults are not and ““sentimen- tal.”” Real adults have left behind the Rambo de- velopmental stage and its preoccupation with machismo. 1 don't mean to deny or even to deni- grate the harder aspects of human nature. The “right stuff” is adaptive in many circumstances, and ambition can be harnessed just as well for the protection of creatures as for their destruction. Nevertheless, I believe that maturity includes the courage to embrace publicly stewardship as a “‘fa- milial” responsibility. Giving succor to the earth is our final and most adult task, our real work (Sny- der, 1980). Some may wonder how can we be effective, let alone charismatic, in communicating these feelings of kinship and concern in our work (Wilson, 1984; Soule, 1988)—our research and teaching—with- out appearing foolish. Perhaps, we can’t. Perhaps, for a real adult, appearing foolish to less mature peers is as inevitable as work itse By “work” I am referring to our careers as well as to our real work, which is to love the earth by preserving its actual and potential diversity. Re- garding career, Freud said some interesting things about work and its relationship to happiness. For him Liebe und Arbeiten, love and work, were not separate compartments. Freud considered the professional work of *'civilized" men and women, with all its grasping for recognition and respect, as sublimation for the giving and receiving of love. If there is some truth in this, if each of us wishes to contribute something of lasting value to the world (giving love), and to be acknowledged for it (re- ceiving love), then we biologists are fortunate in- deed. It is relatively easy for us to express our love of biotic diversity through our research, writing, mentoring, and teaching. And in return, there is not only peace of mind, but also the gentle fellow- ship of coconspirators. For us, conservation biology is the synthesis of love and work. LITERATURE CITED ANSELL, W. F. H. 1971. Order Meester & H. W. Setzer (editors), ie ie An Identification Manua onian Institution Press, з add D. Жейин, S. 1987. Insect pollinators and the diversity г Artiodactyla. In: J. ii of -84. Smith- of dipterocarps. Pp. 277-291 in A. J. G. H. Kos termans (editor), Proc. Third Roid Table ) реў оп гое UNESCO, Jakarta Же, K., B. Ley & O. RYDER. 1980. The technology af Sect propagation. Pp. 225-242 in M оше & В. A. Wilcox (editors), Conservation An Evolutionary-Ecological Perspective. nauer Assocs., Sunderland, Massachusetts eeu: - es À new d for systematibs: the personal component. ASC Newsletter 16: 1-3. BROWNELL, R. L., 1986. Distribution of the To "hocoena T in Mexican waters. Marine mal Sci. 2: -305. CRACRAFT, J. 1983 Species concepts and speciation analysis. Current Ornithology 1: -187. ао. E The design a a nature reserve D stem for ннан New a. Pp. 485-503 п М. E. Soulé (editor), С Win ^ Biology: The bd of Scarcity and der. Sinauer Assocs. Sunderland, Massachuse Енкисн, P. R. 1964. Ses axioms of taxonomy. Syst. Zoo 5 109-123. A. H. EHRLICH. 1981. Extinction: the Caus- es and Consequences of the Disappearance of Species. Random House, New York. ERWIN, T. 1988. The tropical forest y the heart of biotic diversity. Pp. 23-29 in E. O. Wilson & F. M. Peters (editors), Biodiversity. National Academy Press, Washington, D.C. FRANKEL, O. H. 1974. Genetic conservation: our evo- lutionary responsibility. Genetics 78: 53-65. 81. Conservation and Evo- lution. Cambridge Univ. Press, New York. FRANKLIN, I. 1980. Evolutionary dew in small popu ulations. Pp. 135-149 in M. E. Sou B. Wilcox (editors), Conte Biology: M Emu: tionary-Ecological Perspective. Sinauer Assocs., Sun- derland, Massachusetts 1 GENTRY, А. ndemism in Mx versus temperate plant communities. Pp. 153-181 in M. E. Soulé (editor), Conservation Biology: The Science sity. Sinauer phe Sunder- land, Келна BARROWC LOUGH. 1987. Effective pulatio ment (editor), Viable a for Conservation. Cam- bridge Univ. Press, Cambridge € New York Mitts, S. & M. E. SouLé. Updating and = the keystone species concept. E dk 3n bins) N. 1988. Threatened bio "орок" їп trop- ical forests. The Envir ed 8: 1-200. NAS (NATIONAL ACADEMY OF SCIENCE 1980. Re- search Priorities in Tropical Solo. National Acad- emy of Sciences, Washington, D.C. РААВЕ, 5. 1985. Presence of DNA in ancient Egyptian mummies. J. Arch. Sci. 12: 411-418. PERRIN, W. F. (editor). 1988. Dolphins, Porpoises, and Whales: An Action Plan for the Conservation of Biological Diversity: 1988-1992. IUCN, Gland, Switzerland. Rats, K., J. D. BALLOU & A. TEMPLETON. 1988. Es- timates of lethal equivalents and the cost of inbreed- in mammals. Conservation Biol. 2: 185-193. quies P. H. The systematics and evolution of igher plants. /n: The Changing Scene in Natural Annals of the Missouri Botanical Garden Sciences, 1776-1976. 2: y ences, Special Publ 12: 59-83. RYDER, О. А. 1986. Species conservation and system- atics: the кашг of subspecies. Trends in Ecology and Evolut uti : 9. J.H.S w & С. M. WEMMER. 1988. Species, subspecies i. ex situ conservation. Int. Zoo Yearb. (1988) 27: 134 ree SCHONEWALD-COX, S. M. CHAMBERS, F. Mac үү . Tuomas (editors Narr bene and Co servation: A Referenc r Mana Wild ; eee and Plant Populations. ee кыйкы Menlo Park, California. Seat, U. S. 1978. The Noah's Ark problem: ds n- eration management of wild species in captivity. Pp. 3 314 in S. A. Temple (editor), puede Birds уне Techniques for Preservation of Threat- ened Species. Univ. Wisconsin Press, Madison, Wis- consin. SNYDER, G. 1980. The Real Work. New Directions Books, New Yor SouLÉ, M. E. 1980. Threshold B survival: ps ing fitness and evolution atial. Pp. 169 in M. E. Soulé & B. A. Wileo ie 6 onservation y: Ап pa iren Perspective. Sinauer Assocs., Sunderland, Massachusetts 1988. Mind in Ше ко. aad of the of Natural Sci- biosphere. Pp. 465-469 in E. O. Wilson (editor), Biodiversity. National Academy Press, Washington, D 1989. Challenges to conservation biology in the next century. Pp. 297-303 in D. Western & M. Pearl (editors), Conservation Biology for the Next Century. Oxford Univ. Press, New York. є K. Конм (editors). 1989. Research Prior- ities for Conservation Biology. Society for Conser- vation Biology, Special Publ. No. 1 TEMPLETON, A. R. € B. REED. 1984. Factors elimi- nating inbreeding i in a captive herd of Speke's gazelle. Zoo Biol. 3: 177-199. TERBORGH, ГМ а а resources іп the tropical forests. Pp. 330-334 in M. E. Soulé (editor), The Science of Scarcity and ~ rsity. Sinauer unes. Sunderland, Massachu- xus T E. 1986. Greenhouse effect, sea level rise, and coastal zone an Coastal Zone Man- agement J. 14: -Lbfl. WirsoN, E. O. 1984. Biophilia. Harvard Univ. Press, Cambridge, Massachusetts 8. The current state of biological diversity. 3- 18 n К. О. Wilson & F. M. Peter (editors), Biodiversity. National Academy Press, Washington, D.C. THE GENETIC CONSEQUENCES OF HABITAT FRAGMENTATION! Alan R. Templeton,? Kerry Shaw,? Eric Routman,? and Scott K. Davis? ABSTRACT The natural habitats of many species have become fragmented into small “islands,” principally by human activities. In this paper we discuss the long-term genetic and evolutionary consequences o studies on populations that have undergone natural habita the highest land formation found in the midwestern United States. Because o t fragmentation in the O ragmentation as inferred from а most severe genetic с On the positive side, the genetic variation of a fragmented species is not totally lost but is often present as fixe genetic variation at the global level for long periods of time. We discuss the optimal design for a recolonization program to prevent global extinction and to maintain high levels of global genetic variation. As human use of the environment expands, the amount of habitat available for natural communities decreases and the remaining habitat becomes in- creasingly fragmented. That is, habitats become subdivided into “habitat islands” surrounded by different, usually human-altered, environments. Many tropical and temperate habitats have already been extensively fragmented, and the amount of fragmentation will increase substantially in the fore- seeable future. Conservation biologists are concerned about habitat fragmentation because of its potential to increase extinction rates as predicted by island biogeographic theory (MacArthur & Wilson, 1967), because of “Allee” and “edge” effects, and be- cause of its potential to erode genetic variation through genetic drift in small populations and to promote inbreeding depression. Accordingly, there have been several studies on the impact of habitat fragmentation, but most of these have focused on recent, human-induced habitat fragmentations (e.g., Soule et al., 1988). Because conservation biology is concerned with the long-term maintenance of biodiversity, we must deal with the long-term eco- logical and genetic consequences of habitat frag- mentation as well as the short-term ones. Once we have a clearer idea of what these long-term con- sequences are, we can address the question of how best to manage species and communities in frag- mented habitats in order to preserve biodiversity for thousands of years. This paper will focus on the long-term (thousands of years) consequences of habitat fragmentation. In order to study this problem, we cannot use ' We thank the many graduate and undergraduate students who aided us in the collection and genetic surveys of ag the various species discussed in this paper. Our special thanks go to Elizabeth Waters and David Guttman for their ork excellent w on Trimerotropis saxatalis. We also thank George Rogers and two anonymous reviewers for their comments on an earlier draft of this paper. This work was supported by NIH grant RO1-GM31571 and a contract from the Missouri Conservation Commission. * Department of Biology, Washington University, St. Louis, Missouri 63130, U.S.A. * Department of Animal Sciences, Texas A&M University, College Station, Texas 77843, U.S.A. ANN. MISSOURI Вот. GARD. 77: 13-27. 1990. Annals of the Missouri Botanical Garden recent, human-induced, fragmented habitats. In- stead we must look for natural cases of habitat fragmentation that have occurred over the last several thousands of years. THE OZARKS AS A NATURAL LABORATORY OF HABITAT FRAGMENTATION The Ozarks are the highest elevated land mass between the Appalachian and Rocky mountains in North America. The Ozarks are located primarily in southern Missouri and northern Arkansas, and extend into southern Illinois and eastern Oklahoma. Four features make the Ozarks an ideal natural laboratory for study of the long-term consequences of habitat fragmentation. The first feature, geological complexity, provides a mosaic of substrates for natural communities. About 1.5 billion years ago, several volcanoes were formed above a hot-spot in what are now known as the St. Francois Mountains in southeastern Mis- souri (Kisvarsanyi, 1980). Subsequently, this area was periodically invaded by shallow seas, whic were associated with the deposition of various sed- imentary strata, such as limestone and sandstone. s a consequence, the Ozark area has a complex mixture of igneous and sedimentary rocks, all of which are currently exposed at certain locations. The second feature is topographical complexity. After the creation of an extensive plateau by sed- imentary deposits, the Ozark region has been sub- jected to erosion. by surface and underground streams. The surface topography of the Ozarks was thus transformed from a plateau into a series of ridges and valleys through which flow numerous streams, and which are interspersed by karst fea- tures, such as sinkholes and collapsed cave systems. This topographical complexity coupled with the geological complexity mentioned above create much potential habitat diversity in terms of soil types, drainages, ms microc aine The third feature is t t matic changes that have occurred in this region Pani and after 18,000 to 12,000 B.P. (before present), a boreal spruce for- est existed in this area, only to be replaced by a predominantly oak-hickory forest. However, dur- ing the Ipsothermal Maximum of 8,000 to 4,000 B.P., the area was hotter and drier than at present, the Pleistocene Epoch. From about and the oak-hickory forest retreated and was re- placed by prairies and deserts. At the end of the Ipsothermal Maximum, the oak-hickory forest reinvaded as the climate became cooler and wetter (COHMAP Members, During these climatic йын, plant and an- imal communities were free to move in and out of the Ozarks because of its fourth critical feature: the absence of major barriers to dispersal around the Ozarks. Many diverse organisms were able to enter the Ozarks during climatically favorable pe- riods, only to be isolated from their main distri- bution ranges when the climate changed again. Because of the great potential for habitat diversity, relictual populations of many of these species were able to survive in the Ozarks. These relictual pop- ulations are found in naturally fragmented habitats that have been cut off from the remainder of their species distribution for a few to several thousands of years. We have been examining several species with relictual Ozark distributions for our studies on habitat fragmentation. MATERIALS AND METHODS Protein electrophoresis was carried out using horizontal starch gel electrophoresis and the meth- ods outlined in Selander et al. (1971) and Thomp- son « Sites (1986). Gels were prepared by mixing equal volumes of Connaught hydrolyzed starch and Electrostarch, and diluting with gel buffer to a final concentration of 12%. DNA was isolated as described in Hillis & Davis (1986). This procedure isolates total cellular DNA A and nuclear rDNA. The NA extracts were digested with various and thus contains mtDN cellular restriction endonucleases using the reaction con- ditions recommended by the supplier. Agarose gel electrophoresis, Southern blotting, and subsequent hybridizations were carried out as described by Hillis & Davis (1986). Cryptobranchus alleganiensis. Hellbenders were collected by lifting large rocks and grabbing by hand. Animals were anaesthetized in an aqueous solution of ethyl m-aminobenzoate methanesulfo- nate, and blood was collected from an incision in the tip of the tail. Blood samples were diluted in a buffer (0.1 M NaCl, 0.05 M tris, 0.001 M ETDA, pH 7.5) and immediately frozen in liquid nitrogen. All hellbénder were released after regaining con- sciousness except for one or two voucher specimens per river. Blood samples were stored at —80°C until DNA was extracted. Different fractions of the extracted DNA were digested with the following enzymes: BamHI, Bell, ВЕШ, BstEIT, Dral, EcoRI, EcoRV, Neol, PstI, Pvull, SacI, Stul, Xbal, and Xmnl. The mtDNA was probed with two homol- ogous clones that contain all but 3.6 kb of the hellbender mitochondrial genome as well as a clone of the entire mtDNA of the frog Xenopus laevis (pX131, kindly donated by Dr. Igor Dawid). Volume 77, Number 1 1990 Templeton et al. 15 Genetic Consequences of Habitat Fragmentation Crotaphytus collaris. Collared lizards were col- lected by hand or by noosing. At the beginning of our work with this species, it was necessary to sacrifice the animals to obtain genetic data. To perform the isozyme survey, 24 animals were col- lected from across Missouri. Heart, liver, blood, and kidney tissues were used for the isozyme anal- ysis, and skeletal muscle was used for DNA iso- lation. After this initial survey, all subsequent ge- netic data were collected from blood samples obtained nondestructively. Blood was collected by piercing the suborbital sinus with a heparinized capillary tube. The blood used for protein electro- horesis was kept on ice until returned to the laboratory. There, the blood was centrifuged for one minute in a hematocrit centrifuge to separate the serum from the red blood cells. The serum was added to an equal volume of distilled water and then frozen for later electrophoresis. The serum was used to score for three different nonspecific esterase loci, two leucine aminopeptidase loci (E.C. 3.4.11.1), two superoxide dismutase loci (E.C. 1.15.1.1), and three general protein loci. The red blood cells were lysed in 1.5 volumes of distilled water and then frozen for later electrophoresis. The red cell gels were scored for an additional esterase locus, alpha and beta hemoglobin, two lactate de- hydrogenase loci (E.C. 1.1.1.27), and malate de- hydrogenase (E.C. 1.1.1.37). Blood used for DNA extraction was diluted in buffer and kept either on ice or frozen in liquid nitrogen until returned to the laboratory, where it was stored at —80°C. The extracted DNA used for the rDNA and mtDNA studies was digested with Apal, BamHI, Bell, Bgll, BglII, В ЕП, Dral, EcoRI, EcoRV, HincII, HindIII, KpnI, PstI, Руш, SacI, SacII, Stul, XbaI, and XmnI. The mtDNA was probed with nick translated Crotaphytus col- laris mtDNA prepared from intact circular mito- chondrial DNA isolated from the 24 sacrificed an- imals by the methods of Lansman et al. (1981). The rDNA was probed with two clones that con- tained the mouse 185 gene (pF M84) and the mouse 285 gene (PI19), both kindly provided by Dr. Norman Arnheim. Cellular DNA for the analysis of hypervariable VNTR (variable number of tan- dem repeat) loci (“DNA fingerprinting””) was di- gested with НаеШ and probed with a subclone of the human clone 33.15, kindly donated by Dr. A. Jeffreys. Trimerotropis saxatalis. Lichen grasshoppers were caught with nets and placed in a cool con- tainer until returned to the laboratory, where they were stored at —80?C until being processed for electrophoresis and DNA analysis. Processing in- volved removing the digestive tract and grinding the internal tissue of the thorax in buffer or distilled water (for the protein electrophoresis work). The only results reported in this paper deal with the enzyme leucine aminopeptidase (LAP) and mtDNA digested with EcoRI. RESULTS AND DISCUSSION CRYPTOBRANCHUS ALLEGANIENSIS The numerous spring-fed streams that drain the Ozarks provide a habitat for the largest North American salamander, Cryptobranchus allegani- ensis, commonly known as the hellbender. With the exception of the Ozark populations, hellbenders are found only in the rivers that drain the Appa- lachians. Consequently, the Ozark hellbenders rep- resent relictual invaders from the eastern United States. Within the Ozarks, hellbenders are distrib- uted into two disjunct populations. One population inhabits the rivers that drain to the north from the Ozarks and empty into the Missouri River or into the Mississippi River close to its confluence with the Missouri River. Members of this population are classified into the subspecies C. a. alleganiensis, the same subspecies found in the eastern United States. The other Ozark population inhabits the rivers that drain to the south from the Ozarks and eventually empty into the Mississippi River to the south of the state of Missouri. These southern Ozark animals belong to their own subspecies, C. a. bishopi. The hellbender is totally aquatic and is incapable of terrestrial dispersal. As a result, not only are the two Ozark populations disjunct from one another and their eastern ancestors, but the two Ozark populations are further fragmented into different river systems within these two drainages. Even within a river, their distibution is very patchy. Adult hellbenders are usually found only in those portions of the rivers that have large, loose rocks on the bottom. Unfortunately, virtually nothing is known about the distribution and dispersal behavior of the larvae, although the adults are quite sedentary (Nickerson & Mays, 1973). In order to infer the pattern of gene flow in this species, we overlaid genetic surveys onto the geo- graphical distribution of the species. Previous stud- ies (Merkle et al., 1977) using isozyme markers revealed that all hellbenders with a few minor ex- ceptions were monomorphic for the same alleles regardless of capture site. Hence there is simply insufficient genetic resolution with isozymes to ad- dress the issue of gene flow. We therefore used Annals of the Missouri Botanical Garden TABLE l. Variable mitochondrial DNA haplotypes found in various populati of Cryptobranchus alle- ganiensis from Missouri and Tennessee. The ed on the na first eleven variable sites have been ma genome; a "1" refers to the presence of that s e and a “0” its absence. The last three restriction кыин length polymorphisms have not yet been mapped, so small letters refer to distinguishable fragment length patterns. Haplotypes Restriction sites BamH 1-b l ооо 0 оо 0 BgllI-d 1 | 0 1 | l 1 1 BstEII-b 1 1 0 0 0 0 0 0 EcoRV-b 1 | 1 1 l 0 1 1 1 1 0 0 0 0 0 0 -d 1 1 0 0 0 0 0 0 Stul-5 | 1 0 0 0 0 0 0 l 1 1 1 l | 1 0 Xmnl. -c | 1 1 1 0 0 0 0 -d О О 1 0 0 0 0 0 -e 0 0 0 0 0 0 | 0 Bell Pattern a a b b ¢ c с c Ncol Pattern a a b c a a a a Xbal Pattern a a b b b b b b restriction site mapping of mitochondrial DNA (mtDNA) because, at least in mammals, it evolves about 1O times faster than nuclear DNA and ac- cordingly has much higher levels of polymorphism As shown in Table 1, eight € (Brown et al., distinct mitochondrial haplotypes were found hellbenders collected in Missouri and in Tennessee. 'able 2 shows the geographical distribution of these variants within hellbenders, and Figure 1 portrays these data graphically. Figure | also gives a maxium parsimony unrooted cladogram of the eight haplotypes (with a consistency index of 1) as determined by hand. As evident in Figure 1, there is no overlap in the haplotypes found in Tennessee with those found in Missouri. We therefore con- clude that the Missouri population is isolated com- pletely from the eastern portion of the range of the species. Moreover, there are no shared hap- lotypes between the northern and southern drain- ages of the Ozarks, and as the cladogram shows, the northern and southern haplotypes correspond to distinct branches that are well separated from This implies that there is no gene flow between the each other by seven substitutional events. northern and southern drainages within the Ozarks. Figure | also implies that there is restricted gene flow between some of the rivers within the southern and northern drainages. The two southern rivers are fixed for two different haplotypes that define a single branch in the cladogram but that are separated from each other by four substitutional events. This pattern indicates that there is no gene flow between these rivers. In the north, the Niangua River populations are fixed for the B haplotype that is found in only low frequencies in the other northern rivers. However, there are no statistically significant differences in haplotype frequencies among any of the other northern sites, so the possibility of some gene exchange among the Gas- canade, Big Piney, and Meramec rivers cannot be excluded. Data from three rivers with multiple col- lecting sites show no significant differences among sites within a river. This suggests that the patchy distribution of adulte within a river does not cause genetic isolation. [n summary, the pattern shown in Figure 1 implies that Missouri populations are genetically isolated from the eastern species range and that northern and southern Ozark drainages are isolated from one another. In addition, many rivers define genetically isolated populations from nearby rivers, but local populations within a river are not genet- ically isolated from one another. Hence, by overlaying the genetic survey results upon the geographical distribution of the popula- tions, we have been able to infer much about when habitat fragmentation results in genetic isolation and when it does not for this species. This conclu- sion. would be virtually impossible to make from traditional dispersal studies because of the extreme difficulty of following the fate of larvae in this species and because the adults are extremely long- lived, which invalidates the use of short-term stud- ies. These problems are not limited to hellbenders and occur frequently with other species. For ex- ample, one of the better dispersal studies ever done with amphibians involved marking over 25,000 tadpoles and metamorphosing Fowler's toads, which after five years of work resulted in the recapture of 37 adults (Breden, 1987). However, given suf- ficient genetic resolution, the same types of infer- ences about genetic subdivision can be made from a genetic survey of a much smaller number of adults. We believe that using genetic surveys to infer when habitat fragmentation results in complete ge- netic isolation and when it does not represents one of the most important applications of genetics to the problem of habitat fragmentation. The reason for this importance lies in the fact that the long- term ecological and genetic fate of a fragmented population depends critically upon whether or not there is genetic isolation between the habitat is- LI ands, as will be illustrated by our next example. Volume 77, Number 1 1990 Templeton et al. 17 Genetic Consequences of Habitat Fragmentation FIGURE 1. the major river systems in 6 Tennessee H Distribution of mitochondrial DNA haplotypes found in Cryptobranchus alleganiensis. The map shows in Missouri. The pie diagrams indicate the mtDNA haplotype frequencies at various collecting sites from the data given in Table 2. The maximum parsimony cladogram (with a consistency index of 1) that connects the various haplotypes is shown at the bottom. CROTAPHYTUS COLLARIS The main range of collared lizards is in the southwestern United States and in northern Mex- ico. They also reside in highly fragmented popu- lations scattered thoughout the Ozarks. These des- ert animals most likely invaded Missouri during the Ipsothermal Maximum of 8,000 to 4,000 years B.P. With the reinvasion of the oak-hickory forest around 4,000 B.P., these lizards were only able to survive in Missouri in relictual, desertlike habitats known as glades. Glades form on exposed outcrops of sedimentary or igneous rock, usually near tops of ridges with southerly or southwesterly exposures. The poor, rocky soil of glades inhibits tree growth, and the rocky, southerly exposures create a very dry, hot microhabitat. As a consequence, a large number of animals and plants normally found in deserts or dry prairies are also found in Ozark glades, including collared lizards. lades vary in size from less than an acre up to several hundreds of acres (Nelson & Ladd, 1981). The larger glades tend to be on dolomite and have a more prairielike flora and fauna than the more desertlike igneous glades. Collared lizards are one of the few glade endemics that can inhabit both of these major glade types. The dolomitic glades are also more susceptible to forest invasion, and with the advent of forest fire control in historic times, there has been a secondary phase of fragmentation of the dolomitic glades due to forest encroachment (Beilmann & Brenner, 1951). Consequently, with the collared lizards, a primary phase of habitat fragmentation occurred 4,000 years ago followed y a secondary phase on dolomitic glades within the last two centuries. As with the hellbenders, we can surveys to infer when habitat fragmentation results in genetic isolation. Table 3 summarizes the results use genetic of genetic surveys using protein electrophoresis on 17 loci and restriction mapping of nuclear ribo- somal DNA (rDNA) and mtDNA using 19 restric- tion enzymes. The amounts of genetic variation found with any one of these techniques are modest, with the discovery of only one polymorphic protein locus and four haplotypes each for the rDNA and Annals of the Missouri Botanical Garden TABLE 2. haplotypes found in various populations of Cryp- tobranchus alleganiensis from Missouri and Tennessee. Geographical distribution of mitochondrial The haplotype designations are given in Table Numbers of animals with aplotypes: — B € DEFCGE Locations A Northern Missouri Big Piney River Barton Branch 0 1 000000 Boiling Spring 2 1 000000 Slabtown Spring 11 0 00000 о Spring Creek 9 3 ооо 0 0 0 Devils Elbow 11 0 00000 0 Gasconade River 12 0 000000 Niangua River Site 1 о 14 #0 0 0 0 0 0 Site 2 0 3 00 0 0 0 0 Meramec River 21 2 0 0 0 0 0 0 Southern Missouri North Fork of White River Site 1 0 0 14 0 0 0 0 0 Site 2 0 0 3 0 0 0 0 0 Spring River о 0 о 6 0 0 0 0 Tennessee Little River 0 о 00 9 2 1 0 Beaverdam Creek 0 0 00 7 00 1 mtDNA (Table 3). Although no one genetic system detects much variation, the pooled systems provide a reasonable degree of genetic resolution, as shown in Figure 2. The joint pattern strongly suggests that different glades are genetically isolated from one another. As Figure 2 shows, almost all of the genetic variation is found as between-glade differ- ences. Strong genetic differences can exist on even a very local scale. For example, the lizards on Mina Sauk glade are fixed for rDNA and mtDNA hap- lotypes different from those for which the lizards on Proffit Mountain are fixed, despite the fact that these two areas are only about eight miles apart as one would walk along ridge tops (the most prob- able dispersal route for lizards). Even on Proffit Mountain, there seems to be strong genetic differ- entiation between glades only a quarter of a mile from one another, although small sample sizes make this a tentative conclusion. There thus seems to be extreme genetic frag- mentation in the northern Ozarks. There is likewise evidence for extreme genetic fragmentation in southwestern Missouri. The populations on Dewey Bald, Hercules Glade, and Glade Top Trail are all fixed or nearly fixed for different genotypes. The only significant geographical pattern is that the three glade populations on Glade Top Trail all share a unique genotype. Since these three dolomitic glades were undoubtedly a single glade before forest fire control this genetic sharing is not surprising. Moreover, the Lookout population on Glade Top Trail is the only polymorphic population for these genetic systems found on a natural glade (as will be discussed shortly, the Profit Mountain popu- lation is not inhabiting a natural glade). This ob- servation supports the hypothesis that the Glade Top Trail populations were a single, large popu- lation until very recently. As Figure 2 summarizes, the glades isolated at the end of the Ipsothermal Maximum display a genetic pattern that suggests complete genetic iso- lation. The lack of any clear geographical pattern to how this between-glade variation is distributed indicates that the isolation at the end of the Ipso- thermal Maximum was very rapid; that is, all the glades became isolated from one another more or less at the same time, so that there is no evidence for the pattern expected under an isolation by distance model. Once again, we have an example of how genetic surveys can be used to determine the extent to which habitat fragmentation causes genetic frag- mentation. In the case of the collared lizards, this genetic fragmentation appears to be more severe than it is for the hellbenders. The primary eco- logical consequence of complete genetic isolation is demographic fragmentation. By this, we mean that the dynamics of population growth, age struc- ture, etc., within a habitat island exerts no direct influence upon the comparable demographic vari- ables in other habitat islands. Of course, the de- mographic states among habitat islands could be correlated because of a dependence upon some global environmental state, but there are no direct population-level interactions between habitat is- lands. Also, with no genetic interchange, evolution proceeds independently within each habitat island except for indirect correlations caused by global environmental conditions. If the demographically fragmented demes are small in size, genetic drift becomes an important evolutionary force. Ultimately, genetic drift is ex- pected to result in the loss of genetic variation. The rate at which this loss occurs depends upon the variance effective breeding size of the frag- mented subpopulations. As can be seen from Figure 2, most glade populations of collared lizards show Volume 77, Number 1 1990 Templeton et al. 19 Genetic Consequences of Habitat Fragmentation MDH: S&F MtDNA: A -D rDNA: I - III Dro 1400 FIGURE 2. Distribution of isoyme, mtDNA, and rDNA genetic variants found in Crotaphytus collaris. А county outline map of the southern half of Missouri is shown, with pie diagrams indicating the genotype frequencies at various collecting sites from the data given in Table 3. An expanded scale map is shown of the area between Taum Sauk Mountain (on the far right of the expanded map) and Proffit Mountain (on the far left). Glades are indicated by stippled areas on this map. Contour intervals are 100 feet. no polymorphism whatsoever. The only exceptions to this pattern are the Clade Top Trail populations, which were probably fragmented only very re- cently, and the population found on the parking lot near the Upper Taum Sauk Reservoir on Proffit Mountain. The reservoir has its walls reinforced by loose boulders, and this provides a dispersal corridor for lizards between previously isolated nat- ural glades (indeed, lizards have been observed on these walls). Thus, this population is most likely an amalgam of previously isolated populations. Con- sequently, we conclude that most genetic variation in these lizards exists as fixed differences between glades rather than as polymorphisms within glades. Unfortunately, these conclusions are tempered by the small sample sizes of several of these glade populations and by the relatively low levels of over- all genetic variation. To obtain a more accurate quantification of the partitioning of genetic varia- tion between and within glades, we performed an additional genetic survey using hypervariable VNTR loci on five glade populations having sample sizes between 4 and 58. We found a total of 13 variable, high-molecular-weight bands that could be reliably Annals of the Missouri Botanical Garden TABLE 3. in central and western Oklahoma. The only polymorphic enzyme locus, malate dehy« plotypes (designated A а ри р "fast" (F) and “ (S). Four mtDNA ha slow” variation scorable with Bell digests. Four “DNA haplotypes were disc Results of various genetic surveys on Crotaphytus collaris populations found in Missouri glades and had two alleles: rogenase (MDH), were discovered, all due to length — ~ vered as defined by the presence (+) or absence (—) of a Pvull restriction site and by length variants in the nontransc cribed spacer detected by BamHI and by EcoRI. The most common haplotype is designated as І, and is coded as: — — — (i.e., it lacks the Pvull site, lacks the BamHI length variant, and lacks the EcoRI length variant). The other Каа are: IL(- — =), HI (++ =), and IV (= = +). Populations Sample sizes MDH mtDNA rDNA Highway 21 Glade 3 F/F D 1 Highway М Glade 2 F/F A 1 Graniteville Quarry l F/F A I Mina Sauk Glade 12 F/F B П Profit Mountain #4 | F/F A I Proffit ЕЕ #5 2 F/F A 1 Profit Mountain #7 2 F/F C I Profhit a parking lot 9 6: F/F A I 2: F/F C I 1: SF А [ Glade Top Trail saddle 5 F/F A Ш Glade Тор Trail pinnacle 3 F/F A IH Glade Top Trail lookout 5 3: F/F А IH 2: F/F A I Dewey Bald 2 F/F A I] Hercules Glade 2 F/F A 1 Central Oklahoma 3 F/F A IV Western Oklahoma | F/F A П scored (there was obvious variation at some low- molecular-weight bands, but it was difficult to score). The results are summarized in Figure 3. Аз can be seen, a total of 25 different band phenotypes are observed, but only four of these phenotypes are found in more than one glade. We quantified the extent of partitioning of variation at the VNTR loci by estimating the proportion of shared bands within glades ( 4) and between glades (Р, = 0.50). Moreover, we estimated the correlation of band phenotypes within glades relative to the total population to be 0.47. This correlation should be proportional to the standard F, statistic (Roth- man et al., 1974), which unfortunately cannot be estimated directly because the bands must be re- garded only as phenotypes and not genotypes in NTR loci and band homologies. However, this correlation the absence of information about number of V will be greater than О only if F, is greater than zero. This correlation of 0.47 is significantly dif- the 1% level. Hence, F, significantly different from О and rather large in ferent from 0 at is also value. Once again, we conclude that the majority of genetic variation is found as between-glade dif- ferences. The fact that most genetic variation is found between glades not only confirms our conclusion that glades are demographically independent from one another, but it also implies that the variance effective breeding sizes within glades are very small. This inference is consistent with other data. The population on Sandy Ridge in Jefferson County seems to be one of the largest in the northeastern Ozarks. Yet the sample of 58 lizards constitutes a virtual census. Dr. Owen Sexton (pers. comm.) has followed this population in detail and has found that the adult population during the breeding season fluctuated between 21 and 79 individuals 1975- 1985 rapid partitioning of the ancestral genetic variation . Such small population sizes would cause a into the between-glade component. These small population sizes imply that the glade populations should be very prone to local extinc- tion. When populations are this small, inbreeding depression, environmental fluctuations, and de- mographie stochasticity can greatly increase the probability of local extinction (Quinn & Hastings, 87). Given that there is no dispersal between glades, an "extinction ratchet” operates in which each local extinction brings the total population one step closer to global extinction. It is important to note that the rate at which this extinction ratchet operates is primarily a function of local, not global, population size. Hence, populations that are both Volume 77, Number 1 1990 Templeton et al. 21 Genetic Consequences of Habitat Fragmentation " Misplay Glade Karl's Glad FIGURE 3. A county outline map of populations. The number abo Proffit Mountain k L ve, alongside, or below eac Parking Lot Distribution of hypervariable VNTR (“DNA fingerprinting”) phenotypes found in Crotaphytus collaris. Missouri is shown, with pie diagrams indicating the phenotype frequencies in five glade pie diagram is the sample size from that glade. The DNA 5. was cut with НаеШ and probed with the human fingergprinting clone 33.1 fragmented and small should have a very high rate of local extinction and therefore high probabilities of global extinction. There is much circumstantial evidence for local extinction in the lizards. The recorded county dis- tribution for collared lizards in Missouri (Johnson, 1987) excludes many counties that have excellent glade habitat (Nelson & Ladd, 1981) and that are adjacent to counties known to have collared lizards. There is no apparent reason why the lizards should be absent from these counties, especially in view of the distributions of other glade-inhabiting ani- mals and plants. Even within the counties known to have lizards, the authors have never observed collared lizards on many apparently suitable glades despite several trips to these glades and despite the fact that collared lizards are found on nearby glades. For example, Hawn State Park contains some sand- stone glades that appear to be excellent habitat for collared lizards. Just outside the park are several 22 Annals of the Missouri Botanical Garden 1 Kilometer || LAP Fast Allele FIGURE 4. The frequencies of the alleles at the leucine Trimerotropis saxatalis on Lindsey Mountain. very healthy lizard populations, often on glades smaller than those in the park. Indeed, glades in- habited by lizards virtually surround the park. Yet there has never been a collared lizard recorded in the park, despite several deliberate surveys by knowledgeable people (Frank Crimmons, former superintendent of Hawn State Park, pers. comm.). — This biogeographic pattern suggests that collared lizards have already undergone local extinction on many glades and is consistent with the inferences drawn from the genetic surveys. As this example illustrates, genetic surveys with sufficient resolution to partition genetic variation between and within habitat islands can quickly iden- tify those species at high risk for local extinction. The collared lizards are an example in which the genetic partitioning implies a high risk for local extinction. Our next example shows how genetic surveys can identify a fragmented species at a low risk for local extinction. TRIMEROTROPIS SAXATALIS Another glade inhabitant in the Ozarks is the lichen grasshopper, Trimerotropis saxatalis. This species is found primarily on glades with an acidic LAP Slow Allele aminopeptidase (LAP) locus in three glade populations of substrate, such as granite, rhyolite, or sandstone. We had difficulty performing an isozyme surve on this species because its proteins rapidly dena- ture, but we did get good results for leucine amino- peptidase (LAP). Figure 4 survey of 16 individuals each (32 genomes) from shows the results of a three rhyolitic glade populations on Lindsey Moun- tain. There was no significant difference in allele frequency between glades 2 and 3, but there was between glades | vs. 2 or 3 despite the fact that glades 1 and 2 are only separated by about 150 yards of forest. This kind of genetic differentiation indicates that habitat fragmentation once again re- sulted in genetic isolation. Also note from Figure 4 that all three populations are polymorphic for the same LAP alleles. We are currently surveying for genetic variability in rDNA and mtDNA with restriction enzymes. Both DNA systems have re- vealed much genetic variation, but this variation has yet to be mapped. However, it is obvious from the Southern blot patterns that there is extensive polymorphism within all glade populations and that many variants are shared with other widely scat- tered glades- to that observed for the collared lizards. a pattern that is in great contrast Never- Volume 77, Number 1 1990 Templeton et al. Genetic Consequences of Habitat Fragmentation TABLE 4. Distribution of mtDNA variants in four glade populations of Trimerotropis saxatalis. Cutting mtDNA with EcoRI yields five distinct patterns in these populations, labeled А-Е. Glades Russell Profht Steagle Haplo- Sandy Moun- Moun- oun- types Ridge tam—2 tain—12 tain—7 A 2 3 3 0 B 3 6 6 C 0 0 0 2 р 1 3 0 1 E 0 2 0 0 theless, glade populations will frequently have at least some genetic variants not found in other near- by populations, which indicates absence of gene flow. For example, five distinct restriction frag- ment-length patterns are visible when the mtDNA is cut with EcoRI. Table 4 shows the number of individuals bearing these five haplotypes in four glades, and Figure 5 shows the geographical lo- cations of these glades. As can be seen, all popu- lations are polymorphic for mtDNA haplotypes, ~ "a and most haplotypes are found throughout widely scattered areas. Yet, the Russell Mountain, glade 2 population has two mitochondrial haplotypes not found on the nearby (closer than 6 miles or 10 kilometers) Proffit Mountain, glade 12 popula- tion—a difference that is significant at the 1% level. Such a pattern is strongly indicative of no gene flow even between geographically close glades. Although the lichen grasshoppers are highly fragmented into isolated glade populations, the large amount of within-glade polymorphism implies that the numbers within each glade are sufficiently large to insure that genetic drift is weak. This in turn implies that this species is in very little danger of local extinction except from environmental chal- lenges that would cover the entire spatial scale of their habitat. This prediction is consistent with our collecting experiences for this species. Of 85 acidic glades visited that were a quarter of an acre or larger, these grasshoppers were found on all 85. In contrast, collared lizards were only observed on 14 of these 85 glades (however, some of these glades were only visited once, so it is probable that collared lizards аге оп more than 14 of them). These observations support the inference from the Proffit Mtn-12— Sendy Ridge Russell Mtn-2 FIGURE 5. Steagle Mtn-7 Distribution of mtDNA haplotypes found in Trimerotropis saxatalis. À county outline map of Missouri is shown, with pie diagrams indicating d x frequencies in four glade populations. The number above each lade. pie diagram is the sample size from that Annals of the Missouri Botanical Garden genetic data that the local populations of lichen grasshoppers are not at high risk for local extinc- tion. MANAGEMENT RECOMMENDATIONS The ecological and genetic consequences of hab- itat fragmentation depend critically upon whether or not there is dispersal between habitat islands. The examples given in this paper show that genetic surveys are an extremely useful tool for addressing this critical question. The importance of genetic surveys is augmented further by the fact that it is not feasible to study dispersal directly for many species, but genetic surveys can be performed on virtually any species. Even when dispersal studies are feasible, genetic surveys are a more accurate means of inferring demographic fragmentation. For example, Lewis (1982) studied dispersal in the white- browed sparrow weaver, Plocepasser mahali. These birds are colonial breeders, and Lewis discovered that there is much successful immigration of dis- persing birds into the smaller colonies. However, these colonies are not generally reproductively suc- cessful, and most birds in them are effectively dead genetically. Almost all successful reproduction is limited to large groups, and there is very little immigration into these. Hence, there is very little gene flow in this species despite much dispersal. On the other hand, basic population genetic theory (Crow & Kimura, 1970) shows that even very low dispersal rates— particularly, rare long-distance dispersal events—can be sufficient to maintain lo- cal populations as a single genetic unit. It is vir- tually impossible to estimate rare, long-distance dispersal rates in most species. Thus, genetic sur- veys provide a more reliable indicator of gene flow patterns than do dispersal studies. Genetic surveys are therefore a very powerful tool in identifying the pattern of demographic fragmentation caused by habitat fragmentation, and it is recommended that genetic surveys be utilized much more for this purpose. Genetic surveys are also useful in identifying species at high risk for local extinction, given that demographic fragmentation has occurred. As il- lustrated by the contrast between collared lizards and lichen grasshoppers, we can make inferences about the long-term effective population sizes with- in habitat islands by partitioning the genetic vari- ability into within-habitat-island and between-hab- itat-island components. Species displaying small effective sizes are at most risk for inbreeding depression, demographic stochasticity, and extinction through environmen- tal fluctuations. Many species that are endangered to begin with have low population sizes. Therefore, the most-endangered species suffer most from de- mographic fragmentation. Demographic fragmen- tation can therefore greatly accelerate the rate of extinction of an endangered species through the One obvious method of stopping the advance of the extinction operation of the extinction ratchet. ratchet is to recolonize artificially in order to coun- teract local extinction. In conjunction with the Mis- souri Conservation Commission, we have begun such a recolonization program for collared lizards in the Missouri Ozarks in order to study and mon- itor the success of various release strategies. Be- cause the lizards are long-lived and the release program began in 1984, we do not have sufficient data to make evaluations. Nevertheless, our ex- periences in designing this release program have general implications. One of the first decisions to be made is the goal of the release program. We feel that such release programs should preserve the genetic variability of the species while protecting it from extinction. With respect to genetic variation, there is a silver lining in the generally dark cloud of habitat fragmenta- tion. As can be seen from the collared lizard ex- ample, when a population is fragmented into small, demographically independent isolates, there is rap- id partitioning of the available genetic variation from within-habitat to between-habitat. The im- portant point is that this is just a redistribution of the genetic variation, not its elimination. Genetic variation is still present as fixed differences between Moreover, theory (Maruyama, 197 local. populations. population. genetic 2) showed that global ge- netic variation is maintained more efficiently by a fragmented population than by a panmictic pop- ulation, given an equal total size. The primary cause of loss of genetic variation in a finite pop- ulation is genetic drift, but the genetic differences that are fixed between local populations can only be lost by extinction of the entire population. Local extinction, however, can modify this pre- diction. As local extinction occurs, not only is pop- ulation size reduced at the global level, but any unique genetic variants found in that local popu- ation are lost as well. As long as there are many local populations available, the loss in global genetic variation caused by local extinction is very little. But as the extinction ratchet decreases the number of local populations, genetic variation loss will ac- celerate. Accordingly, recolonization intervention is needed to prevent extinction and to insure that a sufficient number of local populations are main- tained so that global genetic variation can be pre- served. However, the goal of preserving genetic Volume 77, Number 1 1990 Templeton et al. 25 Genetic Consequences of Habitat Fragmentation variation places constraints on how the recoloni- zation should be done There are two basic strategies in a recolonization program. One strategy is to obtain all the animals for a release from a single, local, large population. The alternative is to obtain the release animals from several local populations and recolonize with a mixed population. Arguments can be made for both strategies. With a mixed release, high levels of genetic variability can be reestablished at the local popu- lation level. However, since the released population will in general be small, this variation will be rapidly lost due to genetic drift. Nevertheless, even a tem- porary infusion of genetic variability into the local population can be beneficial. Evolution within the local population can occur only if there is genetic variability, so natural selection in the local popu- lation is possible for the first few generations after the release of a mixed population. Hence, a mixed release allows some adaptation to the local envi- ronment. This could be important in increasing the chances of success of the released population if the environment in which the release takes place is not identical to the environments experienced by the source populations. Local adaptation could, however, sometimes fa- vor the strategy of nonmixed release. Frequently, local adaptation in small, isolated populations is achieved by the accumulation of ““coadapted” gene complexes (Templeton, 1986). adapting to the same environment will often achieve that adaptation in genetically distinct and incom- patible fashions. When these coadapted complexes are broken down by recombination, the average fitness of the population could be lowered dramat- “outbreeding ven populations ically—a phenomenon as depression” (Templeton, 1986; Templeton et al., 1986). If severe enough, an outbreeding depression could greatly increase the chances of extinction of the released populations. nnest & Templeton (1978) have experimen- tally monitored the evolutionary and ecological sig- nificance of outbreeding depression in Drosophila populations. They showed that the lowest popula- tion sizes occur during the first generation in which recombination can break up coadapted complexes (generally, the F, or backcross generations). There- fore, extinction due to outbreeding depression gen- erally will occur in the first few generations after release. After that, selection operates to reestablish one of the parental coadapted gene complexes or evolve a new one. Їп either case, the absolute fitness increases and population size rises. Annest & Tem- pleton (1978) found that when a new coadapted complex arose, it generally had superior fitness traits to any of the input parental complexes. Thus, if the population can survive the first few gener- ations, the outbreeding depression will be elimi- nated by the action of natural selection and it is possible that the surviving population will have higher fitness than any parental population. Con- sequently, even if there is an outbreeding depres- sion, mixed releases still might be best in the long run. It would be essential to monitor the released populations closely for at least the first three to five generations. For example, we obtain a blood sample from all collared lizards before release to provide genetic markers to detect outbreeding and / or inbreeding depressions. These genetic markers will also allow us to estimate accurately the vari- ance and inbreeding effective sizes of the released populations by observing the rate of decay of ge- netic variation through time. If severe outbreeding depressions occur, the mixed release strategy might have to be abandoned to insure the survival of the populations in the recolonized areas. n addition to outbreeding depression, the re- leased population may suffer from inbreeding depression. Inbreeding depression is usually caused by the increased incidence of homozygosity for recessive deleterious alleles that occurs with in- breeding. Inbreeding and outbreeding depression are not mutually exclusive since they can involve different genetic systems found in the same organ- isms. When inbreeding occurs because of small population size, deleterious alleles have a finite chance of going to fixation. Hence, the real danger in small, isolated populations is that the inbreeding depression will be fixed by genetic drift. O’Brien al. (1985) have argued that such a fixed in- breeding depression may account for the low re- productive performances of cheetahs. Fixed in- breeding depressions represent a serious problem for the survival of local populations and for the species. We would expect the local populations displaying the most severe inbreeding depressions to go extinct more rapidly. Therefore, as time proceeds, the average severity of fixed inbreeding depression should decrease, but over even long periods of time, we expect fixed inbreeding depres- sion to rise again. When population sizes are very small, even deleterious mutations have a finite chance of fixation (Crow & Kimura, 1970), and once fixed, there is no mechanism to purge the deleterious mutation from the local populations un- der complete genetic isolation. This results in a situation analogous to **Muller's ratchet” in which deleterious alleles will tend to accumulate until the population is driven to extinction (Muller, 1964). Annals йөк nU Garden By carrying out a mixed release, the temporary infusion of heterozygosity will alleviate the inbreed- ing depression for the first few generations after release. As the generations progress, inbreeding will become reestablished in the released popula- tions, and hence inbreeding depression may reap- pear. However, just as outbreeding depressions can be eliminated by the operation of natural selection, so too can inbreeding depressions (Templeton & Read, 1983, 1984). With a mixed release, natural selection has a renewed opportunity to eliminate deleterious alleles. Random drift may once again cause fixation of deleterious genes, but natural selection biases the stochastic. fixation. process against that possibility. Also, this temporary influx of genetic variation allows recombination to be effective at producing new genotypes, and this re- combination can help undo the damage done b "Muller's ratchet” (Muller, 1964). By releasing from a single source population, there is no op- portunity for selection or effective recombination to operate, and therefore any fixed inbreeding depression in the source populations will remain fixed The benefits from selective processes operating and “Muller’s ratchet” will proceed unabated. in released populations increase as the initial amount of genetic variation increases. Accordingly, the best mixed release strategy combines individual popu- lations from a large number of habitat islands. The selective benefits also increase as the number of generations with genetic variation after release in- creases. The number of genetically variable gen- erations can be maximized by making the released population as close as is practical to the ultimate carrying capacity of the habitat island to be re- colonized. Mixed releases also aid in the goal of preserving overall genetic variability. When all individuals for release are drawn from a single source population, it is essential that the source population be very large so that harvesting individuals from it will not endanger the source as well. Such large releases reduce the chances of extinction and increase the opportunity for natural selection to promote local adaptation and eliminate outbreeding and inbreed- ing depressions. Only a few source populations will be sufficiently numerous to support this extensive harvesting. For example, we have identified only four glades that could support a harvesting of 10 collared lizards-—our minimum release size — with- out seriously endangering the survival of the source population. As the collared lizard example shows, the released populations under the single source strategy will often be genetic clones of а small number of source populations. Thus, genetic vari- ation at the global level will be чыш. depleted under this recolonization strate In contrast, the multiple-source recolonization strategy can preserve large amounts of global ge- netic variation. First, because only a few individuals need be taken from any single local population, many more local populations can qualify as source populations, as illustrated by our collared lizard release program. By doing mixed releases in which all animals come from different glades, we need only harvest a single animal from any particular glade. Virtually all natural populations can endure that level of harvesting without ill effect. The re- colonization program this way can tap into much more of the global genetic variability contained in the fragmented species. After release, genetic drift will cause loss of variation. However, different re- leases will undoubtedly become fixed for a different array of genetic variants. Hence, there will be no tendency to “clone” a handful of source popula- tions; instead, the released populations will be ge- netically diverse relative to one another and to their sources. In this manner, very high levels of global genetic variation can be maintained in the fragmented species despite high rates of local ex- tinction. Follow-up monitoring of the released pop- ulations is just as important as the initial release in order to insure that genetic diversity is being preserved and to check for the possibility of out- breeding depressions. SUMMARY OF RECOMMENDATIONS Genetic surveys are an extremely powerful tool for identifying demographically independent hab- itat islands and should be used more extensively to infer the pattern of genetic fragmentation. Indeed, genetic surveys offer a more reliable means of inferring demographic fragmentation than dispers- al studies and can be applied easily to a wide diversity of organisms. Second, genetic surveys, particularly those uti- lizing high-resolution techniques, such as DNA fin- gerprinting, can be used to spot fragmented species at high risk for local extinction. This is accom- plished by quantifying the partitioning of genetic variation between and within demographically in- dependent habitat islands. For species with a high risk of local extinction, recolonization programs are needed to protect the fragmented species against global extinction and to preserve its pool of genetic diversity. In general, a mixed release strategy is best, with the numbers to be released ideally as close as possible to the carrying capacity. Released populations must be Volume 77, Number 1 1990 Templeton et al. 27 Genetic Consequences of Habitat Fragmentation monitored genetically and demographically for the first few generations after release to detect potential inbreeding and/or outbreeding depression and pe- riodically thereafter to insure that genetic diversity is indeed being preserved. LITERATURE CITED ANNEST, J. R. TEMPLETON. 1978. Genetic А and clonal selection in Drosophila L. G. BRENNER. 1951. The recent intrusion of forests in the Ozarks. Ann. Missouri Bot. Gard. 38: 261- BREDEN, F. 1987. The effect of post-metamorphic dis- persal on the population 21 structure of Fowler's Toad. Copeia 1987(2): 386-395. Brown, V. M., E. M. PRAGER, ju Wanc & А. C. WILSON. 1982. олон DNA sequences of primates: tempo and mode of evolution. J. Molec. Evol. 18: 225- J COHMAP MEMBERS. last 18,000 years: observations an tions. Science 241: 1043-1052 F 1970. An Introduction to Population Genetics Theory. Harper & Row, New 1988. Climatic changes of the d model simula- ork. Hius, D. M. & S. K. Davis. 1986. Evolution of ra DNA: fifty million years of recorded | hatary n the frog genus Rana. Evolution 40: 1275-1288. ен Т. К. 1987. The amphibians and reptiles of Missouri. Missouri Dept. of Conservation, Jefferson City, Missouri. 80. Granitic ring complexes and Precambrian hot-spot activity in the St. Francois terrane, midcontinent region, United States. Geology -4T. 8: 4 LANSMAN, К. A., R. O. SHADE, J. F. SHAPIRO & J. AVISE 1981. The use of restriction iE to measure mitochondrial DNA sequence relatedness in natural populations. III. күте and potential applications. J. Molec. Evol. 17: 26. P = . 1982. Dispersal in a ur liue sparrow er population. Condor 84: 306-312. Macast К.Н. & E. О. Witson. 1967. The Theory Island Biogeography. Princeton Univ. Press, Pra New Jersey. MaRUYAMA, Т. 1972. Rate of decrease of genetic vari- ability in a two-dimensional PU population of finite size. Genetics 70: 639-65 MERKLE, D. I. GUTTMAN & М. . NICKERSON. Const uniformity throughout the range of the hellbender, Cryptobrane hus alleganiensis. Co- реа 1977(3): 549-5 MULLER, H. J. 1964. The relation of recombination to mutational advance. Mutat. Res. 1: Missouri glades— pa rt . How many, what kind, and where. Missouriensis -9. NICKERSON, M. A. & C. E. Mavs. 1973. The hellben- ders: North American giant salamanders. Publica- tions in Biology and ме ui 1. Milwaukee Public 1985. Genetic Dada for Ке чагу vulnerability in the cheetah. Science 34. . К. & A. Hastincs. 1987. Extinction in sub- divided Бабаш: Conserv. Biol. 1: 198-208. RoTHMAN, E. D., D. F. Sinc & A. R. TEMPLETON. 1974. A model for analysis of population structure. Genetics 78: 943- SELANDER, R. K., M. SMITH, S. YANG, W. JOHNSON & J. GEN Biochemical polymorphism and netics VI. Univ. of Texas Publ. No. 7103. Pp. 81- 90. SouLÉ, M. E., D. T. BorcER, A. C. ALBERTS, J. WRIGHT, M. Sorice & S. Hitt. 1988. Reconstructed dy- namics of rapid extinctions of chaparral-requiring birds in urban habitat islands. Conserv. Biol. 2: 75- TEMPLETON, А. К. 1986. Coadaptation and outbreeding depression. Pp. 105-116 in М. Soulé (editor), Con- servation Biology: Science of Scarcity and Diversity. Sinauer, Sunderland, Massachusetts B. READ. 1983. The elimination of inbreed- ing depression in a c Pp. 241-261 in C. bers, B. MacBryd U. S. SEaL, W. SuiELDs & D. S. e 1986. ow adap- tation, pur and population boundaries. Zoo 125. Biol. 5: 1984. Factors eliminating inbreed- ing depression i > a captive herd of Speke's gazelle. Zoo Biol. 3: -199. THOMPSON, P. E W. Srres. 1986. Comparison of population structure in chromosomally polytypic and monotypic species of Sceloporus s (Dari: Iquanidae) in relation to i Mente y mediated speciation. Evolution 40: 303-314 CONSERVING BIODIVERSITY IN THE CANARY ISLANDS David Bramwell ABSTRACT The flora of the Canary Islands is rich in endemic species and shows typical insular evolutionary features, such as large-scale adaptive radiation (Echium, Sonchus relict flora of the Ter if lany taxa (e.g., ec onomic plants ог фо ней whose с а Worl agation, and restoration of some ecosyster field are not a The role of local botanic е wn ens in planning, conservation-oriented r Argyranthemum, iid Epoch of the Mediterranean Region and the drier Olea ee er subsp. « s of diet КАНЕ importance. s levels to provide protec tion, especially for dl areas. "Planes Especiales para Protección de Espacios Naturales, ld Heritage Site. This is supported pe intensive research on i endangered species involvin ng studies of genetic Bahe и biology, breedi of the the Aeonium) and interisland vicariance. It is also a regions of Africa, and as such has ‚ Brassica bourgaci) are valuable a This doses major legislation ( LED ystems oretical c practical difficulties encountere esearch, and important field of environmental education is explained, and the value of the garden as the interface between ex situ d. and in situ conservation is considere Since the middle of last century, when Charles Darwin published the remarkable conclusions drawn partly from his observations on the fauna and flora of the Galápagos Archipelago, islands have fasci- nated scientists because they are natural biological and evolutionary laboratories where evolution in isolation has permitted the survival of rare, relict endemic taxa and at the same time promoted the formation of new, often equally rare and interesting ones. Islands, therefore, can provide a novel and in- valuable scientific resource. Particularly when we are concerned with the conservation of small pop- ulations and the protection of very rare species, there are many theoretical and practical conser- vation lessons to be learned from island systems. This is especially true for the fields of population genetics and behavior. such as that carried out on Hawaiian organisms during IBP (International Bi- ological Program) ** Recent research, Integrated Island Ecosystems Program,” suggests that, contrary to widespread popular belief, the genetic properties of island pop- ulations are essentially similar to those of conti- 1981, of this subject). This is confirmed by work on Canar- lan groups, nental ones (see Carson, for full discussion such as Lotus, where the range of karyotype variation and cyanogenic properties closely parallels that reported from continental Ки- ropean species (Ortega, 1978; Urbanska & Willdi, 1975). Thus, concepts derived from the study of the behavior of small isolated populations from islands should be even more applicable to rare and endangered continental species than suspected be- fore the IBP research. Islands, are very vulnerable to eco- logical change, especially the disruptive and de- structive alteration brought about by human in- however, terference in insular ecosystems, the direct effects of humans and those of domestic animals and plants as well as their camp followers, weeds and pests. The [UCN Plants Red Data Book or the recently published Plants in Danger: What Do We Know? (Davis et al., 1986) reveal how many ecosystems and species are currently threatened on islands (Table 1) BIODIVERSITY IN THE CANARY ISLANDS From the islands conservation point of view, the Canaries are no exception. The archipelago of sev- en main islands and several small islets and rocks covers over 7,250 km? with a population of some 1.5 million (approximately 206 people per km’). The total flora, which is taxonomically well known, is about 2,000 species with just over 500 endemics, endemism. The total figure, however, includes a number of weeds and aliens. Endemism is probably between 35% and 45% of the native flora. For such a small area, this is a remarkable ' Jardin Botánico “Viera y Clavijo,” Excmo. Cabildo Insular de Gran Canaria. ANN. Missouni Вот. GARD. 77: 28-37. 1990. Volume 77, Number 1 1990 Bramwell Conserving Biodiversity in the Canary Islands E 1. Small oceanic islands with devastated floras. Ex extinct, E endangered, V vulnerable, R rare, I A indeterminate, K insufficiently known, nt not threatened. Rare or Ex E V R I K nt Total threatened Ascension Island 1 5 = 4 — 1 — 11 10 (91%) Bermuda 3 4 1 © = ? ? ? 14 Norfolk Island 5 11 29 ~ 1 2 — 48 46 (96%) Rodrigues 10 20 8 8 = — 2 48 46 (96%) St. Helena 7 23 — 17 — 2 — 49 47 (96%) degree of diversity and endemism. The total flora is larger than that of the British Isles, where the area is almost 34 times that of the Canary Islands. Although over 67% (383) of the endemic species are rare, threatened, or endangered (Fig. 1), most are extremely tenacious and capable of hanging on in precarious circumstances over long periods. Since the first relatively comprehensive accounts of the flora were published last century (Webb & Berthelot, 1836-1850; Sauer, 1880) only a single species, Solanum nava, has not been found again during the last 20 years. Areas of good natural vegetation, on the other hand, have been decimated over the past 500 years and especially in recent decades. The flora of the Canary Islands is of particular interest for a number of reasons, and the arguments for its conservation are scientifically and econom- ically strong. aleobotanical and biogeographical data suggest that it is historically a relict flora (Bramwell, 1972, 1976, 1985). Its closest relatives are in either the Tethyan-Tertiary Region along the edges of the Tethys Sea, principally in the late Miocene and Pliocene periods, or along the drier margins of eastern and southern Africa, the Arabian Penin- sula, and Socotra as a relict of a once more wide- spread semidesert flora of the pre-Holocene (Que- zel, 1978; Bramwell, 1985). The islands are now isolated from their main source areas by time, by sea, and by desert. This has resulted in a postisola- tion phase of evolution with adaptive radiation and vicariance leading to two main types of endemic taxa: the relicts that have not undergone speciation on a large scale (Dracaena, Bosea), and the active epibiotics (Bramwell, 1972), which at the generic level are probably relicts (4eonium found in Ma- caronesia and East Africa, for example) but have a plethora of local endemic species due to adaptive radiation. In fact, in Echium, Sonchus, Argyr- anthemum, and Aeonium, we have some of the finest examples of plant adaptive radiation on is- 1960; Bramwell, 1972, lands anywhere (Lems, 1975; Humphries, 1976; Aldridge, 1979). These evolutionary models merit conservation in as com- plete a state as possible since every species that becomes extinct breaks a link in the chain and makes our understanding of the processes, pat- terns, and products of their evolution more difficult. any of these Canarian endemics survive in very small populations and have, according to the literature, done so for over a century (Cabrera y Diaz, 1910) so that there may be much to learn from their genetics, population structure, and the roles of dormant individuals held in the natural seed reservoir in maintaining long-term variability and fitness. This, coupled with the scientific interest of the flora, is a valid reason for conservation of such model floras. There are, however, other reasons for concern for the conservation of the Canarian flora. Because - 0 Lo) Ex E V R I K nt FIGURE 1. Threatened endemic plant species of the Canary Islands assigned to IUCN Categories. Ex extinct, E endangered, V vulnerable, R rare, I indeterminate, K insufficiently known, nt not threatened Annals of the Missouri Botanical Garden Azores jaa Canary / Islands P4 [| 1 0f t Cape*, Verde Islands IGURE 2. The geographical situation of the Canary Islands. — Limit of the Macaronesian Region. of its historical association with the old Mediter- ranean floras of the Tethyan- Tertiary, the Canari- an flora has a number of wild relatives of classical Mediterranean economic plants and crops; among the most important of these are Olea europaea subsp. cerasiformis, Phoenix canariensis, Bras- sica bourgaet, Cytisus proliferus, Beta webbiana, Avena canariensis, and Dactylis smithii. To species such as these we can add a long list of potentially valuable ornamental species in such genera as Argyranthemum, Limonium, Lotus, and Senecio sect. Pericallis (endemic to Macaronesia and with the vast majority of species local endemics in the Canaries) which is the Florist’s Cineraria. there are also many local medicinal uses for en- demic species but, although some of these have been the subject of a recent review (Perez de Paz & Medina, 1988), their active constituents have not in most cases been studied in detail. The major vegetation types and plant commu- nities of the islands are endemic to the Canaries or in some cases to the Macaronesian Region (Fig. 2). These include a Fuphorbia scrub zone (domi- nated by Euphorbia broussonetii, E. aphylla, E. canariensis, and other species), which is similar to some North and East African communities; the humid laurel forests in which all the dominant trees are endemic to the Macaronesian Region and are relicts of the Tethyan- Tertiary Region of approx- imately 4 million years ago; montane forests of the locally endemic pine, Pinus canariensis, which has its nearest relatives as Tertiary Mediterranean fos- sils and as the extant species P. roxburghii in the West Himalayas. The subalpine zone of the islands occurs only on Tenerife and La Palma and has over 90% endemism among the perennial plants; it shows morphological and physiological similari- ties to the Afro-Alpine zone of East Africa. CONSERVATION. PROBLEMS IN THE CANARY ISLANDS Before the Spanish occupation in the fifteenth century, the Canary Islands were inhabited by a people, variously known as Guanches or Canarios, with a relatively simple neolithic culture who seem to have lived in relative harmony with their natural environment. After the fifteenth century, however, the situation changed considerably, especially when sugar cane was introduced to the islands and large areas of land were cleared for its cultivation. For- ests were cut down to provide space and to fuel the sugar extraction and refining. As Ayensu et al. (1984) pointed out, "The natural vegetation of these islands was further affected, often totally destroyed, by the of tomatoes an grapevines, the terracing of the slopes for these and other vegetable crops, the widespread culti- vation of bananas in coastal zones, and the intro- duction and spread of many Mediterranean weeds. This all led to a lowering of the water table and increased aridity." This proceeded over 500 years to such an extent that the laurel forest of Gran Canaria now covers less than 1% of its original area (Fig. 3) and on the island of Tenerife, less than 10% of its former extent. From this brief summary of the recent history of the environment in the islands we can see that the main destructive force has been humans through pressure of land use and the destruction of natural resources, but in the second half of the twentieth century a further phase of deterioration is under way due to the massive expansion of tourism in cultivation the islands. Large-scale tourist development (200,000 tour- ist beds on the island of Gran Canaria alone) brings The pressure on coastal region land resources, the con- stant need for building materials, the strain on local ter resources, coastal pollution and disposal of domestic rubbish, and even the damage caused to vehicles with it enormous environmental problems. natural areas by recreational off-road probably pose the biggest threats to biodiversity and natural resources ever. One of the major, but little-considered, impacts is homogenization of the habitat, the destruction of ecologically well-defined vegetation zones, which leads to species swamping by hybridization (see Brochmann, 1984, for further discussion) and continuously tends toward environ- mental uniformity and away from diversity. Volume 77, Number 1 1990 Bramwell 31 Conserving Biodiversity in the Canary Islands Faced with such negative effects of human in- tervention in the environment, the need for an equally interventionist and positive conservation program is obvious. ENVIRONMENTAL CONSERVATION IN THE CANARY ISLANDS Conservation in such circumstances is neces- sarily a multifaceted activity with the following fields of action: legislation (protection of areas, protection of species), planning, research, and ed- ucation. Legislation Protection of the environment is provided for in Spanish law at various levels, from the Constitution to the 1975 Laws on Protected Natural Areas under which the Spanish National Parks were con- solidated, and under minor parts of several other national laws on, for example, hunting, mining and mineral extraction, forestry, and water resources. The autonomous regional governments in Spain also have the powers to legislate locally on envi- ronmental matters. National legislation, at least in theory, protects some species. The hunting laws (Ley de Caza art. 23.2, 1970), for example, state that “Species of scientific interest or in danger of extinction will be the subject of special protectio under this legislation a royal decree (8181/1980) lists nationally protected species and includes most of the native birds, reptiles, and amphibians of the Canary Islands. Following Spain's ratification of the Berne and Bonn conventions on the environment and migra- tory species of wild animals, respectively, many more migratory and visiting birds were added to the list of protected species. Plant conservation legislation is, however, at individual species level rather less developed and is based mainly on forestry legislation and a series of royal decrees protecting tree species. It does cover the majority of laurel forest species but has proven to be difficult to apply to private landown- ers. Spanish law does envisage the protection of individual species, and for mainland Spain and the Balearic Islands two decrees from 1982 and 1984 give protection to some wild endemic plants. In the Canary Islands, however, we do not yet have any legislation specifically protecting species other than the forest trees already mentione We have recently prepared a pa pro- tected species list (Bramwell & Rodrigo, 1982) and this year updated it at the request of the Environ- Q 5 10 km FIGURE 3. ria. Upper — potential distribution. Lower bution Distribution of laurel forest on Gran Cana- — actual distri- mental Department of the Regional Canarian Gov- ernment with a view to preparing new local leg- islation for the protection of Canarian endemic plants; we hope that this will take effect during the next 12 months. Draft legislation for a new com- prehensive law on environmental protection is also being prepared at the moment, and in 1987 a law was passed placing most of the major areas of conservation importance in the islands in a catalog of protected natural areas (Fig. 4). These cataloged areas are now pending the preparation of individual use and management plans, which is the respon- sibility of the Dirección General de Medio-Am- biente (Directorate General of the Environment). Existing law has been put to good use in the Canaries at two levels, both concerned mainly with in situ ecosystem protection At the first level, the Se rational conventions on World Heritage Sites and Man and the Bio- sphere Reserves have been used to conserve two of the best areas of Canarian laurel forest, El Cedro- Garajonay on La Gomera (Fig. 5), which is also a national park, and Monte del Canal on the island of La Palma. The Canary Islands have four national parks, the subalpine zone of Tenerife (Parque Na- 32 Annals of the Missouri Botanical Garden E Protected Landscapes FIGURE : Protected areas of Gran Canaria included in the Ley de Espacios Naturales of the Autonomous Regional Government (1987). Dots— natural. parks. protected landscapes. Natural Parks Lines cional del Teide) covering 13,500 hectares; Parque Nacional de La Caldera de Taburiente on La Palma, which is a pine forest park of 4,690 hectares (ha); the volcano park on Lanzarote (Parque Nacional de Timanfaya) with 5,107 ha; and the already mentioned Parque Nacional de Garajonay with an area of 3,950 ha mainly of laurel forest. 'The second level of action in the Canaries has been promoted by the Autonomous Government and the local individual island councils (Cabildos Insulares). Taking advantage of Spanish Town and Country and Country planning law (Ley de Suelos), a series of planning projects classifying land has been undertaken for each island. I explain the one for Gran Canaria in detail as a possible example of the use of planning legislation for natural eco- system protection on islands. Planning Often the philosophy behind the designation of national parks and nature reserves is "monumen- Vallehermoso . ae * Hermigua valle . Gran Rey San Sebastian o 5 10 km URE 5. National Park of Garajonay, La Gomera, recently accorded the status of World Heritage Site. Laurel forest Euphorbia scrub Olive/Pistacia lentiscus woods Coastal dune vegetation у Pistacia altantica woodland Montana vegetation 4 Pinus/Juniper ecotone FIGURE 6. Canaria. The potential natural vegetation of Gran talis" so that great landscapes and so-called wild areas are given priority, and this was certainly true for the original national parks of the Canary Is- lands, those of Teide, Taburiente, and Timanfaya. These three parks protect only a very small pro- ortion of endangered Canarian endemic species. —. This was a particular problem for Gran Canaria in the center of the archipelago because it had no national park and very few protected areas in 1983 when the Jardin Botanico "Viera y Clavijo” was asked by the Canarian government and the Cabildo Insular de Gran Canaria to form an interdisciplin- ary team of biologists, geographers, lawyers, and architects to prepare a special plan for the pro- tection of natural areas under Spanish planning aw. The methodology we used was to take a wholly scientific approach to the problem and study the entire island rather than to rely on preconceived ideas about which areas of the island should be protected. In part this was done because previous attempts to delimit natural reserves had failed to protect sufficient areas because they did not con- sider restoration of degraded areas as a viable con- servation policy. A summary of our study of the island has already been published (Bramwell et al., 1986), and | give a brief outline of it here. The first phase of preparation involved a review of existing information, bibliography, and the ex- perience of members of the team (which was an important source of data). This enabled us to build up a picture of the island leading to the identifi- cation of major gaps in our knowledge, some of which were then worked on by the team, and others through the cooperation of outside specialists. The following studies made important contributions to the project: Volume 77, Number 1 Bramwell 33 1990 Conserving Biodiversity in the Canary Islands - . N dy E — [^] = 5 һ w »789012345678905123456783012345678901234567892123456 3117 | 4 3117 16 EL 4 Д 16 sh] r Q 16 N y 14 М, 4 13 | Жы эс A ЖЫ ШЕ 4 = и ү "T - а 59 | | 13 12 3 ) LL] 1 AB E Mh. | 12 "ССІ 2 LLLILIS AW ah " 104l S EEP N 10 nd L E als4] [ifa 1 09 3 |5 и аш 09 08 L PEN LL I fot f EE O nl. -Ң 08 07 [ JL fp Ap 12173]. 1 M 07 а, |. ata] [So 2|4|4|]4]5]2 1 á 05 | DECRE BPDDOCUBBE r 0$ p а [2 (+з и [зу 5 |1 ноја 121 [1] ||| |, 04 AO Tt fate folate 04 po 132131 |2 | | јој | Jal Jef [aya AL 9 03 ku af eal MEE 4 02 ott. Stet ft 0 tty 4111111 | L Bh 01 3100 “1312. 11 m. 112 1 y ™ 3100 3099 4 1 2 214[4|4 1 1 7 2008 98 1 1 41717 98 РҮ; ) 214 2 1 [11у 1 7 9? EN 1 4/2 | 1 1 i 4 96 € y Tre z at? = 95 ы 96 94 4 z|2 “74 - 3 2]6|4 5 94 93 1)4|6|1|4| |1 13 11 | 1 314 [4 [1 эз 92 3 1 4 Yat 6 AL 2 92 91 M ЕВЕ Ў 4 4 1 ВЕ ^ 91 90 { 3| |ч Б П 1 SHG вэ J 2|/ , "EE ] өз вв 1 \ se ез 1 \ ќ 1 414 1 2. 1 b 1 87 вв ‚| ], EIN 1 pace 56 E d 1 ИЫ 0 Ж ЖЕ 3 aj Ja 4 1 85 84 | Де [3] NX an ев 83 d 1 4/1 411 E—1$—-t—4 4 1 2 H—31—1—3 83 е \ del faje]? Па! zp || || a2 B4 A f n BI 12 { \ 1 51 A S 81 м F К 4147 : amang ШЕЕ во esv T Ы зе 4 \ 1 > 79 78 114 ў 21214)! 2|2 1 78 77 І а ] í ae "EL TIU n 76 NL + 1 a 76 75 : 7 1 1 76 74 + } 74 73 b Г 72 1 1 П — Concentrations of 1 Т d 70 ге 4 a : ) local endemic species 68 y 3067 LL 2/9*9*9g123*45607.835123466789 512346678972! 0 5 10 km FIGURE 7. The distribution of local endemic species; figures indicate the number of Gran Canarian endemics per i] 1. Existing land use. 2. Population distribution. in a sea of destructive practises," we decided it 3. Geomorphological features, mining, and mineral was necessary to give at least some protection to extraction. 4. Areas of ornithological interest. 5. very large areas even though the number of zones Distribution of native vertebrates. 6. Entomology. ^ needing absolute protection was relatively low. 7. Potential (Fig. 6) and actual natural vegetation. After detailed mapping of boundaries using ae- ainfall and water resources. 9. Areas of ar- rial photographs, the final number of protected chaeological interest. 10. Existing town and coun- areas was 64, varying in size from 0.5 ha to 14,478 try planning. 11. Tourist development. 12. Dis- ha and totalling 109,574 ha or approximately 60% tributions of rare endemic species (Fig. 7). of the area of the island (Fig. 8); these protect the Only when all this information had been analyzed main populations of all the endemic plant species did we preselect areas that were then subjected to as well as the vast majority of birds, reptiles, and intensive field studies and ranked according to their — insects. The levels of п proposed are in- value as potential protected areas. As a result, and dicated in Bramwell et al. 86). The Plan Es- in order to avoid creating what McEwen & McEwen pecial has been provisionally podido: by the Ca- (1982) described as “islets of natural beauty ... bildo Insular de Gran Canaria and has been passed Annals of the Missouri Botanical Garden FIGURE 8. Especial de Areas proposed for protection in the Plan Protección de Espacios Naturales (PEPEN) by the Cabildo Insular de Gran Canaria. on to the Canarian government for definitive ap- proval following which the Directorate General for the Environment will have full responsibility for implementation of the legislation. It is, however, already being used as a guideline document by the Regional Planning Commission for authorization of planning permission. It was also used by the Can- arian government's Environment Department as the basis for their 1987 Law on Protected Areas of Natural Interest, although in this case the total area was reduce lower level of protection in our plan did not fit into the scale of protection envisaged in the national as some of the areas given a Conservation research One of the most important conservation roles for a modern botanical garden involves conserva- tion-oriented research. The fact that many gardens and their staffs have not yet come to grips with this new field does not diminish its importance, as the papers presented at the recent symposium on Botanic Gardens and World Conservation Strategy show (Bramwell et al., 1987). In the Canary Is- lands, the “Viera y Clavijo" garden was originally set up as a center for the study of the local endemic flora, and over the past few years our work has been mainly directed toward conservation problems and particularly the interface between in situ and ex situ conservation. Research on individual species At least 120 of the 500 or so endemic plants of the islands are critically endangered and are potential subjects for conservation-oriented re- rch. Initial studies have generally involved field ob- servation, distribution mapping, population num- bers and size, age structure, and other features leading to a comprehensive picture of the situation of each species in the field. This has been followed by more critical studies of habitat and ecology in many cases. For example, we tried initially to grow the very rare and highly endangered Lotus kun- kelii in tissue culture, but failed until we carried out detailed studies of its natural substrate and found that the pH reached as high as 8.5 when temporary pools formed in the wet season dried out. We now have cultures of this species growing in a special high-pH medium. Reproductive biology of many species has also been studied—for example, pollination in some of our extremely rare red- and orange-flowered plants, such as /soplexis, Lavatera, Scrophularia, Lotus, and Canarina. Most of these, if they grew on the African mainland, would almost certainly be pol- linated by sunbirds (Nectarinidae), but there are no sunbirds in the Canaries. Recent research has shown, however, that several mainly insectivorous birds not usually associated with pollination (Phyl- loscopus and Sylvia species) have taken over the role, and pollen samples collected at random from the heads of such birds have been positively iden- tified as coming from such plants as /soplexis is- abelliana and Canarina canariensis. We have filmed pollination taking place in the natural hab- -. > at. The spectrum of floral structure of bird-polli- nated flowers in the Canaries is relatively narrow and does not include the long, narrow-tubed Aloe type. Observing the behavior of Phylloscopus species on cultivated Aloe, we found that it becomes a nectar-thief breaking into the base of the tube because its beak is too short to reach the nectar by normal means and, therefore, does not effec- tively pollinate long, narrow flowers. The surviving Canary bird-pollinated endemics all have relatively wide throats and short floral tubes. Since the establishment five years ago of a seed bank for the long-term storage of endemic species, we have been involved in the study of seed pro- duction and biology of a considerable number of endangered species. In many cases, for example, Tanacetum ptarmaciflorum, where germination is naturally very low (9%), this can be considerably Volume 77, Number 1 1990 Bramwell 35 Conserving Biodiversity in the Canary Islands improved (to over 60%), by treatment with gib- berellic acid. We now have a large study program on germination and seedling establishment with a view to projects of restocking and reintroduction of species from the seed bank. The seed bank provides material for a number of other research activities. When samples are removed from the bank for routine germination testing, any seedlings obtained can be used for studies that contribute to our knowledge of the species involved. Plants are grown for karyotype analysis and chromosome behavior, and study of variation, especially if the seed originates from very small, wild populations. Cotyledons, for example, can be used to establish tissue cultures, and seed- lings are passed on to the garden nurseries for planting and use in educational displays and other purposes. Recently conservationists have taken up the flag of biodiversity on a large scale, and in the Canaries this trend has been followed via the establishment of two projects under the name CODIGEN (Con- servation of Genetic Diversity). The first of these is a data-collecting project covering crop-plant relatives and known potential resource species in the Canarian flora. The project is being carried out in three phases. 1. An analysis of the actual situation of each species, distribution, biological cycle, potential and actual reproductive capacity, ecology, state of conservation, and classification according to the IUCN Red Data Book System 2. Evaluation of each species as a potential resource and the genetic diversity it represents, closeness of relationship to the crop species, known disease resistance or tolerance, possible use in plant breeding, direct uses and scientific value. ecommendations on conservation measures needed to protect each species, minimum number of populations, types of reserve, the need for ex situ conservation in gene banks or living collections, including possibilities for restocking in the natural habitat The final report is now available and has been handed over to the Canarian government for con- sideration in future conservation policy. The second project is the creation of a conser- vation model for a single species of major value as a natural resource and for this we chose the Ma- caronesian endemic grass Dactylis smithii, which is a member of the Dactylis glomerata (cocksfoot) complex. The aim of the project, which started in May 1988, is to review the distribution and morpho- logical, ecological, and geographical variation with- in the species as a first phase and then attempt to study genetic diversity by means of karyology, protein and isozyme analysis, and then select nat- ural variants for recommendation to plant breeders and evaluate the usefulness of this rare species as a forage grass via experimental cultivation. It could become an important species for dry-zone pastures. At the same time, the conservation situation of each natural population will be monitored and rec- ommendations will be made for conservation. Restoration of vegetation A new field into which the botanical garden research staff members are also moving is what can broadly be termed restoration biology. This involves, at the simplest level, restocking individ- uals into an already existing population to increase numbers and variability in the natural population. This has been done successfully during a pilot proj- ect funded by /SSC for Senecio hadroso- mus; plants were grown in tissue culture from ex- plants collected from a range of the surviving wild individuals, and the reintroduced individuals have been monitored over almost two years. Original reintroduction losses were high for this species but surviving individuals are thriving. There is again, however, a need for specialized research in this field if reintroduction is to become a viable means of conservation. Our attempts to reestablish areas of Canarian laurel forest were successful in that young trees have been reintroduced in former forest areas. Our experience with this showed that in the forest eco- system the development of the understory layer of shrubs, many of these important and rare endemics (e.g., Sideritis discolor, Isoplexis chalcantha), depends on the age structure and life cycle of the dominant trees, and the shrub species seem only to be able to establish themselves when gaps occur in the canopy. We need, therefore, to study the age structure of the dominant communities, species density, spatial organization, specialized niches, competition, effects of weeds and exotics, and a whole host of other factors in order to reconstitute ecosystems. Botanical gardens are probably among the most appropriate places for experimental work in this field Ecosystem reconstruction is an important sub- ject for conservationists concerned with the floras of small islands, and there is an urgent need for research on restoration biology in island ecosys- tems. Finally, I would like to mention micropropaga- tion and in vitro cultivation of endangered species. 36 Annals of the Missouri Botanical Garden The techniques are now well known and are the subject of specialist literature, but their application to propagation of rare plants warrants some com- ment. Cloning does not appear to be one of the most appropriate means of conserving genetic diversity, but under certain circumstances it can be a viable tool. In our approach we have considered in vitro or micropropagation as useful in the following cir- cumstances: 1. For species with chronically very small nat- ural populations where the removal of seed or other propagating material may further damage natural capacity for reproduction and maintenance of di- versity. There are in the Canaries several species known only from single localities and with popu- lations of fewer than fifty individuals, for example, Globularia ascanii, Helianthe- mum bystropogophyllum (a very distinct species Lotus kunkelii, with a known population of less that 10 individuals), and we have successfully propagated several of these in vitro. 2. For species with low natural fertility, either because of genetic problems (pollen and seed via- bility) or susceptibility to natural predators of pollen or seed. For example, the Flor de Mayo Lenosa (Senecio hadrosomus) loses most of its naturally produced seed to larvae of the dipteran fly Oedos- phenella canariensis (Suárez, 19 3. For propagation of species from remote areas when conventional propagating material, such as seed or cuttings, is not readily available. 4. For maintaining dioecious species in culti- vation and enabling propagation when plants of only one sex are available. This situation is not infrequent in botanical gardens, and stock can be maintained and propagated for distribution to other gardens via in vitro cultivation. 5. As a means of maintaining germ plasm, es- pecially of species with recalcitrant seeds, over long periods of time (Bramwell, in press 6. For mass production of horticulturally valu- able plants to relieve pressure on wild populations. An example of this is the very rare Fuphorbia landen from the island of Fuerteventura, which is highly desirable for succulent collectors and has been overcollected in the field. We have the species in tissue culture and have propagated it in quantity, and we hope to make plants available to the suc- culent trade. We may also be able to restock as the natural habitat of the species is now included in the proposed reserves established under the new legislation mentioned previously. Education Environmental education is one of the most im- portant aspects of our conservation activities, and we have a substantial program at the garden via organized garden visits and activities and through teacher-training courses, publicity, and extension activities. Two full-time teachers are seconded to the garden by the Education Department of the Canarian government and are responsible for daily visits, audio-visual shows, and a permanent exhibit on different aspects of the flora and fauna of the islands. (Currently there are two ma Canarian trees and birds of the Canary Islands.) Teacher-training courses, particularly, are a very successful means of introducing awareness of en- jor themes, vironmental problems and conservation of local ecosystems and species into the local education system. Our work in this field was reviewed by Navarro (1987) in Botanic Gardens and the World Conservation Strategy. Open days and public involvement with such projects as tree planting and the use of Canarian plants in local horticulture are also considered to be very productive in making the general public aware of the local flora and its value. From our nursery we supply plants to schools so that they can establish their own small Canarian gardens and to local councils for use in public parks and open spaces. During the next year we are planning to organize a tree planters club among school children to promote small-scale local planting of Canary palms (Phoenix canariensis), dragon trees (Dra- caena draco), laurels (Laurus azorica), and other species. CONCLUDING REMARKS Conservation of biodiversity in the Canary Is- lands is a substantial subject for a single paper, and | have attempted to give a general picture of the situation and how the activities of one botanic garden fit into it at each level, from providing the scientific basis for legislation and planning to an extensive education program. Such activities re- uire a dedicated team of workers, political and financial support, and the help of national and international conservation organizations. In order Q to coordinate this, we incorporated our conserva- tion activities into a single project we called **Plan- which could be presented as a pack- age to various funding bodies and for local support. tas y Futuro," The project covers mainly the field of research, including seed bank, micropropagation, a conser- vation data base, and the education program. It Volume 77, Number 1 1990 Bramwell 37 Conserving Biodiversity in the Canary Islands was successful in obtaining funding at local (Cabildo Insular de Gran Canaria), regional (Gobierno de Canarias), national (Dirección General de Medio- Ambiente del Ministerio de Obras Publicas, Ma- drid), and international levels through the WWF/ IUCN Plants Program, as well as from a local private foundation (Los Palmitos), who donated a microcomputer and a research fellowship for work on Euphorbia handiensis. Much more still needs to be done before we can claim that the future of the Canarian flora and fauna 15 secured, but we seem to be moving in the right direction. Unfortunately this is still not enough, nor is it on a time-scale that will result in the conservation of many species and ecosystems. However, knowing this should stimulate an even greater effort in the future. LITERATURE CITED ALDRIDGE A. 1979. Evolution within a single genus: Sonchus in Macaronesia. Pp. 279-292 in D. Bram- well (editor), Plants and Islands. Academic Press, London € New ise D, G. Lucas & R. DEFILIPPS. 1984. Our E ee Living World: The Wisdom to Save It. Cambridge Univ. Press, Cambridge, Lon- don & New York. BRAMWELL, D. 1972. Endemism in the flora of the Canary Islands. Pp. 141-159 гп D. Н. Valentine (editor), Taxonomy, Phytogeography and Evolution. Academic Press, London & New York. 1 Some Eni aspects of the adaptive r radiation of Canary Islands Echium species. Anales ‚ Bot. Cavanilles 32: 241-254. . The endemic flora of the Canary Is- lands, Pp. 207-240 in С. Kunkel (editor), Biogeog- raphy and Ecology in the Canary Islands. W. Junk, The Hague. 1985. Contribución a la че de las 7А Canarias. Bot. Масаг. 14 34. The role of “in vitro” ged ead in the con- servation of enda "dt species. (In press. & DRIGO. 1982. Prioridades para la con- servación n de la div bd genéticos en la flora de las Islas Canarias. Bot. Macar : 3- W. BELTRAN, V. MONTELONGO & C. Rios. 1986. Plan especial de protección de los espacios naturales de Gran Canaria (P.E.P.E.N.). Bot. Macar. 15: 1-72. ‚ О. МАК H. HEvwoop & Н. Syn 1987. Botanic Gardens and the World Coon Strategy. Academic Press, London. BROCHMANN, C. 1984. Hybridization d distribution of Argyranthemum coronopifoliu m (Asteraceae Аъ themideae) in the a а Nord. J. Bo CABRERA Y Diaz, A. La extincion de varias es- pecies de la flora Canaria. Bol. Soc. Esp. Hist. Nat. 0: -424. Carson, Н. L. 1981. Microevolution in insular ecosys- tems. Pp. 471-482 in D. Mueller-Dombois, K. Bridges & Н. Carson (editors), Island Ecosystems. Hutchinson Ross, Stroudsburg & Woods Hole. Davis, S. AND COLLABORATORS. 1986. Plants in Danger: a We Know? Internationa for Con- servation of Nature and Natural Resources. (IUCN) Gland & Cambridge. Humpuries, C. J. 1976. Evolution and endemism in Botanical notes on the Canary Islands . The evolution ч жк forms in the Islands: Aeo- nium. Ecology 4 McEwen, A. & Me :EwEN. 1982. National Parks: Conservation or Cosmetics? George Allen & Unwin, ondon. Navarro, B. 1987. The botanical garden as a vehicle for environmental education. Pp. 59-65 in D. Bram- well, O. Hamann, V. H. Heywood & H. Synge (ed- itors), Botanic Gardens and the a Conservation ‚ Lon Citogenética ke үне Lotus еп Macaronesia Ш: Variación en el contenido de glu- cósidos cianogénicos en Lotus be las Islas Canarias Madeira. Bot. Macar. 5: 9- PEREZ DE Paz, Р. L. & I. MEDINA. 1988. Catálogo de las Plantas Medicinales de la Flora Canaria. Gobierno de Canarias/Instituto de Estudios Canarios. Canary Islands. QUEZEL, Р. 1978. Analysis of the flora of Mediterranean and Saharan Africa. Ann. Missouri Bot. Gard. 65: 80. Catalogus plantarum in Canariensibus Insulis sponte et subsponte crescentium. Halle. SUÁREZ, C. 1982. Aportaciones a la Distribución y Ecolo- gía de Senecio Ате . fil var. preauxiana . Bi omus Svent. en Gran S Hist stoire ан turelle des Iles de ba Ш. are i 2. РҺуїор phia Canariensis. Par INTEGRATED STRATEGIES FOR CONSERVING PLANT GENETIC DIVERSITY" Donald A. Falk? ABSTRACT As the intensity and magnitude of threats to biological diversity increase worldwide, conservation strategies must develop accordingly i in three respects. First, conservation efforts must address all levels of biological organization, not simply species diversity; an overemphasis on species-level conservation jen missing biologically significant diversity at higher and lower levels of organization. Second, conservation measures st be conceived in response to particular threats or impacts to the biological entity of concern. Finally, the full фоли of conservation resources must esources in these three areas, a higher de d in a coordinated manner without exclusive reliance on any : single approach. By integrating information, gree of protection is ken for plant diversity than is possible by any ый strategy. It is this inherently multidisciplinary approach that characterizes integrated conservation strategies Current threats to the survival of global plant diversity, and the biological and economic conse- quences of its decline, have been well-documented and publicized (Myers, 1979, 1987; Davis et al., 1986; Brown, 1987; Raven, 1987; Wilson, 1988). The threats appear as diverse as the plants them- selves: habitats are fragmented, ranges reduced, and populations extirpated by virtually every known human use of land, from subsistence agriculture in the Third World tropics to resource extraction and recreation in developed nation Enormous investments are ГИР еасһ уеаг їп programs intended to conserve natural diversity їп order to head off biological disaster. For instance, by 1985 the cumulative expenditures of the World Wildlife Fund, a private conservation organization, reached over $100 million on 4,000 projects in 130 countries ш 1986); The Nature Соп- servancy, IUCN, and many other conservation or- ganizations maintain similarly extensive programs. Nonetheless, the pace of species decline and habitat destruction continues virtually unabated on a global scale, suggesting an unprecedented loss of biolog- ical diversity (Myers, 1979; Raven, 1987; Wilson, 1988). Despite persistent efforts by public and private conservation organizations, the pressures of population growth and demands on natural re- sources appear to be far greater than the ability to contain or redirect them. Under such circumstances it is vitally important to set priorities clearly and to allocate scarce con- servation resources carefully; no single method will suffice to provide adequate protection for biological diversity. Given the persistence and complexity of threats to biological systems, conservation cannot rely exclusively on any one approach, such as legal protection for species or acquisition of land, to ac- complish its broadest aims (Civen, 1987; Falk, 1987, 1989b; Jenkins, 1989). The increasing pace of destruction, continued shortage of re- sources for conservation, and the exigencies of resource development and population growth de- mand new models that draw on a wide range of tools and have the potential to build broad coalitions of support. A trend toward more integrative, synthetic modes of conservation exists in the programs of such organizations as the International Union for Con- servation of Nature and Natural Resources (IUCN, 1980; Bramwell et al., 1987), The Nature Con- servancy (Morse, 1987; Jenkins, 1989), Conser- vation International (1988), World Wildlife Fund (Phillips, 1986), Center for Plant Conservation (Falk, 1987, 1989b; Falk & McMahan, 1988) These agencies increasingly emphasize reciprocal interactions of conservation and economic devel- opment, and collaborative, multi-agency projects. These efforts thus illustrate an emerging model of integrated conservation strategies that show con- siderable promise for protecting biological diversity in coming decades. This paper will explore several elements of an integrated conservation approach, ' The author acknowledges the support and assistance of Linda R. McMah n and Ker y S. Walter of the Center r Plant Conservation in d oping the ideas contained in this paper, and of Linda S -DeBiom of the Center in editing and preparing the manuscri ? Center for Plant UE ur 125 Arborway, Jamaica Plain, Massachusetts 02130, U.S.A. ANN. Missouni Bor. GARD. 77: 38-47. 1990. Volume 77, Number 1 1990 Falk Conserving Plant Genetic Diversity examining as an example efforts to prevent rare plant extinction in the United States. THE INTEGRATED CONSERVATION MODEL The integrated conservation model is based on the assessment and synthesis of three basic sets of information (Falk, 1989b). These include (i) de- termination of the biological entity of concern, in- cluding definition of the target level of biological organization; (ii) identification of the threats to this entity; and (iii) consideration of the full range of conservation resources that may be brought to bear on the problem. A salient characteristic of inte- grated conservation approaches, then, is that they deliberately seek a broad base of information about a conservation problem and employ a wide range of complementary tools to accomplish a given ob- jective. DEFINING A TARGET LEVEL OF BIOLOGICAL ORGANIZATION The term “biodiversity” is often used in the general conservation literature without clear defi- nition. Diversity means a range of variety or dif- ference, and as such exists among alleles, individ- uals, populations, ecotypes, taxa, communities, and ecosystems. “Biological diversity” is frequently used as a shorthand reference to species diversity, but diversity at all levels of organization is of ecological and evolutionary consequence, and is therefore of concern to conservation (cf. MacMahon et al., 1978). Considerations of biological hierarchy in- fluence both the feasibility of a particular conser- vation project and its cost-effectiveness. For ex- ample, a strategy well suited to capture genetic variation in a single plant population may be inef- fective in protecting a whole ecosystem; converse- ly, acquiring and managing an entire ecosystem may be an expensive diversion of funds if all that is required in a specific instance is a sample of allelic variation at a particular locus. A later section of this paper will consider the conservation of infra- specific genetic variation in rare plant species as an example of the importance of this dimension. Taxon-oriented legal mandates such as the U.S. Endangered Species Act can be applied to conser- vation at multiple levels of diversity, although the application is occasionally difficult (Sidle & Bow- man, 1988). For plants, the Definitions of the Act identify species and subspecies as the target level of hierarchy [Endangered Species Act of 1973, as amended, 16 U.S.C. 1531 et seg., Section 3(16)]. In some respects, programs such as those of The Nature Conservancy, which are designed to func- tion at several levels of biological hierarchy si- multaneously, may thus be preferable. An “element of diversity” in the Conservancy’s system may be anything from a population to an ecosystem (Jen- kins, 1988). The latter approach is hampered pri- marily by the lack of commonly accepted classi- fication schemes for communities and other higher orders of biological organization, although progress is being made in this direction (S. Buttrick, pers. comm.). In extreme cases, a number of species have been reduced to a single individual; McMahan (1989b) estimated that there are four such taxa in the continental United States and two in Hawaii. Con- servation in these extreme cases, where each in- dividual plant is significant, must follow a different strategy than would be applied to a population. In some cases cuttings have been propagated to in- crease the number of ramets, but it is not known if species can survive such a severe bottleneck (Barrett & Kohn, in press). DEFINING SPECIFIC THREATS TO THE ENTITY OF CONCERN Having specified a target level of biological or- ganization, it is next necessary to determine the threat to that entity as precisely as possible. Al- though outright destruction or conversion of habitat is unquestionably the largest single cause of en- dangerment, many other factors—for example, in- vasion of exotic species or alteration of successional influences such as fire—contribute to the decline of populations, extinction of species, and disruption of communities. Other causes of endangerment include competition with invasive species, over- grazing by feral or commercial herbivores, suppres- sion of fire and other disruptions of the natural successional process, loss of pollinators or sym- bionts, alteration of surface and groundwater hy- drology, reduction of population size below mini- mum viable population level, and long-term climatic changes. Endangerment factors such as these may not respect property ownership boundaries or legal mandates. Thus, while legal protection for land and species is a necessary condition for conservation, specific threats to species and community diversity may require a long-term commitment to site man- agement and stewardship (Jenkins, 1989). This approach assumes that the causality of threat is fully known; in fact, there is frequently only an- ecdotal information or casual observation of trends or causes of decline, especially for communities. Recovery plans developed by the United States Fish and Wildlife Services and Biological Manage- 40 Annals of the Missouri Botanical Garden ment Abstracts of The Nature Conservancy pro- vide a growing data base on threats, but the ac- cumulation of such information is slow in relation to the overall scale of endangerment. For instance, as of June 1989 there were 85 recovery plans completed for the 208 federally threatened and endangered plant species (U.S. Fish and Wildlife Service, 1989), out of a total of approximately 5,000 rare or threatened taxa in the United States. Illegal collecting illustrates a cause of endan- germent that may not be controlled simply by ac- quisition of land. For many U.S. plants, overcol- lecting and vandalism represent a serious component of endangerment; examples include Pediocactus knowltonii L. Benson, Lilium occidentale Purdy, Lilium grayi S. Wats., Ancistrocactus tobuschii (W. Т. Marsh.) W. Т. Marsh. ex Backeberg, and Betula uber (Ashe) Fern. (Center for Plant Con- servation, 1987b; Olwell et al., 1987; Stafford, 1989). Herbivore browsing represents a particularly in- tractable source of endangerment, especially graz- ing and seed predation by feral and domestic cattle, sheep, goats, deer, and rodents. In many instances, this is the primary cause of decline, including Alec- tryon macrococcus Radlk., Phacelia argillacea Atwood, Agave arizonica Gentry & J eber, A. murpheyi F. Gibson, Pritchardia munroii Rock, Zizania texana A.S. Hitchc., and Hibiscadelphus distans Bishop & Herbst (DeLamater & Hodgson, 1987; Center for Plant Conservation, 1988a, THE CONSERVATION SPECTRUM The use of a wide range of conservation methods is compelled largely by the diversity in objects of conservation and threats they encounter. The in- tegrated conservation method is thus highly site- specific and situational, in contrast to traditional approaches that they have tended to greater uni- formity by stressing legal protection or fee-simple land acquisition in every case (Falk, 1989a). The early history of conservation in the United States was characterized by the setting aside of vast tracts of land —the Adirondacks Forest Pre- serve, Yellowstone and Yosemite national parks. The era of massive protection and acquisition on this scale in the United States probably concluded with passage of the Alaska National Interest Land Conservation Act in 1980; although some sub- stantial areas remain unprotected, most ac quisition is now on a smaller scale. For instance, in 1985- 1987 the average land acquisition project of The Nature Conservancy was approximately 423 acres (The Nature Conservancy national office data). Moreover, the total land area protected by private conservation organizations (between 3.5 and 4.0 million acres) is dwarfed by other types of own- ership such as Native American tribal lands, which include more than 25 million acres in the Southwest alone (Nabhan, 1988). urthermore, in the United States west of the Rocky Mountains, approximately 47% of the total land area is owned by the federal government, with another 21% owned by state and local governments (United States Geological Survey, 1970). Thus, in the United States, much of this land is managed by the Bureau of Land Management, United States Forest Service, and other agencies with multiple- use mandates that include resource extraction and recreation as well as conservation. A further consideration from a strategic point of view is the extent of human disruption in the functioning of natural systems and the distribution of species. True wilderness—that is, a natural sys- tem unaffected by human activity—is declining steadily worldwide. In the continental United States it is restricted almost entirely to Alaska and parts of the Great Basin, and even in the rest of North America wilderness is found primarily in the boreal forests and tundra of northern Canada. Virtually all other land has been affected by human activity, however slightly, whether by outright destruction or conversion of habitat, or by removal of preda- tors, pollinators, disruption of migratory pathways, introduction of exotic species, suppression of fire, or alteration of basic biogeochemical cycles and environmental parameters. Even fencing a habitat refuge may alter movement of animal species, and for wide-ranging species a protected area may en- compass only a portion of their total range (see Janzen, 1989). These influences have precipitated a dramatic shift away from the “‘laissez-faire” ethic that dominated conservation philosophy earlier in this century (Hays, 1959; Sober, 1986) toward an acceptance of biological management, which may be defined as a calculated intervention into the functioning of a natural system in order to attain a stated biological construction or goal (see also Western & Pear the palette of conservation practice has expanded As a consequence, considerably in recent decades, and now includes botanic gardens, arboreta, and seed banks (Bram- 1987; Jenkins, 1989), as well as more extensive programs of park management (Hales, 9) and restoration (Jordan et al., 1987). Each of these conservation methods is most effective at well et al., particular levels of biological hierarchy. For ex- ample, seed banks are well suited to conserve allelic diversity within a population, but inherently incapa- Volume 77, Number 1 1990 Falk Conserving Plant Genetic Diversity ble of conserving communities or ecosystems. Nonetheless, they may play a key role in an overall integrated strategy to address diversity at multiple levels of organization. In many respects these different approaches to conservation are distinguished more by degree than by kind. For instance, successional management of a fire-adapted ecosystem, such as prairie or savanna, may involve fencing, site preparation, controlled burns, and reseeding with native species. uch a management regime may be distinguished from a reintroduction program only by the number of years during which a particular species was absent from the site, or from ecological restoration only by the extent of reconstruction of a whole community. Modern land conservation manage- ment is essentially management of ecological succession, whether practiced as natural area stew- ardship, community restoration, or species rein- troduction. Ironically, often the tools of destruc- tion—bulldozer, herbicide, chainsaw, and fire— are, in different hands, the tools of conservation. The underlying continuity among these modes of conservation resolves the obsolete dichotomy be- tween ex situ and in situ conservation; neither exists in a pure form. Їп place of polarized alternatives, integrated conservation substitutes a spectrum of compatible, mutually reinforcing methods. CASE STUDIES IN INTEGRATED CONSERVATION The integrated approach is being used with in- creasing frequency in the conservation of rare plant species in the United States (Falk, 1987, 1988; Falk & McMahan, 1988). The following examples illustrate its utility and the diversity of situations in which it may be applied. Other examples o integrated conservation of endangered plant species are given in Table 1 Florida torreya (Torreya taxifolia Arn.) is en- demic to cool microclimate refugia in the Apa- lachicola River basin of Florida and Georgia. The species occurs largely on protected land in Florida’s Torreya State Park, the Apalachicola Bluffs and Ravines Preserve of The Nature Conservancy, and land owned by the U.S. Corps of Engineers and a city park department (U.S. Fish and Wildlife Ser- vice, 1986; McMahan, 1989a), but remains en- dangered by a severe fungal infection that prevents trees from reaching reproductive age. The primary conservation target is the population and species level, since the habitat is essentially protected. Con- servation measures in a program being undertaken by the Center for Plant Conservation in cooperation with The Nature Conservancy and other agencies include isolation and identification of native and exotic fungal pathogens, determination of natural ecological factors, such as fire, that may inhibit fungal growth and reproduction, and placement of experimental populations in more northerly loca- tions to determine optimal climate and potential range. In many cases the causes of population and species decline are largely ecological. For example, Malheur wire-lettuce (Stephanomeria malheuren- sis Gottlieb) was extirpated from its only known locality in east-central Oregon by overgrazing, competition, and fire (Brauner, 1988). Recovery was targeted at the species and population level, with an effort to reestablish the species in its original habitat using proposals collected by Gottlieb and germinated at the Berry Botanic Garden in Port- land, Oregon. The recovery project, which was funded by the United States Fish and Wildlife Service and the Bureau of Land Management, also included a study of community dynamics, by test- ing the success of S. malheurensis seedlings with five dominant ground covers (Center for Plant Con- servation, 1987a; U.S. Fish and Wildlife Service, 1982). An interesting case of conservation below the species level is Catalina Mountain mahogany (Cer- cocarpus traskiae Eastw.) endemic to a single can- yon on Santa Catalina Island in California’s Chan- nel Islands, and perhaps the rarest tree on the continent (Rieseberg, 1988). The taxon has been drastically reduced by grazing and uprooting by feral and introduced goats, pigs, sheep, mule deer, and bison. Earlier impacts included heavy cutting by settlers in the last century for use in making fenceposts. Many of these impacts were eliminated or moderated by acquisition of the land and fencing of the С. traskiae population in the 1970s by the Santa Catalina Conservancy. This protection, how- ever, revealed a more insidious threat to the species by hybridization with the more widespread C. be- tuloides Nutt. ex T. & G. var. blanchae (C. K. Schneid) Little, which overlaps С. traskiae in range. lectrophoretic and morphological analysis by Rieseberg et al. (1989b) determined that as many as five of the seven remaining adult trees were hybrids, as were several of the new seedlings within the fenced area, suggesting that swamping of the C. traskiae gene pool by C. betuloides var. blan- chae is occurring. Recommendations for biological management included removing adjacent stands of C. betuloides var. blanchae, transplant of C. tras- kiae, to protected sites el on Santa Catalina Island, and removal of nonnative herbivores from the area 42 Annals of the Missouri Botanical Garden TaBLE 1. Recent studies of integrated conservation and management of rare U.S. plant taxa. Taxa References Agave arizonica Arctostaphylos pallida Arctosta Banara vanderbiltii los uva-ursi subsp. leobreweri Cercocarpus traskiae Chrysopsis floridana Erysimum menziesii Goetzia ele Helianthus schweinitzii ans Center Hemizonia arida Hibiscus dasycalyx McM Шатпа corei Monardella linodes subsp. viminea Pediocactus knowltonii Penstemon barrettiae Spigelia gentianoides bancada malheurensis Amme & Havli Reid & Wal Popenoe d үөн. (19 Wallace & McMahan ña & Smith (1987) DeLamater & Hodgson (1987) (1987) sh (1987) 88); Rieseberg et al. (1989a) 1988 t Conservation (1988a) for Plan South Carolina Wildlife & Marine Resources Dept. (1989) Faull (1987) d (1987 Milne (1987); Olwell et al. (1987); Stafford (1989) Schwartz (1 Rogers (1988) U.S. Fish and Wildlife Service (1982); Center for Plant Conser- 988); Thompson (1988b); Kierstead (in press) vation oo Brauner (1988) 1х tex Trifolium colon rum Torreya taxifolia Warea amplexifolia Zizania texana x (1987) к. (1985); Pickering (1989) U.S. Fish and Wildlife Service (1986); McMahan (1989а) Center for Plant Conservation (1988a); Wallace & McMahan (1988 Center for Plant Conservation (1988a) CONSERVATION OF [NFRASPECIFIC GENETIC DIVERSITY IN RARE SPECIES Effective conservation of biological diversity re- quires a sound basis in scientific understanding of the entities being protected; this may be viewed as a fundamental axiom of conservation biology (Lacy, 1987). Nowhere is this principle more important than for conservation efforts for rare and endan- gered plants, many of which are highly endemic or reduced in range and facing serious threats to their survival. Conservation and management of these species often have little margin for error, and likewise little opportunity for extensive field ex- perimentation to determine optimal management techniques. Biological management, moreover, is an infor- mation-intensive undertaking, requiring detailed in- formation about the ecology, population biology, genetics, and representative biology of the target species. Such information, however, is in short supply in the literature; Table 2 lists recent pub- lished research on M population biology and ge- netics of rare U.S. plants. The paucity of infor- mation is in part p to the nature of the subject species themselves; endangered species are often restricted to one or two sites in remote areas, with poorly documented phenology and population his- tory. Ironically, we often have the least biological information on the very species about which we need to know the most An understanding of patterns of genetic varia- tion in rare plants is of particular importance for conservation strategies that involve offsite collec- tion, maintenance, and reintroduction of living plant germplasm. The distribution of genetic variation within and among populations and factors that may influence that distribution are crucial to the estab- lishment of genetically representative offsite col- lections. Similarly, inbreeding, drift, and methods of assessing genetic diversity within species are important considerations in long-term management of these collections for conservation and research (see Millar, 1987; Center for Plant Conservation, 1986; Kruckeberg & Rabinowitz, 1985 A case in point is the conservation and man- agement of genetic variation within rare species, which are frequently of the most critical conser- vation concern (see Huenneke & Holsinger, 1986; Lacy, 1988). Studies of genetic variation in rare plants are still infrequent in the literature. For example, a series of papers on the rare endemic Pedicularis furbishiae S. Wats. has contributed useful knowledge to its conservation (Macior, 1980; Menges et al., 1985; Gawler et al., 7; Waller et al., 1987), but few rare taxa have been studied Several other recent studies of the genetics of Volume 77, Number 1 1990 Falk Conserving Plant Genetic Diversity 43 TABLE 2. Studies of the population biology, genetics, and ecology of rare North American plant taxa. Taxa References Abies bracteata Aeschynomene virginica Astragalus osterhoutii and A. linofolius Astragalus tennesseensis Calochortus spp Cercocarpus traskiae ена, iowense Clarkia бека lala Collomia rawsoniana Helianthus exilis A. Gray Helianthus paradoxus Heiser Helianthus praecox subsp. hirtus Howellia aquatilis Hymenoxys acaulis var. glabra Leavenworthia spp. Lespedeza ен Mimulus gutta Oenothera organensis Orcuttia Ledig (1987) Carulli & Fairbrothers (1988) Karron (1987a); Karron et al. (1988) Bowles et al. (1987b) Fiedler (1986, 1987) Rieseberg et al. (1989b); Rieseberg (in press) Schwartz (1985 Gottlieb (1973, 1974) Erickson (1945); Learn & Schaal (1987) Taylor et al. (1987) Conkle (1987) Wiens et al. (1987, 1988) Berg (1987 Rieseberg et al. (1988); Rieseberg (in press) Rieseberg et al. (submitted 1989a); Rieseberg (in press) Rieseberg & Doyle (1989 Lesica et al. (1988) DeMauro (1989) Solbrig (1972) Smith (1987) Ritland & Ganders (1987a, b) Levin et al. (1979) Griggs & Jain (1973) Pedicularis үт Macior (1980); Menges et al. (1985); Gawler et al. (1987); Waller et al. (1987) Pinus radiata Pinus torreyana Plantago cordata E Sioa шш Silene inv Warea ш ЖОН Millar et al. (1988) Ledig & Conkle (1983) Meagher et al. (1978); Primack (1980); Bowles et al. (1987a) Kubetin € Schaal (1979) Taylor & Palmer (1987) Wallace & McMahan (1988) rare species have contributed valuable insight into the biology of th , including the studies listed in Table 2. Nonetheless, considering that there are upward of 5,000 plant taxa of conser- vation concern in the United States alone—rep- resenting close to 20% of the entire national flora— the overall extent of research into the population biology and genetics of rare species is disappoint- ingly small. One illustration of the need for population bio- logical data for conservation is the clasping warea (Warea amplexifolia (Nutt.)), endemic to central peninsular Florida. The two remaining populations are composed of fewer than 100 individuals each; as the species is annual, population sizes fluctuate dramatically from year to year (Center for Plant Conservation, 1988a; Wallace & McMahan, 1988). In addition to the dangers of demographic stochasticity, seed set has been extremely low in recent years, suggesting that the populations may be below a minimum viable population threshold and hence subjected to intensified levels of inbreed- ing. Karron (1987a; Karron et al., 1988) compared levels of genetic polymorphism in common an widespread taxa within a genus (Astragalus, Fa- baceae) and observed that while some rare species are genetically depauperate, others exhibit as much polymorphism as more widespread congeners (see also Karron, 1987b). Other studies of genetic vari- ation in rare species such as those of Pinus radiata (Millar et al., 1988) indicate unexpectedly high levels of polymorphism even among severely re- stricted taxa (see also Ledig & Conkle, 1983; Con- kle, 1987; Ledig, 1987). Thus, while some species conform to Hamrick’s findings (1983; Hamrick et al., 1979, 1981) that endemic species generally exhibit lower diversity than widespread species, many other species do not. Historical, reproduc- tive, and ecological factors may account for much of the observed variation in rare species, but the number of comparative studies in general is still small. The ecological and evolutionary significance of genetic variability in rare species is still a matter of significant debate. Huenneke (in press), for in- 44 Annals of the Missouri Botanical Garden stance, argues that the loss of even low-frequenc y alleles in species that are already genetically re- duced may result in significantly lower evolutionary fitness. By contrast, there is some evidence that plants may be able to survive and re-radiate fol- lowing severe genetic bottleneck events (Templeton & Reid, 1983; Barrett & Kohn, in press; see also Barton & Charlesworth, 1984). Many species are susceptible to inbreeding at low effective population numbers, but Menges (in press) has discussed a variety of reproductive strategies and other life history characteristics that may tend to buffer plants from the effect of low population size. The consequences for conservation of an inad- equate base of population genetic research should not be underestimated. For instance, land acqui- sition programs for habitat conservation or field sampling for offsite germplasm collections are often undertaken to help ensure the survival of rare species and to prevent anthropogenic extinctions. But without an understanding of the partitioning of genetic variation in populations of rare species, r in the ecological and evolutionary impact of inbreeding in small populations, there may be in- sufficient criteria to guide the scientific selection, design, and management of reserves or offsite germplasm sampling or reintroduction programs. In any event, since rare species often occur in small, isolated populations characterized by intense inbreeding at low effective population sizes, con- servation of intraspecific genetic diversity must be recognized as critically important in the manage- ment of these taxa. CONCLUSION Rare plant species present a particularly acute set of problems in the conservation of biological diversity. Because of their restricted geographic ranges, frequently narrow ecological amplitudes, and vulnerabilities to drift, inbreeding, and sto- chastic events, endangered species represent crit- ical priorities in biodiversity conservation. There is a pressing need for improved knowledge of their biology, especially in the areas of population bi- ology, genetics, and ecology. In many cases, an integrated conservation ap- proach offers the only realistic hope of preventing extinction. By focusing on a defined level of bio- logical hierarchy, integrated strategies help clarify the objects of conservation. By articulating the threats specific to that entity, such strategies iden- tify the proximate causes of decline; this enables the use of the widest range of conservation methods from the spectrum of available resources. And, by integrating these basic elements into an action plan that utilizes a wide variety of conservation re- plant biodiversity —from the allele to the may be best protected. sources, community LITERATURE CITED AMME, D. & N. Haviik. 1987. Assessment and man- agement and 4rc к pallida Eastwood. Pp. їп (editor), Conservation and M fornia Native Plant Society, Sac ramento, n BARRETT, S. C. H. € H a eneti and e и tionary y consequences of sma ize in plant ms for conser valon: In: DA Fal & К. аге . Press, BARTCIS, R. 1985. Banon of Trifolium stoloni- ferum Muhl. ex A. Eaton. Rhodora 87: 425-429 BARTON, N. . CHARLESWORTH. 1984. Genetic revolutions, ndes effects, and speciation. Ann. Rev 15: 133-164. Berc, K. S. 1987. Population dynamics of Erysimum menziesii, a facultative biennial mustard. Pp. 351- 8 in T. S. Elias (editor), Conservation and Man- ы of Rare and Endangered Plants. = Native Plant Society, Sacramento, Califo Bow Les, M. K., R. BATCHELL & M. M. DEMA Uno. "19872. Status and restoration of Plantago cordata in the southern lake region. Morton Arboretum, Lisle, Il- ols y, ——, ——, L. С. Sykora & L. R. Bau- 1987Ь. Propagation o used in es- TISTA tablishing a greenhouse population of Astragalus tennesseensis Gray. Morton oa Lisle, Ili- nois. D., O. Hamann, V. Heywoop & H. S BRAMWELL, 1987. Botanic Gardens and the oe Contain Strategy. ma Press, Londor BRAUNER, 5. Malheur wire- an (Stephano- meria пант biology and interactions wit ss: 1987 study results and recommendations M agement, Burns, Oregon District, U.S. Dept. of In- terior. Brown, L. R. о 1987. State of the World 1987. W. W. Norton, New York. Сакош, J. Р . E. FAIRBROTHERS. 1988. Allozyme variation in жс ее eastern United States 5 m of Aeschynomene (Fabaceae), including the rare А. virginica. Syst. Bot. 13: 559-: CENTER FOR PLANT CONSERVATION. jm Recommen- dations for the Collection and Ex Situ Management of Germplasm Resources from Wild Plants. Jamaica Plain, Massachusetts. 87a. Experimental reintroduction shows success. Pl. Conserv. 2(4): 4. 987b. Rare Texas cactus poached. Pl. Con- LE 9 Five аша of endangered plants. Pl. Canter 3(4): 2 1988b dint adt Fund supports Hawai- ian plants, Pl. Conserv. 4(1):7. CoNkKLE, M. T. 1987. Electrophoretic analysis of vari- ation in native Monterey cypress (Cupressus macro- Volume 77, Number 1 90 Falk 45 Conserving Plant Genetic Diversity carpa Hartw.). Pp. 249-256 in T. S. Elias (editor), Conservation and Management of Rare and Endan- pai e California Native Plant Society, Sac- o, California. Conservation INTERNATIONAL. 1988. Ecosystem Con- vation: Context = Biological Conservation and ment. Conservation International, Chasing the wild Texas snowbells. Pl. Conserv. 2(4): 1, 8. Davis, S. D., S. J. M. Dr ROOP, P. GREGERSON, L. HENSON, C. J. LEON, J. L. Vitta-Lopos, Н. SYNGE & J. ZANTOVSKA. 1986. Plants in Danger: e Know? International Union for Conservation of Мыр иге and Natural Resources, Gland, Switzerland. DELAMATER, & W. HopcsoN. 1987. Agave ari- zonica: an endangered species, a hybrid, or does it matter? Pp. 305-310 in T. S. Elias (editor), Con- servation and Management of Rare and Endangered Plants. California Native Plant Society, Sacramento, California. DEMauno, M. 1989. ology of the endangered Hy ys acaulis var. то implications for Eu Masters Thesis. Univ. of Illinois, E s Illinois. Erickson, В. O. 1945 e Clematis fremontii var. riehlii population in the Denis Ann. Missouri Вог ard. 32: 413-460. FALK, D. A. 1 Integrated conservation be for erroe plants. Nat. Areas J. 7(3): 118-123. elping to ensure a future in the wild. PI. mer 3(2): 3. 89a. Integrated conservation strategies for eI forest resources. Forest Diversity (in press). . 1989b. The theory of integrated conservation strategies for biological diversity. /n: Proc. of Nat. Areas Assoc., Syracuse, New York, 6-9 June 1988. Nat. Areas Assoc., Rockford, Illinois (in press). — — — & L. R. McManan. 1988. Endangered species: managing for diversity. Nat. Areas J. 8(2): 91-99. FauLL, M. R. 19 Management of Hemizonia avida (Asteraceae) by the Er dede Department of Parks 440 in T. S. Elias (editor), ment of Rare and Endan- iA Plants. California Native Plant Society, Sac- ento, Californi iB. & S. Bone: 1987. Methods of i px native AE acr of Erysimum menziesii. Pp. 50 511 ¿n T. S. Elias (editor), Conservation and 2 екы of Rare and Endangered Plants. California Native Plant Society, Sacramento, California. FIEDLER, Р. 1986. Concepts of rarity in vascular plant species, with special reference to the genus Calo- сенш» Taxon 35: 502-518. . 1987. Life history and population dynamics of rare and common Mariposa lilies (Calochortus Pursh: Liliaceae). J. Ecol. 75: 977-995. GAWLER, S. C., D. M. WALTER & E. М. MENGES. 1987. Environmental factors affecting establishment and growth of Pedicularis furbishiae, a rare endemic of the St. John River, Maine. Bull. Torrey Bot. Club 114: 280-292 87. What a conservationist requires of ex situ collections. Pp. 105-118 in D. Bramwell, O. Hamann, V. Heywood & H. Synge (editors), Bo- tanic Gardens and the World Conservation Strategy. Academic Press, London. Aspects of the SAS bi- GOTTLIEB, L. D. 1973. Enzyme NA and phy- loge = in Clarkia franciscana, C. rubicunda and C. ena. Evolution 205-21 1974. Genetic confirmation of the origin of Clarkia lingulata. Evolution 28: 244-25 СА 197 Скіссѕ, Е. Т. Conservation of vernal pool plants in California II. Population biology a rare and unique grass genus Orcuttia. Biol. Conserv. 27 1-193. Hates, D. 1989. Changing санаи of national parks. Pp. 139-144 in D. West M. Pearl yc Conservation for E ‘Twenty first Century. Oxfor Univ. Press, New Hamrick, J. L. 1983. The distribution of genetic vari- ation within and among plant populations. Pp. 335- in С. . M. Chambers, F. MacBryde & L. Thomas (editors), Genetics and Con- servation: A Reference for Managing Wild Animal and Plant Populations. benjamin ИШЕНИШ Pub- lishing, Menlo е "pe Ў В. LINHART & J. В. ON. lationships between "life hist a characteristics ба олунар ama cie gene ariation in plants. Ann. Rev. Ecol. ‚ LO: 173- 200. " Т B. Mitton & Y. В. > mad 1981. Levels of genetic variation in trees: influence of life history cti Pp. 35-42 in М. Т. Conkle (editor), Isozymes o North American уак se and Forest Exp. Sta. Techn. Conservation and the Gospel of Effi- ciency: The Progressive Conservation Movement. Harvard Univ. Press, Cambridge, Massachusetts. HUENNEKE, L. F. Ecological implications of genetic vari- ation in plant populations. /n: D. A. Falk & K. E Holsinger (editors), The Conservation of Rare Plants: Biology and Genetics. Oxford Univ. Press, New York (in press). K. E. HorsiNGER. 1986. Plant population biology and the к of viable plant рори- lations. Pp. 1 3 in B. A. Wilcox, P. F. Brussard & . Marcot e e Management of бы pee. Theory, Applications, and Case ies. Center for Conservation Biology, Stanford, Califor. nia. ет UNION FOR CONSERVATION OF NATURE AND Na L RESOURCES. . World Conservation Кы. IUCN/ United Nations Environ gramme/ World Wildlife Fund, Gland, иат Й. JANZEN, D. Н. The evolutionary biology of na- tional parks. Сеше: Biol. 3(2): 109-112. JENKINS, R. E. 1988. Information management for the conservation of biodiversity. Pp. 231-239 in E. O. Wilson da Biodiversity. National Academy Press, Washin mental Pro 1589, Long-term conservation and preserve complexes. The Nature Conservancy Mag. 39: 4-7. JORDAN, W., III, M. GILPIN & J. D. ABER (editors). 1987. Restoration Ecology: A Synthetic но to Eco- logical Research. Cambridge Univ. Pre Yor Karron, J. D. 1987a. The pollination ie of co- occurring geographically restricted and widespread species of Astragalus. Biol. Conserv. 39: 179-193. 1987b. A comparison of levels of genetic polymorphism and self- a in A Ev cally i estr icted and t ol. Ecol. 46 Annals of the Missouri Botanical Garden , Y. B. LinHarT, C. A. CHAULK & С. ROBERTSON. 1988. Genetic structure of populations of geographically restricted and widespread species of Astragalus (Fabaceae). Amer. J. Bot. 75(8): 1114- 1119. KiERSTEAD, J. R. Reestablishment of Barrett's DE Penstemon жана Gray) at Merian Columbia Gorge, Oregon. In: Proc. of Soc. Ecol. Rest. Management, Oakland, California, 16-20 Jan. ~ 1989. Soc. Ecol. Restoration Managem., Madison, Wisconsin бё ргезз). KRUCKEBERG, А. R. & D. RABINOWITZ. 1985. m al = ts of endemism in higher plants. Ann. Rev. Ecc Syst. 16: 447-479. KUBETIN, W. R. & B. A. SCHAAL. 1979. Apportionment of isozyme variability in Polygonum pensylvanicum (Polygonaceae). Syst. Bot. 4: 148-156. Lacy, В. С. 1987. Loss of genetic diversity from man- aged populations: interacting effects of drift, muta- tion, immigration, selection, and population subdi- vision. Conserv. Biol. 1(2): 143- report on population genetics in conservation. Conserv. Biol. 2(3): 245-247 LEARN, G. H., Jr. & B. A. ScHAAL. 1987. Population subdivision. for ribosomal DNA repeat variants Clematis md Evolution 41: 433-438. Lepic, К. T. 1987. Genetic structure and the conser- vation of С 'alifornia's — Boe near-e endemic со- des C ne Native Plant Society, Sacramento, Califorr & N E CONKLE. 1983. Gene diversity and genetic structure in a narrow endemic, Torrey pine (Pinus torreya Parry ex Carr). Evolution 37: 85. LESICA, . Leary, F. W. ALLENDORF & D. E. BILDE RBAC к. 1988. Lack of genetic diversity within and among populations of an endangered plant, How- ellia aquatilis. Conserv. vae 2(3): 275-282. Levin, D . ELLSTRAND. 1979. Protein polymorphism in le narrow Md Oe- nothera organensis. Evolution 33: 534-542. Macion, L. W. 1980. The population a of Fur- ts.). bish’s lousewort И furbishiae 5. Wat Rhodora 82: 105 MacMaHon vas A., D. L. PHILLIPS, J. V. oe x D. J. SCHIMPF. 1978. Levels of biol an organism-centered approach. BioScience 28: 700- 704. МСМАНАМ, L. R. 1988. Neches River rose mallow. Pl. Conserv. 3(4): 7 ——. 1989 Conservationists join forces to save Florida torreya. Pl. Conserv. 4(1): 1, 8. 1989b. Rarest U. S: plant are literally one aa kind. Pl. Conserv. 4(2): МЕАСНЕВ, T. R., J. i К. Primack. 1978. Experimental ecological genetics in Plantago. II. Genetic variation and demography in relation to sur- vival of Plantago cordata, a rare species. Biol. Con serv. 14: 243-257. Mences, E. 5. The application of minimum viable pop- ulation theory to plants. /n: D. A. Falk & K. E. Holsinger (editors), The Conservation of Rare Plants: Biology and Genetics. Oxford Univ. Press, New York (in press). ]. GAWLER & D. M. WALLER. 1985. Pop- m Dolus of the endemic plant Furbish's louse- rt (Pedicularis furbishiae). Report to the U.S. Fish and Wildlife Service. ша Research In- stitute, Indianapolis, Indian MiLLaR, C. I. 1987, The California germplasm con- servation project. Conserv. er 1(3): 191-193. . STRAUSS, M E & R. D. WESTFALL. 1988. Alloo ее но aa biosystematics of the Californian closed-cone pines (Pinus subsect. Oocarpae). Syst. Bot. 13: 351-370 MILNE, M 1987. Conserving some of the world's tiniest cacti. Pl. Conserv. ни 1, 8. Morse, n E. 1987. lant protection, Conservancy style. The Nature С аи News 37(5): 10-15. Myers, М. 1979. The Sinking Ark. a Press, Oxford dt 87. The extinction spasm impending: syn- ergisms at work. Conserv. Biol. 1(1): 14-21. NABHAN, G. 1988. Southwest project cuts across cul- tural, us vereque Р, Сопзегу. 3(2): 1, 8. OLWELL, P., А. CULLY & S. BRAC 1987. Pediocactus knowltonii adus efforts. In: T. S. Elias (editor), Conservation and Management of Rare and Endan- gered Plants. макшар Native Plant ко Sac- ramento, Califorr PHILLIPS, С. 1986 VWF Yearbook 1985/86. World Wildlife Fund, Gland, Switzerland. PICKERING, 1989 eenen efforts boost hopes for rare clover. Pl. Conserv. 4(2): 3. POPENOE, J. One of М world's rarest species. ): 980. Phenotypic variation of rare and о species of Plantago. Rhodora 82: 87- RAVEN, "v. H. 1987. The a wis plant conservation problem world-wide. Pp. in D. Bramwell, O. Hamann, V. Heywood E = ги о irl Gardens and the World Conservation Strateg ac emic Press, London. Кыр, T. S. & R. C. Warsa. 1987. Habitat reclamation for endangered species on San Bruno Mountain. Pp. 500 in T. S. Elias (editor), Conservation and M of Rare and Endangered Plants. Cali- fornia Native Plant Society, Sacramento, California. RIESEBERG, L. Saving California’s rarest tree. 1. Conserv. 3(1): 1, 8. yb ar in rare plants. ра from case ME in 1 Helia and Cercocarpus. In: D. A. a . E. fer (editors), The Co nservation of Rare Plante: Biology and Genetics. Oxford Un Press, New Y ss). —— & M. 1980 Allozyme variation іп Helianthus s praeco x ssp. hirtus, a rare sunflower from ers Texas. Aliso 12: 379 ; , R. CaRTER & S. ZONA. 19893. Molecular tests of the beni hybrid origin of two diploid Helianthus species (Asteraceae). Evolution (submit- ted). s & J. D. PALMER. 1988. A mo lec alee reexamination of introgression between He- lianthus a H. bolanderi (Compositae). Evolution 42: 227-238. ONA, L. ABERBOM & T. MARTIN. 1989b. Hybridization in the E endemie, Catalina ma- hogany. Conserv. Biol. 3(1): 1- Volume 77, Number 1 1990 Falk 47 Conserving Plant Genetic Diversity RITLAND, К. & К. R. GANDERS. 1987a. Covariation of selfing rates with parental gene fixation indices within populations of Mimulus guttatus. Evolution 41: 760- pest ————. 1987b. Crossability of Mimulus guttatus in relation to components of gene fixation. Evolution 41: 772-786. 1988. Spigelia jc ces a ш i , 8. monardella in San Fein County: site selection for transplants. Pp. 501-5 d T ias (editor), Conservation and Manage nt of Rar e and Endan- ramento, Сад. . 1988. Banking on seeds to avert ex- tinction. Audubon Mag. Jan. 1988: 22-25. SCHWARTZ, O. A. 1985. Lack of protein shail phum in the endemic relict Chrysosplenium iowense (Sax 031-2034. . B. Bowman. 1988. Habitat protection under the "Trad Species Act. Conserv. Biol. 2(1): 116-118. SMrTH, W. R. 1987. Studies of the population biology Pp. Soper, E. 1986. Philosophical problems for environ- mentalism. Pp. 173- n B. G. Norton (editor), he Preservation a Species Princeton Univ. Press, ARINE RESOURCES DE- are sunflower planted on South Carolina p line. Pl. Conserv. 4(1): 2. STAFFORD, R. 1989. Botanists find cactus better neigh- hood. Pl. Conserv. 4(1): 3 W. & В. E. PALMER. 1987. Ecology and enda — status of Silene invisa populations in the central Sierra Nevada, California. Pp. 321-328 in ES Elias Ао) Conservation ап nd Man na ement dangere . STEBBINS & W. В. 1987. En- dangered status of Collomia rawsoniana (Polemo- niaceae), Western Sierra Nevada, California. Pp. 225-232 in Т. 5. Elias (editor), Conservation and Management of Rare and Endangered Plants. Cali- fornia Native Plant Society, ve c California. TEMPLETON, А. R. & B. REID. 198 elimination of inbreeding ORA YER in a captive herd of Speke's gazelle. Ы 4 -246 in С M. Cha Cummings Publishing Co., Menlo P THOMPSON, P. J. 1988a. Peter' 5 та рге mallow. Pl. ея 3(4): 7 88b. Pio barrettiae, renewed plant the af ck. Pl. Conserv. 3(2): Dar adi FISH AND WILDLIFE SERVICE. 1982. Mal- wire-lettuce listed with critical habitat. Endan- gered тем Techn. Bull. 7(12): 1 ——. 1986. Florida torreya (Torreya taxifolia) re- capers plan. Southeast Regional Office, Atlanta, Mie 1989. Box score of listings E recom plans. Endan ngered Species Techn. Bull. : 12. UNITED STATES GEOLOGICAL SURVEY. po The Na- tional Atlas of the United States of America. U.S. G 1988. A place in WALLACE, S. В. . McMaHAN. the sun for the plants. не а 20-23. WaLLER, D. M., D. M. O” T da GAWLER. 1987. Genetic variation endemic agit Pedicularis furbishiae (Scrophularia Con- serv. Biol. 1(4): 335-340. WESTERN, D. & M. PEARL (editors). 1989. Conservation for the Twenty-First Century. Oxford Univ. Press, New York. Wiens, D., C. I. Davern & С. L. CaLvin. 1988. Ex- ceptionally low seed set in Dedeckera eurekensis: is tion Symposium, Bish- op, -r Univ. of California Press, Berkeley Califor T ГА CaLvIN, С. A. WILSON, С. I. DAVERN, D. FRANK & 5. К. SEAVEY. 1987. поо suc- cess, spontaneous embryo abortion, and genetic load : 501-509. 988. Biodiversity. National Academy Press, Washington, D.C. CONSERVING BOTANICAL David Given* DIVERSITY ON A GLOBAL SCALE! ABSTRACT e world moves toward the twenty-first century, it seems ill prepared to cope with increasing social, economic, and dune cibis problems. It is not surprisi ng that loss of biodiversity i is not perceived as a problem by many people. Current depletion rates are = serious proportions. Attention is eri focused on losses to moist tropical forests and the need to monitor this at a variety of levels: area and quality of habitat, species extinction, and e of genetic variability. However, йз эжене i of tropical forest must not be at us expense of other habitats Alarming losses are occurring in many other systems, such as wetlands and isa shrublands. The a nts of even such recently pae regions as the Antarctic are under pressure from human influences. Future scenarios are not easy to construct as there are many uncertainties. Major deleterious зш could be: global climatic changes especially through the greenhouse effect and alteration in the ozone layer around the earth; m epletion, genetic loss, and climatic changes resulting from tropical forest loss; increase in the human population; economic instability and imbalances in distribution of wealth; and decreased resources for research and biological « conservation. n view of this, there is need to question whether conventional strategies chiefly involving protected natural areas, botanic models. Mane ement 9 modified landscapes 15 с to ae nid singen in the future but needs to be ALIO dec cisions that may be unpopu ular with some нен of conservation but may be necessary for the long- term well- being. of the biosphere. An increased level of cooperation is necessary; currently this is hampered in some countries a "free market" philosophy of Vea du funding. Adequate resources for conservation are essential, a this means establishing a better case for research and ылкы лы funding, clearer definition of objectives, and ater accountability by scientists. Perhaps, above e all a new ethic is long overdue, marked by a return to the concept of Дом and regional commons with recognition of inerdependence rather than independence. There is danger in the view that “everyone is a conse ervationist at heart” unless there is clear understa nding of what this means in that conservation research and management must not ¢ aniy | þe directed for ihe good of plants and animals, but must be communicated to people living alongside then If we sell you our land, love it as we've loved it. this: Is biological conservation an important issue Care for it as we cared for it. Hold in your mind today? If the answer is negative, then we are prob- the e of the land as it is when you take it А : : ' with all iile strength, with all your mind ably wasting our time being here today! Norman with : all your "m = erve it for your children — Myers (1979, p. 3) succinctly expressed the view Chief Seattle, of many people: Recently, І кү the head of science ata Ask a man in the street what he thinks of the problem local high school. During the conversation 1 asked of disappearing species, = he may well re ply that why there is so little conservation in the senior it would be a pity 1 tiger or the blue whale disappeared. But he may well add that it would be science syllabus. He thought and answered, “Well, E no big deal, not as compared ee cris ses of energy, conservation of plants and animals was an issue in PEE : ie nos food and pollution eal’ problems. the 1970s, but it isn't really an issue now* One In other words, he cares about мейде, species, of the questions that we need to address is exactly but he cares about many other issues more: he simply ! [ am grateful to the staff of the Missouri Botanical Garden, especially Peter Raven and Gerrit Davidse, for the opportunity to present this paper and to Warwick Harris, Director of Botany Division, DSIR, for his support. t. Numerous colleagues commented on sections of the manuscript and Colin Webb, Murray Parsons, and Colin Meurk provided perceptive discussion on the whole text. Discussions and correspondence worldwide, in the course of my work for IUCN and WWF International, have contributed to the ideas expressed here. Consequently, I dedicate this paper to the many dedicated conservationists whose tireless but often unacknowledged efforts are devoted to making this a better world for future generatio 2 Bot tany Division, DSIR, сама, New Zealand. ANN. MISSOURI BOT. GARD. 77: 48-62. 1990. Volume 77, Number 1 1990 Given Conserving Biological Diversity does not see it as a critical issue. If the tiger were to go extinct tonight, the sun would still come up tomorrow morning. Myers pointed out that his man in the street is quite correct in one respect —by tomorrow morn- ing there is likely to be at least one fewer species on planet Earth than there was this morning. Diversity is often expressed as numbers of taxa, and commonly we speak of conserving “species.” But diversity can be expressed in many other ways such as morphology, size, color, function, range of habitats, and use by people. It is useful to distin- guish three principal levels of diversity. Ecological diversity is expressed as diversity of habitats and ecosystems, the diversity we see in a mosaic of forest, grassland, wetland, and other habitat types throughout a landscape. Species diversity is ex- pressed as numbers of taxa, reaching maximum levels in regions with high numbers of locally en- demic species. Genetic diversity occurs at the level of genes, the building blocks of life which make one individual and population different from another Whatever the measure of diversity, there is little doubt that the world of plants is immensely varied. There are about 235,000 flowering and 270,000 nonflowering extant plant species uus et al., many different habitats from the tropics to polar regions, and from 100 m under the sea to over 6,000 m altitude. An exciting outcome of polar botany in recent years has been the discovery of plants in the Antarctic growing under environ- mental extremes formerly considered uninhabit- able. This includes mosses growing on m d soils of active volcanoes (Broady et al., 1987) an endolithic algae of the dry valleys of southern is toria Land (Friedmann, 1982). The latter plants live in the interstices just below the surface of coarse-grained rock, demonstrating the persistence of life under the most adverse of environments. Plant diversity is essential to the Earth's biosphere as we know it. With the exception of some sulphur bacteria, all animals are ultimately dependent on the Earth's green mantle of photosynthetic plants. WHAT ARE THE Losses? All species have a limited life in geological terms, whether it be measured in hundreds of thousands or tens of millions of years. But this is poor jus- tification for the frequently offered counter to con- servation that: “Because extinctions have always been occurring there is surely nothing to be con- cerned about. Does it matter if a few more species disappear?” This argument ignores the balance between extinction of some taxa and the evolution of others. What causes concern is that current rates of extinction are many times the natural background rate. The present accelerating loss of species has been likened to mass extinctions of the geological past, where five particularly prominent **megaspasms" of extinction have been identified from the fossil record (Jablonski, 1986). Present extinction rates are at least several hundred times the current nat- ural background rate, and it is argued that they will rise several orders of magnitude by early next century to constitute the greatest mass extinction for at least 65 million years (Myers, 1979; Ehrlich & Ehrlich, 1981; Jablonski, 1986). A unique fea- ture of the present mass extinction is that it is primarily due to the impact of a single species, Homo sapiens. Extinction rates are surprisingly hard to mea- sure, especially on a global scale. What can now е measured reasonably accurately is the rate of loss of habitat. Remote sensing backed by ground surveys provides the key to this. The Global En- vironment Monitoring System (GEMS) and espe- cially the Global Resources Information data base permit measurement on a continuing basis of hab- itat loss. The emerging picture is disturbing. Taking ev- ergreen, moist tropical forests as an example, there were probably about 1.6 billion hectares worldwide before widescale human-caused deforestation be- gan to accelerate. Current deforestation rates have been conservatively estimated at about 30 hectares each minute or 15 million hectares annually; other estimates range up to over 26 million hectares eac year (Roche & Dourojeanni, 1984; Myers, 1984a; Gradwohl & Greenberg, 1988). Of this, about 4.5 million hectares are disrupted by commercial log- ging, 2.5 million hectares by nonsustainable gath- ering of fuelwood, and at least 2 million hectares as a result of cattle ranching. But one of the prin- cipal forms of deforestation is the wave of slash and burn agriculture foll of forests by roads. Loss of forest due " ai. scale culti- vation involves perhaps as many as 250-300 mil- lion people occupying about 22% of present-day moist tropical forests (Myers, 1986a; Gradwohl & Greenberg, 1988). There has already been a loss of 27-37% of these forests, and a further 12- 25% is likely to be lost before the end of this century (Simberloff, 1986) owever, not just loss of area is important, but also loss of quality. What to the layperson is intact forest may be botanically depauperate. Moist trop- ical forests, along with Mediterranean-type eco- 50 Annals of the Missouri Botanical Garden systems, are noted for their high level of biodiversi- ty and for complex networks of interacting animals and plants. Modern technology permits habitat conversion and timber extraction at rates undreamed of in the past. The tropical lumber industry has been likened to mining with extraction rates in in excess of sustainable yields (Whitmore, 1980). For some smaller areas, such as the Brazilian most cases well T state of Rondónia, quite precise data are available (Myers, 1986a; Prance, 1986, 1987). Rondónia has an area of 243,000 square kilometers, and before 1975, 1,200 square kilometers had been cleared of forest. At that stage, the population averaged about two people per square kilometer. By the late 1970s the population was increasing by almost 16% each year. Between the end of the 1970s and 1985 it doubled to over one million, an increase of over 1,000% since 1968. As the population increased so did deforestation, with more than 10,000 square kilometers cleared by 1980 and 17,000 square kilometers by 1985. Today there is a vast network of roads and farms through- out the forest, and removal of forest continues. The tragedy is that the rich soils that attracted farmers to Rondónia also made it home to a rich assemblage of indigenous animals and plants, many endemic to the region. As the forests disappear, so do their component species. Worldwide, an alarming number of species are down to only a few individuals. An extreme example is the St. Helena olive (Nesiota elliptica), which is a monotypic genus and is only known from a single plant. Despite moderately successful ini- tiatives to save the St. Helena flora, the St. Helena olive is still critically endangered (Cronk, 1987). Seed rarely sets and cuttings seem nearly impos- — sible to strike. The remaining tree is old and vul- nerable. Similar examples come from the Pacific lowland of Ecuador, regarded by Myers (1988b) as one of the three most critical tropical forest “hot spots" where biodiversity is under threat. Since 1960 the original rainforest has been almost totally elimi- nated and converted to cash crops. А small remnant at Río Palenque of less than one square kilometer is now the only remaining site for 43 plant species, of which a good number, including Dicliptera dod- sonii (an attractive vine known only from one plant) and the useful timber tree Persea theobromifolia are known from very few individuals (Gentry, 1977; Lucas & Synge, 1978; Myers, 1988b). The ad- jacent Centinella Ridge once supported 100 en- demic plant species which were eliminated by clear- ance for agriculture between 1980 and 1984 (Gentry, 1986; Myers, 1988b Loss of species, especially those with human appeal, sometimes attracts publicity, but loss of genetic variability within species is harder for the layperson to appreciate. Yet, in terms of usefulness to mankind and potential for evolution, loss of variability within some species may be just as sig- nificant as loss of some taxa. Loss of variability in wild populations of the Himalayan Dioscorea del- toidea, an important source of cortico-steroids, has direct commercial implications. Analyses of past collections indicated up to 6% of the steroid dios- genin, but recent analyses following intensive har- vesting of wild populations have failed to reveal plants with more than 1% of this chemical (Gupta & Sethi, 1 ) Genetic erosion, which is “the loss of genes from a gene pool due to the elimination of populations because of such factors as the adoption of modern varieties and land clearing" (Plucknett et al., 1987), is an increasing problem. As farmers have adopted modern crop varieties and agricultural practices, they have tended to shift to monoculture and ge- netic simplification of farmlands. The genetic base of crops entering world trade has narrowed because of this. Simultaneously, there has been genetic erosion of wild genetic resources and landscapes through reduction in habitat quality and quantity, and rs of many populations containing distinct genot The significance of genetic erosion is demon- strated by the battle against grassy stunt virus in rice. This proved a serious pest of rice throughout much of South and Southeast Asia between the 1960s and late 1970s. The International Rice Re- search Institute (IRRI) tested thousands of breed- ing lines and wild species samples for resistance. Only a single sample, collected from a population of Indian Oryza nivara їп 1963, showed resis- tance, and even this was restricted to three plants out of the sample of 30. From these three plants came the virus-resistant gene which was bred into cultivar IR36 and is now found in every high- yielding cultivar of rice grown in Ба Asia (Hoyt, 1988; Plucknett et al., 1987 Losses are not confined to tropical forests. Alarming depletion rates are encountered in many ecosystems around the world. One-third of the planet's land area is semiarid or arid. In the arid tropics, desertification is a major problem, chiefly through unsustainable levels of agriculture, exces- sive removal of woody plants for fuel, faulty irri- gation, and poor range management. Each year about 12 million hectares of this land deteriorates to agricultural worthlessness (Myers, 1984b). Med- iterranean-type vegetation of some arid and semi- arid regions is often highly diverse, sometimes with Volume 77, Number 1 1990 Given Conserving Biological Diversity many extremely local endemic species. However, as despised so-called nonproductive “‘scrub,”’ frequently at risk through clearance for farmland and housing. The highly distinctive Mediterranean- type shrublands of the South African Cape are T i p diverse plant communities known, yet of their component species are d т, (bur ed & Veldhuis, 1985), and perhaps as many as 350 plant extinctions may occur in the next 30—50 years unless major action is taken to rescue species (A. V. Hall, pers. comm. ) Wetlands are particularly at risk, and increasing numbers of wetland species are found in lists of threatened species. Particularly alarming is pol- lution by new ultratoxic compounds formed through the dumping and subsequent mixing of substances such as chlorine and thiocyanates (Kaul, A complication in the management of wetlands is that they are often subject to natural periodic fluc- tuations in water level. These can be extremely difficult to replicate, especially when wildfowl shoot- ers, fishermen, boat-owners, and other users de- mand different water levels to suit their particular requirements (Keddy & Reznicek, 1986; Spence, 1982). Edward Maltby (1986) pointed out that with the extirpation of wetland species we lose many opportunities to study such processes as anaerobic metabolism, salt tolerance, and natural detoxifi- it is cation. Even in apparently pristine regions there is loss of habitat along with local extirpation of plants. The Antarctic is often cited as the last untouched wilderness, but even here, where less than 1% of the continent is ice-free, plants compete with Ant- arctic bases and camps for suitable habitat along the coast (Given, 1987). This raises many questions regarding the adequacy of environmental assess- ment and protection procedures in such regions, as well as the need for more stringently managed protected areas. It is not enough to shut people out of a protected area. David Ehrenfeld (1986a) pointed out that you may be able to keep people out but you cannot fence out their introduced plants, acid rain, ozone, insecticide residues, drifting herbicide, heavy met- als, and atmospheric particulates. He gave the ex- ample of Huchison Forest at Rutgers University, where entry is limited and the site is used only for "unobtrusive" ecological research. Yet the forest is increasingly invaded by alien animals and plants. PRESENT EFFORTS AND FUTURE FocaL POINTS An impressive effort is being made at present to conserve plant diversity worldwide. In the 1960s this started to gain momentum, leading to inven- tories, Red Data Books, population monitoring for conservation purposes, and setting aside reserves primarily for conservation of plant taxa and gene pools. The most widespread and perhaps the most cost- effective means of conserving plants is by pres- ervation of natural habitats. Many thousands of protected areas play a valuable role in preserving plants. They range from large national parks to small intensively managed nature reserves and remnant strips along roads and railways. ome reserves require precise and constant management. In the fens of Cambridgeshire in England maintenance of specific water levels is critical to the survival of vegetation and rare species. In an increasing number of instances, depleted pop- ulations of plants are being reinforced by replanting programs, or there is deliberate habitat rehabili- tation. However, this is usually costly and time consuming; too often management is minimal and best described as benign neglect. Too many pro- tected areas exist only on paper. They have not been surveyed, gazetted, fenced, or properly iden- tified on the ground, or their management is such that the features for which they were set up are no longer protected. A second major approach to conservation is through botanic gardens or arboreta. There are probably about 1,500 botanic gardens worldwide, but their geographic distribution is far from satis- factory. About half are in temperate Europe. Of the remainder, only 30 are in tropical Africa, about 60 in South America, and just over 120 in tropical Asia—all areas of high biodiversity. Many gardens maintain large, botanically valuable collections and some have a long history of involvement in con- servation. Some of the earlier tropical gardens were established as acclimatization gardens. The botanic garden at Bogor in Indonesia is the repository for large collections of species from Southeast Asia, including many of economic importance. Some gar- dens concentrate exclusively on indigenous species; examples are the National Botanic Gardens in Can- berra, Australia and Kirstenbosch near Cape Town, South Africa. Both of these gardens have labora- tories to carry out conservation research on rare species, involving study of propagation and repro- ductive biology, ethnobotany, and habitat require- nts. The 1980s have seen interest generated in set- ting up botanic garden networks; for instance, the Center for Plant Conservation with a network span- ning North America. The National Collections Centre based at Wisley in England has identified several hundred collections of national significance Annals of the Missouri Botanical Garden in the United Kingdom. Following initiatives at the Las Palmas botanic gardens conference in 1985, 1987 saw the launching of the Botanic Gardens Conservation Secretariat, which now incorporates over 140 gardens in 26 countries. Systematic storage of seed and other propagat- ing material in gene banks plays a vital role in conservation of crops and their wild relatives (Plucknett et al., 1987). Thirteen agricultural re- search centers are coordinated through the Con- sultative Group on International Agricultural Re- search, including the network coordinated by the International Board for Plant Genetic Resources PGR) and centers such as the Rice Research Institute. There are many problems associated with long-term storage of wild species. Some species have very short-lived seeds, while in others seeds are recalcitrant or require specific germination pro- cedures, which are difficult to simulate. Small num- bers of available propagules can cause problems for gene banks used to dealing in hundreds of seeds or tubers in each sample. Some conservationists have concluded that seedbanks are expensive and vulnerable, and inadequate to preserve more than a fraction of even the recorded varieties of crop plants (Ehrenfeld, 1986a). Nevertheless, despite some shortcomings, gene bank storage should be attempted as an integral part of any long-term endangered species program. By the year 2000 over 90% of remaining variations of major crops should have been collected, stored, and evaluated, and many close wild relatives of crops will have been collected. By then the role of cryopreservation will have been assessed, and collecting and handling will have 1987 Conservation is moving away rapidly from total reliance on traditional techniques to utilize new technology. In western Australia satellite data were used to locate habitat of the rarely seen subter- ranean orchid Rhizanthella gardneri—the data techniques for vegetative germ plasm ~x been markedly improved (Plucknett et al., showed the orchid to be associated with a specific host and more widespread than formerly thought (Dixon & Pate, 1984). Electrophoresis and, less often, DNA analysis are being used to assess pop- ulation variation and to elucidate probable breeding systems. Isozyme analysis of Californian conifers and Australian narrow-range endemics, e.g., Fu- calyptus caesia, is indicating considerable a differences between species that otherwise appear similar and assists in estimating minimum viable population sizes. Tissue culture is being used for some orchids and other plants that are difficult to propagate by conventional means. Callus tissue can also serve to propagate plants in large numbers at predetermined times of the year for reestablish- = — ment in the wi Legislation and education are vital aspects of threatened species programs. In the international arena, the Convention on International Trade in Endangered Species of Fauna and Flora (CITES) plays a major regulatory role. Many countries have enacted legislation to protect rare and endangered species and their habitats, although often this is applicable only to the plants found on state-owned lands or to taking of plants for commercial pur- poses. Legislation and regulation are starting to give greater acknowledgment to traditional uses of and attitudes toward plants; this also involves rec- ognition of traditional forms of resource ownership such as common property regimes. A vital role of botanic gardens in particular must be to create an awareness of the need for plants and their conservation. The proportion of urban people in most countries is steadily rising; in 1980 half the world was urbanized, but by the year 2000 75% % of people will live in cities. For these people botanic gardens provide a window into nature. For protected areas, too, education and awareness are important components of management. Study of a rural community's attitudes to a nearby conser- vation area in Natal showed that positive attitudes were generally correlated with greater affluence and education, and particularly with direct expe- rience of the benefits of the conservation area (Infield, 1988). This suggests that it is important to involve local people in conservation programs. A remarkable example of this is found at Guana- caste, Costa Rica, where Dan Janzen is involving local people in management, “hands-on” educa- tion, and interpretation. Janzen’s philosophy is that the remnant forests of Guanacaste are a library, an archive, and a classroom for the indigenous people, as well as serving valuable scientific and conservation functions (Janzen, 1986 The 1980s have seen increasing stress on dem- onstrating the value of biodiversity and the use- fulness of plants to people. To some extent this has been a process of rediscovery by the developed nations of what was well known to many traditional societies in less-developed parts of the world. Eth- nobotany has acquired a new respectability by dem- onstrating the degree to which wild plants are used (e.g., Myers, 1984a; Hanks, 1984; Plotkin, 1986; 1987). Quantitative studies of the use of trees by four tribes in Amazonia show that Prance et al., these rainforests contain an exceptionally large number of useful trees; up to 76% of tree species on inventory sites can be “useful” (Prance et al., 7). About 75% of the medical needs of the Volume 77, Number 1 1990 Given 53 Conserving Biological Diversity Third World are met by traditional herbal remedies (Adesiwojo et al., 1984). Scores of locally culti- vated crop plants have potential to supplement the 20 plants that contribute 9076 of the world's food. These include more than 20 root crops, legumes, grains, and fruits found in South America, which e “lost crops of the Incas” (Vietmeyer, 1986). There are conflicts: should ethnobotany stress *'dis- covery” and commercialization of plants for world- wide use, or should it encourage a retention of traditional use; and should the emphasis be on conservation of plants and their habitats because they are “useful”” to mankin Documentation, inventory, КР monitoring are vital to all conservation strategies. It is essential to know not only what taxa are at risk, but the trends and priorities, and whether strategies are success- ful. At the global level key roles are played by organizations such as IUCN's monitoring center at Kew, the specialist groups of the IUCN Species Survival Commission, The Nature Conservancy, Conservation International, and WWF. Their role is crucial. But they cannot do all the work. There is urgent need to enhance existing national and regional data bases and to establish new ones. Inventories are showing that in many countries 10-15% of the vascular flora is likely to be at risk. Unfortunately, only sketchy data are usually avail- able for cryptogams, and there is urgent need to assess the risk for these. There is also an urgent need to develop a monitoring system at the pop- ulation level that is useful to the local manager of a protected area and can contribute to global es- timates of depletion. The Plantwatch concept—as it has been christened by IUCN— must be simple and quick to operate while asking the relevant questions. An important project is the identification and assessment of centers of plant diversity, some of which are global “ої spots” under a high degree of threat (Myers, 1986b, 1988b). In just three of e “hot spots” for tropical forests (Madagascar, the Atlantic coast of Brazil, and western Ecuador), Myers (1988b) estimated that some 6,200 plant and 124,000 animal species could become extinct. A pilot scheme developed by IUCN at Kew has identified almost 160 centers of plant diversity around the world. Greatest numbers are in tropical South America, Asia, and Africa, including oceanic islands suc , Mauritius, St. Helena, and the Chatham к. (IUCN, 1987). These are key sites to monitor the state of biodiversity and must be priorities for recovery projects espite past and present achievements, there must be some doubt as to whether they will stem the tide of biodiversity loss. As we move toward the twenty-first century, several focal points emerge. These are the need for a more biological approach with greater understanding of persistence and ex- tinction processes, development of management strategies appropriate to human-disturbed ecosys- tems, adequate resources for conservation, greater innovation and boldness, and development of a renewed conservation ethic. UNDERSTANDING PERSISTENCE AND EXTINCTION Prevention of extinction and maximizing persis- tence are major complementary aims of conser- vation. Yet we are far from a full understanding of either persistence or the precise mechanisms of extinction and their relationship to such factors as viable population size. A major change of perspec- tive has been the adoption of a systems approach in the early 1980s distinguishing deterministic ex- tinction from chance or stochastic extinction. It shows that four separate forces make key contri- butions to population extinction (Schaeffer, 1981). These are: demographic stochasticity (variation in population size leading to unacceptably low num- bers), genetic stochasticity (excessive inbreeding and loss of selective variation), environmental sto- chasticity (environmental shocks received by all members of a population), and catastrophes. This work has been extended by Gilpin & Soulé and incorporated into a conceptual model that rec- ognizes three components (Gilpin & Soulé, 1986): "population phenotype," *' and "population structure and fitness." Environmental changes set up feedback loops of biological and environmental interactions that impact the popu- lation negatively. Gilpin & Soulé referred to these event trains as *'extinction vortices" that can lead to extinction. They identified four distinct vortices triggered by various combinations of chance de- „ 2° environment, crease in population size, decrease in genetic ef- fective population size, decrease in population growth rate, and increase in the variance of pop- ulation growth rate. The vortices can operate over different time scales and can interact in different combinations. An important advance has been to distinguish deterministic extinction (when something essential is removed, such as habitat or food, or when some- thing lethal is introduced, such as predation) from stochastic extinction resulting from random changes or environmental perturbation. The Gilpin & Soulé model has several practical implications. It identifies demographic stochasticity as often being the immediate precursor to extinc- 54 Annals of the Missouri Botanical Garden tion, justifying concern for the future of small pop- ulations. It also indicates that two important factors in long-term persistence are the maintenance of relatively large areas of habitat and sufficiently large populations to ensure that random effects do not result in loss. Genetic aspects of extinction and especially the effects of loss in heterozygosity are often empha- sized. Concentration on loss in genetic variability alone may overlook other stochastic factors or the possibility of loss through catastrophe. Especially in highly dynamic ecosystems, such extirpation may result from other effects long before loss of fitness is significant Well-documented case studies of extinction in regions of high risk are urgently needed. One on- going project that should provide some insights is the Minimum Critical Size of Ecosystems Project in Amazonia sponsored by WWE(US) and Brazil’s National Institute for Amazon Research (Lovejoy et al., 1986). This is a long-term attempt to de- termine the effects of fragmentation on habitat and species. It takes advantage of the decree that when land is cleared for farming, 50% must be left as standing jungle. At the project site the pattern of clearance is designed to leave remnants of varying size from one hectare up to 10,000 hectares. Results of the effects of size and area are starting to emerge; for instance, the decline of some animal populations in remnants up to 10 ha (Lovejoy et 86). As one species is lost, populations of other species may also decline. The smallest Am- azonian reserves are not large enough to support populations of army ants. As they disappear, so do birds dependent on the ants for food. The loss of these birds is likely to lead to loss of interacting species that are dispersal agents or pollinators for plants. Loss of the ants also affects decomposition of plant matter and the recycling of nutrients. It is not sufficient to assume that once a pop- ulation has reached a predetermined “safe” size it will survive for the forseeable future. Persistence is concerned with the probability that under certain conditions and with a population of particular size, there is X probability of existence for Y number of years. But the selection of persistence times has an ethical aspect. Some people might be content with a low probability of persistence for 100 years whereas others might demand a very high proba- bility for 1,000 years. An important aspect of extinction theory is that processes for plants may not be quite the same as those for at least some animals. Examination of palaeobotanical data and comparison with data on animal fossils shows that, although there are com- parable bimodal extinction rates, major changes in the composition of vascular plant floras have not coincided with similar events in the animal kingdom (Knoll, stances plant extinctions appear to be related to 1986). Knoll suggested that in many in- competition with newly evolved taxa or to climatic change rather than other extrinsic events that have been suggested as causative factors in some animal extinctions. Above all, a more rigorous biological approach to conservation is needed. Even such elementary data as broad classes of breeding system, pollination and dispersal vectors, and germination require- ments are generally unknown for threatened plants. This seriously hinders development of effective management programs for plant species. There is a severe shortage of experienced ecologists who can make management decisions for large areas of rainforest, wetlands, and shrublands. There seem to be no lack of students who want to be ecologists (or taxonomists and reproductive biologists), but as Les Kaufman vividly expressed the problem (Kauf- man, 1986): "Ecologists somehow never quite ‘made it’ alongside those professionals whose ser- vices are viewed as essential to society. ... But what about a messed-up ecosystem? We have yet to produce a school of competent ecological en- gineers. MANAGING IN MODIFIED HABITATS If one thing can be guaranteed into the early part of next century it is expansion of man-modified landscapes. This will be a natural consequence of increases in numbers of humans and the conse- quent need for land to provide more living space, food, and fuel. Our species has just reached 5,000 million in- dividuals and by the year 2000 should stand at about 6,000 million. A peak of perhaps as many as 15,000 million early in the twenty-second cen- y has been frequently forecast, although the С slobal 2000 study озуна a peak as high as 30 billion (Barney, 1 . Technology may allow pro- duction of more rae and fuel on less land and may reduce pollution levels in the future, thus slowing Methods for habitat re- habilitation are being developed. But there will be the rate of habitat loss. an unavoidable time lag in application of such tech- nology. A significant factor will be the availability and cost of appropriate technology. Largest pop- ulation increases are in the many countries that total only an 18% share in world expenditure on research and development (Myers, 1984b). Eco- nomics may dictate hard choices regarding the Volume 77, Number 1 1990 Given 55 Conserving Biological Diversity amount of pristine habitat and wilderness that each country can afford. In regions that are highly modified through hu- man influence, five particular factors can profound- ly affect wild populations. 1) First, the landscape becomes “‘frozen in place” as attempts are made to control such natural pro- cesses as floods, fires, predation, and erosion so that their effects become minimal in a static pattern of towns, roads, farms, orchards, and remnant na- ture reserves. This can have serious consequences for many species. They can no longer move from one site to another in a continual pattern of local extinction and recolonization. Species become at- tuned to patterns and magnitudes of natural dis- turbance and stress, which are likely to be dras- tically altered in highly modified ecosystems. Orothamnus zeyheri—a monotypic genus of the Proteaceae from South Africa—is just one example. For many years botanists had condemned the destructive effects of fire on the Cape Flora, resulting in fire protection. Investigation showed that this policy resulted in the decline of many species, including Orothamnus, which was reduced to 34 plants in four sites by 1968. Research showed that the species has natural cyclical fluctuations dependent on intervals between fires, with an op- timum periodicity of 15 years for firing (Boucher, 1981) One of the consequences of fragmentation of the landscape into a static pattern of land use is that conservation must become an integral part of land planning. There is also need for adequate mitigation processes and flexible options for pres- ervation. Conservation of islandlike patches of hab- itat with narrow-range endemics in the San Ber- nardino Mountains of California, U.S.A., (Krantz, 1987), involved negotiation with five governmental agencies over ten years as well as detailed biological assessment and identification of key sites. A key device in this exercise is a regional biotic resources map, which assists selection of the best and most manageable sites for protection. ) The second factor is isolation of patches of suitable habitat. Connecting corridors that facilitate gene flow, seasonal movement, and migration are lost. 3) The third factor is reduction in size of suitable areas of habitat. This has two main consequences. Some sites become so small that they are dominated by edge effects, and no undisturbed core habitat remains. Species in need of such core habitat die out and their place is taken by opportunist plants. The second consequence is reduction in population size so that random stochastic forces place the population at ris 4) The fourth factor is increased competition with weeds. In a consultant's report to the last IUCN/WWF Plant Advisory Group meeting, al- though Cronk (1988) identified invasive weeds as an extremely serious environmental problem, he pointed out that “there is an extraordinary infor- mation vacuum where invasive plants are con- cerned.” Similar conclusions were reached 198 by a workshop on botanical management of the Galapagos Islands. There, such plants as Cinchona succirubra (quinine), Lantana, and the common guava threaten endemic plants of these unique oceanic islands (Adsersen, 1989). 5) Related to this is the fifth factor: predation. The effects, sometimes disastrous, of exotic her- bivores on island floras is well documented. Today, predation by a wide range of animals, including goats, pigs, deer, chamois, and feral sheep and cattle, is a serious problem in many parts of the world. But smaller and sometimes overlooked or- ganisms, such as fungi and viruses, can also have significant effects on plant populations. Disease transference from cultivation crops or nursery- propagated stock to wild plants must be guarded against, especially when executing recovery pro- grams on oceanic islands. Cucumber mosaic virus has been detected in cultivated plants of the notable Chatham Islands endemic Myosotidium hortensia. Transfer of the virus to wild populations could have serious consequences for the species. Similarly, transfer of Phytophthora cinnamonii into previ- ously uninfected areas can be a risk without strin- gent hyg In К highly modified by human activity there is special need for a series of key indicators of population health. These could collectively sug- gest which extinction vortices are involved, and the immediate and long-term remedial action re- quired to prevent extinction. Halting population decline to low levels will be a critical aspect of this. Plantations of exotic plants for timber and wind shelter have been neglected as potential havens for other plants. Yet, sometimes they can contain a surprisingly diverse and rich flora. In South Africa substantial populations of endangered members of the Cape fynbos flora have been found in pine plantations. In New Zealand the Iwitahi Reserve has been established in a mature stand of Pinus nigra. Twenty-seven species of native orchids are found here, including Chiloglottis gunnii, which is classed as vulnerable in New Zealand (Gibbs, 56 Annals of the Missouri Botanical Garden 1988). Such habitats must not be regarded as necessary substitutes for undisturbed sites. But their potential role in conservation, especially in highly modified, peri-urban lowland regions, has been underestimated. In highly modified landscapes interventionist management is unavoidable, and the objectives of such intervention need to be unambiguous. For many plant communities —especially in temperate regions —high апа low levels of disturbance are likely to lead eventually to low levels of species diversity, which may be the natural state. However, if biodiversity is maximized in a particular region (especially given the fragmentary nature of many habitats), communities may have to be perturbed to promote the development of “unnatural” species- richness. It may be that some to be sacrificed in the short term in order to max- imize biodiversity. “naturalness” has ADEQUATE AND APPROPRIATE RESOURCES Global expenditure on science and technology is very unevenly distributed. Approximately 82% of the effort on science and technology is in the developed nations with 20% of the population. The global communications system is controlled by those same countries (Myers, 1984b). Regions with the greatest need for scientific expertise may have least access to it. A critical evaluation of floristic knowl- edge in Latin America and the Caribbean by Toledo (1985) pointed out that this region has perhaps the world’s greatest botanical diversity. Yet it lacks the indigenous resources to inventory fully and conserve that diversity. The 27 countries involved have over 900 resident botanists, but this contrasts with the 1,500 amateur and professional botanists in Great Britain alone. Colombia has an estimated 45,000 vascular plant species but only 47 resident botanists in 1985, and in 1981 had fewer than 350,000 plant specimens in its herbaria. A counter to this is the considerable botanical investment in Latin America by agencies such as Missouri Bo- tanical Garden, Field Museum, New York Botan- ical Garden, and W WF (US). Nevertheless, Toledo concluded that the urgent need to inventory the great wealth of New World plants, the accelerated rate of habitat destruction, and limited resources — implies: ““collecting and correctly inventorying the greatest number of species possible in the least amount of time, that is, an endeavour which Fos- berg once called a ‘salvage botany'.' A second problem is that we are inundated by a flood of knowledge. Words flood the world and the entry of data processing in the 1970s seems to have done little to control and systematize the flow of information. Most scientists and managers know the problem of always seeming out of date, even in the most limited subject. Despite the most rigorous library and data base searches, there are always new papers and bulletins being published, journals that one cannot access, and incidentally discovered unpublished reports. New journals and newsletters are being proliferated at a rate that taxes the ability of most libraries to keep up to date. Language provides a barrier, and there is too little contact between scientists with different pri- mary languages. The effects of the information flood are com- pounded by a third problem—the subjection of science to marketplace economics. In the most ruthless form of free-market enterprise, science becomes a tradable commodity expected to return a profit to its financiers. This can result in a scram- ble for funds with competition between formerly cooperating departments of state, universities, and research organizations. The “free market” philos- ophy in science has ““somehow changed the role of scientific enquiry in our society to a materially driven pursuit which is indistinguishable from mar- keting, labour relations, product development or pollution control. It is a key component of an industrial society. But it is much more than that. [...] it is very important that our citizens—par- ticularly our young people—appreciate the cul- tural and intellectual significance of scientific en- quiry” (Upton, Transformative values in science have been a driving force in scientific investigation throughout civilized history. Conservation biology is not just to determine the best way to preserve biodiversity so that we can use it better, faster, and more efficiently —it also serves to open to our minds a world that is infinitely varied and has inspired poets, writers, and thinkers through the ages. Stringent curbs on conservation expenditure, such as are occurring in many countries, can place the scientist and manager in an invidious position. Which species are allocated money for research and preservation? Can predator and weed control be undertaken? Are botanic gardens a luxury? Should limited resources be used to buy areas of wild habitat? When the choice is jobs or species preservation, which goes first? Problems of resource allocation are unlikely to improve in the near future unless there is a drastic change in attitudes toward nonhuman species. Increasing numbers of people and greater demands on space will mean less space for wild plants and animals, and diminished re- sources for conservation. Volume 77, Number 1 1990 Given Conserving Biological Diversity Conservationists of the future may be faced with the problem of triage: having to determine which species will slide toward extinction because there are insufficient resources to save them, which are not yet in need of urgent attention, and which can be treated with existing resources. It is tempting to put large amounts of time and money into a small number of spectacularly endangered species with public appeal while allocating little to species with less appeal or that are becoming critically endangered. [n terms of conservation of overall biodiversity, it might be better to spend more on preventing species from reaching critical levels of endangerment. COMMUNICATION AND COOPERATION Cooperation means assessing the strengths and weaknesses of contributing people and organiza- tions, and then making the best use of each other's strengths and countering their weaknesses. It means maximizing use of resources, especially where these are scarce. At the simplest level it may also mean using the same vehicle to get researchers from different agencies onto field sites; at more complex levels it may mean closely integrating research and management programs and sharing major re- sources, such as herbaria, gardens, and laborato- ries. The need for cooperation exists at all levels from local and regional to national and interna- tional. The trend toward an increasing degree of spe- cialization and a view that “only engineers are qualified to talk about engineering, only biologists can talk about biology n result in a dearth of people who can ac such disciplines together cooperatively. There can also be unwill- ingness for people to speak to others outside their own speciality or to question the assumptions and conclusions of these other specialists. Yet we all need to question and understand viewpoints from other disciplines and cultures. National and international organizations have a communication. The setting up of the joint IUCN. WWF P one of several positive moves to achieve this, but lant Advisory Group is just it is essential that such groups be in touch with what is actually happening in the “real” world. The urgent need in conservation is hands-on bot- anists and managers rather than desk-top admin- istrators and prophets. Cooperation, especially in tropical countries, means involvement of local peo- ple, "grassroots support," rather than grandiose schemes run by expatriots. It is important that 9. indigenous people and their knowledge be involved in setting up and managing nature reserves, res- toration of habitat, and maintenance of sustainable forms of agriculture. Examples in Gradwohl & Greenberg (1988) show that through communi- cation and cooperation, wise use and preservation of biodiversity can be achieved. or most of the last two years I have been working on a global synthesis of the principles and practice of plant conservation. The project is both challenging and frustrating. One of the greatest pleasures has been to communicate with and learn from many hundreds of people, some working in the most out of the way places. It is a humbling experience, yet frustrating because many of these people with such good ideas and relevant experi- ences do not get the chance to share with others as they should. In many instances they do not have the resources to attend conferences and workshops outside their region; in other instances they are too busy performing practical conservation. INNOVATION AND BOLDNESS Two things that may determine much of the effectiveness of future conservation are innovation and boldness. If the situation in the early part of next century approaches worst-case scenarios, every bit of innovation and boldness may be needed to avoid vast numbers of extinctions, especially in the tropics. A novel extension of the botanic garden concept is the idea of setting up a “Noah's ” refuge for species, especially for the establishment of plan- tations of northern hemisphere trees threatened by acid rain. This has been suggested several times in the last five years for isolated countries, such as New Zealand, where pollution is relatively low, and land is available. There are problems: are whole communities or ecosystems to be shifted to their new island home? Who pays? Will some intro- ductions adversely affect indigenous plants through competition or introduction of new predators? Will expenditure on such a “Noah's Ark" downgrade efforts to conserve indigenous species? An overall problem is the question of “how long" will the ark have to accommodate species from defunct habi- tats? It is one thing to assume that ex situ con- servation on a massive scale is permanent, but quite another to regard it as a temporary expedient for several decades (or even a century) until suitable habitats can be rehabilitated or created. stablishment of new populations of plants in quasi-natural sites where they have not formerly occurred is similar to the **Noah's Ark” concept. 58 Annals of the Missouri Botanical Garden It may be justified where predation makes it im- possible to preserve a species in its former habitat. The transfer of animals and plants to goat- and rat-free off-shore islands is an example of this. Another situation where it may be justified is where each population is reduced to only one sexual state, and the most reasonable chance for sexual repro- duction is to establish a completely new population with a mixture of plants from remaining sites. New Zealand's Gunnera hamiltonii provides an ex- ample, being known from only four single-sex pop- ulations. In some extreme instances the genome of ex- ceedingly rare species might be transferred into related and more common species, either through hybridization or genetic engineering. This brings into question the ultimate goal of conservation: whether the preservation of species or of biodiversi- ty. If the latter, then one can argue that it matters less how valuable genes are preserved than that they be preserved somewhere and somehow. There are various ways in which in situ and ex situ approaches can be combined to make persis- tence of a small population more likely. One tech- nique might involve maintenance of some areas of suitable habitat and harvesting of seed, sampled to maximize genetic diversity. Some of this seed would be banked for long-term storage as a precautionary measure against catastrophe, while some would be used for garden propagation and replanting back into the wild site. In effect, population size is main- tained above a critical level by a combination of in situ preservation in the wild and ex situ pres- ervation in gene banks, laboratories, and botanic gardens. If models of global warming through the “green- house effect” may be needed to conserve biodiversity in the twen- are valid, every scrap of ingenuity ty-first century. Although some aspects such as sea-level rise and fluctuations in ocean and at- mospheric circulation are hotly debated, it is not too early to consider possible scenarios and their implications for conservation. Climatic shifts will have a number of serious consequences (Peters, 1988). Changes in species distribution can be ex- pected as conditions become locally unsuitable for persistence. In some long-lived species, although adult plants may remain, this will result in a re- generation gap, especially where germination or seedling growth requires seasonal chilling. Spec- tacular initial losses are likely to involve arctic- alpine species at relictual sites, such as Rhododen- dron lapponicum at Wisconsin Dells or the post- glacial disjuncts along the north shore of Lake Superior (Given & Soper, 1981). For many species the choice will be to migrate or be extirpated. But suggested rapidity of climate change means that migration rates required will exceed natural migration rates for many species. Today’s fragmented landscape offers a formidable obstacle course to migrating species. Peters (1988) gave the example of Picea engelmanii, which would require over 1,000 years to adjust its range to a new climatic regime To mitigate climatic change requires several courses of action (Peters, 1988; Myers, 1984b). Refinement of "greenhouse" models is urgently needed, and selected species in regions where early effects can be detected should be monitored without delay. Preservation of representative populations throughout the whole range of species is another strategy; often there has been a tendency to be satisfied with one reserve for a species. In all re- serves there need to be contingency plans for mod- ification of drainage and irrigation to allow for changes in moisture regimes. Changes in weed and predator distribution are likely and could profound- ly affect some species. This requires early identi- fication of such problems. It makes sense to locate new reserves at the poleward end of species ranges (because of poleward retreat of temperature-sen- sitive species). Particular reserves should maximize diversity in habitats and altitudinal gradients. At least in the short term, an important role must be assumed by botanic gardens and gene banks to ensure that plant diversity that may have difficulty surviving in the wild is preserved. This challenge needs to be taken up at all levels, from the local garden and research center to interna- tional organizations such as IUCN and GR The limitations of ex situ conservation have been pointed out by Foose (1986) with respect to ani- mals. He argued that the capacity of the “zoo ark" to cope with at least 1,500 mammals, birds, rep- tiles, and amphibians expected to be endangered by the middle of next century is very limited. As a consequence, the American Associaton of Zoo- logical Parks and Aquariums has designated through its Species Survival Plan 37 priority taxa for ex situ preservation. The eventual aim is to have up to 1,000 species designated. However, this will still fall far short of requirements for the next century. Similarly, it is debatable whether botanical gar- dens and gene banks can eventually guarantee the persistence under ex situ conditions of their share of the world's biota, at least without a great deal of reorganization. If the number of botanic gardens could be doubled and each persuaded to take prime responsibility for ten different plant taxa under threat, we would be a long way toward shepherding Volume 77, Number 1 1990 Given 59 Conserving Biological Diversity the flora of the world through the problems of next century. There have been numerous criticisms of botanic gardens and gene banks in recent years and whether their long-term role for conservation is outweighed by their deficiencies. Most of the problems cited, such as poor documentation and lack of adequate regenerative procedures for seeds, are not a reflection of fundamental problems so much as the result of poor management, unclear objectives, and inadequate finance. These are all capable of correction, provided there is a change in the fundamental attitudes of society to biological conservation. REDISCOVERING AN ETHIC It is easy to make facile appeals to new tech- nology to solve environmental problems, and there is no doubting the contribution of such advances as satellite imagery to conservation today. But no amount of technology or innovation will sustain plant diversity into the future unless there are fundamental changes in attitudes to biodiversity. This means moving away from exploitation and toward sustainable systems, changes in economic systems dominated by benefit-cost analysis (BCA), acknowledging the place of nonutilitarian values such as intrinsic and transformative values (Norton, 988), greater social justice, and equity in resource distribution. Progressively more intensive appli- cations of science and technology to Western so- ciety, requiring increased levels of energy con- sumption as well as larger and more centralized structures and institutions, have increased stress in environmental and social systems in ways not reflected in conventional analyses (Dahlberg, 1987). Although lip service is paid to conservation, “In the pursuit of economic gain, most people do not want to be bothered by questions of biodiversity" (Cobb, 1988) Is biodiversity misunderstood even by members of the scientific community who ought to be its champions? David Ehrenfeld (1986b) suggested that ecologists have been “‘co-opted by economists" because it is difficult for the former not to respond to arguments based on fundamental scientific mis- conceptions and, more important, because of the compelling power in Western society of arguments based upon numbers and resources. He contended that “most of us, apart from a few hard-core phi- losophers, suffer from a deep-seated fear of econ- and challenged scientists themselves to "recapture the love of diversity for its own sake.” We live in a global commons where there needs to be mutual interdependence and not indepen- Ы ээ puso dence. We need a new understanding of natural sources and resources based on a hierarchy where "genetic and biological diversity are more funda- mental than renewable resources, which in turn are more basic than nonrenewable resources" (Dahlberg, 1987). It is too tempting for individuals and institutions, especially those of the developed Western countries, to be “free riders” in the global system, ignoring limits to growth and assuming that just around the corner yet another bit of technology will solve the present crises. The irony is that some of the indigenous people who may be displaced from their traditional lands by the “free riders" are the very people who have learned to live in harmony with their environment, utilizing yet con- serving the plants around them. No culture can afford to be so arrogant that it ignores the cultural and technological systems of others. Do doomsday messages cut any ice? A few days ago a letter appeared in the Christchurch Press following the suggestion by geologists that there was a high risk of a severe earthquake: Sir [. . .] This is the second, in as many days, of this kind of depressing report. It L. na npn to be bombarded daily with the h lities of present unemployment, nation crime, etc., but to have to hear of the futu not give us articles about the positive things “likely” to Жа This has been described as the Cassandra prob- lem, an allusion to the Greek prophetess Cassandra whose prophecies were true but not believed by their hearers. Les Kaufman (1986) pointed out that: “To some people, these Cassandras, as they call themselves, are professional doomsayers, in- tellectual terrorists who should not be encouraged, supported, or believed. What seems to have es- caped such doubters, however, is that the whole point of being a doomsayer is to agitate the world into proving you wrong or into doing something about it if you are right.” There is no lack of publications arguing the case for a change of attitude to natural resources and a new stewardship of biological diversity. A re- search agenda suggested by Myers (1986b) out- lines nine topics as priorities for research, including extinction links, vulnerability of systems and taxa, rates of recovery from depletion, and the econom- ics Of threatened species. A study of ecological aspects of development in the humid tropics (Na- tional Research Council-US 1982) sets out a similar agenda for use of tropical resources. That govern- Annals of the Missouri Botanical Garden ments and institutions seem to have difficulty im- plementing many such agendas seems to suggest that we do have a Cassandra problem, and that the difficulty is not so much one of ignorance but lack of will and commitment. It is tempting to yield to “doom and gloom scenarios and to give up. On the other hand, the gathering of nearly 800 participants in November 1986 to discuss conservation and management of rare and endangered plants in California, U.S.A., shows that interest in plant conservation can and should be generated (Elias, 1987). Myers (1988a) has pointed out that, despite the gloomy prognosis for many of the world’s species, given the will, the means are available to stem the tide of extinction. He calculated the cost of an Action Plan to pro- mote intensified stable agriculture, halt spread of deserts and depletion of rainforests, supply family planning services throughout the Third World, and provide clean water to 1.5 billion people as $67 billion (U.S.) a year. This is less than twice as much as current development aid. Norman Myers asks t “Can we afford to do it?” but “Нох can we afford not to do it?’ It seems reasonable to express optimism that the worst scenarios are unlikely to occur and we can hope for sanity and “соттоп sense” to pre- vail. Is it too much to hope that by the year 2100, McNeely’s (1986) scenario of a world no longer needing national parks will be a reality? As he pointed out, national parks and other protected areas are an admission that people have not learned to live with nature or are forced to live with nature. A change to a conservation-oriented world means a commitment to social justice and equity on a global scale. It means (Raven, 1986): ness of and compassion for all life, and an appre- ciation of the fact that we are all part of a living world. [. . .] we should not allow ourselves to re- spond only when the crises that we have caused “ап aware- o are so extensive that they threaten our lives. do so is to be thoroughly immoral in the fullest meaning of the word.” To change means opening people's eyes to the positive aspects of nature; yet it also means being honest and realistic about the risks to biodiversity. The greatest danger may not be that biodiversity will be lost in a sudden cata- clysm precipitated by Homo sapiens but that we will allow planet Earth to die by attrition. In James A. Michener’s novel Caravans, the question is asked of a ruined desert city: “Why did it die?” (Michener, 1964). We are told: “‘[. . .] this used to be the world’s foremost example of irrigation. [. . .] But people got lazy. They didn’t keep working on it. They felt that what had worked for a hundred years was good enough for the next hundred years. They stopped building the ditches [. . .] built no new dams. They guessed right. For a hundred years, no trouble. But the death warrant had been signed." We need to ensure that the death warrant is not signed for Earth— even by default. Conservation has a message for the world about plant diversity. But for the world to take notice may depend on heeding the advice of General Booth, founder of the Salvation Army, who pointed out that if you present good news to a starving man then you must wrap it in a sandwich. The tragedy we face through the loss of plant diversity may mean that very soon we may have tens of millions of starving people but very few sandwiches. LITERATURE CITED ADESIWOJO, S. S., S. TJOKRONEG ORO & R. TJOKRONEGORO. 1984 JCN Commission on Ecology Devel- opment Paper 7: pe l rare plants of the Galápagos Islands and their conservation. Biol. Conservation 47: 49-77. BARNEY, С. O. (editor). 1980. The Global 2000 Report to the President of the U.S. Pergamon, New York. Autecological and population stud- ies of e А in the Саре of South Africa. Pp. 34 in H. Synge (editor), The Bi- ological Aene | hes Plant Conservation. Wiley, Chiches Broapy, P., D. С IVEN, L. GREENFIELD & К. ‘THOMPSON. 1987. The biota and environment of fumeroles on Mount Melbourne, Northern Victoria Land. Polar Biol. 17: 97-113. as M 1988. A Christian view of biodiversity. Pp. -485 in E. O. Wilson (editor), Biodiversity. Na- em Academy Press, Washington, D.C. CRONK, Q. C. B. The plight of the St. Helena olive, Nesiota elliptica. Bot. Gard. Conservation News 1: 30-32. — 1 Combatting the invasive Jod threat. Report to IUC N/WWF Plant Advisory DAHLBERG, K. A. 1987. Redefining lun prior- ities: genetic бк and agroecodevelopment. Con servation Biol. 1-321. Dixon, К. W. « " E ATE. 1984. Biology and dis- tribution status B Rhizanthella gardneri Rogers. Kings Park Res. Notes 9: 1-54. EHRENFELD, D. 1986a. Life in the next millenium: who will be left in Earth’s community? Pp. 167-186 in L. Kaufman & К. Mallory (editors), d Last Ex- D ed Cambridge, Massachuse —— . Thirty million cheers for diversity. New S Sci. de June 1986: 38-43. EHRLICH, Р. & A. EHRLICH. 1981. Extinction. Random House, New York. Erias, T. S. (editor). 1987. Conservation and Man ment of Rare and Endangered Plants. California Na. tive Plant Society, Sacramento, California. ESTACIÓN CIENTIFICA CHARLES DAR WI IN. 1987. Com- Volume 77, Number 1 1990 Given Conserving Biological Diversity bined recommendations from the workshop “Воїап- ical Research and Management in Galápagos, April 1987.” 1 8, FOOSE, Т. 1986. Riders of the last ark: the role of pt р. 145- 161 in L. Kaufman & К. Mallory (editors), ш Last the Аагене as du. Science 215: 1045-105 3. . S. Elias (editors), Ex- tinction is Forever. New Tork Bot. Gard., New York. 1986 ant communities. Pp. 153-181 in M. оше (editor), Conservation Biology. Sinauer, Sunderland, assac GIBBS, М. M. 1988. Iwitahi: a native orchid reserve in exotic pine forest, New Zealand. The Orchidian 93 ): 49-51. GILPIN, M. E. & M. E. Sour£. аи ае тя populations: E of species extinctio ‚ 19- 34 in М. оше (editor), Comer тайн Biology. mera | Massachus GIVEN, D. R. 1987, Plants in he y os IUCN Bulletin 18(10-12): 1 & . ОРЕК. 1981. ment of he vascular flora | Lake Su , Publ. Bot. 10: 1-70. L, Ex quet. 1988. Saving the Tropical Forests. Earthscan, London. Gupta, R. & K. L. SETHI. 1983. Conservation of me- e es resources in the Himalayan region. Pp. 01-109 i n & K. L. Mehra (editors), он, of Tropical Plant Resources. Botanical Survey of India, Howrah, India. HarL, A. V. & H. A. VELHUIs. 1985. South African = Data Book: Plants— Fynbos and er Biomes. South African Nat. Sci. Prog. Rep. 117. CSIR. HANKS, |. (editor). 1984. Traditional het. Con- servation and Rural Development. IUCN Commission Ecology Development Paper 7. Eus M. 1988. Attitudes of a rural community to- s a local conservation area in Natal, South Af- rica. Biol. Conservation 45: 21-46. IUCN. 1987 Centres of Plant Diversity. A Guide and ore? for Their е IUCN мин s Unit, е жаш Janos D. Mass extinctions: new answers, ew question . 43-62 i Kaufman & K. Mallory waki boa Last Extincion. MIT, Cam- bridge, Massachu JANZEN, D. H. 86. un National Park: Trop- ical, не i EM Cultural Restoration. EUED, San osé, Costa T E; 1986. О ы ark is sinking. Pp. 1-42 The Arctic- -alpine ele- uperior. Nat. habitats of Kashmir. Pp. 11 K. L. Mehra (editors), Conservation of Tropical Plant Resources. Botanical Survey of India, Howrah, India. Keppy, P. A. & A. A. REZNICEK. 1986. Great Lakes vegetation dynamics: the role of fluctuating water levels and buried seeds. J. Great Lakes Res. 12(1): 25-36 KNOLL, A. Н. 1986. Plant extinctions. Threatened Plants Newsletter 17: 2-3. Krantz, T. 1987. Island biogeography and preserve d esign of an insular rare plant community. P in Lepic, F. Т. 1986. eL heterosis and fitness in out-breeding plants. Pp. 77-104 in M. E. (editor), rendu Biclopy. Sinauer, Sunderland, Massachuse Lovejoy, T. E., RO BIERREGAARD, A. B. RYLANDS, T. E 1986. Edge and other effects of isolation on Amazon forest fragments. p. 257-285 in M. E. Soulé (editor), Conservation Biology. Sinauer, Sunderland, Massachusetts. Lucas, С. € Н. Syn 1978. The IUCN Plant Red Data Book. IUCN, E^ Switzerland. & The assessment and con servation of mA. plants around the world. Pp. -18 in H. Synge (editor), The Biological Aspects of Rare Plant Conservation. Wiley, Chichester, U.K. MALTBY, ak 1986. Waterlogged Wealth. Earthscan, London & "Wer ae мы Í 1986. Protected areas and human ecology: is national parks can contribute to sus- taining societies of the 22nd century. Paper to Con- 21 = 100—Session on Conservation Manage- Melee J. A. 1964. Caravans. Secker & Warburg. MILLER, C. L. 1987. The Californian forest germ plasm project: a case for genetic conservation of dev sje tree species. Conservation Biol. 1(3): 191-193. Moran, G. F. & S. D. H 1987. Oe 7А population structure of eucalypts and the conserva- tion of their genetic resources. Pp. 151- W. Arnold, A. A. Burbidge & A. J. M. (editors), Nature Conservation: the Role of Remnants id Native Vegetation. Surrey Beatty, Chipping Nor- n, New South Wales, Australia. MENS N. 9. The Sinking Ark. Pergamon, Oxford. ns т Primary Source. Norton, New York 6 Гоп (editor). Pos. The Gaia Atlas of Planet Man- agement. Anchor Press/Doubleday, Garden City, New York. 1986a. Tropical deforestation and a mega- extinction spasm. Pp. 394-409 гл M. E. Soulé (ed- itor), Conservation Biology. Sinauer, Sunderland, Massachusetts. 1986 Nature conservation at global level: the scientific i issues. Inaugural lecture for new chair and visiting WE ra Univ. Utrecht, Nether- lands, ү e gos D pu problem, splendid opportunity: EIL 2: 205-215 88b. Threatened biotas: “hot “pak in те peat The En vironmentalist 8(3): 1 NATIONAL RESEARCH COUNCIL (U.S.). 1982. Ecological NORTON, B. G hy Preserve Natural Variety. Princeton Univ. Press, Princeton, New Jerse PETERS, R. L. 1988. The effect of global climatic Май on natural communities. Pp. 450-464 in E. O. Wil- 62 Annals of th Missouri Bos Garden son (editor), ешш National Academy Press, Washington, D. PLOTKIN, М. 1986. байк Е ethnobotany in tropical South America. т US. PLUCKNETT, D. L., N. J. H. SMITH, б Т. WILLIAMS & N. NISHETTY. 1987. Gene Banks and the World’s Food. Princeton Univ. Press, Princeton, New Jersey. PRANCE, С. A 986. The e dui lost? Pp. 62-106 in L. Kaufman & К. Mallory (editors), The Las Extinction. MIT, a Massachusetts. 1987. The conservation and utilization of the densa rain forest. Revista Colomb. Ci. Exact. Fis Nat. 16(61): 117-128. W.B м & R. L. CARNEIRO. 1987. Quantitative ie any and the case for conservation in Amazonia. Conservation Biol. 1: 296- Raven, P. H. 1986. We're killing our world. Keynote address, American ru rd for the Advancement of Science, Chicag A Feb. 1986. R. F T S. E. EicHoRN. 1986. Biology New Yor 1984. A ES to in of Plants, 4th edition. Worth, КоснЕ, L. & M. J. DoUROJEANNI. situ c 'onserv ation of о resources of tropical woody FAO Forest Res. Div. Re /2. L. 1 Minimum population sizes for species conservation. BioScience 31: 131-134. SIMBERLOFF, D. 1986. Are we on the edge of a mass extinction in tropical rain forests. Pp. 165-182 in D. K. Elliott (editor), Dynamics of Extinction. Wiley, New York. SPENCE, D. H. N. The zonation of plants in freshwater lakes. ‘Advances Ecol. Res. 12: 37-125 TorEpo, V. 1985. A Critical Evaluation of the Floristic Knowledge in Latin America and the Caribbean. Re- port to The Nature A Washington, D.C. International Progr Upton, S. 1988 Unpublidicd address to Science Fac- ulty, Otago Univ., New Zealand, 14 July. VIETMEYER, М. Lesser known plants of potential use in agriculture and forestry. Science 232: 1379 1384. WHITMORE, Т. С. ere The conservation of tropical rain forest. Pp. 303-318 in M. E. Soulé & B. A. Wilcox (editors), узы е Biology. Sinauer, Sun- derland, Massachusetts INTRODUCTION A SYMPOSIUM ON THE BIOLOGICAL DIVERSITY AND EVOLUTION OF THE TARWEEDS The tarweed subtribe Madiinae of the sunflower tribe Heliantheae is a highly diversified, yet co- hesive group of 17 genera and 127 species, with centers of diversity in California and Hawaii. In California, the tarweeds received the attention of such pioneers in biosystematics as H. Hall, E. Bab- cock, and the team of J. Clausen, D. Keck, and W. Hiesey. The results of these early workers significantly influenced the way succeeding gen- erations of botanists viewed the evolutionary di- vergence of plant populations. In Hawaii, the tar- weeds became known for their spectacular adaptive radiation into a tremendous variety of habitats and growth forms. Thus, the tarweeds have provided an excellent means for studying the processes of biological diversification in continental and insular ecosystems. А symposium exploring the biological diversity and evolution of the tarweeds was spon- sored by the American Society of Plant Taxono- mists and the Botanical Society of America and was held in Davis, California, at the American Institute of Biological Sciences meeting in August 1988. The symposium was dedicated to Sherwin Carlquist in recognition of his outstanding contri- butions to tarweed research. The following papers stem from the symposium and summarize current research aimed at understanding diverse aspects of this fascinating group of plants. ANN. MISSOURI Вот. GARD. 77: 63. 1990. ADAPTIVE RADIATION OF THE HAWAIIAN SILVERSWORD ALLIANCE (COMPOSITAE-MADIINAE): ECOLOGICAL, MORPHOLOGICAL, AND PHYSIOLOGICAL DIVERSITY" Robert Н. Robichaux,’ Gerald D. Carr? Matt Liebman, and Robert W. Pearcy’ ABSTRACT The ecological, morphological, and Li bape al а of species in the Hawaiian silversword alliance is excep- tional. The 28 species, which belon o the endemic ra Argyroxiphium, Dubautia, and Wilkesia, have a wide чаб of geographical distributions and a ranges within the archipelago. They grow in habitats as varied as dry se crub and woodland, wet scrub an phological and E ecd diver r utia Lc exhibiting feo variation in leaf turgor maintenance capac cities. y is also evident amon ite of р в sympatric species. Át а в ‚ А. за л. Ae D. menziesii exhibit different suites of morphological and physiological traits nabling them si die with the severe environmental conditions. The patterns of diversity and the est, cinder and lava, and dn Ecological diversity is also evident among e, D. ciliolata and D. scabra are restricted y on the island of e genomic relationships in the 28 species suggest that a variety of factors may have played important roles in their adaptive radiation. The Hawaiian silversword alliance is a premier example of adaptive radiation in plants (Carlquist, 1980; Carr et al., 1 he alliance includes 28 species in three endemic genera: Argyroxiphium, Dubautia, and Wilkesia (Carr, 1985). The species grow in a wide range of habitats and have a wide variety of growth forms. They are also closely related, as evidenced by the high frequency of spontaneous interspecific and intergeneric hybrids in nature, coupled with the ease of production of artificial hybrids in the laboratory (Carr & Kyhos, 1981, 1986). The detailed analysis of the hybrids and parental taxa using cytogenetic, electropho- retic, and molecular approaches has provided com- pelling evidence that the silversword alliance is a genetically cohesive group whose origin and di- versification probably trace to a single colonizing 1988; Carr et al., 1989). Our objective in this review is to provide insight ancestor (Baldwin et al., into the ecological, morphological, and physiolog- ical diversity of species in the silversword alliance. With respect to ecological diversity, we compare the geographical distributions, habitats, and ele- vational ranges of the 28 species, then analyze the local distributions of several sympatric Dubautia species. With respect to morphological and phys- iological diversity, we compare the growth forms, leaf sizes, and leaf shapes of the 28 species, then examine the turgor maintenance capacities of the Dubautia species and the water and temperature balances of two sympatric Argyroxiphium and Du- bautia species. Our primary theme is that the alliance. The research was supported by NS to ihe: senior author. We assistance in the field. We also thank the staff of critical logistical support, and J. Canfield and N. Friedm * Department of Ecology & Evolutionary Biology, University of Arizona, Tucson, ae чачты Hawaii 06822, U.S.A. awaii, ANN. Missouni Bor. GARD. 77: 64-72. 1990. Майы, ' This paper is dedicated to Dr. Sherwin Ert in honor of his pioneering research on the Hawaiian silversword F Grant DEB-8206411 and a gift from the Atlantic Richfield Foundation thank Lani Stemmermann for invaluable companionship, botanical insight, and technical Haleakala National Park, particularly R. Naga an for Jia photographic assistance a and L. Loope, for Arizona 85721, U.S Or rono, ia Ln U.S.A. ,U.S Volume 77, Number 1 1990 Robichaux et al. 65 Adaptive Radiation of Hawaiian Silversword Alliance Т | 22 Kauai С) 4 Оаһи CN Molokai z = Maui 2 Lana) = 20F Е 50km Hawaii 160 158 156 Longitude (°W) FIGURE 1. ecological, morphological, and physiological diver- sity of the species is exceptional. Understanding the pattern and significance of this diversity, to- gether with the factors giving rise to it, provides a key to understanding adaptive radiation in plants. ECOLOGICAL DIVERSITY The Hawaiian archipelago includes six major islands (Fig. 1). Maui, Lanai, and Molokai, which are currently separated by shallow channels, were united into one large island during periods of lower sea level in the Pleistocene (Macdonald & Abbott, 1970). Thus, they function as a single biogeo- graphic unit known as the Maui complex (Carson & Kaneshiro, 1976; Simon, 1987). The ages of the islands increase progressively from southeast to northwest. Based on potassium-argon dating, the ages of older sections of the islands, in millions of years, are: Hawaii—0.7, Maui— 1.3, Lanai— 1.5, Molokai— 1.8, Oahu— 3.3, and Kauai— 5.6 (Macdonald & Abbott, 1970). Species in the silversword alliance have signifi- cantly different geographical distributions within the archipelago (Table 1). All Argyroxiphium species, most 13-paired Dubautia species, and one l4-paired Dubautia species are confined to Ha- waii, the Maui complex, or both. Most 14-paired Dubautia species and both Wilkesia species, in contrast, are confined to Kauai. Two 13-paired Dubautia species are restricted to Oahu, and two 14-paired species grow on Kauai, Oahu, and at least one of the younger islands. Of the latter species, only D. plantaginea grows on all major islands. Major islands of the Hawaiian archipelago. The archipelago supports a wide variety of ter- restrial habitats. The habitats containing species in the silversword alliance are grouped into four broad categories in Table 1. The dry scrub and woodland habitat has low annual rainfall with a prolonged dry season. The vegetation ranges from low, open scrub dominated by deciduous and evergreen shrubs to taller woodland with scattered deciduous and evergreen trees. The wet scrub and forest habitat has medium to high annual rainfall without a pro- longed dry season. The vegetation ranges from tall, closed forest with a high diversity of evergreen trees to lower, more open forest and scrub domi- nated by the evergreen Metrosideros polymorpha (Myrtaceae) and rich in epiphytic bryophytes and pteridophytes. The bog habitat has very high an- nual rainfall, waterlogged, acidic soils, and stunted vegetation. The cinder and lava habitat has low or high annual rainfall, exposed volcanic substrates with very limited soil development, and sparse vegetation. The range in annual rainfall among the four habitats, from less than 400 mm to more than 12, mm, is exceptional. Species in the silversword alliance have signifi- cantly different distributions among the four hab- itats (Table 1). One Argyroxiphium species, six 13-paired Dubautia species, and both Wilkesia species grow in the dry scrub and woodland habitat. Most Argyroxiphium species, four 13-paired Du- bautia species, and all 1 4-paired Dubautia species, in contrast, grow in the wet scrub and forest hab- itat, the bog habitat, or both. In addition to growing in other habitats, five species grow in the cinder and lava habitat, where they are among the first Annals of the Missouri Botanical Garden Ecological and morphological characteristics of Argyroxiphium, Dubautia, and Wilkesia species. Miner and diploid chromosome pairs der the analysis of To 2,0 are: H = Hawaii, K = Kauai, M = e from Ca 985). Ecological and morphological data derive from nds 000 n penne (С агг, 1985) and from extensive field research. Isla Maui complex, and O = Oahu. Habitats are: b = og, с = cinder and lava, d = dry scrub and woodland, d w = wet scrub and forest. Growth forms are: | = liana, г = rosette shrub, s = shrub, and t = tree. Leaf lengths are ranges for mature leaves from dried specimens. Chromo- some Elevational Growth Leaf length Species pairs Island Habitat range (m) form (mm) Argyroxiphium caliginis 14 M b 1,350-1,650 r 35-150 A. grayanum 14 M b 1,200-2,050 г 50-31 А. kauense 14 H b 1,625-1,900 r 200-400 A. sandwicense 14 Н, М d, c 2,125-3,750 r 130-390 4. virescens == М w 1,600-2,300 f 170-300 Dubautia arborea 13 H d 2,125-3,100 s, t 30-90 D. ciliolata 13 H d, € 900-3,200 s 5-30 D. dolosa 13 M w 1,525-2,275 s 40-120 D. herbstobatae 13 О d 580-925 5 20-55 D. linearis 13 H, M d, w, € 450-2,500 s 10-7 D. menziesii 13 M d,c 1,800-3,075 5 20-50 D. platyphylla 13 M d 1,725-2,7 5 40-90 D. reticulata 13 M w 1,575-2,300 s, t 30-70 D. sherffiana 13 О w 600-1,150 s 25-100 D. imbricata 14 K w, b 700-1,550 5 60-150 D. knudsenit 14 K w 550- 1,37: s, t 50-220 D. laevigata 14 K w 575-1,250 s 70-240 D. latifolia 14 К w 975-1,200 l 80-17 D. laxa 14 М, О, К w, b 350-1,700 s 40-200 D. microcephala 14 K w 825-1,275 5 100-240 D. paleata 14 K b 1,100-1,550 5 35-200 D. pauciflorula 14 K w 700-725 5 80-210 D. plantaginea 14 Н, М, О, К w 300-2 200 5 80-260 D. raillardioides 14 K w 600-1,37 5 -25 D. scabra 14 H, M w, c 15-2, 500 s 10-90 D. waialealae 14 K b 1,450-1,600 s 10-30 Wilkesia gymnoxiphium 14 K d 425-1,100 r 150-500 W. hobdyi 14 K d 275-400 r 100-200 vascular plants to colonize the exposed volcanic substrates. The most widespread species ecologi- cally is the 13-paired D. linearis, which grows in the dry scrub and woodland habitat, the wet scrub and forest habitat, and the cinder and lava habitat. Elevations in the archipelago extend from sea level to 4,206 m (Mueller- Dombois, 1981). Species in the silversword alliance collectively span most of this elevational gradient, growing from 75 m to 3,750 m. evational ranges, The species have a wide variety of el- however, with no two species having the same range (Table 1). Dubautia pau- ciflorula and D. waialealae, for example, have narrow ranges, whereas D. ciliolata and D. scabra In addition, D. herbstobatae hobd yi have ranges restricted to low ele- have wide ranges. and JF. vations, whereas 4. sandwicense and D. arborea have ranges restricted to high elevations. At low elevations minimal daily temperatures rarely drop below 15°С. At high elevations, in contrast, minimal daily temperatures may drop below 0°C for ex- tended periods, especially during winter months. At many localities in the archipelago, species in the silversword alliance grow sympatrically. In sev- eral instances, the sympatric species exhibit local distributions that differ markedly. Dubautia cil- iolata and D. scabra, for example, grow sym- patrically at a site on the upper slopes of Mauna Loa, Hawaii. The site is covered by exposed lava from past eruptions of Mauna Loa. Most of the site is covered by pahoehoe lava from a 1935 eruption. This younger flow is discontinuous in places, however, such that lava from a prehistoric flow is exposed. The older flow consists of a mixture Volume 77, Number 1 1990 Robichaux et al. 67 Adaptive Radiation of Hawaiian Silversword Alliance Adults Juveniles Number of individuals S 150 Distance along transect (m) ооа bs adults and juveniles of 200 Lava L 0 FIGURE 2. Dubautia ciliolata a ra at a site » bes on Mauna Loa, Hawaii. The was located at 1,980 m near Puu Huluhulu a de Saddle Road (Robichaux, 1984). The 200-m-long, 10-m-wide transect crossed two lava flows. The first 100 m crossed an older flow Ж: pahoehoe and aa lava; the second 100 т crossed a other was very abrupt. Adults and Pan were defined as individuals with crown mber of individuals along the transect, the unshaded per shaded bars denote D. cil- iolata and D. scabra, respectively. For the brin. of lava along the transect, the u көк and shaded 5 r flows, coc The species distributions are йау different for adults and for juveniles with chi-square tests at P < 0.001 of pahoehoe and aa lava. At points of contact between the two flows, the transition is very abrupt, occurring over a distance of 1-2 mm. The pockets, or kipukas, of the older flow vary in size from less than 1 m? to more than 10,000 m?. Thus, the landscape at the site is a mosaic of the two flows. ough both Dubautia species are common at the site, their local distributions differ significantly (Fig. 2, Table 2). Dubautia ciliolata is almost completely restricted to the older flow, whereas D. scabra is completely restricted to the younger flow. TABLE 2. Distributions of adults of Dubautia ciliol- ata, D. scabra, and their natural hybrid at a site of sympatry on Mauna Loa, Hawaii. The site is the same as in Figure 2. The oval-shaped sample area was approxi- mately 350 m long and 150 m wide. It included a large pocket, or kipuka, of the older lava flow that was ap- proximately 250 m long and 50 m wide. It also included e section of the younger lava flow that co pletely surrounded the kipuka. Number of adults Taxon Older flow Younger flow D. ciliolata 1,692 28 D. scabra 0 487 Hybrid 0 29 With regard to the small proportion of adult in- dividuals of D. ciliolata that grow on the younger flow, over 80% are small plants that occur within 4 m of the boundary between the two flows. In this zone, the younger flow tends to decrease in thickness. Hybrids between the two species are also common at the site. Like D. scabra, they are completely restricted to the younger flow (Table The differential restriction of D. ciliolata and D. scabra to the two flows occurs for juvenile and adult plants (Fig. 2), suggesting that factors leading to the differential restriction may operate primarily during the stages of seed dispersal and seedling establishment. Both species produce very large quantities of small, light, wind-dispersed seeds each year, which appear to be readily dispersed across the two flows. Hence, seedling establishment may be the more critical stage. The local distributions of D. ciliolata and D. scabra also differ at other sites of sympatry on Hawaii. At 1,125 m near Keanakakoi Crater on the slopes of Kilauea, for example, D. ciliolata is restricted to a 1959 cinder substrate and D. scabra is restricted to an adjacent 1974 pahoehoe lava flow (L. Stemmermann, pers. comm.). Other sympatric species in the silversword alli- ance also exhibit different local distributions. Du- bautia paleata and D. raillardioides, for example, grow in the Alakai Swamp region of Kauai. This large, dissected, upland plateau receives 6,000- 10,000 mm of rainfall per year. The vegetation is composed primarily of wet forest, with a mosaic of bogs scattered over the more level areas. Though both species are common at sites of sympatry, D. paleata is largely confined to the bogs, whereas D. raillardioides is always restricted to the wet forest (Canfield, 1986) Annals of the Missouri Botanical Garden MORPHOLOGICAL AND PHYSIOLOGICAL DIVERSITY Species in the silversword alliance have a wide variety of growth forms (Table 1). All Argyroxiph- ium and Wilkesia species are rosette shrubs, with some species being primarily monocarpic (e.g., sandwicense and W. gymnoxiphium) and others being commonly polycarpic (e.g., 4. grayanum and W. hobdyi). The rosettes are sessile in some species (e.g., 4. caliginis) and elevated on = stems up to 5 m tall in other species (e.g., W. g iphium). Most Dubautia ке ан аге shrubs, though they vary from small, spreading forms (e.g., D. herbstobatae and D. scabra) to large, woody, erect forms (e.g., D. dolosa and D. plantaginea). Three Dubautia species commonly grow as small trees 5-8 m tall, with the woody trunks of large individuals of D. arborea and D. reticulata reaching 0.4-0.5 m in diameter. Du- bautia latifolia is a liana that climbs into the canopies of large trees. Stems of D. latifolia often exceed 8 m in length and may reach 70 mm in ywvmnoYxi diameter near the base. Species in the silversword alliance also have a wide range of leaf sizes and shapes. In the extremes, leaf lengths differ by two orders of magnitude, ranging from 5 mm to 500 mm (Table 1). All Argyroxiphium and Wilkesia species have long, narrowly ligulate to linear leaves, with the leaves of 4. caliginis being the shortest on average. With the exception of D. scabra and D. waialealae, the 14-paired Dubautia species tend to have longer leaves than the 13-paired species. Leaf shapes among the Dubautia species include elliptic, lan- ceolate, linear, ovate, oblanceolate, oblong, and obovate (Carr, 198: In addition to morphological diversity, species in the silversword alliance exhibit significant phys- iological diversity. Among the Dubautia species, for example, differences in chromosome number and habitat water availability are strongly corre- lated with differences in leaf turgor maintenance capacity (Robichaux, 1984, 1985; Robichaux & Canfield, 1985). The 13-paired species from the dry scrub and woodland habitat, such as D. cil- iolata, D. menziesii, and D. platyphylla, have much greater capacities for maintaining high tur- gor pressures as tissue water content decreases than the 1 4-paired species from the wet scrub and forest habitat, such as D. knudsenii, D. planta- ginea, and D. raillardioides. Their greater turgor maintenance capacities may play a key role in enabling the 13-paired species from the dry scrub and woodland habitat to tolerate conditions of low soil water availability. Morphological and physiological diversity is also evident among species that grow in the same hab- itat. Argyroxiphium sandwicense and D. men- ziesii, for example, grow sympatrically in the al- pine cinder and lava habitat of Haleakala, Maui, where environmental conditions are severe. Under the high solar irradiances characteristic of this hab- itat in summer, substrate temperatures are often very high during the day. Coupled with the low annual rainfall of 600—700 mm, these conditions appear to limit plant growth significantly, resulting in a landscape with very low vegetative cover. The growth forms and leaves of 4. sandwicense and D. menziesii differ significantly (Figs. 3-6). Argyroxiphium sandwicense is a monocarpic, ro- sette shrub with long, narrow, densely pubescent leaves, whereas D. menziesii is a polycarpic, woody, diffusely branched shrub with small, glabrous leaves. The difference in leaf pubescence results in a large difference in leaf absorptance. Leaf absorptances 0.4-0.7 um) measured with an integrating sphere are 0.39 + 0.01 (mean + 1 S.E.; N A. sandwicense and 0.80 + 0.01 (N = 4) in D. menziesii. (The leaf absorptances are significantly different with a f-test at P < 0.001 The leaf water balances of 4. sandwicense and D. menziesii during the middle of the summer growing season also differ significantly and are markedly influenced by the prevailing atmospheric conditions at Haleakala. On most July days, when the moisture-laden tradewinds blow from the north- east, a massive cloud bank forms around the moun- tain. The altitude of the cloud bank varies consid- erably from day to day, however, such that the atmospheric humidities to which A. sandwicense and D. menziesii are exposed also vary consid- erably. On 1 July 1985, for example, the upper ceiling of the cloud bank was 1,750-1,850 m throughout the day. As a result, atmospheric hu- midities at the study site (2,820 m) were very low, with the leaf-to-air vapor pressure gradient reach- ing 29 mPa Pa ' in both species at midday (Table 3). On 5 July 1985, in contrast, the upper ceiling of the cloud bank was 2,650-2,750 m during much of the day. As a result, atmospheric humidities at the study site were significantly higher, with the vapor pressure gradient reaching only 15 mPa T T 3; P < 0.001 for the difference between days in each species). ~ a ' in both species at midday (Ta The large differences in vapor pressure gradients are paralleled by significant differences in leaf con- ductances to water vapor in both species (Table 3; P « 0.001 for the difference between days in each species). In 4. sandwicense and D. menziesii, midday leaf conductances on 1 July 1985 were Volume 77, Number 1 1990 Robichaux et al. Adaptive Radiation of Hawaiian Silversword Alliance Lo / /j EI SES M i A \ X T WA У i FIGURES 3-6.— 3. Argyroxiphium sandwicense. —4. and D. menziesii (lower), with a nickel coin for scale. —6. Maui. For both species, the average height of mature individuals in the photograph is 0.6-0.9 т. 67.5% and 68.7%, respectively, of those on 5 July 1985. Thus in both species, a higher midday leaf conductance is correlated with a lower midday va- por pressure gradient. The correlations suggest a direct stomatal response to changes in the vapor pressure gradient, as has been reported in a large number of species from a wide variety of habitats (Schulze, 1986). Such stomatal responses enable plants to restrict water loss before severe water deficits develop. Though the two species exhibit similar corre- lations between leaf conductances and vapor pres- sure gradients, midday leaf conductances are sig- nificantly higher in A. sandwicense than in D. menziesii (Table 3; P < 0.001 for the difference between species on each day). On 1 and 5 July 1985, midday leaf conductances in A. sandwic- ense were approximately twice as high as in D. menziesii. With its higher leaf conductances, 4. ДШ, Dubautia menziesii. —5. Leaves of A. sandwicense (upper Habitat of + y жг, А. sandwicense апі D. menziesii on Haleakala, sandwicense also exhibits significantly higher mid- day and daily leaf transpiration rates than D. men- ziesii (Table 3; 0.001 for the difference between species on each day). Thus, the total daily water use per unit leaf area of A. sandwicense appears to be much greater than that of D. men- ziesii in July. Despite its higher transpiration rates, 4. sand- wicense does not exhibit markedly lower midday water potentials than D. menziesii (Table 3). This suggests that the two species may differ in rooting depths, and thus in access to available soil water supplies. Alternatively, they may differ in hydraulic resistances or capacitances. In contrast to D. men- гіеѕи, А. sandwicense accumulates very large amounts of extracellular polysaccharide in its ma- ture leaves (Carlquist, 1957; Carr, 1985). In the related species, 4. grayanum, the presence of this polysaccharide is correlated with a large increase 70 Annals of the Missouri Botanical Garden TaBLE 3. Water and temperature balance parameters of Argyroxiphium sandwicense and Dubautia menziesii on 1 and 5 July 1985 at a site of sympatry on Haleakala, Maui. The site was located at 2,820 m along the summit road. Clear skies prevailed at the site on both days, with midday solar irradiances exceeding 1,020 W m". Standard errors of the means are given in parentheses. Results of the relevant t-tests are provided in the text. А. sandwicense D. menziesii Parameter' l July 5 July l July 5 July Midday vapor pressure gradient (mPa Pa ') 29.1 15,5 29.1 15.3 (0.1) (0.1) (0.4) (0.3) Midday leaf conductance (mmol m ° ѕ !) 332.7 493.1 170.0 247.6 (4.5) (4.3) (11.6) (7.8) Midday leaf transpiration rate (mmol m * s !) 8.7 6.6 4.7 3.5 (0.1) (0.1) (0.3) (0.1) Daily leaf transpiration rate (mol m + d ') 276.2 234.8 158.0 113.3 (5.1) (4.1) (12.4) (4.2) Midday water potential (MPa) —1.1 —0.9 —1.0 —0.8 (0.1) (0.1) (0.1) (0.1) Midday leaf temperature (°С) 19.6 17.7 19.5 17.6 (0.1) (0.1) (0.2) (0.2) ! Parameters were measured for five individuals per species, with the ten individuals growing epa | in an area of approximately 10 m radius. Except for water potentials, parameters were measured for 2-4 recently mature leaves per individual. Leaf and air temperatures were measured with 0.127-mm copper-constantan roca connected to a Wescor сл model TH-65 digital thermocouple thermometer. Leaf conductances to w or were аса. with a Б Corp: model EE 1600. stead -state à no Чез species were amp istomatous, with the measured DAE conductances included the effects of the e appressed сан ieee Lea BAG күлөм were measured at 2-3-hr. intervals throughout the day, with each measurement sequence taking 0.3-0.7 hr. Midday leaf pei were he al leaf conductances in both species. Boundary layer conductances to water vapor an leaf transpiration rates were calculated according to the equations in Nobel (1983). Wind speeds were measured with a hot-wire anemometer and averaged | m s ' during the sampling periods. ceca dici humidities were measured with the humidity probe in the porometer. Intercellular-air-space humidities were assumed to be saturate and were calculated from e measured leaf temperatures. Estimates of daily leaf ош иге rates were obtained by pe eli the diurnal curves for each species. Hydrostatic pressures in the tissue apoplasm were measured with a Co. model 600 pressure chamber fitted with a 0-3-MPa, 150-mm-diameter gauge. The hydrostatic pressures in п tissue apoplasm were assumed to equal the water AE in the tissue symplasm (Koide et al., 1989). Measurements were made on the terminal 130 mm of recently mature leaves of 4. sandwicense and on terminal shoots of D. menziesii. in leaf capacitance (Robichaux & Morse, 1990). young leaves of 4. sandwicense experience very The midday temperatures of recently mature high temperatures for a significant portion of the leaves of A. sandwicense and D. menziesii in July day (Fig. 7). The young leaves are located at the do not differ significantly (Table 3). Both species bottom of the cone-shaped depression in the center exhibit leaf temperatures of 17-20°С, which is of the rosette. During the early morning, when the quite moderate given that midday solar irradiances young leaves are shaded from direct solar irradi- at the study site in July typically exceed 1,020 W — ance, their temperatures are lower than those of m-?, while midday substrate temperatures exceed the exposed, recently mature leaves. Once direct 55°C. Thus, recently mature leaves of A. sand- solar irradiance penetrates the cone-shaped depres- wicense and D. menziesii appear to be equally sion, however, the young leaves experience a rapid effective at decreasing heat energy inputs and in- increase in temperature and reach maximal tem- creasing heat energy outputs. In A. sandwicense, peratures in excess of 37°С by early afternoon. In steep leaf angles and low leaf absorptances (Figs. D. menziesii, which lacks the rosette growth form, 3, 5) may aid in decreasing inputs, while high leaf midday temperatures of young and recently mature transpiration rates (Table 3) may aid in increasing leaves are not significantly different. outputs. In D. menziesii, steep leaf angles (Fig. 4) Argyroxiphium sandwicense and D. menziesii may aid in decreasing inputs, while small leaf sizes thus illustrate the contrasting modes of morpho- (Fig. 5) may aid in increasing outputs. logical and physiological adaptation that have In marked contrast to recently mature leaves, evolved within the silversword alliance. Though the Volume 77, Number 1 1990 Robichaux et al. 71 Adaptive Radiation of Hawaiian Silversword Alliance two species grow in the same habitat, they exhibit different suites of traits enabling them to cope with the severe environmental conditions. ADAPTIVE RADIATION The ecological, morphological, and physiological diversity of species in the silversword alliance is exceptional. The patterns of diversity and genomic relationships among the species suggest that a variety of factors may have played important roles in their adaptive radiation. First, the periodic origin of new islands in the archipelago may have offered repeated opportu- nities for interisland dispersal and colonization. At least fourteen major dispersal and colonization events appear to have occurred during the evo- lutionary history of the silversword alliance (Carr et al., 1989). Founder effects associated with these events may have resulted in the evolution of new morphological and physiological traits. econd, the wide variety of habitats and wide range of elevations in the archipelago may have facilitated rapid ecological divergence, which is ev- ident in the modern species at grand and local scales. Rapid divergence may have been further enhanced by a genetic system that allowed for a high degree of morphological and physiological flexibility. Striking evidence for such flexibility is provided by А. sandwicense and D. menziesii, two species that differ radically in morphology and physiology, yet readily produce vigorous, fertile hybrids in nature (Carr & Kyhos, 1981) Third, chromosomal repatterning at the diploid level may have promoted reproductive isolation between species and may have contributed to the origin of novel gene complexes. Based on cyto- genetic evidence, at least eight major genomic ar- rangements exist among species in the silversword alliance (Carr & Kyhos, 1986; Carr et al., 1989). The genomic arrangements are structurally differ- entiated via reciprocal chromosome translocation or aneuploid reduction. The degree of postzygotic reproductive isolation between species, as mea- sured by the pollen stainability of hybrids, is strong- ly correlated with the degree of chromosomal dif- ferentiation. With respect to the origin of novel gene complexes, the aneuploid reduction in chro- mosome number among the Dubautia species is correlated with several major ecological, morpho- logical, and physiological differences. Relative to the 14-paired Dubautia species, the 13-paired species extend into drier environments, have small- er leaves, and have greater turgor maintenance capacities. Thus, the aneuploid reduction in chro- Y Y 40 o Young leaves > e Mature leaves o A Bo о 30r - o b pS 2 B a 20- oe У Б i ышы 5 *——) 5 Pi . / 10 n y o т) P 0 L L 6 10 14 18 Local time (h) FIGURE 7. Temperatures of young and recently ma- ture leaves of an individual of Argyroxiphium sandwic- ense on 18 July 1982 on Haleakala, Maui. Young leaves were less than 40 mm long. Air temperature was 7.4°C at 0700 hr. and 19.220 at 1400 hr. Midday solar irra- diance exceeded 1,020 W тг. At 1400 hr. on 2 July 1985, young and recently mature leaves of six individuals respectively. The leaf temperatures on 2 July 1985 | are significantly dif- ferent with a t-test at Р < 0.001. The site and mea- surement techniques were the same as in Table 3. mosome number may have been particularly im- portant in the evolutionary history of the silver- sword alliance. Understanding the specific roles played by these d other factors in the adaptive radiation of the silversword alliance will require more detailed anal- yses. Whatever their roles, it is clear that their combined effect has resulted in an extraordinary degree of diversification. LITERATURE CITED W. Күноѕ & J. Dvorak. 1988. m : CANFIELD, J. Я The cn of даре Factors and Plant Water иран ics in Plant Distribution in the Bog/Wet Forest Complex of Alakai Swamp, sperd Ee Ph.D Dissertation, Univ. of Hawaii, Hon Pireo. 5. 1957. Leaf anatomy and ontogeny in Argyroxiphium and Wilkesia (Compositae). Amer. J. Bot. 44: 696-705. Annals of the Missouri Botanical Garden 980. Hawaii: A Natural History. Pacific Tropical Botanical Garden, Lawai, Kauai. Carr, G. D. 19 Monograph of the é diinae (Asteraceae): y der лшщ Dubautia, and Wilkesia. Allertonia 4: 1-12 YHos. 1981. poem radiation in the Hawaiian silversword alliance (Compositae- Ma- diinae). I. oo of spontaneous hybrids. Evo- lution 35: 543-556. & 86. Adaptive radiation in the Hawaiian к alliance (Compositae-Madi- ytogenetics of artificial and natural hy- Күп 40: 959-97 ; . ROBICHAUX, eS WITTER & D. Күноз. 1989. Adaptive raadon of the b silversword alliance (Compositae- Madiinae): a com- parison with Hawaiian picture-winged Drosophila. Pp. 79-97 in L. V. Giddings, K. Y. Kaneshiro W. W. Anderson (editors), Genetics, Speciation, and the Founder Principle. Oxford Univ. d. CARSON, Н. L. & К. Y. KANESHIRO. 1976. Drosophila of Hawaii: ee) and ы genetics. Апп. Rev. Ecol. S 31 Koipe, R. T., эн. о HAUX, s. R. Morse & C. M. SMITH. 1989. Plant water status, hydraulic resis- tance, and capacitance. Pp. 161- cul in R. W. Pearc hleringer, yy, J. К. & P. W. Rundel (шо), Plant Physiological e Field Methods and Instrumentation. Chapman « Hall, Londor ea С. А. & A. Т. Аввотт. 1970. Volcanoes in the Sea: The Geology of Hawaii. Univ. of Hawaii . Some bioenvironmental nditions m ES general design of IBP research in Hawaii. Pp. 3-32 in D. Mueller-Dombois, K. W. Bridges & H. L. Carson (editors), Island Ecosystems: Biological Organization of Selected Hawaiian Com- munities. Hutchinson Ross, Stroudsburg NonpEL, P. S. 1983. S ipie ju учоону and Ecology. W. H. Freeman, San Fra ROBICHAUX, R. H. 1984. Variation in Pw tissue water relations of two sympatric Hawaiian Dubautia нш and their natural hybrid. Oecologia 65: 75- 19 Tissue elastic properties of a mesic facet fawn Dubautia species with 13 pairs of chromosomes. Pacific Sci. 39: 191-194. J. E. CANFIELD. 1985. Tissue elastic prop- erties of eight Hawaiian Dubautia species that differ in habitat and diploid chromosome number. Oecologia 7-80. — & S. R. MORSE. 1990. Extracellular polysac- yis roxiphium grayanum | (Compositae- Madiinae). Amer. J. Bot. 77: 134-13 ScHULZE, E. D. 1986. Carbon IT and water vapor exchange in response to drought in the atmosphere and in the soil. Ann. Rev. Plant Physiol. 37: 247- il SIMON, ( . Hawaiian ise biology: an introduction. Trends Ecol. Evol. 2 FLAVONOID DIVERSITY IN RELATION TO SYSTEMATICS AND EVOLUTION OF THE TARWEEDS'? William J. Crins? and Bruce A. Bohm* ABSTRACT The tarweeds produce a diverse array of flavonoids, including glycosylated compounds and methylated aglycones. Extra hydrox ylati ion at positions 6 and/or 8, with fre the view investigations suggest ing on the evolution of its taxa. Flavonoids have the dynamics of К: in dude d In Hemizonia, multivariate analyses jos uent O- methylati ion at these and other ositions, is preva alent. ans of assessing arpha, Lay en exhibit distinctive substitution and абоза бов tendencies. onary may be helpful in developing and testing phylogenetic hypotheses in the Madiinae The North American Madiinae (Asteraceae: He- liantheae) consist of about a dozen genera distrib- uted from Baja California to British Columbia, with a major concentration of taxa in California. Layia Hook. & Arn. and Madia Molina, two of the larger genera, contribute conspicuously to spring floral displays in California. The Hawaiian Madiinae con- sist of three genera: Argyroxiphium DC. with five species, Dubautia Gaudich. with 21 species (33 taxa in all), and Wilkesia A. Gray with only two species. Arguably the most spectacular members of the island taxa are the subspecies of 4. sand- wicense: subsp. sandwicense grows on Mauna Kea on the island of Hawaii; subsp. macrocephalum (A. Gray) Meyrat, grows on Haleakala on Maui. Both of these have suffered serious predation and are maintained only with special effort. Species of Dubautia occur on all major islands of the archi- pelago. They exhibit a very wide range of mor- phological forms and are ecologically the most di- verse within the Hawaiian tarweeds. Wilkesia consists of two species, W. gymnoxiphium A. Gray and W. hobdyi H. St. John, both of which are restricted to western Kauai. The former is much more common, occurring in comparatively large populations along the western edge of Waimea Canyon. Wilkesia hobdyi is known from only two sites on the far western ends of two ridges and consists of perhaps no more than 200 individuals; it is under severe predation from feral goats and may be on the verge of extirpation. The Madiinae provide ideal opportunities for the study of evolutionary divergence. In a few genera, hypotheses of species formation have been pro- posed, and recently tests of these ideas have been conducted using electrophoretic data (Warwick & Gottlieb, 1985; Witter & Carr, 1988). However, it is surprising that in a group as well studied as the tarweeds, so few explicit hypotheses of phy- logenetic relationships have been proposed. There now exists a large enough body of data that such hypotheses could be constructed, and these could serve as frameworks for assessment of congruence among different types of data and for testing models of speciation. Flavonoid data could serve in both roles. Since the genetic bases of their biosynthesis are well known (Heller, 1986), these compounds ultimately can be used to construct phylogenies and to assess phylogenies based on other data. SURVEY AND BIOSYNTHETIC TRENDS Before dealing with the flavonoids from a phy- logenetic viewpoint, we will discuss their structural ! This Е еф poa of Canada to research was supported by operating and equipment grants from the Natural Sciences and Engineering B.A.B. emosystematic studies of the tarweeds (Asteraceae: Heliantheae: Wi. qune 6. ? New York State Museum, Biological Survey, Albany, New York 12 ' Department of Botany, University of British Columbia, Vancouver, [seem cra Canada V6T 2В1. ANN. Missouni Bor. GARD. 77: 73-83. 1990. 74 Annals of the Missouri Botanical Garden ^ R R3 s o Ro R о R4 o OH O FIGURE 1. Structures of tarweed flavonoid aglycones. Upper structure represents flavanones and dihydrofla- vonols. Flavones and flavonols are represented by the middle and lower structures, respectively. See Tables 1- 3 for substitution patterns (R = H, OH, or OMe) and occurrence of individual ade! in the tarweed gen- variation and accumulation in the tarweeds. The variety of substitution patterns observed is based upon four flavonoid classes, flavanones, dihydro- flavonols, flavones, and flavonols (Fig. 1). The glan- dular exudates of Adenothamnus validus (Bran- degee) Keck and all species of Holocarpha (DC.) Greene contain all four classes (Crins & Bohm, 1987, 1988a, b). Some of the n = 13 (glandular) species of Dubautia, such as D. ciliolata (DC.) and fla- vonols in their resins, and two classes (flavanones Keck, accumulate flavanones, flavones, and flavonols) among their vacuolar constituents (Crins et al., 1988a). All species of Hemizonia DC. examined so far accumulate flavones and flavonols, and all but one of these also accumulate flavanones (Tanowitz et al., 1987). None of the taxa of Ar- gyroxiphium, Calycadenia DC., Lagophylla Nutt., Layia, or Wilkesia examined so far accu- mulate flavanones. Taxa from these genera that do accumulate flavonoids exhibit flavones and fla- vonols. Two of the species of Layia examined, L. carnosa (Nutt.) T. 8 G. and L. chrysanthemoides (DC.) A. Gray, lack flavonoids externally, but both secrete large amounts of scopoletin (a coumarin) 1988 glycosylated flavonoids. Ta (Crins et al., b). All tarweeds contain vacuolar rweed flavonoids exhibit a wide variety of substitution patterns. Tables 1-3 present the known arrays of flavonoid aglycones in leaf exudates. It is immediately apparent that tarweeds have a pro- pensity for O-methylation. Another prominent fea- ture is the abundance of extra hydroxylation at positions 6 and/or 8, with these hydroxyls fre- quently being methylated. These series of com- pounds can be examined in light of biosynthetic considerations, O-Methylation of flavanones can occur at positions 7 and/or 4”. Dihydroflavonols may be formed from their flavanone precursors either before or after O-methylation has occurred. The same is true for later biosynthetic steps (formation of flavones from and some generalizations arise. flavanones or flavonols from dihydroflavonols). However, the series of compounds within a given plant generally suggests that O-methylation at po- sition 7 occurs early in the biosynthetic pathway and that extra hydroxylation occurs after the bond between carbons 2 and 3 has become unsaturated. We have found no evidence of 6-hydroxyflava- nones or 6-hydroxydihydroflavonols. [The report of 6-hydroxyeriodictyol 7-methyl ether by Crins & Bohm (1987) must be corrected to 7-O-methyl- dihydroquercetin in light of new evidence on struc- tural differentiation of flavanones and dihydrofla- vonols using mass spectrometry (Balza et al., 1988).] The affinities of the 6- (and 8-) O-methyltransferase enzymes for their sub- strates must be very high, as 6- and 8-hydroxy derivatives are only infrequently encountered. Several taxa lack external flavonoid aglycones. In Layia chrysanthemoides, loss of glandular structures is correlated with the inability to secrete these compounds, although the leaf wash did con- tain the coumarin scopoletin. In the n = 14 Du- bautia taxa, glands are absent from the foliage and stem, although there may be some glands on 1985). In most cases insufficient flowering material was available the pedicels and corollas (Carr, to test for flavonoids. However, in the case of D. scabra (DC.) Keck, where large amounts of flow- ering material was available, flavonoids were not detected (Crins et al., 1988a). In several mainland Volume 77, Number 1 Crins 8 Bohm 75 1990 Flavonoid Diversity of Tarweeds TABLE 1. Occurrence and substitution patterns of flavanones and dihydroflavonols in tarweed genera. R-groups on model structure correspond to R-groups in columns one to four. Н = hydrogen; OH = hydroxyl group; OMe = methoxyl group; Ad = Adenothamnus; D = Dubautia; Hm = Hemizonia; Ho = Holocarpha; M = Madia. Data for Madia are based on an incomplete survey of M. elegans D. Don. Substitution pattern Genus R, R, R, R, Ad D Hm Ho M H OH H OH + + - - H OH H OMe + — — — — H OMe H OH — T — - H OMe H OMe + — — + — H OH OH OH — — + - — H OMe OH OH + + + + 4 H OMe OH OMe + — — — — ОН ОМе ОН ОН + Е - + + TABLE 2. Occurrence and substitution patterns of flavones in the tarweed genera. R-groups on model structure correspond to R-groups in columns one to five. H = hydrogen; OH = hydroxyl; OMe = methoxyl; Ad = Adenothamnus; Ar = Argyroxiphium; С = Calycadenia; D = Dubautia; Hm = Hemizonia; Ho = Holocarpha; Lg = Lagophylla; Ly = Layia; M = Madia; W = Wilkesia. Ra Ra R2 À в, он о Substitution pattern Genus R, R, R, R, R, Ad Ar C D Hm Ho lg Ly M W H OH H H OH — + — + — * — + + - H OMe H H OH — — - + — = ai E. = ES OH OH H H OH — - - + Е = = ES 3 b. OMe OH H H OH T - - + — — — + + 2 OMe OMe H H OH * + - + = = + + = + ОМе ОН H H OMe — — — — = = = HE E = OH OMe H OH - - – ip = EE = = " de OMe OMe H OH — + — = — = = ES ES + OMe OH OMe H OH - — - = = = = = T OMe OMe OMe H OH — + — — — = ЗЕ = = + OH OH OH - -— — _ + zd = E — 3 H OMe H OH OH + - — + — + A РА a + H OH H OH OMe ~ = — + — = = = 5 3 H OH OMe OH OH — — - — - ES x: = = ji OMe OH H OH OH + — - — = = = 4 — E OMe OH H OMe OH — — - — - WE ae & = zi OMe OH H OMe OMe = — - — — – - de = x. OMe OMe H OH — — — + 4. = = = - + OMe OMe H OMe OMe — — — — — = + = = = OMe OH OH OH -— — — = + = = = — — OMe OH OMe OH OH — — — — = - = = = 4 ОМе OMe OMe OH OH — — — — — - En + Annals of the Missouri Botanical Garden TABLE 3. correspond to R-groups in columns one to six. Ar = Argyroxiphium; С = Calycadenia; D = Dubautia; Hm = Occurrence and substitution patterns of flavonols in tarweed genera. R-groups on model structure H = hydrogen; OH = hydroxyl; OMe = methoxyl; Ad = Adenothamnus; Hemizonia; Ho = Holocarpha; Lg = Lagophylla; Ly = Layia; М = Майа; W = Wilkesia. в, Rs Ra o " Ra R, OH о Substitution pattern Genus R, R, R, R, R R, Ad Ar € D Hm Ho lg Ly M W OH H OH H H OH — — — + — — — — — OMe H OH H H OH + — — — — + — — — + ОН Н ОМе Н Н ОН - — — * - + — — — OH H OH H H OMe — — — — + - — — — — OMe OMe OH H H OH + — + - — - — + — OMe OMe OMe OMe H OH — — + - - - — — — - OH H OH H OH OH — - — 4 + - m - - — OMe H OH H OH OH — — — E * + — -— 4 — OMe H OH H OMe OH — — — — — + — — + — OH H OMe H OH OH + — — - — 4 — — — — OMe H OMe H OH OH — - — — + + — — — OMe OH OMe H OH OH — — — — — + — — - — OH OMe OH H OH OH — - — — + — - — — OMe OMe OH H OH OH — = = — + + — — — — OMe OMe OH H OMe OH — — — — — — + — — - OMe OMe OMe H OH OH — - - + — + — — — OMe OMe OMe H OH OMe — — = — - — + — — = ОМе ОН ОМе ОМе ОН ОН — Е + — — — — — — — OMe OMe OH OMe OH OH - - - + — — — — — OMe OMe OMe OMe OH OH — — + - — m — — + + ОМе ОМе ОМе ОМе ОМе ОН — — + — — — Е = = OMe H OH OMe OH OH — - - - + - — — — — OMe H OMe OMe OH OH — — + — + — — - — — tarweeds, po are present but no flavonoid agly- cones : Achyrachaena mollis Schauer, о. scaber Hook., Lagophylla mi- nor (Keck) and Lay- {а carnosa. Coumarins were observed in some of these, however (Crins et al., 1988b). Many of these plants have highly reduced or modified floral and / or inflorescence structures compared with their Achyra- chaena mollis is clearly divergent, with its reduced, eck, L. ramosissima Nutt., congeners, or the subtribe as a whole. yellow-orange rays and reduced inflorescence. Within Lagophylla, L. minor has few glands, and L. ramosissima has greatly reduced capitula, com- pared with such species as L. glandulosa A. Gray, which yields copious flavonoid-rich resin. Layia carnosa is Clearly differentiated from other Layia species by its reduced white rays and specialized habitat preference. Some trends are beginning to emerge in terms of accumulation tendencies and substitution pat- terns among the ten genera for which we have data. Adenothamnus validus, Hemizonia species, and Holocarpha species accumulate flavanones, as do several xerophytic species of Dubautia: D. arborea (A. Gray) Keck, D. ciliolata, and D. linea- ris (Gaudich.) Keck subsp. hillebrandii (Н. Mann) G. Carr. the island of Hawaii. In contrast, D. linearis subsp. linearis from Lanai, and D. sherffiana Fosb. from Oahu (both л = 13) have much less diverse arrays of external flavonoids, and they do not accumulate 'These three taxa of Dubautia occur on flavanones. This trend is further emphasized since all n = 14 taxa lack external flavonoids and they all occur on the older islands. Thus, flavonoid di- versity parallels cytological and morphological ra- diation in. Dubautia (Carr, 1985; Crins et al., 1988a; Crins & Bohm, unpublished data). The capacity of a particular taxon to accumulate flavanones and dihydroflavonols must be used ju- diciously in considering relationships since these classes of compound serve as precursors of flavones and flavonols, respectively (Heller, 1986). How- ever, if flavanones or dihydroflavonols exhibit sub- stitution patterns that are not seen in flavones or Volume 77, Number 1 1990 Crins & Bohm 77 Flavonoid Diversity of Tarweeds flavonols from the same taxon, then information of phylogenetic interest can be gained. Methylation at any hydroxylated position on the flavonoid nu- cleus requires a specific O-methyltransferase; a high level of site selectivity has also been shown for O-glycosyltransferases (Heller, 1986; Ibrahim et al., 1987). Assuming that such enzyme specificity is general in plants, relationships of taxa within genera and perhaps among genera in the Madiinae can be examined. In Adenothamnus and Holocarpha, one fla- vanone (7,4'-dimethylnaringenin), present in large quantities, has no counterpart among suites of flavones and flavonols. Thus it appears that the 4'-O-methylation step is unrelated to subsequent biosynthetic steps, and this flavanone is an accu- mulation product in its own right. This is also true of 4'-methylnaringenin and persicogenin in Ád- enothamnus. On the generic level some common tendencies in substitution patterns are apparent among fla- vones and flavonols. Argyroxiphium appears to produce only flavones, but only one species has been examined. Four genera have 3-unsubstituted flavonols in their pigment profiles, Adenothamnus, Dubautia, Hemizonia, and Holocarpha. Of this group only Dubautia lacks the corresponding 3-O-methyl derivatives. All genera are capable of 7-O-methylation although each genus also contains compounds in which the 7-hydroxyl group remains unsubstituted. All genera produce at least one 6-O-methylated compound. Less common is 8-0- methylation, which was observed in Argyro- xiphium, Calycadenia, Dubautia, Hemizonia, Madia, and Wilkesia. Within the Madiinae as a whole there is a strong tendency for the hydroxyl groups at positions 6 and/or 8 to be methylated. Only three of ten genera studied exhibit 6-hy- droxyflavonoids (Calycadenia, Dubautia, and Holocarpha), and only Hemizonia contains 8-hy- droxyflavonoids. B-Ring O-methylation of flavones and flavonols is also restricted in distribution: 3'-O-methylation is seen in Calycadenia, Holo- carpha, Lagophylla, Layia, and Madia, and 4'-O- methylation is known only in Dubautia, Hemizon- ia, Lagophylla, and Layia. WES glan- dulosa and Layia hieracioides (DC.) Hook. & Arn. are alone in exhibiting 3',4'-di-O- ecc rm 1988b). (Crins et al., EvoLUTIONARY TRENDS ANALYSIS OF HYBRIDS IN DUBAUTIA We begin our discussion of the application of flavonoid data to tarweed systematics by examining a case of natural hybridization between two species of Dubautia. On the island of Hawaii, lava flows from two of its major volcanoes, Mauna Loa and Mauna Kea, meet in an area known as the “Saddle.” Dubautia scabra (n = 14) and D. ciliolata (n = 13) co-occur here. Dubautia scabra exhibits very high chromosomal homology with all of the n — 13 species and is considered to be their nearest living ancestor (Carr, 1985). Natural hybrids have developed in areas of marginal or intermediate habitat in the contact zone. Carr & Kyhos (1981) suggested that these hybrids were generally F,s with a few higher-generation recombinants perhaps also present. Flavonoid data clarify the nature of the hybrid population. The flavonoid glycosides of the parental species differ, with D. ciliolata producing kaemp- ferol and quercetin 3-O-monoglucosides and 3-O- rhamnosylgalactosides and D. scabra producing quercetin 3-O-glucoside and 6-O-methoxyflavonol mono- and diglycosides. The hybrids share 3-0- monoglucosides with both parents, but all of their diglycosides are “hybrid”? compounds, with rham- nose being derived from D. ciliolata and glucose from D. scabra (Crins et al., 1988a). All of the hybrid plants contain the same array of glycosides. Glandular resins allow further resolution of the structure of the hybrid population. Dubautia sca- bra does not produce external flavonoids; the resin of D. ciliolata contains several flavones, two fla- vanones, and one flavonol. Quercetin and kaemp- ferol appear to be restricted to the hybrids possibly on account of recombination or other disruption of biosynthesis. The varying distributions of compounds among hybrid plants indicate differing degrees of disrup- tion of biosynthetic pathways, which suggests the presence of higher-generation recombinants. When considering the entire set of hybrids in the sample, gradation in the complexity of flavonoid aglycone arrays emerges. The field-collected hybrids show considerable variation in their arrays. However, synthetic F, hybrids produced under controlled conditions were chemically uniform. Comparison of the flavonoid chemistry of the synthetic and natural hybrids strongly suggests that all of the natural hybrids are at least second-generation crosses, and the complexities of some of their pro- files suggests introgression toward D. ciliolata. POPULATION ANALYSIS IN CALYCADENIA Resin flavonoids have also served usefully as markers of inter- and intrapopulational differences in Calycadenia ciliosa Greene (Emerson et al., 1986). Calycadenia ciliosa consists of at least five structurally distinct chromosome races (Carr & 78 Annals of the Missouri Botanical Garden Carr, 1983), and high degrees of structural het- erozygosity in some populations complicate the sit- uation. The Dry Creek race is the most variable morphologic ally, the most widespread geographi- cally (Carr & Carr, 1983), and the most diverse chemically (Emerson et al., 1986). In fact, there is a striking correlation among morphological vari- ability, geographical extent, and chemical diversity in all races of C. ciliosa. For the most part, fla- vonoid data concur with the cytogenetic results. However, it seems that the Dry Creek race is most likely to be the primitive race, rather than Ciliosa, as concluded earlier (Emerson et al., 1986), since the former has the most complex array of flavo- noids, including the only compound with а mono- hydroxylated B-ring. The Lewiston and Corning races both accumulate 5,3',4'-trihydroxy-3,6,7,8- tetramethoxyflavone along with one of its precur- sors unsubstituted at position 8. None of the pre- cursors remain in the Ciliosa and Pillsbury races, suggesting increased efficiency in substrate utili- zation. FLAVONOID PATTERNS WITHIN DUBAUTIA Several interesting evolutionary problems can be found in the genus Dubautia, whose 21 species (33 taxa in all) are ordered into three sections. Section Dubautia consists of ten species (18 taxa and is characterized by having n — 14 and, relevant to our interests, being eglandular and thus lacking resinous flavonoids. Section Venoso-reticulatae (А. Gray) G. Carr consists of only D. latifolia (A. Gray Keck which has n = 14, is eglandular, and, in- terestingly, is a liana. The remaining ten species (14 taxa) comprise sect. Railliardia, which is het- erogeneous with regard to chromosome number and leaf surface chemistry. All taxa in this last section have n = 13 except D. scabra, which has n = 14 in common with the other two sections. It also lacks leaf surface flavonoids in common with the other two sections. Dubautia scabra is con- sidered to be the nearest extant relative to the n = 13 species, as indicated by enzyme electropho- retic data (Witter & Carr, 1988). * flavonoid aglycones in Dubautia herbstobatae is more difficult to explain. It may have resulted from a reversal to a nonglandular state, or this may be — — ‘he absence of a more primitive taxon than has previously been suspected. Dubautia herbstobatae might have been the first offshoot of the п = 13 lineage, before glandularity and flavonoid secretion became prom- inent features of the group, a contention supported by its occurrence on Oahu, the oldest island, on which п = 13 species presently occur (Fig. 2). Within the л = 13 glandular species it is possible to distinguish two groups based on comparative complexity of flavonoid profiles. Dubautia linearis subsp. linearis, D. reticulata (Sherff) Keck, and D. sherffiana have simple aglycone profiles con- sisting of three compounds each. A close relation- ship between D. linearis subsp. linearis and D. sherffiana is suggested by their identical profiles. This is the only case in Dubautia where profiles of two different taxa were identical. The other three species in the n = 13 glandular group that we have examined, /). glutinosa G. Carr, and D. linearis subsp. brandii, exhibit much more complex arrays of flavonoids in their resins. Each of these latter three taxa has its own peculiar profile. One of the more interesting things to emerge from flavonoid analysis of this group of taxa is the dissimilarity between these two subspecies of D. linearis. Dubautia line- aris subsp. linearis occurs on Maui and Lanai (but not on Hawaii), while D. linearis subsp. hille- brandii occurs on the island of Hawaii. (Dubautia linearis subsp. opposita (Sherff) G. Carr) occurs on Molokai and was not available for study.) Du- bautia linearis subsp. linearis has a flavonoid pat- tern identical to that of D. sherffiana. This rela- tionship is supported by close similarities in vegetative anatomy (Carlquist, 1959) and mor- phology (Carr, 1985). Dubautia linearis subsp. hillebrandii differs substantially from subsp. linea- ris and may be more closely related to other taxa from the islands of Maui and Hawaii, contrary to its current placement in D. linearis. In fact, Carlquist (1959) pointed out the anatomical and morphological similarities of subsp. hillebrandii with D. arborea and D. menziesii (A. Gray) Keck. Furthermore, Witter & Carr (1988) showed that subsp. hillebrandii had higher genetic identities with D. reticulata, D. dolosa (Degener & Sherff) G. Carr, D. platyphylla (A. Gray) Keck, D. men- ziesii, and D. arborea than it did with D. sherf- fiana. (Data for D. linearis subsp. linearis were not available for comparison.) Further studies of this system will be necessary to gain a better un- derstanding of relationships. Flavonoid glycoside patterns may likewise pro- vide indications of relationships within Dubautia. On a relatively coarse scale, the presence of an eriodictyol glycoside correlates well with the xe- rophytic habit, although at least two mesophytes (D. laxa Hook. & Arn. subsp. hirsuta (Hillebrand) G. Carr and D. microcephala Skottsb.) also contain this compound. On the other hand, several of the mesophytes (D. knudsenii Hillebrand subsp. knud- senii, D. laevigata A. Gray, D. latifolia (A. Gray) Volume 77, Number 1 1990 Crins & Bohm 79 Flavonoid Diversity of Tarweeds Keck, D. plantaginea Gaudich. subsp. planta- ginea, and D. raillardioides Hillebrand) lack this compound. It is unclear whether the occurrence of this compound signifies relationship, or whether this is a case of parallelism (see discussions of Hemizonia and Holocarpha below), since the sug- gested relationships of these mesophytes to the xerophytes do not conform to those presented in the scheme of Witter & Carr (1988). MULTIVARIATE ANALYSIS OF HEMIZONIA FLAVONOIDS Tanowitz and associates (Proksch et al., 1984; Tanowitz et al., 1987) have examined the flavonoid aglycones of several species of Hemizonia. In their work we can also look for indications of relation- ships at the macroevolutionary level. They at- tempted to elucidate relationships by comparing the flavonoid aglycone arrays of 18 taxa of Hemi- гота, including three perennials (H. greeneana Palmer subsp. peninsularis Moran, H. minthornii Jepson, and Н. streetsii A. Gray) using phenetic methods. Their ordination and clustering results suggest that each of these taxa belongs to a dif- ferent subgroup within a heterogeneous assemblage of taxa, including Н. fitchii A. Gray of sect. Cen- tromadia Keck, and an assortment of species from sect. Madiomeris Nutt as emended by Tanowitz. The cluster analysis also split sect. Madiomeris into two groups. If we consider the phenetic relationships within sect. Madiomeris in a flavonoid biosynthetic con- text, several contrasts are evident. Before pro- ceeding it is important to remember that the pres- ence of a compound in a profile requires that all precursors in its pathway must have been present at some time so that, although the absence of a compound will affect the computation of distance measures, it will not necessarily affect an estimate of relationship based on biosynthetic principles. After recoding the data to account for precur- sors (only unique end products are useful for phy- logenetic analysis) there are only six compounds that can potentially yield useful phylogenetic in- formation and one additional compound that serves as an autapomorphy. When this data set is analyzed using Penny’s branch-and-bound method to deter- mine the most parsimonious trees (using PHYLIP), two topologies are found. (Since a few taxa from this section are not included in the data set owing to lack of flavonoid data, the group analyzed is paraphyletic.) биеш Кей, the topologies provide novel frameworks for testing ideas about relation- ships within the section. Figure 3 illustrates the topologies and a consensus tree based on them. D. linearis subsp. linearis other n=14 taxa D. scabra D. herbstobatae D. sherffiana D. reticulata other n=13 taxa. RE 2. Partial resolution of phylogenetic relation- i within a sect. Railliardia, based on fla- vonoid aglycon Although the consensus tree is very poorly re- solved, it suggests that two subgroups exist within sect. Madiomeris (excluding the perennials). These subgroups are not entirely consistent with the groupings found in the phenetic analyses reported by Tanowitz et al. (1987). In their group II, halliana Keck is included with H. arida Keck, H. fasciculata (DC.) T. & G., H. pallida Keck, and Н. paniculata. Hemizonia pentactis (Keck) Keck is placed in their group III. The group placements Н. halliana and Н. pentactis are reversed in our analysis relative to their positions in the phe- netic analysis of Tanowitz et al. (1 en the data for the three perennials are included in the analysis, almost 1,500 most par- simonious trees are possible. It has not been possible to examine more than a small number of these trees, but a few features are worthy of note. In the sample of 50 trees examined, H. arida, Н. fasciculata, H. pallida, H. paniculata, and H. pentactis are always maintained as a group. How- ever, the distinction between these five taxa and the other group found when perennials were ex- cluded (consensus tree, Fig. 3) is not always upheld. Little can be said about the relationships of the perennials, except that Н. streetsii is always grouped with Н. lobbii in the 50 trees. Thus, the flavonoid Annals of the Missouri Botanical Garden E c — — lo æ = ШС > ~ 4 " S o5 © Б seaefkane ер e c - YN = б © s £ S а € FIGURE 3 tree. Hal eir consensus fasciculata; ~ ~ data provide general indications of relationship Madiomeris, but fine-scale resolution does not seem possible. within sect. PHYLOGENETIC ANALYSIS OF HOLOCARPHA USING FLAVONOIDS Whenever possible, an attempt should be made to integrate all available information into a clas- sification scheme or a phylogenetic hypothesis. In a sense, this has been done with //emizonia, where a recently revised genus has been studied in terms of its flavonoid chemistry, and further suggestions about relationships have been set forth. However, data are incomplete for many taxa, and no explicit о for the genus has been proposed against the chemical data could be examined. Since hat is no consensus about the way in which chem- ical changes occur ын the process of evolution (Gornall & Bohm, 1978), a test of congruence of chemical data against a phylogeny derived from other sources (morphology, cytology, etc.) would Alternative phylogenetic trees of the taxa of Hemizonia sect. Madiomeris n which flavonoid aglycone é ey to species Ari = H. halliana; lobbii: Pal = Н. pallida; Pan = H. paniculata; Pen = Н. E £x -— = æ = Y - » d © = б SBS aax ЕЕ? ELO ч > а & 2 E £€ 8 H. arida; increscens; 19 = f = H. conjugens; Cor = H. "H kelloggii, Lob = pentac tis. Kel = H. be more prudent. In the case of Holocarpha, a genus of four summer-flowering, self-incompatible annuals, we generated several alternative clado- grams using various algorithms (PHYLIP version 2.8, Felsenstein, 1985). Three cladograms were based on 12 discrete-state morphological charac- ters, one ‘cladogram’ was based on a scenario where two independent aneuploid events were hypothe- sized. The most parsimonious cladogram for the morphological data set was supported by four dif- ferent algorithms (Crins & Bohm, 1988b). Fla- vonoid aglycone biosynthetic data were then su- perimposed on these topologies. Fourteen compounds have been characterized from species of Holocarpha. The presence or ab- sence of terminal steps in the biosynthetic network are more or less species-specific. For example, the capacity to hydroxylate position 6 in 3,7-di-O- methylquercetin is lost in H. macradenia (DC.) Greene, and only H. heermannii (Greene) Keck can dehydrogenate the 2-3 bond and subsequently methylate the hydroxyl group at carbon-3. Volume 77, Number 1 1990 Crins 8 Bohm 81 Flavonoid Diversity of Tarweeds The three cladograms yield similar results in terms of their congruence with the flavonoid data. All three require parallel losses of the capacity to produce 7-O-methylluteolin in Н. heermannii and Н. macradenia, and 3-O-methylquercetin and 3,3'- di-O-methylquercetin in H. obconica (Clausen & Keck) Keck and H. macradenia. All other losses or gains in biosynthetic capacity are autapomor- phies. The least parsimonious tree for morpholog- ical characters requires only a single parallelism, for the loss of 7-O-methylluteolin. The loss of meth- ylated quercetin derivatives is а synapomorphy uniting H. macradenia and H. obconica on this tree. The tree most consistent with the chemical data is the least parsimonious on morphological and cytological grounds (Crins & Bohm, 1988b). It should be noted that the difference (based on flavonoid data) is a single step, however. In the present case, where morphological, cytological, and ecological (Palmer, 1982) data all support a par- simonious hypothesis of evolution, we must suggest that flavonoid aglycone biosynthesis has undergone some degree of parallelism in terms of losses in different lineages. We cannot ignore the bulk of evidence in favor of a more parsimonious inter- pretation of chemical changes. SOME OBSERVATIONS ON LAYIA In Layia, Clausen et al. (1941) defined several species on morphological and cytological grounds. Its 15 species are considered to have diverged through specialization of peripheral populations on novel substrates or in new habitats (geographical speciation). Warwick & Gottlieb (1985) obtained evidence from enzyme electrophoretic studies that supported the geographical speciation hypothesis for the species with n — 7. Flavonoid aglycones have also provided some support for this hypoth- esis. The identical nonpolar phenolic profile of L. platyglossa (F. & M.) A. Gray (n = 7) and L. glandulosa (Hook.) Hook. & Arn. (n = 8) suggests that they are close relatives, as did the hybridization experiments of Clausen et al. (1941). This evidence suggests aneuploidy with little subsequent differ- entiation in the genome. Although more taxa must be examined, a pattern is emerging in which flavonoid aglycone diversity increases from total absence in L. chrysanthe- moides (n = 7) to the profiles of L. glandulosa and L. platyglossa (n = 7), which contain a small number of compounds, to the richer profiles of the 8)-L. paniculata Keck (п = 16) group. Layia septentrionalis Keck (n = 8) L. hieracioides (п = is an outlier in that it lacks the flavonol aglycone biosynthetic pathway, and L. carnosa (n = 8) is also an outlier, lacking flavonoid aglycones entirely. It seems that as each derivative species arose, independent chemical changes occurred. There an overall trend toward increasing comple E in the flavonoid aglycone composition of the resin, assuming that n = 7 is the basic chromosome number of the genus (Gottlieb, 1987), but individ- ual offshoots (L. paniculata, L. septentrionalis, L. carnosa) have taken different, occasionally di- vergent, routes in the process. Аз more species are examined, the flavonoid data should facilitate our understanding of the phylogenetic history of the genus. FUTURE DIRECTIONS Studies of flavonoid occurrence and variation within the tarweeds have begun to provide new insights into the modes and directions of evolution in this diverse group: Some of the numerous prob- e been mentioned above ш Sa hypotheses, especially for the larger genera such as Calycadenia, Hemizonia, and Ma- dia can serve as templates for the assessment of congruence with chemical and other data sets. There are numerous taxa in these and other genera for which no flavonoid or other chemical data are available. Table 4 summarizes the current state of the flavonoid surveys. We have concentrated large- ly on flavonoid aglycones, but our examination of flavonoid glycosides in two species of Dubautia and their natural hybrids indicates that these polar compounds will also be useful in tracking micro- evolution (Crins et al., 1988a). Preliminary chro- matographic examinations of glycosides in a wide range of tarweed taxa reveal substantial differences in the occurrence of compounds that may be useful in macroevolutionary studies. These sorts of anal- yses provide a feedback mechanism whereby we learn more about the scale at which different types of data are useful in tracking evolution. In some cases, flavonoid suites suggest that biosynthetic parallelisms have occurred (e.g., Hemizonia and Holocarpha); in other cases, losses or gains in biosynthetic capability appear to be species-specific (e.g., Holocarpha and Layia); and in still others, flavonoids serve as accurate markers of population differentiation (e.g., Calycadenia). Little is known of the functional significance of flavonoids in tarweeds or in most other groups of plants. Because of the rich and variable arrays of compounds produced by different species of tar- weeds, these plants may be useful for studying flavonoids as ultraviolet radiation shields, as anti- 82 Annals of the Missouri Botanical Garden ABLE 4. Summary of progress in tarweed flavonoid survey. 2D TLC = chromatographic evidence for flavonoid glycosides. +/— = some species produce exudate flavonoids, some do not. The “ ?" indicates that no data exist at the moment. num- specie bers of exam- Flavonoid Flavonoid Genera species ined glycosides aglycones Sources of data Achyrachaena 1 1 + (2D TLC) none Crins & Bohm ee data) Adenothamnus 1 1 + Crins & Bohm (19 Argyroxiphium 5 2 +} + Bohm & Fong eee data) Blepharippapus 1 1 + (2D TLC) none Crins & Bohm (unpublished data) Blepharizonia 1 0 ? T Calycadenia 11 5 + + Emerson et al. (1986) Dubautia 21 18 + +/- Crins et al. (1988a) Hemizonia 32 17 ? + Proksch et al. (1984); Tanowitz et al. (1987) Holocarpha 4 4 ? + Crins & Bohm (1987); Crins & Bohm (unpublished data) Holozonia 1 0 ? ? Lagophylla 5 3 ? +/= Crins € Bohm (unpublished data) Гауга 7 + (2D TLC) +/— Crins et al. (1988b) Madi 20 8 ? + Bohm et al. (unpublished data) Raillardella 5 1 + (2D TLC) none Crins & Bohm (unpublished data) Wilkesia 2 2 ? + Bohm & Fong (unpublished data) herbivore defense compounds, or as factors in the CARLQUIST, S. 1959. Vegetative anatomy of Dubautia, floral biology of the tarweeds. For example, Du- sla ele dui Wilkesia (Compositae). Pacific bautia contains species in virtually all of the avail- Eu C x 1985 po he Ми: able habitats in the Hawaiian Islands, and flavonoid dinae (Astera ае): Argy тори Dubautia, and aglycone diversity is correlated with habitat as well Wilkesia. Mee 4: 1- as the age of individual islands. This system might ‚ KyHos. 1981. ш radiation in provide an excellent opportunity to сата ihe the Най silversword alliance (Compositae-Ma- lati ; for UV s : iara diinae). I. Cytogenetics of spontaneous hybrids. Evo- relative requirements for screening, with species иий 35: 54 3-556 in mesic habitats presumably requiring far less pro- Carr, R. L. & С. D. CARR. 1983. Chromosomal races tection than species in xeric environments. The and struc d heterozygosity d ciliosa relative effectiveness of different flavonoids in UV В Greene (Astera aci Amer. J. Bot. 70: 744-755. M Pando a CrAUsEN, J., D. D. Keck & W. M. Hersey. 1941. Sc СТБ очата з : саа oque Carnegie Inst. Wash. Year Experiments addressing the question of why such Book 40: 160-170. diverse arrays of compounds are produced by some Crins, W. J. & B. A. Вонм. 1987. Flavonoid aglycones species are also needed. Any study of coevolution of Holocarpha obconica. Phytochemistry 20: 2128- 9129 that attempts to inc rate the role of flavonoids | ние оаа E uM = & —— 1988a. Flavonoids of Adeno- A кош 899 * in ограп1ып» with wc thamnus A d 27: 2647-2649. the tarweeds interact. Flavonoids (at least agly- ———— & 988b. Chemotaxonomy and its cones) probably contribute to antiherbivore defen- Me кс to Muri hypotheses in Holocar- ses, just as other secreted natural products, such ав = liantheae: Madiinae). Canad. Bot. р 2097 cand ahh s. do Ён & /Canad. S c. Pl. Physiol. Joint Meeting, 5-9 аз Оеп мө bayo ate iud s 1988, Victoria British Columbia. TORT Rodriguez, 1983), but no conclusive experimental ARR. 1988a. Flavonoids evidence is available. There is also a lack of data as ae 'ators of hybridization d in mixed population of about natural herbivores, parasites, and pathogens lava-colonizing Hawaiian tarweeds (Asteraceae: He- at татуеёйв liantheae: bera Syst. Bot. 13: 567-571 s & B. A. Boum. 1988b. Non-polar vhenolics Ta seven T of Layia (Asteraceae). Biochem. Syst. Ecol. 16: 467-469 LITERATURE CITED „т dE W. J. Crins, В. A. не Н. N.T Emerson, J. K., R. L. Carr, S. McCormick & В. А. Mass spectrometry in the e dieran ч Вонм. 1986. 8-О- Methylated flavones from Са- "bs anones and fihydrofiavonols. p e erar lycadenia ciliosa (Compositae): inter- am nar dis 2715-2717 lational variation. Biochem. Syst. Ecol. 9-32. Volume 77, Number 1 1990 Crins 8 Bohm 83 Flavonoid Diversity of Tarweeds FELSENSTEIN, J. P 1 ( 1985. PHYLIP— Phylogeny Inference ge (version 2.8). Univ. of Washington, Seattle, ashington. Guam, R. J. & B. A. Boum. 1978. Angiosperm 353- flavonoid evolution: a reappraisal. Syst. Bot. 3: 68. 987. ish a dard od and iso- ations in Layia -15. GOTTLIEB, L. D. 1 citrate dehydrogenase gene a a Amer. HELLER, W. 1986. Пека biosynthesis, an overview. ‚ 25-42 in V. Cody, E. Middleton & J. В. Har- Бове (editors), Plant Flavonoids in Biology and Med- icine: Biochemical, Pharmacological, and Structure- Activity Relationships. Alan R. Liss, Inc., New York. Luca, Н. Kouni, L. LATCHINIAN, 1987. Enzymology d compartmentation of polymethylated flavonol glucosides in Chrysosplenium americanum. Phyto- 1982. Ecological and evolutionary pat- terns in Hobo ocarpha (Compositae, Madiinae). Ph.D. Disseration. Univ. of California, Davis, California. Ркокѕсн, P. & E. RODRIGUEZ. 1983. Chromenes and benzofurans of the Asteraceae, their chemistry and biological significance. Phytochemistry 22: 2335- 2348. ‚ Н. Bupzikiewicz, В. D. Tanowitz & D. М. SMITH. 1984. Flavonoids from the external leaf resins of four Hemizonia species (Asteraceae). Phy- tochemistry 23: 679-680. TaNowrrz, B. D. Taxonomy of Hemizonia sect. Madiomeris (Asteraceae: Madiinae). Syst. Bot. 7: те. 1987. R, Р. N. Ross & P. PROKSCH. 1985. Genetic ¡ m 36-1241 1988. od radiation and genetic differentiation in the aiian silver- sword alliance (Compositae- Madiinae). Evolutiol 42: 1278-1287 BIODIVERSITY AND CYTOGENETICS OF THE TARWEEDS (ASTERACEAE: HELIANTHEAE-MADIINAE) Donald W. Kyhos,* Gerald D. Carr,? and Bruce G. Baldwin? ABSTRACT Because the tarweeds (Madiinae) attracted the attention of early pioneers in the areas of са апа biosys- tematics, this агарган ада of plants has been studied more intensively than most comparably sized gro comprise 127 tarweeds is extrem species in 17 e. The Hawaiian members genera, with centers of diversity in California and Haw alone include herbaceous mat-for s. The tarweeds ап. The biolozical ieee of the rming plants, cushion plants, monocarpic and polycarpic еа trees, and lianas. Collectively, the group includes self-compatible and self- пас annuals a perennials. An а conspicuous component of the dive and 34 are represented by one or more taxa. pe aun: of 3 uch that gametic number n = 4-14, 16, 17, 24 this variation suggests Det the ancestral chromosome bar in the Madii Madiinae are summarized. Amon and nine intergeneric hybrid combinations known in the 'rsification of tarweeds has been chr ” inae is n = 7. e numierous infrageneric the latter hybrids, one indicates a close relationship between Raillardiopsis muirii and Madia bolanderi, and another establishes a genetic link between the mainland and Hawaiian representatives of the group. The tarweed subtribe (Madiinae) of the sunflower tribe Heliantheae has been subjected to intensive biosystematic investigations for over 75 years. The tarweeds first captured the attention of the inno- vative experimental systematist H. M. Hall in 1912. In 1915, Hall enlisted the efforts of E. B. Babcock in a comprehensive treatment of the hay-field tar- weeds (Babcock & Hall, 1924). The early tarweed studies were carried out at the University of Cal- ifornia, but Hall joined the Carnegie Institution of Washington in 1919, and the center of tarweed research eventually shifted to Stanford. Before his death, Hall was able to welcome Jens Clausen to the growing staff of the Division of Plant Biology of the Carnegie Institution at Stanford. Clausen, working with D. . M. Hiesey, ini- tiated a series of extensive investigations into the biology of the dozen or so genera of tarweeds they recognized. These investigations continued to pro- duce tremendously valuable and voluminous data over the next two decades. Most of this information is summarized in Clausen's Stages in the Evolution of Plant Species (1951 Although little biosystematic research on the tarweeds was conducted in the 20 years following — the peak of activity by Clausen, Keck, and Hiesey, another researcher, Sherwin Carlquist, began to study tarweeds from а different perspective. Carlquist's studies emphasized. anatomical obser- vations that led him to interpret the tarweeds as a rather closely knit group that includes a number of genera not previously considered to belong to the Madiinae, i.e., Raillardella (including Rail- piu from western North America and Ár- gyroxi m, Dubautia, and Wilkesia from Ha- хап (Carlquist, 1959). Carlquist’s position on the last-mentioned three genera is of special signifi- cance because Keck (1936a, b) had specifically denied any close relationship between the Hawaiian genera and the Pacific coast tarweeds. Carlquist further stimulated interest in the Hawaiian genera in his Hawaii, A Natural History (1970). Beginning about 20 years ago, biosystematic interest in the Madiinae was again on the rise. Research during this period produced a consider- able amount of new biosystematic information, par- ticularly in the Hawaiian genera and in the main- land genera Hemizonia, Holocarpha, Lagophylla, and Layia (e.g., Carr, 1975a, b, 1977, 1985a; Carr & Carr, 1983; Carr & Kyhos, 1981, 1986; Gottlieb & Ford, 1987; Palmer, 1982; Tanowitz, 1977, 1978, 1982, 1985; Tanowitz & Adams, 1987; Thompson, 1983). The purpose of this paper is to survey briefly the di- Calycadenia, versity of the Madiinae, t and integrate existing biosystematic information (some from un- published dissertations), and to provide a preview of the results of ongoing biosystematic efforts within ! This work has been supported in part by NSF grant BSR 8615046 to Ж + Department of Botany, University of California, Davis, California 95616 ile ‘Department of Botany, University of Hawaii, 3190 Ma Way, Honolulu, nee 96822, U.S.A. ANN. Missouni Bor. Garb. 77: 84-95. 1990. Volume 77, Number 1 Kyhos et 1990 85 Tarweed Biaversit and Cytogenetics TABLE 1. Characteristics of the genera of Madiinae. Numbers Genera of species Haploid chromosome numbers! Descriptions? PERENNIAL Adenothamnus 1 14 Small evergreen subshrubs, SI Argyroxiphium 5 14 Polycarpic and monocarpic rosette shrubs, Dubautia 21 13(9), 14(12) Subherbaceous mats, cushions, shrubs, trees, lianas, SI, SC? Holozonia 1 14 Mesophytic herbs with fleshy rhizomes, SI Raillardella 3 17(1), 17, 1821), 34(1) Mat-forming or spreading scapose rhizoma- tous herbs of high elevations, Raillardiopsis 2 7(1), 8(1) Tufted to mat-forming rhizomatous herbs of high elevations, SI, SC? Wilkesia 2 14 Polycarpic and monocarpic rosette shrubs, SI TRANSITIONAL* Hemizonia 6P, 27A 94), 10(2), 11(4), 12(11), 13(3), Mostly late-flowering xerophytic herbs, the 14(7) perennials subshrubby or rhizomatous, SI Madia 2P,18A 6(1), 7(1), 8(6), 9(3), 14(1), Mesophytic to xerophytic herbs, the perenni- 16(5), 16, 24(1), 24(1) als with woody rhizomes, SC, S ANNUAL Achyrachaena 1 8 Small vernal mesophytic herbs, SC Blepharipappus 1 8 Small xerophytic herbs, SI Blepharizonia 1 4 Stout, xerophytic, late-flowering herbs, SI? Calycadenia 11 4(1), 5, 6(1), 6(4), 7(5) Mostly late-flowering xerophytic herbs, Sl, SC Holocarpha 4 4(1), 4, 52(1), 4?, 6(1), 6(1) Mostly late-flowering xerophytic herbs, SI Lagophylla 4 7 Mesophytic to xerophytic herbs, SI, SC Layia 16 7(6), 8(8), 16(1) Vernal herbs, SI, SC Osmadenia 1 9 Highly branched herbs, SI ' The information on chromosome numbers is a compilation of those given by Carlquist, 1959; Carr, 1975a, b, 1977, 1978, 1985; Clausen, 1951; Clausen et al., 1933, 1934, 1935, 1936, 1937, 1940, 1941, 1945; 1932, Johansen, 1933; Keck, 1949, 1958, 1959; Kyhos (in Carlquist, 1959) Strother (in iip des 1978), 1983; Tanowitz, 1982; Venkatesh, 1958; and previously unreported counts of п = 12 for Hemizonia frutescens A. Gray based on root tip a of pa material collected from a Island, Mxico by Bruce Baldwin (Baldwin & Beauchamp 688 in DAV), a = 9 for Madia doris-nilesiae T. W. Nelso elson from bud material collected from Trinity саш, California, by Barbara Williams (Wi illiams 518). In genera pc Eod is variation in chromosome minbera, the number of species with a particular chromosome constitu tion is giv n parentheses 1 = self- ое си SC? = self-compatible ог partially so; SI = self-i eaters SI? = self -incompatible or partially so T he two > genera in this category contain perennial (P) and annual (A) species as indicated in the second column. this fascinating group of plants. Details of other current research on the tarweeds are summarized elsewhere (e.g., flavonoid chemistry—Crins & Bohm, this volume; enzyme electrophoresis— Wit- ter, this volume; physiological ecology—Robi- chaux et al., this volume; chloroplast DNA evo- lution—Baldwin et al., this volume). TARWEED CHARACTERISTICS The name tarweed refers to the often copious sticky glandular secretions produced on the sur- faces of these plants, especially in the regions of the capitulescences. As viewed herein, the tarweed group (Madiinae) comprise 127 species in 17 gen- era, six of which are monotypic (Table 1). Seven genera are wholly perennial, eight are wholly an- nual, and two include annual and perennial species. While all of the 28 Hawaiian species are perennial, only 15 perennials are found among the 99 species from the Pacific coast of the Americas. The annual species of mainland tarweeds range from about 2 cm in Madia minima (A. Gray) Keck to 2.5 m tall in M. elegans D. Don ex Lindley subsp. densifolia (E. Greene) Keck (Keck, 1959). The perennial mainland taxa comprise small ev- ergreen or deciduous subshrubs and herbaceous rhizomatous taxa, some of which form extensive mats (e.g., Raillardella species). In contrast, the Annals of the Missouri Botanical Garden Hawaiian taxa are exceedingly diverse and include a mat-forming subshrub, Dubautia scabra (DC.) Keck; a cushion plant, D. waialealae Rock; mono- carpic and polycarpic rosette shrubs, e.g., Argy- roxiphium sandwicense DC. subsp. macrocepha- lum (A. Gray) Meyrat and Wilkesia hobdyi Н. St. John; woody shrubs, e.g., D. plantaginea Gau- dich.; trees, e.g., D. id ran (Sherff) Keck; and a liana, D. latifolia (A. Gray) Keck (Carr, 1985a). Although there are some exceptions, the leaves М of mainland tarweeds аге typically narrow and somewhat linear, often grasslike in general ap- pearance. The Hawaiian assemblage includes this type but in many species the leaves are much broader (Carr et al., 1989 quently exhibit a somewhat parallel orientation that he leaf veins fre- becomes extreme and exceedingly grasslike in Wil- kesia species. Howev latifolia exhibit a highly isodiametrically reticulate er, the leaves of Dubautia venation that contrasts sharply with the situation in Wilkesia (Carr, 1985a). In general, the leaves of vernally flowering mainland taxa such as Layia or plants of comparatively wet habitats in Hawaii are larger and more mesomorphic compared to those of late-flowering mainland taxa such as Holo- carpha or Hawaiian taxa occurring in dry habitats (Carr et al., 1989). [n surface area the leaves range from about 0.1 to 75 ст“. The flowering heads of tarweeds range from small and inconspicuous (4 mm high and less than 2 mm broad) in Calycadenia hooveri G. Carr and Dubautia pauciflorula H. St. John & G. Carr, with two or three florets per capitulum, to large and showy (35 mm tall and 60 mm broad) in prse RN validus (Brandegee) Keck and Ar- gyroxiphium sandwicense subsp. macroceph- alum, the latter with up to 650 florets per capit- ulum. Almost all of the mainland taxa have radiate flowering heads, whereas in Hawaii, rays are found in the heads of only five of the 28 species. Flower colors are white, red, orange, yellow, or combi- nations thereof. he mainland tarweeds occur almost exclusively in the western U.S., with their distributional center in the central valley of California. A few species extend into Mexico, and a few others are restricted to the Mexican mainland or offshore islands. Two highly disjunct species are found in South America (Chile and Argentina). The mainland perennial species are mostly quite rare and restricted in dis- tribution, suggesting that they may be relicts of an ancestral plexus. A second center of diversity of tarweeds is the Hawaiian archipelago, where representatives of the group occur on all of the major islands including Hawaii, Maui, Lanai, Molokai, Oahu, and Kauai. All of the tarweed species in the Hawaiian Islands are included in the Hawaiian endemic genera Ar- gyroxiphium, Dubautia, and Wilkesia. The mainland tarweeds are at home in lowland, sometimes quite xeric habitats; however, species are found above timberline (e.g., Raillar- della argentea); and Madia bolanderi (A. Gray) A. Gray typically occurs in very wet mid-elevation habitats. The Hawaiian representatives occur in tremendously diverse habitats, including recent lava flows, cinder cones, dry scrub, dry forests, mesic forests, rainforests, and bogs. Annual precipitation in these habitats ranges from about 35 cm to over 12 m. Hawaiian tarweed sites are found from near sea level to about 3,750 m elevation. some The annual mainland species generally germi- nate during the period of winter rains, persisting as rosettes until bolting and flowering occurs. In many species this takes place in late summer or fall, and it is not uncommon to find tarweeds flow- ering during October and November, or even De- cember, before the winter rains of the following season. At the other extreme, Achyrachaena mol- lis Schauer and most species of Layia, among others, are vernal in their growth and flowering responses. The perennials are less predictable, al- though even in Hawaii they are mostly seasonal in their flowering, with summer flowering most com- mon. A few species, such as Dubautia knudsenii Hillebrand, flower more or less continuously throughout the year. As a result of the long period of intensive biosys- tematic research on tarweeds, numbers are known for 122 (over 96%) of the 127 species. The only species for which no chro- mosome information is available are Argyroxiph- ium virescens Hillebrand (presumed extinct), Hemizonia martirensis Keck. H. streetsii A. Gray, Layia ziegleri Munz, and Madia stebbinsii T. W. Nelson & J. P. Nelson. Calycadenia, Madia, and Hemizonia are the most diverse in chromosome number, and together with Raillardella, include all of the different numbers found in the entire subtribe (Table 1). Eight of the genera have only a single confirmed chromosome number. Among the perennials, chromosome numbers are n = 6, 7,8, 12, 13, 14, 17, and 34, whereas the annuals are somewhat more diverse in chromosome num- bers with n = 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 16, and 24 (Tables 1, 2). Some species have more than one chromosome number, such as Ca- lycadenia pauciflora A. Gray (п = 5, Ө; Carr, 1975b) and Madia gracilis (Smith) Keck (n = 16, 24; Clausen et al., 1945). Mixed chromosome Volume 77, Number 1 1990 Kyhos et al. Tarweed Biodiversity and Cytogenetics TABLE 2. Distribution of Madiinae taxa in relation to gametic chromosome numbers.' п = 4 5 6 7 8 9 10 11 12 13 14 15 16 17 24 34 P Р ALL SPECIES P A A A A A A A A A 18PH A A P Р А А Р Р А А Р А А А А Р А Р А А А Р А А А А А А А А А А А А А А А А А А А А А 9ph A A A A A A A A A P A A A A A A A A A A A A A Р А А А А А А A A А А А А А Р А {ТР 3PH ALL GENERA? P P P AP AP P AP A A AP AP A A A A AP ph AP A A А А A A A AP AP AP AP A AP P А ME PERENNIAL 3PH GENERA? AP ph P AP Р P AP AP Р Р Р п = 4 5 6 7 8 9 10 11 12 13 14 15 16 17 24 34 derived ual, Р = perennial, AP = both annual and p PH = perennial Hawaiian taxon, ph = sie entries represent o zonia. than perennial Hawaiian taxon; boldfac ? Several genera are represented in one catego gor * Includes perennial component of uie with both annual ra perennial species. Some genera are represented in ry. more than one catego numbers showing apparent polyploid relationships within genera have turned up in Layia, Madia, and Raillardella. One report of n — 18 for Raillardella argentea A. Gray from San Gorgonio Pass in southern Cal- ifornia (Kyhos, in Carlquist, 1959) differs from those determined for this species from other lo- cations in the Sierra Nevada (n = 17; Strother, 1983; n — 17; Kyhos, unpublished) and represents a number not otherwise known in this genus. The erroneous report of n — 16 by Powell & Powell (1978) for Raillardiopsis muirii (A. Gray) Rydb. (as Raillardella muirii) was apparently based on a miscommunication of the much earlier deter- mination of 2n — 16 by Kyhos (in Carlquist, 1959). 18 for Raillardella argentea by the same authors was apparently based Similarly, the report of n — on the earlier determination for this taxon by Kyhos (in Carlquist, 1959). Powell & Powell (1978) pro- vided the locality and voucher information for the material оп which Kyhos's determinations were made. Idiograms depicting n = 5 in Holocarpha virgata (A. Gray) Keck and n = 4 in Н. obconica (Clausen & Keck) Keck by Clausen (1951) rep- resent numbers not otherwise known for these species and need further documentation. Very few plant groups of comparable size are better known cytologically. An examination of the pattern of distribution of the 122 species in Ma- dünae with known chromosome numbers reveals very prominent modes at n = 7 and n = 8, lesser modes at n — 12 and n — 14 (if one deemphasizes the monophyletic Hawaiian taxa), and minor peaks at n — 6, 9, and 16 (Table 2, top). This extensive and somewhat complex distribution pattern is open to various interpretations. Annals of the Missouri Botanical Garden The pattern is simpler and less ambiguous if the chromosome number variation is considered at the generic level, where the several modes observed at the species level condense to a relatively un- complicated pattern (Table 2, middle). This sim- plification is due in large measure to elimination of the influence of single-chromosome-number clus- ters of closely related species. This perspective at the generic level reveals more clearly how chro- mosome numbers correlate with major units of morphological diversity and therefore probably provides a more accurate evolutionary and system- atic understanding of the Madiinae. t the generic level the prominent modes at n = 7 and n = 8 are still present, and the only other well-developed mode is at n = 14 (Table 2, middle). This n = 14 mode consists of five quite distinct mainland genera in addition to three mutually closely related Hawaiian genera. In contrast, the n = mode, consisting of a small cluster of closely related annual autogamous species of Майа and Layia, represents much less diversity. Of the two genera at п = 13, one is Hawaiian (Dubautia) and consists of species known to be derived from members of Hawaiian Madiinae with n = 14 (Carr & Kyhos, 1981, 1986) and the other is Hemizonia, repre- sented by three fairly closely related species. With these three exceptions, each gametic chromosome number from n = 10 only a single genus Thus, the relatively simple picture that emerges upward is represented by at the generic level supports the view that the majority of the polyploid taxa within Madiinae are = 7 and that although n = 8 is well represented at the diploid level, its polyploid de- based on n rivatives are few. 7 is о greater antiquity in the subtribe and perhaps de- picts the ancestral chromosome number for Ma- This suggests that n = diinae. When Hemizonia is omitted from the total array of chromosome numbers (Table 2) a very large gap appears in the overall distribution of chro- mosome numbers, such that no taxa are repre- sented at n = 10, 11, 12, and 13 (except the Hawaiian derivatives). All of the remaining species or genera fall into two quite discrete groups, one consisting of apparent diploids with gametic chro- mosome numbers ranging continuously from n = 4 ton = 9 and a second group composed of what appear to be polyploid taxa with gametic chro- mosome numbers of n = (13), 14, 16, 17, 24, and 34 (Table 2). With its 33 species, Hemizonia is the largest genus in Madiinae, making up almost 26% of the subtribe. On this basis alone, one might predict that its removal is likely to have a pro- nounced effect on the overall pattern of chromo- some number distribution in Madiinae. Neverthe- less, it seems quite significant that the removal of no other single genus or even any combination of genera cleaves the rest of the tribe into separate diploid and polyploid groups. Hemizonia includes a continuous series of gametic chromosome num- bers ranging from n = 9 to n = 14, and it modally centers at л = 12. This may indicate that the genus is primitively polyploid and that gametic chromosome numbers above and below n = 12 represent aneuploid derivatives from this polyploid mode. The distribution pattern of gametic chromosome numbers of the perennial genera within Madiinae is quite similar to that of the entire subtribe (Table 2, bottom). When the perennial taxa are examined at either the specific or generic level, a small mode appears at п = 7 and a more prominent mode is present at п = 14. If one accepts the reasonable assumption that the ancestral taxa to the tribe are more likely to have been perennials, it seems par- ticularly significant that the chromosome number distribution of the extant perennials conforms to the previous pattern observed for the entire sub- tribe in that the polyploids are concentrated at n = 14 and thus appear to be based principally on n = 7, not on л = 8. This interpretation is supported further by the total absence of perennial taxa with n = 16. On the other hand, the possibility cannot be dismissed that the prominence of n = 14 taxa within the subtribe may at least in some cases be the result of the combination of diploid species with п = 6 and п = 8 to produce polyploids at п = This possibility is strengthened by preliminary evi- dence obtained from hybridization experiments in- volving the Hawaiian and mainland Madiinae (Bald- win, 1989; Baldwin et al., unpublished). INFRAGENERIC CYTOGENETIC RELATIONSHIPS With the exceptions of Wilkesia and monospe- cific genera, one or more interspecific hybrids have been produced in all of the tarweed genera. The cytogenetic information on these hybrids is sum- marized below according to the sequence of genera in Table 1 GENERA WITH PERENNIAL SPECIES ONLY Argyroxiphium. five perennial species, all with n = 14 (Table 1). Argyroxiphium comprises Only the interspecific combination 4. grayanum (Hillebrand) Degener x A. sandwicense has been produced (Carr sandwicense subsp. & Kyhos, 1986). The two F, individuals available for analysis Volume 77, Number 1 1990 Kyhos et al. 89 Tarweed Biodiversity and Cytogenetics averaged 63% pollen stainability and consistently produced 12 pairs and one chain or ring of four chromosomes at meiosis, indicating heterozygosity for a single reciprocal chromosome translocation. In spite of its heterozygosity, this hybrid functioned well as a male and as a female parent in the production of an intergeneric recombinant involv- ing Dubautia linearis (Gaudich.) Keck. Self-in- compatibility has been demonstrated in A. sand- wicense (Carr et al., 1986). Dubautia. The wholly perennial genus Du- bautia includes nine species with n = 13 and 12 species with n = 14 (Table 1). A total of 41 different interspecific hybrid combinations, 11 of them oc- curring in nature, have been available for study (cf. Carr & Kyhos, 1986). Chromosome structural stability among the 13-paired species of the genus is indicated by the consistent meiotic display of 13 pairs of chromosomes and high pollen stainability (96% or greater) of all interspecific hybrid com- binations involving these species. The close rela- tionship between the genome of D. scabra (n = 14) and the 13-paired genome is demonstrated by consistent production of 12 pairs and one chain of three chromosomes during meiosis and the high pollen stainability averaging 81% in 29 individuals of seven different interspecific combinations in- volving D. scabra and 13-paired species. Cyto- genetic, biogeographic, and morphological evi- dence, as well as information from isozyme studies (Witter, this volume; Witter & Carr, 1988) sup- port the conclusion that the 13-paired species rep- resent a monophyletic group comparatively re- cently derived via aneuploid reduction of chromosomes from the n — 14 genome possessed by Dubautia scabra. Unlike the 13-paired genome, the 14-paired enome is not structurally uniform. Biosystematic studies have so far revealed four п = 1 mosome structural cytotypes comprising one or more highly morphologically differentiated species that differ from one another by a minimum of one or two reciprocal translocations, as indicated by the meiotic configurations in different interspecific hybrid combinations. Pollen stainability in these hybrids ranges from 27% to 42% and appears to be determined largely by the complexity and be- havior of the multiple chromosome associations. А single univalent chromosome resulting from chias- ma failure in one of the chromosome multiples may be seen in a low frequency of meiocytes in some of these hybrids. The cytogenetic evidence indi- cates that a limited number of chromosomes have been repeatedly involved in separate translocation events to produce the observed spectrum of cy- totypes. It has been possible to create a number of hybrid products beyond the F, generation. In one instance, two interspecific hybrids involving species with the same chromosome structural arrangement were crossed to produce a hybrid product combining the four species Dubautia knudsenii, D. laxa Hook. & Arn., D. microcephala Skottsb., and D. plan- taginea. In another example, three chromosomally differentiated species, D. knudsenii (n = 14), D. latifolia (n — 14), and D. sherffiana Fosb. (n — 13) were combined in two successive hybridiza- tions. Clearly, results to date suggest that it is possible to hybridize and potentially to recombine virtually any combination of species in the genus regardless of their exceedingly diverse morpholog- ical and chromosomal attributes. Self-incompati- bility and self-compatibility occur in the genus (Carr et al., 1986). Raillardella. "Three perennial species com- prise the genus Raillardella (Table 1). Two in- terspecific hybrid combinations, R. pringlei E. Greene (n — 17) x R. scaposa A. Gray (n — 34) and R. pringlei x R. argentea (A. Gray) A. Gray (n — 17), have been produced (Baldwin, 1989). At meiosis the first hybrid exhibited a mixture of univalents, bivalents, trivalents, and possibly larger chromosome multiples, whereas the second hybrid exhibited only normal meiotic bivalents. Experi- ence with cultivated plants suggests that all three species in the genus are self-incompatible. Raillardiopsis. | Raillardiopsis consists of two perennial species. One hybrid between R. тийїї (n — 8) and R. scabrida (Eastw.) Rydb. (n — 7) exhibited mostly univalents and chains of chro- mosomes at meiosis. The spectrum of multiple as- sociations observed in the hybrid indicates that the parental ноев але differentiates P a ka three 89). Stain- able pollen grains in this hybrid i A were mostly four-colporate, suggesting a derivation from un reduced meiocytes. Experimental manipulations in- dicate that R. muirii is self-incompatible (Baldwin & Kyhos, in press) and suggest that R. scabrida may have limited self-compatibility. GENERA WITH PERENNIAL AND ANNUAL SPECIES Hemizonia. The largest genus of Madiinae is Hemizonia with 27 annual and six perennial species (Table 1). The perennial species are all highly re- stricted in distribution and most are quite rare. They occur principally on the offshore islands of Annals of the Missouri Botanical Garden southern California and Baja California Norte, al- though Н. minthornii Jepson and Н. perennis (К. Greene) Keck are confined to the mainland in coastal and near-coastal habitats. The otherwise insular Н. greeneana Rose has a narrowly restricted exten- sion on the mainland of Baja California at Punta Banda. Nearly all of the annual species of Hemi- zonia are confined to the mainland of California, with a few extending into the adjacent western states and Mexico. In contrast to the character- istically rare and restricted perennial species, which appear to be relictual, many of the annual species of Hemizonia are quite common, relatively wide- spread, and thus apparently vigorous, evolved entities. Most of what is known biosystematically about Hemizonia has been derived from the extensive research conducted by Clausen, Keck, and Hiesey in the first half of this century (Clausen, 1951). Subsequently, Tanowitz (1977, 1978, 1982, 1985) generated additional biosystematic information for recently the genus. The species of Hemizonia appear to form four rather natural species groups or sections, which have been given various names by different taxonomists. Section Hemizonia (formerly sect. Euhemizonia) consists of seven species, each with 14 pairs of chromosomes, which sets them apart from all other species in the genus. The Carnegie research team found that these seven species could all be intercrossed easily and that their hybrids represent the most fertile interspecific combina- Section Centromadia, monly known as the spikeweeds, is an obviously tions in Hemizonia. com- natural group of four or so species that can all be intercrossed readily to produce hybrids of moderate fertility in most cases. The Carnegie research team studied four annual species of this section. These few species were found to possess a surprisingly extensive series of chromosome numbers (n = 9, 11, 12, a H. perennis, which apparently was unknown to d 13). One extremely rare spikeweed, Clausen's research group, is a rhizomatous peren- nial and has subsequently been investigated ex- perimentally. It has 13 pairs of chromosomes and the ability to cross with the annual spikeweeds to produce vigorous hybrids generally with rather low fertility (Kyhos, unpublished). The third species group within Hemizonia consists of all the re- maining annual species. These have gametic chro- mosome numbers that range continuously from n to n = 13. The fourth species group consists of five of the six perennial species (excluding Н. perennis) discussed above. This group is charac- terized by the chromosome number п = The view that emerges from these experimental studies is that most interspecific crosses within this genus produce hybrids with low to extremely low fertility (see fig. 62, Clausen, 1951). Interspecific hybrids with normal to near normal fertility are exceptional. Often the reduced fertility appears to result from chromosome differences between the species, but in some instances genetic factors are likely to be responsible. Almost half of the more than 55 interspecific combinations produced by the 'arnegie research effort were highly sterile hybrids. ~ Most of the remaining hybrid combinations had substantially depressed fertility. Intersectional hy- brids typically were extremely sterile, with but a single exception involving H. pungens (Hook. & Arn.) Torrey & A. Gray and H. ramosissima Benth. (H. fasciculata). Some 20 interspecific hybrid combinations attempted by the Carnegie research team failed. Thus experimental evidence reveals strong reproductive barriers between most Hemi- zonia species. A high degree of chromosome repatterning with- in Hemizonia is clearly indicated by an extensive series of chromosome numbers, which range con- tinuously from n = 9 to n = 14. The annual species show the greatest diversity with their members spanning this entire series. The perennials are much more limited, with five of their six species having ı = 12, and the remaining species, H. perennis, with n = 13. Unquestionably, chromosome evo- lution has played a major role in the evolution of the genus. Madia. Madia includes some 20 species (18 annuals). The modal chromosome number is n = 8, but numbers of n = 9, 14, 16, and 24 are also found among the SE species (Table 1). The two rhizomatous perennial species, M. madioides (Nutt.) E. Greene (n = 7) and M. bolanderi (n = 6), are chromosomally unique within the genus. Аз with Layia, nearly all in-depth biosystematic knowledge of Madia stems from Clausen, Keck, and Hiesey (Clausen, 1951). They produced 22 interspecific hybrid combinations among the 14 species in their crossing program. Their research revealed that the diploid species generally are strongly isolated from one another reproductively; this isolation often seems to be associated with chromosome repatterning, as is evident from the diversity of chromosome numbers in Madia. Some polyploid species retain substantial ability to ex- change genes. GENERA WITH ANNUAL SPECIES ONLY This genus comprises 11 annual Calycadenia. species with chromosome numbers ranging from n Volume 77, Number 1 1990 Kyhos et al. 91 Tarweed Biodiversity and Cytogenetics = 4 to n = 7 (Table 1). Seventeen interspecific hybrid combinations have been examined (Carr, 1975a, 1977). Hybrids among the seven-paired species С. mollis A. Gray, С. truncata DC., and С. villosa DC. are very sterile, exhibiting less than 1% stainable pollen grains, and are characterized by very low meiotic pairing of chromosomes. The hybrid between the two seven-paired species C. villosa and C. hooveri also has less than 1% stain- able pollen but often produces five pairs and one ring or chain of four chromosomes, indicating het- erozygosity for a reciprocal chromosome translo- cation. This hybrid was also heterozygous for at least one paracentric inversion, as indicated by bridge and fragment configurations in 26% of the meiocytes. Hybrids among С. ciliosa E. Greene (л = 6), C. hispida (E. Greene) E. Greene (n = 6), С. multiglandulosa DC. (п = 6), С. pauciflora (п = 5, 6), C. oppositifolia (E. Greene) E. Greene (n = 7), and C. spicata (E. Greene) E. Greene (n = 4) are characterized by strict meiotic bivalent formation or, more frequently, large multiple as- sociations of chromosomes, indicating structural heterozygosity for three or more reciprocal chro- mosome translocations in some cases. Pollen stain- ability in these hybrids ranged from less than 1% to 69%. A single hybrid between the seven-paired С. villosa and a six-paired representative of С. pauciflora had a pollen stainability of less than 1%. Based on a maximum meiotic association of оой of bridge and fragment configurations in 66% of the meio- cytes of this hybrid, it was on for at least three reciprocal chromosome translocations and nine chromosomes and the a probably multiple paracentric inversions. Two closely related species complexes, Caly- cadenia hispida-C. multiglandulosa and C. cilio- sa—-C. pauciflora, exhibit very different modes of evolution. The former complex is made up of five morphologically differentiated but chromosomally stable and highly interfertile taxa, while the latter complex comprises at least eleven mostly somewhat morphologically cryptic chromosome races differ- entiated from one another by reciprocal chromo- some translocations and in some cases at least one pericentric inversion (Carr, 1975b, 1977; Carr & Carr, here appear to be no significant differences in reproductive biology, numbers of chromosome mutations detected in field popula- tions, habitats, flowering times, or chiasma fre- quencies that might provide a basis for the very different modes of evolution observed in these two species complexes. These differences may be at- tributable to the importance of gene position (pat- tern) effects and/or gene linkage effects in pro- ducing optimum genomic arrangements in these plants (Kyhos & Carr, unpublishe Over 500 interpopulational hybrids in more than 300 progenies have led to the recognition of four homoploid races (n — 6) and one heteroploid race (n = 5) in Calycadenia pauciflora and five homo- ploid races in C. ciliosa (n = 6) (Carr, 1975b; Carr & Carr, 1983). Ongoing research with ad- ditional hybrids indicates further unresolved com- plexity in C. ciliosa and one additional race (n — ) of C. pauciflora (Carr & Carr, unpublished). The evidence suggests that a six-paired race of C. pauciflora arose from C. ciliosa, underwent ad- ditional chromosome repatterning, and ultimately produced the five-paired race by aneuploid reduc- tion. The research with Calycadenia ciliosa has re- vealed extremely high levels of chromosome struc- tural heterozygosity among individuals of three populations. In one population, 30% of the plants sampled were heterozygous for one or two trans- locations or a pericentric inversion. Їп another population, morphologically indistinguishable plants within of one another were differentiated y a minimum of four reciprocal chromosome translocations (Carr & Carr, 1983). Other current research reveals the presence of at least three structurally differentiated genomes in C. truncata. Extensive biosystematic manipu- lations with all species of the genus have demon- strated strict self-incompatibility in all but C. Aoo- veri and one population of C. truncata. Oo o> Holocarpha. Despite being a small genus of four annual species, the taxa of Holocarpha pres- ent a very complex situation in terms of chromo- some repatterning and reproductive isolation. This extreme complexity was first revealed by Clausen, Keck, and Hiesey in the 1930s and 1940s (Clau- sen, 1951). The initial indication of the complex- ities within this genus was the discovery that H. macradenia (DC.) E. Greene and H. virgata have a chromosome number of n = 4, whereas H. ob- conica and Н. heermannii (E. Greene) Keck have n = 6. It was further recognized that three of these species, exempting H. macradenia, displayed con- siderable intraspecific karyotypic variation, indic- ative of substantial chromosome evolution among and within the species. Clausen (1951) expressed the opinion that the rare H. macradenia had prob- ably already become extinct, which may be the reason it was not studied more thoroughly. However, subsequent research by Palmer (1982) revealed that H. macradenia, although near ex- tinction, survived in seven small, widely scattered Annals of the Missouri Botanical Garden populations in the San Francisco Bay and Monterey Bay regions. Five of these relict populations remain today. Palmer showed that the then-extant seven relict populations were cross-compatible, interfer- tile, and in possession of the same chromosome arrangement, with the exception of a single pop- ulation near Santa Cruz. This exceptional popu- lation differs from the others primarily by a chro- mosome translocation, with a second translocation also possibly present in some members of the pop- ulation. All tested individuals of H. macradenia are self-incompatible, as are the other three species of Holocarpha. Some populations of the chro- mosomally diverse H. virgata differ from the modal chromosome arrangement of H. macradenia by only a single translocation and thus produce in- terspecific hybrids with H. macradenia that are upward of 80% fertile. Palmer’s (1982) research also revealed a fairly close relationship between Н. macradenia (п = 4) and Н. heermannii (п = 6) in that their interspe- cific hybrid at meiosis forms a chromosome pairing configuration of a multiple of six and multiple of four. Similarly, Palmer turned up a close relation- ship between Н. macradenia and Н. obconica (п = 6), their interspecific hybrid producing a con- figuration of three pairs and a multiple of four chromosomes at meiosis. Lagophylla. This is a small genus of four an- nual species, all with a chromosome number of n = 7. Biosystematic investigations (Thompson, 1983) showed these species to be differentiated from one another by one or two reciprocal chromosome translocations and in some instances by inversions. This relatively modest amount of chromosome re- patterning seems to be responsible for the reduced fertility observed in all interspecific hybrids. Fer- tility in these hybrids modally ranges from values as low as 15% to 25% in some interspecific com- binations to as high as 40% to 60% in others. Lagophylla dichotoma Benth., L. glandulosa А. Gray, and L. minor (Keck) Keck are self-incom- patible, whereas L. ramosissima Nutt. readily selfs in cultivation and presumably in nature. Correlated with these breeding system differences, L. ramo- sissima failed to serve as a successful pollen parent in experimental hybridizations but did function quite well as a pistillate parent when care was taken to prevent selfing. This sort of unilateral interspecific incompatibility has been repeatedly observed be- tween self-incompatible and selfing taxa among members of Compositae. Species of Lagophylla appear to have the most specialized floral features of the entire subtribe. All populations of Lagophylla always possess 11 flo- rets per capitulum, five of which are ray florets, and the remaining six are disk florets arranged as a concentric ring of five florets encircling a single central disk floret. The ray florets are strictly pis- tillate, whereas the disk florets are only pollen fertile, never producing viable fruits. The flowers of Lagophylla species open in the morning and are typically closed by midday. Within the entire Madiinae this constellation of floral features occurs only in Lagophylla. Layia. This entirely annual genus contains some 16 species with chromosome number = 7, 8, and 16 (Table 1). Most of the en biosystematic knowledge of this genus was accu- mulated by Clausen, Keck, and Hiesey (Clausen, 1951). hybrid combinations among 13 of t They produced 24 different interspecific e 14 species in their crossing program. Although very little de- tailed information was presented, they attributed the broad spectrum of reduced fertilities of the hybrids to the presence of variable numbers of unpaired chromosomes. Tanowitz & Adams (1986) studied naturally oc- curring hybrids between the polytypic, obligately outcrossing Layia glandulosa (Hook.) Hook. & Arn. (n = 8) and the relatively uniform, self-com- patible L. paniculata Keck (п = 16). As expected, this triploid hybrid has a low fertility of less than 6% and appears incapable of exchanging genes with its two parental species. The modal meiotic chromosome configuration of the triploid hybrid is eight bivalents plus eight univalents, which was interpreted as evidence that the two parental species share a common genome. An alternative interpre- tation based on the possibility of autosyndetic pair- ing among the chromosomes of L. paniculata is also tenable. In-depth comparative investigations by Ford and Gottlieb involving research into the genetics and associated developmental processes of the rare ser- pentine endemic L. discoidea Keck and its prob- able ancestor L. glandulosa greatly extend the original research of Clausen, Keck, and Hiesey (Gottlieb & Ford, 1987; Ford & Gottlieb, 1989). INTERGENERIC CYTOGENETIC RELATIONSHIPS Argyroxiphium X Dubautia. Four different intergeneric F, hybrid combinations involving ge- nomes | and 2 of Argyroxiphium (n = 14) and genomes 1, 2 (both л = 14), and 5 (л = 13) of Dubautia have been analyzed (Carr & Kyhos, 1986). Meiosis in each of these hybrids is char- acterized by a very high frequency of normal chro- mosome pairing and one or more multiple chro- Volume 77, Number 1 1990 Kyhos et al. 93 Tarweed Biodiversity and Cytogenetics mosome associations. The results indicate that the parental genomes are differentiated by two or three translocations, and by aneuploidy in the hybrids involving Dubautia genome 5. The mean pollen stainabilities of these hybrids range from 11% to 29%. The least fertile of these, A. sandwicense subsp. macrocephalum (п = 14) x D. menziesii (A. Gray) Keck (n = 13), occurs in nature and produces backcrosses under field conditions. A syn- thetic intergeneric hybrid combining the genomes of D. knudsenii, D. laxa, and A. sandwicense subsp. macrocephalum was produced and subse- quently crossed with D. scabra to yield a hybrid product combining the genomes of four species (Carr & Kyhos, unpublished). In another instance, a hybrid between genomes 1 and 2 of Argyroxiph- ium (п = 14) was successfully crossed with genome 5 (п = 13) of Dubautia (Carr € Kyhos, 1986). Argyroxiphium х Wilkesia. Two hybrid com- binations have been produced (A. sandwicense subsp. macrocephalum X W. gymnoxiphium A. Gray and A. grayanum х W. hobdyi), but neither has yet furnished material for meiotic analysis. Calycadenia X Osmadenia. Osmadenia te- nella Nutt. has been successfully crossed with Ca- lycadenia truncata, С. mollis, and С. villosa (Carr, 1977). Observations of meiosis in the last-men- tioned two combinations demonstrate essentially complete lack of pairing of the parental genomes. The chromosomes of Osmadenia (n = 9) are much smaller than those of the seven-paired species of Calycadenia, and these size differences were very apparent at meiosis in the two intergeneric hybrid combinations examined. The hybrids exhibited less than 1% pollen stainability. Dubautia x Raillardiopsis. Recently, D. lae- vigata A. Gray (n = 14) was successfully crossed with R. muirii (п = 8) (Baldwin, 1989; Baldwin et al., unpublished). This represents the first hybrid produced between Hawaiian and mainland tar- weeds. The cross was easy to make and the hybrids are vigorous. Analysis of root tips of the hybrids indicates that 2n = 22 and verifies their parentage. Dubautia x Wilkesia. Hybrids between W. gymnoxiphium or W. hobdyi and Dubautia ge- nomes l and 4 have consistently yielded a high frequency of normal pairs and either two chains of three or one chain of six chromosomes (Carr & Kyhos, 1986). The mean pollen stainability is about 29%. One or occasionally two univalents may be found in cells exhibiting incomplete synapsis of the members of the chains of three chromosomes. Uni- valents of a similar origin were fairly commonly observed in a small sample of meiotic material from a hybrid between W. gymnoxiphium and D. herb- stobatae С. Carr (n = 13). Maximum chromosome associations in this hybrid indicate that at least two translocations differentiate the parents, as is the case in the previous examples. However, an ad- ditional chain of three chromosomes represents the aneuploid relationship of the parents in this in- stance. Pollen stainability of this hybrid was 16% (Carr & Kyhos, unpublished). Hemizonia х Holozonia. The very brief report by Clausen et al. (1937) of this intergeneric com- bination based оп Hemizonia arida Keck (n = 12) x Holozonia filipes (Hook. & Arn.) E. Greene (n — 14) appears to have been overlooked by most recent tarweed workers. Aside from noting that it is a remarkable combination involving an annual and a perennial species, Clausen et al. (1937, p. 212) did little more than characterize it as “а vigorous but sterile hybrid grown in 1937.” Layia х Madia. Two hybrid combinations have been made (Clausen, 1951). The first, M. elegans (n — 8) x L. platyglossa (Fischer & C. Meyer) A. Gray (n — 7), was extremely weak and apparently did not yield cytogenetic data. The sec- ond, M. sativa Molina (n — 16) x L. platyglossa (n = 7), although fairly vigorous, was highly sterile as a result of its being a triploid with 23 somatic chromosomes. Madia x Raillardiopsis. Only one combina- tion, M. bolanderi (n = 6) x К. muirii (n = 8) has been produced (Baldwin, 1989). Hybrids of this combination exhibit a mixture of chromosome pairs, multiples, and univalents during meiosis. The multiples observed indicate that the parental ge- nomes are differentiated by at least two reciprocal chromosome translocations. Pollen stainability of two individuals averaged 7%. Raillardella x Raillardiopsis. The single hy- brid combination produced, Raillardella pringlei (n = 17) x Raillardiopsis muirii (n = 8), exhib- ited mostly univalents at meiosis. Only a low fre- quency of cells yielded one or two pairs of chro- mosomes. The chromosomes fall roughly into two size classes with about eight (presumably those of Raillardiopsis muirii) in the large size class. The pollen stainability of this hybrid is 1% (Baldwin, 1989). DISCUSSION Infrageneric biosystematic and cyt tic stud ies indicate that variation in chromosome number and/or structure occurs in all ten of the genera 94 Annals of the Missouri Botanical Garden with more than one species that have been inves- tigated. It is clear that repatterning of chromo- somes has been a very significant aspect of the overall differentiation of the tarweeds, but chro- mosomal and morphological diversification have frequently proceeded at very different rates. In some cases, as in Calycadenia, chromosomal dif- ferentiation has been pervasive and has accrued even among forms that are otherwise exceedingly similar, if not indistinguishable (Carr, 1977; Carr & Carr, 1983). Шш involved in translocations їп Calyca- denia (Carr, 1975b; Carr € Carr, 1983) and Du- bautia (Carr & Kyhos, 1986). By comparison, the chromosomes of Lagophylla exhibit only modest infrageneric variation in structure (Thompson, 983). Overall, the kinds of chromosome evolution that have been documented within the Madiinae include aneuploidy, polyploidy, and structural dif- ferentiation by way of reciprocal translocations and pericentric and paracentric inversions. Nine different intergeneric hybrid combinations have been produced. Of the six that have been examined, pollen stainability ranges from less than 1% to 29%. Meiotic chromosome associations among the intergeneric hybrids range from essen- “ertain chromosomes have been tially all univalents in most meiocytes of Calyca- denia X Osmadenia to exclusively pairs and mul- tiples in most meiocytes of Argyroxiphium X Dubautia. There appears to be very little potential for intergeneric gene flow, especially under field conditions, except between the Hawaiian genera Argyroxiphium and Dubautia. The relatively high fertility among the Hawaiian taxa could be used to argue for their treatment as congeners. However, the three Hawaiian genera recognized here are easily distinguished and con- form well to generic concepts used elsewhere in the Asteraceae (Carr, 1985b). Moreover, the ar- gument used to reduce these taxa to a single genus could logically be extended to reduce them to a single species, ап extension that even the most strict adherents to the biological species concept would have to question. Two of the intergeneric hybrids briefly reported here establish a closeness of relationship not pre- viously perceived for the genera involved. Fur- thermore, the degree of chromosome association and fertility in the first of these hybrids, Майа х aillardiopsis, suggests that a taxonomic realign- ment may prove necessary. The second of these hybrids, Dubautia laevigata x Raillardiopsis muirii, establishes a direct genetic link between the mainland and Hawaiian tarweeds. Attempts to produce both hybrids were encouraged by results from a cpDNA study that indicated very close relationship among Madia bolanderi, Raillar- diopsis muirii, and the Hawaiian taxa (Baldwin, 1989; Baldwin et al., this volume). These hybrids provide the opportunity to explore the derivation of what may very well be the ancestral genomes of the entire assemblage of Hawaiian Madiinae. This may be an exciting new chapter in the rich history of biosystematic and other experimental research on the tarweeds. As Jens Clausen used to say, “We have only just begun to scratch the surface." LITERATURE CITED Bancock, E. B. & Н. M. Наш. 1924. Hemizonia congesta: a genetic, ecologic, and taxonomic study of the xr field tarweeds. Univ. Calif. Publ. Bot. 13: 15- BALDWIN, В. G. 1989. Chloroplast DNA Phylogenetics and Biosystematic Studies in Madiinae (Asteraceae) Ph.D. Dissertation. Univ. of California, Davis, Cali- fornia. W. Күноѕ. 1990. А systematic and geographic review of Raillardiopsis [Raillardel- la) muirii (Madiinae- Asteraceae), with special ref- isjunct California Coast Range popu- on. Madroño (in press IST, S. 9 on Madinae: anatomy, cytology, and evolutionary relationships. Aliso 4: 171- 230. 1970. Hawaii, a Natural History. Natural ену Press, New Yor CARR, G. D. 1975a. Calycadénia hooveri (Asteraceae), a new tarweed from California. Brittonia 27: 136- 1 975b. Chromosome evolution and aneuploid md in Calycadenia pauciflora (Asteraceae). Evolution 29: 681-699. 977. A cytological conspectus of the genu Calycadenia (Asteraceae): an exa ж ча сан modes of evolution. gy es J. Bot. 64: 694-703 1978 Chr romosome dieron к: Hawaiian па- f A in selected tive plants and taxa. Amer. J. Bot. 65: 236-2 1985а. А з. "m Hawaiian Ma- diinae (Asteraceae): pic a a Dubautia, and Wilkesia. Allertonia 4: 1-123. neris variation in the Hawaiian Ma- its relevance to generic con- cepts in the Compte Taxon 34: 22-25. ———— & D. W. Күноѕ. 1981. Adaptive radiation in the Hawaiian ese alliance (Compositae- Ma- diinae). I. Cytogenetics of spontaneous hybrids. Evo- lution 35: 543-5 Hali 1986. Adaptive оа. in the Hawaiian silversw ord alliance (Com ae-Madi- inae). П. ogenetics of artificial un ER hy- brids. Кейн in 959-976 o D. W. Күноз. 1986. Self- inc ompatibility і in ie Hawaiian Madiinae (Composi- tae): an iii to Baker's Rule. Evolution 40: 430- ча RoBicHAUX, M. S. WITTER & D. VE KYHos. дп Adaptive radiation of the Hawaiian Volume 77, Number 1 1990 Kyhos et al. Tarweed Biodiversity and Cytogenetics silversword alliance (Compositae-Madiinae): a com- ‚ V. Gid a үу. W. yore m p ei Speciatio on and the ipa Principle. Cambridge Univ. Press, Cam- bri Carr, К. L & G. D. CanR. 1983. Chromosome races "ib structural блар Ашы in Calycadenia ciliosa ene (Astera Amer. J. Bot. 70: 744-755 Nes J. 19 gritos in the Evolution of Plant pecies. Cornell Univ. Press, Ithaca, New York. ‚ D. D. Keck & W. M. Heusi. 1932. Exper- imental taxonomy. Carnegie Inst. Washington Year Book 31: 201-205 & 3. Experimental tax- onomy. Carnegie Inst. Washington Year Book 32: 192-196. & —— ——. 1934. Experimental tax- onomy. ee Inst. Washington Year Book 3 173- — ——— [W. M. Hiesey]. 1935. Experimental taxonomy. Carnegie Inst. Washington Year Book 34: 201-206. 1936. Experimental tax- onomy. oo Inst. Washington Year Book 35: 208-2 3 & ——— . Experimental tax- onomy. IM Inst. Washington Year Book 36: 209-214 З & ————. 1940. Experimental tax- onomy. Carnegie Inst. Washington Year Book 39: 158-163. Experimental tax- onomy. Carnegie Inst. Washington Year Book 40: 160-170. & 945. Experimental studies on the nature of species. II. Plant evolution through amphiploidy and autoploidy, with examples 2 m the Madiinae. Publ. Carnegie Inst. Wash. 564: -174 Eos W. J. 1978. Notes on two rare, endemic species from the Klamath Region of northern Cali- fornia, Phacelia dalesiana (Hydrophyllaceae) and ннн pringlei (Compositae). Madroño 2 | FR Y. S. & L. D. GOTTLIEB. 1989. i ipu evolution in Layia (Compositae): T bination in hybrids between L. discoidea e a glandulosa. Syst. Bot. 14: 284 GOTTLIEB, L. D. & V. S. Еовр. 1987. Genetic and developmental studies of the absence of ray florets in Layia discoidea. Pp. 1-17 in H. Thomas & D. Grierson (editors), bios Mutants in Higher Plants. Society of i ie Biology. Cambridge Univ. Press, Cambridge. iia D. A. 1933. Cytology of the tribe Madinae, family Compositae. Bot. Gaz. (Crawfordsville) 95: 77-208. rgyroxi ne ium. Occas. Pap. Bernice Pau- ahi bs Mus. 11: 1-38. 9 The s й of Hawaii. Serv, Bull. Candie Inst. Wash. 4: 75-78. 19 Hemizonella becomes a Madia. Ma- News ГЕЯ 10: 22. 958. Taxonomic notes on the California flora. 14. Ala 4: 101-1 ——. 195 Madiinae. Pp. 1106-1129 in P. A. Munz (editor), A California Flora. Univ. California 1982. Ecological and Evolutionary Pat- terns in Holocarpha (Compositae, Madiinae). Ph.D. Dissertation. Univ. of California, Davis, California. PowELL, A. М. € S. A. PowELL. . Chromosome numbers in е рц Madroño 25: 160-169 STROTHER, J. L. More chromosome studies in i 24. An indu hybrid in v a (Compositae: Madiinae). Madroño 24: -61. 978. Hemizonia conjugens (Compositae): distribution, chromosome number, and relationships. EI o: Té of Hemizonia sect. Ma- es (Asteraceae; Madiinae). Syst. Bot. 7: 314- 339. 1985. Systematic studies in Hemizonia (As- ER Madiinae): hybridization of H. fasciculata with H. ne and H. minthornii. Syst. Bot. 18. 10: 110 & J. v. Adams. 1986. Natural hybridization between Layia glandulosa and L. paniculata (As- teraceae: ее. Madroño 33: 244-252. THOMPSON, W 983. biosystematic study of La- к» (Compostar Heliantheae) and related gen- ssertation. Univ. of California, Davis, A cyto-genetic and evolution- udy of Heaton. section Centromadia. Amer. 4. . D. CARR. 1988. = TET. and genetic differentiation in the Hawaiian silve sword alliance (Compositae: Madiinae). Evolutidn 42: 1278-1287 CHLOROPLAST DNA EVOLUTION AND ADAPTIVE RADIATION IN THE HAWAIIAN SILVERSWORD ALLIANCE (ASTERACEAE-MADIINAE) Bruce G. Baldwin,’ Donald W. Kyhos,’ and Jan Dvorak? ABSTRACT Chloroplast DNA (с DNA) restriction site variation among 24 of 27 extant species of the Hawaiian silversword alliance (Argy. roxiphium, Dubautia, Wilkesia) was studied to enhance understanding of phylogenetic pattern ns and evolutionary processes in this outstanding example of insular adaptive radiation. Analysis with 16 restriction endonucle- ases revealed a total of comparisons amon hese taxa, cpDNA нса ата of the silversword alliance, in accord with 55 restriction site differences, 36 being shared by a subset of at least two taxa. In pairwise divergence ranged from 0 to 0.46% of nucleotides. Шш DNA divergence within Dubautia deca that of all intergeneric comparisons. Á strict ariation was topologically ae with evolutionary resolutions from Dollo parsimony and the between these outcomes and results from a phenetic analysis a cpDNA miplesular clock is not ope ae among these taxa. The cpD biosystematic and с consensus tree from Wagner parsimony togenetic evidence. T suggest: (1) rapid radiation of lineages early in the history of the silve rsword alliance; (2) few successful interisland migrations that led to new species lineages; he n = 13 Dubautia group; (4) among Wilkesia and five Kauaian Dubau of nuclear DNA to genetically cohesive group of plants. (3) a unidirectional, older-to- -younger ania pa striking dese between res rates of m o tia species; and (5) h cert a Argyroxiphium and Dubautia taxa. These phylogenetic се require testing with clarify the extent to which hybridization has influenced the evolution of this reproductively and ttern of migration within DNA evolution Prominent examples of insular adaptive radia- tion are widely acknowledged as exceptional dem- onstrations of organismic evolution. Indeed, these grou[ characterized by levels of phenotypic diversity oth- ys of seemingly closely related species are y y р erwise associated with higher taxa (e.g., Hawaiian Drepanidae, Galápagos Geospizinae). Evolutionary studies of such systems hold great potential to provide unique insights into mechanisms of evo- lution, origins of major organismic lineages, and genetic bases of conspicuous or fundamental char- acteristics. Despite extensive documentation of island plant life (e.g., Carlquist, 1965, 1974), few biosyste- matic or ” data from insular floras are available. For example, only a small fraction of the Hawaiian archipelago flora has been thus examined to any degree (Carr, 1987). This is un- fortunate, given the increasingly precarious posi- tion of island ecosystems and the importance of understanding phylogeny for evolutionary inter- pretations. In plants, molecular phylogenetic approaches employing DNA appear to be especially suited to the study of insular adaptive radiation. Previous investigations of continental plant groups using chloroplast DNA (cpDNA) have revealed convinc- ing evidence of relationships obscured by morpho- logical evolution (reviewed in Sytsma & Smith, 1988). Molecular insight is needed in studies of island species given their ap- p 8; Palmer et al., parent potential for rapid morphological change, parallelism, and hybridization. һе Hawaiian silversword alliance represents an ' Supported in part by grants from the National Science Foundation (В5К-87 15270: to B. Baldwin and D. Kyhos) and the е nal Academy of Sciences, 2d dn Gagné, К. Hobdy, A. Medeiros, J. Obata, S. Perlma for A assistance; R. Jansen, J. reviews; S. zum с ooperatior Xi (to , E. Powell, R. a haux, L. Stemmermann, and L. Walker Palmer, nd К. ы for cpDNA clor rtens for computer assistance; and G. Carr for field assistance, review of the manuscript, and generous . Baldwin). We thank S. Bainbridge, J. Canfield, В. es; J. Strother and K. Sytsma for helpful D epu of Botany, rae of California, Davis, California 95616, U.S.A. Please send correspondence regarding this manuscript to Bruce Baldwin at this a ress. ‘Department of Agronomy ШШ Range Science, University of California, Davis, California 95616, U.S ANN. Missouni Bor. GARD. 77: 96-109. 1990. Volume 77, Number 1 1990 Baldwin et al. 97 Chloroplast Evolution in the Hawaiian Silversword Alliance ideal island plant assemblage for DNA systematic study. Comprising only 28 species in three dis- tinctive genera (Argyroxiphium, Dubautia, and Wilkesia), this manageably small group spans an exceptional range of diversity in morphological, anatomical, and ecological characteristics (Carr, 1985). In fact, Carlquist (1974) regarded the sil- versword alliance as the most outstanding example of angiosperm adaptive radiation in the Hawaiian archipelago. The extensive background of evolutionary re- search on these plants includes anatomical (Carl- quist, 1957, 1958a, b, 1959a, b; Kim, 1987), cytogenetic (Carr € Kyhos, 1981, 1986), isozymic (Witter, 1986; Witter & Carr, 1988), ecophysi- ological (Robichaux, 1984, 1985; Robichaux & Canfield, 1985), chemosystematic (Crins et al., 1988), and breeding system (Carr et al., 1986 investigations. These studies clearly delimit Hawai- ian Madiinae and indicate their evolutionary co- hesiveness. For example, artificial and natural hy- bridization have linked nearly all species, the F, generations being of at least marginal fertility in every instance (Carr & Kyhos, 1981, 1986). Most importantly for DNA phylogenetic analysis, a closely related lineage that is ancestral to the silversword alliance has been identified by chloroplast DNA investigations and interspecific hybridization in- volving perennial North American Madiinae (Bald- win, 1989). These studies led to recognition of two outgroup species with close affinities to Hawaiian Madiinae: Madia bolanderi and Raillardiopsis тигїї. In this paper, we present results from а DNA systematic study of restriction site polymorphisms in the silversword alliance. Chloroplast DNA (cpDNA) was selected for analysis for three rea- — sons: (1) evolutionary interpretation of uniparen- tally inherited cpDNA is essentially free from the complexities resulting from sexual processes en- countered in nuclear DNA (reviewed in Birky, 1988); (2) unlike plant mitochondrial DNA, cpDNA is in most cases structurally conservative and evolves more rapidly by point mutations (Palmer, 1985), facilitating restriction site analysis; and (3) all pre- vious genetic studies of the silversword alliance have focused on the nuclear genome or its expres- sion. А uniparental phylogeny was desired to re- solve better the role of hybridization in the evolution of these interfertile species. MATERIALS AND METHODS Fresh leaves were field- or greenhouse-collected from representatives of one to five populations of 24 of the 27 extant silversword alliance species and from two outgroup taxa of North American Madiinae, Madia bolanderi and Raillardiopsis muirti (Table 1). Leaves were ground in liquid N, with mortar and pestle. Total DNA was isolated from these leaf N, powders and purified twice on cesium chloride gradients using a modified proce- dure of Palmer (1986), omitting separation of or- ganelles. Mercaptoethanol was increased up to 1% in the extraction buffer to increase the solubility of mucilaginous carbohydrates. DNA aliquots from one or two populations of each species were di- gested with each of 16 restriction endonucleases recognizing four, five, or six nucleotide-pair sites (Table 3). “Six-cutter” enzymes with AT-rich rec- ognition sites were chosen to maximize cleavage of the AT-rich epDNA molecule. Resulting cpDNA fragments were resolved in 1.25-4% agarose gels by Southern blot hybridization with *P-labeled cpDNA sequences. Probes were prepared by ex- tension of random primers (Amersham) annealed to gel-isolated inserts from clones of a Sac I Lac- tuca cpDNA library, generously provided by R. К. Jansen and J. D. Palmer, and a 9.0 kb Pst I Petunia fragment (J. D. балы, and E. Clark), obtained through K. ytsma. Together these probes represented a de the lettuce cpDNA molecule (Fig. 1). Partial restriction mapping of cpDNA from four Hawaiian species with three enzymes revealed no deviations in cpDNA size or structure relative to lettuce (Baldwin et al., unpublished). Restriction site changes were inter- preted with reference to standard markers of triple- digested (Bam HI, Eco RI, Hind III) and single- digested (Hind III) lambda phage DNA. Length mutations were distinguished from restriction site mutations near preexisting sites by comparisons of all digests at the region in question. Both Wagner and Dollo parsimony were em- ployed in cladistic phylogenetic analyses of the restriction enzyme data set. Wagner parsimony treats all mutations equivalently, whereas Dollo parsimony prohibits certain evolutionary parallel- isms—parallel restriction site gains and loss/ gains—the least probable homoplasious changes (DeBry & Slade, 1985). A presence/absence ma- trix of all restriction sites shared by a subset of two or more taxa, including the outgroup species, was analyzed by Wagner parsimony using the PAUP package, version 2.4 (D. L. Swofford, Illinois Nat- ural History Survey). The GLOBAL and MUL- PARS branch-swapping options and the branch- and-bound (BANDB) option of PAUP were used to search for the shortest topologies. A strict con- sensus tree of the maximally parsimonious reso- Annals of the Missouri Botanical Garden TABLE 1. plast DNA restriction site polymorphisms. DNA samples Collections of species analyzed for chloro- from a given loc ation comprised one individual unless indicated by “( = pooled sample of two to ten indi- viduals from one population. Lowercase letter designa- tions following voucher numbers indicate separate c lusters of individuals within a population. Collection numbers in parentheses indicate accessions screened only for select aterials and B. G. mutations found in the main analysis (see * Methods"). Collecte MA et al; GD = 6. D. . Vouchers are at AV (BGB and A. Medeiros брен) or HAW (other о unless specifically indicated. ' (BISH, HAW, US). * (PTBG). * (BISH, F, US). * (BISH, F, HAW, US). 5 (BISH) r abbreviations are: BGB = = Argyroxiphium caliginis C. Forbes — BGB 660, Pwu Kukui, West Maui. Argyroxiphium grayanum (Hillebrand) Degener BGB 661, Pwu Kukui, West Maui; 4. Medeiros East Maui. Arg yroxiphium sandwicense DC. subsp. macrocepha- lum (A. Gray) Meyrat —(GDC 1239), Haleakala, East Maui. Arg vroxiphium sandwicense DC. subsp. sandwic- s.n., Haleakala, ense В 657 (P), upper Wailuku drainage, Mauna Kea, Hawaii. Dubautia arborea (A. Gray) Keck BGB 527 (P), Pu'u La’au, Mauna Kea, Hawaii; ane 056), upper Wailuku drainage, Mauna Kea, Haw Dubautia ciliolata (DC.) Keck subsp. cial 529 (P), Crater Rim Trail, Kilauea Dubautia ciliolata (DC.) Keck subsp. pone ia G. (BGB 525) (P), near Pwu Huluhulu, Hawaii; (BGB did upper Wailuku drainage, Mauna Kea, Hawaii BGB жо herbstobatae б. Carr —GDC 1244, Ohikilolo idge, Waianae Range, Oahu. Dubautia imbric ata H. ^ John & € bricata 1B 667. . Carr subsp. im- Wahiawa Menon Kauai. Dubautia knudse nii Hillebrand subsp. filiformis G. ` 1234), Makaleha Mountain, Kauai. Dubautia knudsenii Hillebrand pes knudsenti— GDC 1047, Awaawapui Trail near Kokee Road, Waimea, Kauai; n DR EL Kahuamaa Flat, Kaua Dubautia knudsenii Hillebrand subsp. nagatae (H. 5t. John) G. Carr —(GDC 1322a, b) (P), Pihea Trail, Alakai Swamp, Kauai. Dubautia Mp cn A. m 671, Kaluapuhi Trailhe Kokee Park, Kaua Dubautia “latifolia (A. Gray) Keck BGB 675, (Flynn 120%), Makaha Ridge, Kokee Park, Kauai. Dubautia laxa Hook. & Arn. subsp. hirsuta (Hille- brand) G. Carr —(GDC 833'a, b) (P), Mt. Kaala, Waianae Range, Oahu; (Flynn 3201") (P), Pihea Trail, Alakai Swamp, Kauai. Dubautia laxa Hook. & Arn. subsp. laxa Pu'u Kukui, West Maui. — BGB 002, TABLE 1. Continued. Dubautia linearis (Gaudich.) Keck subsp. hillebrandii ann) G. Carr — BGB 531 (P), Pohakuloa Mili- tary Reservation, Hawaii. Dubautia linearis (Gaudich.) Keck subsp. linearis — (BGB 516) (P), Piilani Highway E of Ulupalakua, East Maui. Dubautia menziesii (А. Gray) Keck — BGB 522 (P), (BGB 521) (P), west slope of Haleakala, East Maui. Dubautia microcephala Skottsb. — GDC 1044'a (b) cahuamaa Flat, Kokee Park, Kauai; (BGB 671), Kaluapuhi Trailhead, dpi Park, Kauai. Dubautia paleata A. Gray —GDC 13750 (a, b, d) (P), ihea Trail, Alakai Swamp, E Dubautia pauciflorula H. St. John & G. Carr — BGB 668, Wahiawa Mountains, Kauai. Dubautia plantaginea Gaudich. subsp. humilus— (GDC 1183), Black Gorge, West Ma pie plantaginea Gaudich. subsp. pm. ¿DC 1180, Castle Trail, Koolau Range, Oahu; ae v b, c) (P), Mt. Kaala, Waianae Range, Jahu. коо platyphylla (A. Gray) Keck — BGB 524, (BGB 523) (P), west slope of Haleakala, East Maui. Dubautia raillardioides Hillebrand —— BGB 670 (P), border of Wahiawa Bog, Kauai; (GDC 1373a, b, с) Trail, Alakai Swamp, Kauai. Dubautia reticulata (Sherff) Кеск BGB 664, Koolau Gap, Haleakala Crater, East Maui. Dubautia scabra (DC.) Keck subsp. scabra — BGB 530 (P), Crater Rim Trail, Kilauea, Hawaii; (BGB P), near Pu'u Huluhulu, Hawaii. BGB 515 (P) Kamaileu- nu Ridge, Waianae Range, Oahu; (GD 837a, b, c, d) (Р), Mt. Kaala, Waianae Range Oahu. Madia bolanderi (A. Gray) A. Gray Tahoe, Sierra Nevada, Califorr ty oa Dubautia sherffiana Fosb. BGB 509, Lake Raillardiopsis muirii (A. Gray) Rydb. BGB 618 (P), ucia Range, California. Il ilkesia mud ue A. Gray — Char 76.022c (a, b) (P), Hiau Loop Trail, Waimea Canyon rim, Kauai: (GDC 11572, b, c) (P), Waimea Canyon Rd., aual. Wilkesia hobdyi H. St. John idge, Kauai. GDC 1150, Polihale lutions was created using the PAUP CONTREE program. Confidence intervals were derived for each resolved node of the consensus tree using the same data matrix and the PHYLIP statistical pack- age, version 3.1 (J. Felsenstein, Univ. of Wash- ington) bootstrap (BOOT) option (100 replicate runs). Dollo parsimony analysis was conducted with the PHYLIP DOLLOP program from a similar data matrix that indicated double and single fragment profiles for each mutation. We conducted two genetic distance analyses of Volume 77, Number 1 1990 Baldwin et al. 99 Chloroplast Evolution in the Hawaiian Silversword Alliance the restriction enzyme data. Chloroplast DNA di- vergence values (p values) for all pairwise species combinations were calculated from restriction site variation following Nei & Li (1979). These esti- mates of nucleotide substitutions per nucleotide position were used to produce (1) a Fitch & Mar- goliash (1967) dendrogram (FITCH program of PHYLIP) and (2) a phenetic tree, from a modifi- cation of the Fitch-Margoliash algorithm (KITSCH program of PHYLIP) that assumes clocklike evo- lution of nucleotide sequences. A preliminary survey of intraspecific cpDNA variation was conducted based on results from the above analyses of one to two populations per species. Additional populations (Table 1) were screened for one or more restriction site mutation(s) unique to a given species in the primary analysis or, where no such mutations existed, for restriction sites most diagnostic for that taxon. RESULTS Approximately 875 restriction sites were ex- amined in each cpDNA included in the main anal- ysis. This accounted for 3.7% of the nucleotide sequence in each cpDNA. One of the two copies of the inverted-repeat cpDNA sequence was ex- cluded from these estimates. Fifty-five restriction site differences were detected among the Hawaiian species, 36 (65.5%) of these being shared by subset of two or more taxa (Table 2). Six additional mutations were common to all species of the sil- versword alliance and separated them from the two orth American outgroup taxa. Chloroplast DNA divergence within the silver- sword alliance ranged from 0 to 0.46% of nucleo- tides (p values; Table 3). For the samples used in the main restriction site analyses, complete inter- specific identity in cpDNA was apparent in four instances: Argyroxiphium caliginis/ A. grayan- um (West Maui); Dubautia imbricata/D. pau- ciflorula/ D. raillardioides; D. ciliolata/ D. sca- bra; and D. menziesii/ D. platyphylla / D. reticulata. Maximum cpDNA divergence was found within Dubautia, exceeding any intergeneric value among the Hawaiian species (Table 4). Genetic distances between the Kauaian rainforest shrub Dubautia laevigata and the Mauian bog rosette plants Argyroxiphium caliginis and A. grayanum (p = 0.15%) were the lowest of any intergeneric cpDNA comparisons, closely approaching the max- imum divergence found within the genus Argy- roxiphium (p = 0.13%), which is conservative in it. Chloroplast DNA restriction sites were most ge single-copy region; SSC — small single- еру: region. Sizes ¡ in kilobases of these frag- ments are: Sla = 12.3; Slb = 8.3; = 9.9; $3 = 3.5; S inverted repeat; LSC = lar = 10. 6; S13 = = А 6; S14 = 5.4; S15 = 6.3. Additionally, gel-isolated probes from digests of S6 (x Kpn I) and S12 (X Xho I) were used. *P10*" represents only that portion of the 9.0-kb Pst 1 Petunia fragment that bridges the 56. inverted repeat and highly concentrated in the vicinity of the ribulose bisphosphate carboxylase, large-subunit gene (rbcL; 2) and near the inverted repeat-proximal end of the Compositae inversion (S6; Jansen € Palmer, 1987). Variation in both regions approached one mutation per kilobase (Table 5). Lower occurrences of mutations were found throughout the remainder of the large single-copy (LSC) and small single- copy (SSC; S5) regions (Fig. 1). Only two mutations clearly involved restriction sites within the inverted repeat (Fig. 1), in accord with its marked sequence conservation in other higher plant groups (Palmer & Zamir, 1982; Palmer et al., 1983a, b; Clegg , 1984; Sytsma & Gottlieb, 1986). Each of these: two site differences yielded alternative frag- ment profiles with on band and two others differing to the same extent in opposing digests (Table 2) Intraspecific cpDNA variation was detected in eight of 15 species screened for specific site mu- tations found in the main analysis. The following samples lacked mutations unique to their species 100 Annals of the Missouri Botanical Garden TABLE 2. Restriction site mutations from ca. 875 restriction sites per cpDNA sampled in the primary analysis of Hawaiian Madiinae an outgroup taxa Table D. мша 1-43 were shared by a subset of two or more species (synapomorphies, cline outgroup). The remainir were unique to a ш Hawaiian accession in the primary survey (autapomorphies). ind nique to the outgroup taxa are in Baldwin (1989). In three instances of 2 conflict (mutations 9, 14, and 43) the derived condition has been designated based on the resolutions of Wagner and Dollo parsimony with > outgroups included. ' No mutations were detected with Hae I, i l, or Sca I. The derived fragment condition (in kilobases) appears to the right of the hyphen, жуы the h ancestral state. ' See Figure 1 for molecular position. Uncertain location of a mutation is indicated by slashes (/) between regions. Hyphens (-) between regions indicate the common domain of both seque ‘l= Reilly muirii; 2 = Madia bolanderi; 3 = Wilkesia gymnoxiphium; 4 = W. hobdyt 5 za ds knudsenii; 6 — imbricata; 7 = D. pauciflorula; 8 = D. raillardioides; 9 = D. paleata; 10 = Argyroxiphium sandwicense; T — 4. de nd (E. Maui); 12 = A. caliginis; 13 = A. grayanum (W. Маш); 14 = e pend 15 = D. laxa; 16 = D. scabra; 17 = D. ciliolata; 18 = D. linearis, 19 = D. arborea, 20 = D. menziesii, 21 = e 99 = D. reticulata; 23 = D. herbstobatae; 24 = D. sherffiana; 25 = D. microcephala; 26 = laevigata; 27 = D. latifolia. * Additional fragment profile from a single mutation occurring within the inverted — wN I JI near the LSC borders. Enzyme' Site mutation” Region’ Mutated cpDNA' 1. Ва 1.7 + 0.7-2.4 S13 3-27 2. BstNI 2.8-1.7 + 1.1 51 25-27 3. BstNI 2.5-2.1 + 0.4 S6 3-27 і. BstNI 0.7 + 0.7-1. SII 3,4 BstNI 0.8 + 0.1-0.9 S12 25, 26 6. Dral 1.1 + 0.7-1.8 P10-S6 14-2 7. Dral 1.2 + 1.8-3.0 P10-S6 14-17 8. Dral 1.2-0.7 + 0.5 P10-S6 3-9 9. Dral 1.2 + 1.5-2.7 S6 2,10 10. Dral 1.5-1.1 + 0.4 S6 14, 27 ll. Dral 0.9 + 0.2-1.1 S7 5-9, 14 12. Dral 2.5-2.4 + 0.1 58 14-24 13. Dral 0.3 + 2.1-2.4 S15 10, 11, 14, 15 14. EcoRI 2.6-1.4 4 1.2 S6 ‚1 15. EcoRI 1.4 + 0.7-2.1 S6 14-24 16. EcoRI 2.6 + 2.4-5.0 S6 25, 26 17. EcoRI 4.4 + 2.2-6.6 S7 16-19 18. EcoRI 3.7-2.5 + 1.2 512 3, 19. EcoRI 1.6-1.0 + 0.6 813 12-15 20. EcoRI 1.4 1.1-2.5 S5 25, 26 21. EcoRV 1.0 + 7.5-8.5 S7 16-19 22. EcoRV 7.5-4.2 + 3.3 58 12,13 23. EcoRV 4.9-4.5 + 0.4 $12/813 3—21 24. Hinecll 2.4 + 4.4-6.8 S11 5-9 25. HindMl 1.6 + 1.0-2.6 37/8 3-27 26. Hindlll 4.5 + 6.6-11.1 $12 14-24 27. Hpal 3.6 + 0.3-3.9 S5 5-9 28. Прай 1.0 + 1.0-2.0 S6 3-21 29. Hpall 0.4 + 0.5-0.9 S10 14, 15 30. Hphl I7 + 015-32 56 14-24 31. Hphl 3.9 + 0.9-4.8 58 23, 24 32. Hphl 1.1 + 0.7-1.8 512 16, 17 33. Sspl 1.9 + 1.0-2.9 51 5-9, 27 34. Sspl 4.0 + 0.1-4.1 Sl 16-22 35. Sspl 3.8-3.6 + 0.2 $3/4/5 3-9 36. Sspl 0.4 + 0.4-0.8 P10-S6 3-27 37. Sspl 1.2-1.0 + 0.2 S6 5-8 38. Nbal 32.5 + 6.7-39.2 55 25-27 39. Nbal 2.8 + 2.4-5.2 P10 5-9, 19 40. Abal 2.3-1.3 + 1.0 S6 20-22 41. Xbal 2.3 + 1.5-3.8 S12 5-9 Volume 77, Number 1 1990 Baldwin et al. 101 Chloroplast Evolution in the Hawaiian Silversword Alliance TABLE 2. Continued. Enzyme' Site mutation* Region” Mutated cpDNA' 42. Xmnl 0.5 + 1.3-1.8 51 3-9 43. Xmnl 6.3 + 0.7-7.0 Sl 2, 3, 10, 20-22, 25 (4.9 + 0.7-5.6) 44. Bglll 5.7-4.6 + 1.1 S5 25 45. BstNI 1.8 + 0.8-2.6 S6 3 46. BstNI LE+ 0.2-1.3 S9 19 47. BstNI 1.1-1.0 + 0.1 S9 9 48. BstNI 0.2 + 0.1-0.3 $12 4 49. Dral LZ 1.1-2,3 P10-S6 27 50. Dral 3.1 + 0.5-3.6 58 24 51. Dral 8.2 + 0.7-8.9 S10 24 52. EcoRI 3.7-2.1 + 1.6 $12 9 53. EcoRl 0.9 + 0,2-1.1 S15 11 54. Hpall 2.6-1.5 + 1.1 S12 4 55. Hphl 3.9-2.6 + 1.3 55 4 56. Ndel 7.9-4.8 + 3.1 S6 10 57. Ndel 1.0 + 1.0-2.0 S14 24 58. Sspl 0.9 + 1.3-2.2 S12 15 59. Sspl 1.0-0.5 + 0.5 EZ 27 60. Xbal 4.3 + 1.3-5.6 512/513 9 61. Xmnl 8.6-7.0 + 1.6 1 24 (7.9-6.3 + 1.6) in the primary survey, most of these taxa being different subspecies than those analyzed in depth (number in parentheses = mutation designation in left column of Table 2): Argyroxiphium sandwic- ense subsp. macrocephalum GDC 1239 (56); Du- bautia arborea BGB 656 (46); D. laxa subsp. hirsuta GDC 833a, b, Flynn 3201 (58); Р. knud- senit subsp. filiformis GDC 1234 (52); D. knud- senii subsp. nagatae GDC 1322a, b (52); and Wilkesia gymnoxiphium GDC 1157a (45). The last sample of W. gymnoxiphium, however, indi- cates intrapopulational cpDNA variation in this taxon. In addition, populations of three species— again, mostly subspecies not sampled in the main analysis—lacked mutations shared between their species and one or more others in the primary survey: Dubautia ciliolata subsp. glutinosa BGB 525, BGB 659 (21, 32); D. linearis subsp. lin- earis BGB 516 (21); D. plantaginea subsp. hu- milis GDC 1183 (6, 7, 10); and D. plantaginea subsp. plantaginea GDC 838a, b, с (10). No in- traspecific cpDNA variation was observed in Du- bautia latifolia, Р. menziesii, D. microcephala, D. paleata, D. platyphylla, D. raillardioides, or D. sherffiana. Four length polymorphisms resulting from in- sertions or deletions of ca. 50-250 nucleotides were resolved within the silversword alliance (Table 6). Three of these four mutations occurred in the LSC region near the inverted repeat ($1 and S6; Fig. 1), regions noted for high incidence of length variation in many plants (Palmer, 1985). The re- maining mutation occurred in the SSC region (S5; Fig. 1). Smaller-length polymorphisms could not be well resolved by the gel-systems used. Involve- ment of the deletion shared by Dubautia plan- taginea and D. laxa in an unscored Ssp I site loss in these species was inferred from diminished size of the resulting fragment. This length mutation was absent in the sole accession of D. plantaginea subsp. humilis (Table 1). agner parsimony analysis of the entire data set (including outgroups) using the PAUP package branch-and-bound (BANDB) option found 1,035 maximally parsimonious trees at 53 steps. The high number of maximally parsimonious trees was at- tributable to arbitrary resolutions of zero-length branches by PAUP. A PAUP Wagner analysis of these data using the GLOBAL and MULPARS options also found minimum-length trees of 53 steps. Strict consensus trees from the upper limit of 100 trees retained in both PAUP analyses were topologically congruent. The common tree from both approaches is presented in Figure 2. A PAUP Wagner branch-and-bound analysis of the two fully resolved, Kauaian sublineages of this tree (i.e., Wilkesia/ Dubautia sister group; Dubautia lae- vigata/ D. latifolia/ D. microcephala) indicated that each represented the most parsimonious to- pology. Wagner branch-and-bound analysis of the 102 Annals of the Missouri Botanical Garden TABLE 3. 1). Values in the upper right half of matrix are numb ers of m Pairwise cpDNA divergence among primary Hawaiian Madiinae samples and outgroup taxa (see Table utations separating two cpDNAs from ca. 875 restriction 1 sites sampled in each. Numbers of nucleotide substitutions per nucleotide position, or p values (Nei & Li, 1979), appear in the lower left half of matrix as 100p. Samples top matrix borders are: 1 = кезка тигїї; corresponding to the number pen along the left and Madi g xiphium; 4 = W. hobdyi; 5 = Dubautia knudsenii; 6 = D. imbricata/D. pauciflorula/D. raillardioides; T = D. paleata; 8 = l 2 3 4 5 6 7 8 9 10 | 19 2 26 29 28 29 22 20 20 2 0.398 — 1 21 24 23 24 13 15 15 3 0.525 0.377 — 5 12 11 12 9 9 9 + 0.546 0.440 0.104 ^ 13 12 13 12 10 10 5 0.611 0.504 0.250 0.271 1 4 15 13 13 6 0.589 0.483 0,229 0.250 0.021 3 14 12 12 7 0.611 0.504 0.250 0.271 0.083 0.062 15 13 13 8 0.461 0.271 0.187 0.250 0.313 0,292 0.313 == 4 © 9 0.419 0.313 0.187 0.208 0.271 0.250 0.271 0.083 4 10 0.419 0.313 0.187 0.208 0.271 0.250 0.271 0.125 0.083 = 11 0.589 0.525 0.398 0.419 0.440 0.419 0.440 0.292 0.250 Ape 12 0.568 0.504 0.377 0.398 0.461 0.440 0.461 0.27 0.229 229 13 0.568 0.504 0.377 0.398 0.461 0.440 0.461 0.313 0.271 1 271 14 0.525 0.461 0.334 0.355 0.419 0.398 0.419 0.27 0,229 0,229 15 0.568 0.504 0.377 0.398 0.419 0.398 0.419 0.313 0.271 0.271 16 0.525 0.419 0.202 0.355 0.419 0.398 0.419 0.229 0.229 0.229 17 0.483 0.419 0.292 0.313 0.377 0.355 0.377 0.229 0.187 0.187 18 0.568 0.504 0.377 0.398 0.461 0.440 0.461 0.313 0.27 0.271 19 0.525 0.377 0.250 0.313 0.377 0.355 0.37 0.187 0.187 0.187 20 0.483 0.377 0.250 0.271 0.334 0.313 0.334 0.187 0.146 0.146 21 0.504 0.398 0.271 0.292 0.313 0.292 0.313 0.208 0.166 0.166 D. plantaginea/D. laxa/ D. scabra/n = 13 Du- program) analyses all yielded results topologically bautia sublineage yielded nine maximally parsi- monious resolutions, all concordant with the con- sensus topology. Bootstrap (PHYLIP BOOT program) confidence intervals for resolved consen- sus tree branches ranged from 59% to 100% (Fig. 2) The consensus phylogeny, excluding the out- groups, had a consistency index (C.I.; Kluge & Farris, 1969) of 0.85 (15% parallelism) with 48 site losses and 24 site gains, including nine parallel losses and two parallel gains. Exclusion of two highly homoplasious characters (13 and 43; Table 2) resulted in the same consensus resolution with a C.I. of 0.91 (9% inc lude autapomorphies. parallelism). These С.І. values e high incidence of parallelism largely in- Ne mutation "hotspot" centers at both ends of the LSC region, near the inverted repeat (Palmer, 1985). The predominance of parallelism involving the Dubautia plantaginea and D. laxa clade is not readily explained by length polymorphisms, introgression, or weak placement in the tree. Ab- sence of a parallel site gain (10; Table 2) in ad- ditional populations of D. plantaginea suggests that this mutation indeed arose twice Dollo parsimony (PHYLIP package, DOLLOP congruent with the strict consensus tree (Fig. 2). These trees, however: (1) consistently placed the D. plantaginea/ D. laxa branch basal to D. herb- stobatae and D. sherffiana, thereby resolving the Dubautia species and D. scabra as monophyletic group, and (2) placed A. sandwi- cense in a sister-group relationship with East Mauian A. grayanum Fitch-Margoliash genetic distance analysis yield- both the Wagner strict consensus tree (Fig. 2) and the Dollo a ed results completely harmonious with parsimony results. The dendrogram produced was the best topology of all trees examined, i.e., in being most compatible with measured species di- É 3). The average "standard de- of this outcome was 4.53%. Though fully resolved, this result was only slightly better than vergences (Table viation” alternative conformations differing at nodes unre- solved in the consensus tree. Accordingly, the en- tire dendrogram topology is not presented. Phenetic analysis of the genetic distance values using the PHYLIP package, KITSCH program gave results contrary to the consensus tree (Fig. 2). Unlike resolution (average all other analyses, the best phenetic “standard deviation” = 26.06%) placed the following species, separated Baldwin et al. Chloroplast Evolution in the Hawaiian Silversword Alliance Volume 77, Number 1 103 1990 TABLE 3. Extended. Argyroxiphium sandwicense; 9 = A. grayanum (East Maui); 10 = А. caliginis/ A. grayanum (West Maui); 11 = Dubautia plantaginea; 12 = D. laxa; 13 = D. scabra/D. ciliolata; 14 = D. linearis; 15 = D. arborea; 16 = D. menziesii/ D. platyphylla/ D. reticulata; 17 = D. herbstobatae; 18 = D. sherffiana; 19 = D. microcephala; 20 — D. laevigata; 21 — D. latifolia. 11 12 13 14 15 16 17 18 19 20 21 28 Pun 2 25 27 25 23 27 25 23 24 25 24 24 22 24 20 20 24 18 18 19 19 18 18 16 18 14 14 18 12 12 13 20 19 19 17 19 17 15 19 15 13 14 21 22 22 20 20 20 18 22 18 16 15 20 21 21 19 19 19 17 21 17 15 14 21 22 22 20 20 20 18 22 18 16 15 14 13 15 13 15 11 11 15 9 9 10 12 11 13 11 13 11 9 13 9 7 8 12 11 13 11 13 11 9 13 9 7 8 = 3 9 9 11 9 7 11 19 17 16 0.062 8 8 10 8 6 10 18 16 17 0.187 0.166 ES 2 4 6 6 10 18 16 17 0.187 0.166 0.041 == 2 4 1 8 16 14 15 0.229 0.208 0.083 0.041 к= © 6 10 18 16 17 0.187 0.166 0.125 0.083 0.125 = 1 8 14 14 15 0.146 0.125 0.125 0.083 0.125 0.083 = 4 14 12 13 0.229 0.208 0.208 0.166 0.208 0.166 0.083 = 18 16 17 0.398 0.377 0.377 0.334 0.377 0.292 0.292 0.377 2 9 0.355 0.334 0.334 0.292 0.334 0.292 0.250 0.334 0.041 E 7 0.334 0.355 0.355 0.313 0.355 0.313 0.271 0.355 0.187 0.146 =< by slash marks, in sister-group relationships: (1) the species of Ar y phium/Dubautia laevi- gata; (2) Dubautia menziesii, Р. platyphylla, Р. reticulata/ D. sherffiana; and (3) Dubautia herb- stobatae/ D. arborea, D. ciliolata, D. linearis, D. scabra. DISCUSSION CHLOROPLAST DNA DIVERGENCE Chloroplast DNA nucleotide sequence diver- gence within the silversword alliance (р = 0-0.46%) TABLE 4. Limits of intrageneric and intergeneric chlo- was similar to that of the least divergent continental genera that have been documented; e.g., Helian- thus (p = 0.3%-0.4%; Rieseberg et al., 1988), Lycopersicon (p = 0-0.7%; Palmer & Zamir, TABLE 5. Molecular distribution of chloroplast DNA restriction site mutations in Hawaiian Madiinae. Regions refer to probe sequences flanked by Sac 1 or Pst 1 re- striction sites (see Fig. 1). ' Region of the 9.0-kb Pst 1 sequence between the inverted repeat S * Underestimated owing to unscored Dra I mutations not well resolved with S5 probe. Numbers of Mutations/ roplast DNA divergence detected in Hawaiian Madiinae. Regions mutations kilobase | Divergence values (p — nucleotide substitutions per nu- S6 (14.7 kb) 1 0.95 cleotide position) are from Table 3. S12 (10.6 kb) 10-11 0.94-1.04 - к 58 (6.7 КЬ) 4 0.60. Minimum Maximum S9 (3.8 kb) 2 0.53 (p) (р) S1 (12.3 kb) 6 0.49 Argyroxiphium 0 0.13% S7 (7.0 kb) 3 0.43 Dubautia 0 0.46% S15 (6.3 kb) 2 0.32 “ilkesi 0.10% 0.10% S10 (6.9 kb) 2 .29 Argyroxiphium/ Dubautia 0.15% 0.31% S5 (18.8 kb) 5-6 0.27-0.32? Argyroxiphium/ Wilkesia 0.19% 0.25% S11 (7.7 kb) 2 .26 Dubautia/ Wilkesia 0.23% 0.42% S13 (4.6 kb) 1-2 0.22-0.44 0 0.46% P10' (7.1 kb) 1 0.14 104 Annals of the Missouri Botanical Garden TABLE 6. Chloroplast DNA length mutations within results from the two genetic distance programs the silversword alliance. Additional insertions and deletions smaller than 100 nucleotide pairs were not clea re- ' Samples sid the ble 1). itgroup comparison. solved using the gel system employed. ЛЕ restriction site analysis (Ta See Figure ' Direction of mutation based on ot yes approximate map position but visibly larger in- sertion in D. herbstobatae relative to D. sherffiana. Type of Size of Taxon' Region’ mutation! mutation D. plantaginea axa 51 deletion 100-250 bp D. paleata S5 insertion 50-150 bp D. herbstobatae' | P10/S6 insertion 50-100 bp D. sherffiana' P10/S6 insertion 50-100 bp 1982), Lisianthius(p =0-0.3%; Sytsma & Schaal, 1985), and Pennisetum (p = 0-0.6%; Clegg et al, 1984). Overall allozymic variation Hawaiian Madiinae also approximated that of some continental genera (Witter & Carr, 1988). Both measures of genetic differentiation, however, in- among dicate higher levels of molecular divergence in the silversword alliance than in previously investigated autochthonous oceanic plant groups. For example, preliminary surveys of cpDNA variation in Hawai- ian Tetramolopium (Asteraceae) and Dendroseris (Asteraceae) from the Juan Fernandez Islands re- 1987), though fewer species and restriction sites were ex- vealed no polymorphism (Crawford et al., amined than in the present study. Comparisons of cpDNA divergence values pre- sented here with those measured among Nort American Madiinae (Baldwin, 1989) revealed ex- pected and surprising patterns. Higher overall levels of divergence were found among the continental perennial taxa relative to within the silversword alliance. This accords with evidence indicating the mainland group is ancestral to the Hawaiian species (Baldwin, 1989). Maximum cpDNA divergence within Dubautia (p = 0.46%), however, was sim- ilar to that between both North American outgroup species in this study (Майа bolanderi and Rail- M ше muirii) and Hawaiian Madiinae. In fact, :pDNA genetic distances between North American Madia bolanderi and species of Argyroxiphium (р = 0.27%-0.31%) were considerably less than in some comparisons within Dubautia (Table 7). These results suggest closer relationship of North American Madiinae to the silversword alliance than indicated by phenotypic considerations. Patterns of sequence divergence in the silver- sword alliance (Fig. 2) suggest that cpDNA evo- lution has been irregular in this group. Conflicting FITCH and KITSCH indeed provide evidence against operation of a cpDNA molecular clock among these species. In particular, lack of cpDNA sequence divergence was apparent in Argyroxiph- ium, comprising the silverswords and greenswords. Allozymic divergence within Argyroxiphium was likewise low (Witter, 1986). The mechanics of cpDNA evolution are still too little understood (Bir- ky, 1988) to address the lack of cpDNA divergence in Argyroxiphium, i.e., does highly irregular or long-delayed sexual reproduction in this genus in- fluence the rate of cpDNA evolution? COMPREHENSIVE PATTERN OF ADAPTIVE RADIATION Origin of the silversword alliance from a single colonizing event to the Hawaiian arc ы ог at most a few events closely spaced in time, is strongly suggested by the high confidence (89%) 2). This is completely concordant with the interfertility and that this group is monophyletic (Fig. cytogenetic cohesiveness of these species. Further, the basal, unresolved hexachotomy of the consen- sus tree is consistent with a rapid radiation of major lineages early in the history of the alliance. This pattern could alternatively be explained by an in- crease in the rate of cpDNA evolution through time or a period of extensive hybridization and introgression that led to fixation of one cp type throughout Hawaiian Madiinae. The wealth of systematic data for the silversword alliance, how- ever, supports close genetic affinities among all members despite extreme phenotypic and ecolog- ical differences, exactly as would be expected in a group that underwent explosive diversification. ‘This explanation has also been proposed for similar un- resolved cpDNA polychotomies among sections of Clarkia (Sytsma & Smith, 1988) and in Oncidiinae 1988). (reviewed in Palmer et al., BIOGEOGRAPHIC CONSIDERATIONS A consistent feature of the cpDNA consensus tree topology is the strong positive correlation be- tween species affinity and geographic proximity. In almost every case where resolution was available, individual species were most closely и with others occurring on the same island. alone suggests that few successful interisland col- onizations led to subsequent species radiations. Sec- ond, extensive introgressive hybridization or hybrid Volume 77, Number 1 Baldwin et al. Chloroplast Evolution in the 105 of Hawaiian Madiinae (see Table 1). fe ge ы polari American tarwe eds, Madia bolanderi and Rail owing evolutionary parallelism or genetic distance (p) Dun and those in Table 3. Boo branches as per . Br W.- Wi sn. sia; D. im ium; ilkes laty/retic = Dubautia menziesii, D. divides; D. menz/ T number: К = Kau hu; M = Maui; EM = p cies (А. vi riis D. laxa, D. linearis, D. pla ны islands (Carr, 1985). speciation on a given island may have resulted in the observed cpDNA relationships. The most wide- spread species in the alliance, Dubautia planta- ginea and D. laxa, found on most of the Hawaiian Islands, are important for assessing these hypoth- eses. On different islands these species could be introgressed with or ancestral to local endemics. The consensus tree topology (Fig. 2) does not pre- clude one of these two taxa from being an old, genetically polymorphic species that founded ex- isting lineages. Preliminary evidence of consider- able intraspecific cpDNA variation between D. LOpsis е ii. oken lines indicate pisa polychotomies. Taxo :/ rail = 1990 Hawaiian Silversword Alliance HE W. gymnoxiphium (K14) E W. hobdyi (кла) 8° 35° x^ knudsenii k. (K14) (11) 24 a7 EN OU 4t D. imbr/pauc/rail (K14) 47* 60 109% Lp, paleata (K14) 9 (13) (43) 56* — A. sandwicense S. (H14) SLE. A. grayanum (EM14) T. A. caliginis/grayanum (WM14) (10*)(11) [ren EO AE ЛЕ. 96% = D. laxa І. (M14) 89% ma р. ееси 6 12* 14 15 26 30 D. linearis h. PM = (39) 46 94% D. arborea (H13) D. menz/platy/retic (M13) 31] D. herbstobatae (013) | 50 5 : c D. sherffiana (013) 43)44* em D. microcephala (k14) D. laevigata (k14) 59* D. latifolia (K14) 0.25% 0.20% 0.15% 0.10% 0.05% 0 L П П П П П Ї T T T T 1 Genetic Distance (p) FIGURE 2. Strict consensus tree from Wagner wo analyses of a chloroplast DNA restriction site survey sing two North was determined by global parsimony u ite gains; parentheses homoplasy. This parallelism accounts for slight deviation otstrap confidence intervals are given below the ho rizonta | ona abbreviations are: A. = ubautia imbricata, D. pau platyphylla, an s data. А following taxa names give the ¿land of sample da followed y the haploid chromosom East Mau = We ui; аш; Н = Hawaii (Big Island). Only ntaginea, D. нй have known distributions on multiple plantaginea on Maui and Oahu is therefore in- triguing. Another biogeographically interesting cpDNA pattern was resolved strictly among the Dubautia species having a chromosome number of n = 13 (hereafter “n = 13 Dubautia"). Carr & Kyhos (1986) argued from cytology, morphology, and distribution that these species form a monophyletic assemblage derived from an ancestral n = 14 Du- bautia genome by a single reciprocal translocation followed by dysploidy. Close relationship of all n 13 Dubautia taxa is also evident from their 106 Annals of the Missouri Botanical Garden TABLE 7. Lowest chloroplast DNA divergence values between silversword alliance species and the North American tarweed outgroup taxa, Madia bolanderi and Raillardiopsis muirii. These genetic distances (p values, Table 3) are equivalent to or actually exceeded by some among Hawaiian Madiina Divergence from Madia bolanderi (p) Divergence from Raillardiopsis muirii (p) Argyroxiphium sandwicense 0.27% Argyroxiphium caliginis 0.31% Argyroxiphium grayanum 0.31% Wilkesia gy о i 0.38% Dubautia laevigat 0.38% Dubautia жо ЫЫ 0.38% Dubautia latifolia 0.40% Dubautia reticulata 0.42% Dubautia menziesii 0.42% Dubautia platyphylla 0.42% Dubautia herbstobatae 0.42% Wilkesia hobdyi 0.44% Dubautia linearis 0.40% Argyroxiphium caliginis 0.42% Argyroxiphium grayanum 0.42% Argyroxiphium sandwicense 0.46% average identity in allozymes, comparable to that of a single mainland species (Witter & Carr, 1988). Accordingly, except for the position of D. scabra see below), all Dollo е analyses of the cpDNA data resolved n = ubautia as mono- phyletic. Aside from the то toa am- biguity, the cpDNA consensus tree (Fig. 2) is sim- ilarly consistent with monophylesis of n = 1 Dubautia. A unidirectional, west-to-east pattern of colo- nization from Oahu to progressively younger is- lands by members of the n — 13 Dubautia lineage is suggested by cpDNA phylogenetic data (Fig. 2), albeit with limited statistical confidence from boot- T strap analysis. This migratory trend was proposed for the silversword alliance in general by Carr et al. (1989) and is supported by allozymic data at the intraspecific level in some Dubautia taxa (Wit- ter, 1986). The same pattern of migration was inferred from polytene chromosome analysis in Hawaiian picture-winged Drosophila (Carson, Instances of back-migration from younger to older islands cannot, however, be dismissed for n = 13 Dubautia. Hybridization between such im- migrants and established л = 13 species could result in fixation of a native cpDNA genome in the alien taxon. Natural interspecific hybridization among п = 13 Dubautia, resulting in highly fertile hybrids, has been documented for eight species combinations (Carr, 1985). Comparisons of phy- logenies based on biparentally and uniparentally inherited characters are critical to understanding the influence of hybridization on these species. lozymic (Witter, 1986; Witter & Carr, 1988) and cpDNA data both indicated monophylesis of the two Oahu 7 D. herbstobatae and D. sherffiana, but differed slightly in resolution of = 13 species, Mauian and Hawaiian л = 13 Dubautia relation- ships. Dubautia linearis subsp. hillebrandet was placed with D. arborea and D. ciliolata in the NA consensus tree (Fig. 2) but was exclude from the allozyme-defined clade comprising D. ar- borea, Р. ciliolata, D. menziesii, and D. platy- phylla. This conflict could indicate that the cpDNA of D. linearis subsp. hillebrandei was obtained from another Hawaiian Dubautia taxon by hy- bridization. Such hybridization may explain (1) ex- treme disparity in flavonoids between D. linearis subsp. hillebrandei on Hawaii and subsp. linearis on Maui(W. J. Crins and В. A. Bohm, pers. comm.) and (2) anatomical similarities of D. linearis subsp. hillebrandei to D. arborea (Carlquist, 1959a). In- traspecific cpDNA polymorphism within n = 13 Dubautia species, including D. linearis, lends promise that additional sampling and cpDNA anal- yses accompanied by nuclear DNA studies will enhance resolution of biogeographic relationships and hybridization among these species. PHYLOGENETIC IMPLICATIONS I. Wilkesia and Kauaian Dubautia A well-resolved cpDNA sublineage within the silversword alliance supports monophylesis of Wil- kesia and of five Kauaian Dubautia species (D. knudsenit, D. imbricata, D. paleata, D. pauci- florula, and D. raillardioides). Wilkesia is also delimited by unique alleles "i nen ips in itter, 1986; Witter & C Among Kauaian Dubautia, рен ин p s allozymes ( allozymes suggests close relationship between P. Volume 77, Number 1 1990 Baldwin et al. 107 Chloroplast Evolution in the Hawaiian Silversword Alliance knudsenii and D. pauciflorula and between D. paleata and D. raillardioides (Witter & Carr, 1988). Surprisingly, Wilkesia and the above Dubautia species together comprise a single lineage. Both species in Wilkesia, W. gymnoxiphium and W. hobdyi, have large, yuccalike rosettes born on narrow, naked stems. Leaf venation in W. gym noxiphium closely approximates the parallel. -vein condition seen in many monocots (Carlquist, 1959a; Kim, 1987). Unlike these unusual Kauaian xero- phytes, members of the cpDNA sister group of Wilkesia comprise a Dubautia lineage of Kauaian rainforest trees and shrubs having relatively typical dicot habits and leaf venation. The extensive phe- notypic contrasts between these two groups were reviewed by Carr (1985). Close cpDNA affinity between Wilkesia and these Dubautia species rep- resents a striki ple of inconsistency between relative rates of “cpDNA and phenotypic evolution. The taxonomic implication of this finding is that Dubautia is either paraphyletic (i.e., Wilkesia is derived from Dubautia) or polyphyletic, assuming that ancient hybridization does not account for this cpDNA pattern. The distribution of genome 1 species in Dubautia (discussed below), however, argues against polyphylesis of this genus (Carr & Kyhos, 1986). The other lineage of endemic Kauaian Dubautia taxa, comprising D. latifolia, D. laevigata, and D. microcephala, is well delineated by two basal restriction site gains. These rainforest species are to varying degrees sympatric with members of the previous Dubautia group. This resolution places D. latifolia, the only liana in the silversword al- liance, in close relationship with D. laevigata, a large shrub, and D. microcephala, a small tree. II. Argyroxiphium Resolved patterns of cpDNA relationship within Argyroxiphium (silverswords and greenswords) strongly suggest that either hybridization has been an active force in this genus or extreme evolu- tionary parallelism in morphology or cpDNA has occurred. These data place a bog greensword (А. grayanum) from West Maui in closer relationship to the sympatric bog silversword A. caliginis than to a presumably conspecific greensword on East Maui. In addition, a weaker cpDNA sister-species relationship between East Mauian А. grayanum and the parapatric silversword A. sandwicense was consistently resolved by Dollo, but not Wagner, parsimony. Intergeneric hybrid origin of the allopatric and somewhat morphologically distinct greenswords on East and West Maui via different stabilized com- binations of silversword and Dubautia species would account for these cpDNA patterns. This hypothesis is supported by the remarkably greensword-like appearance of the artificial F, hybrid Argyroxiph- ium sandwicense X Dubautia plantaginea (Carr & Kyhos, unpublished). The latter species occurs sympatrically with greenswords in both re- gions (Carr, 1985 Alternatively, these cpDNA patterns may indi- cate extensive introgression resulting in movement of cpDNA from silversword to greensword or per- haps the reverse. Natural hybrids have been re- ported between 4. caliginis and A. grayanum at Pwu Kukui on West Maui (Carr, 1985), where these taxa were sampled. In addition, contact be- tween A. sandwicense and A. grayanum on East Maui may have occurred until recently, prior to the decimation of these plants by feral mammals. MI. Dubautia scabra and п = 13 Dubautia A compelling argument for introgressive cpDNA transfer can be made from the cpDNA relationship between D. scabra and n = 13 Dubautia. In contrast with incomplete cytogenetic knowledge of Wilkesia, Argyroxiphium, and Kauian Dubautia taxa, extensive chromosomal data indicate genomic uniformity among n = 13 Dubautia species (Carr & Kyhos, 1986). These last taxa appear to be derived from the n = 14 D. scabra by a reciprocal chromosomal translocation and can be linke through D. scabra to other n = 14 genomes. In addition, D. scabra and n = 13 Dubautia are markedly similar in morphology and distribution, reflected by their exclusive taxonomic segregation as Dubautia sect. Railliardia. This evidence sug- gests D. scabra is a likely sister taxon to n = 13 Dubautia. Terminal placement of D. scabra in the 13 Dubautia cpDNA lineage and cpDNA identity of D. scabra and D. ciliolata at Kilauea are therefore unexpected. If these cpDNA and n= cytogenetic results both reflect organismic phylog- eny, one of two improbable scenarios must have occurred: (1) a centromere and two chromosome arms similar to those originally lost by the n = 13 group arose, concomitant with a reciprocal trans- location, to re-create the n = 14 D. scabra genome by ascending dysploidy (i.e., D. scabra is derived from n = 13 Dubautia), or (2) multiple indepen- 13 genome occurred from D. scabra or a genomically identical progenitor dent origins of the n = (i.e., D. scabra or an immediate predecessor is ancestral to n = 13 Dubautia). Field studies indicate that in nature D. scabra 108 Annals of the Missouri Botanical Garden hybri with n = 13 Dubautia species which D. plantaginea had high genetic identity in J to жата е highly fertile F, generations (Carr, 1985). Six species combinations of this type are known. Dubautia scabra and D. ciliolata occur in sym- patry in the two areas where the former was sam- pled. In addition to apparent F, hybrids occupying intermediate habitats at both sites, Crins et al. (1988) reported what appear to be introgressed flavonoid profiles from individuals at the Pu'u Hu- luhulu site. Uniformity of cpDNA between these species at Kilauea could have resulted from such introgression. ш a cpDNA afhnity of D. scabra (n = 14) ton = 13 Dubautia taxa may, however, be the al of earlier hybridization on Hawaii. Intraspecific variation between D. ciliolata subsp. glutinosa at Pwu Huluhulu and D. ciliolata subsp. ciliolata at Kilauea is not mirrored by D. scabra, which is uniform at examined restriction sites in all of several plants from both localities. Chloroplast DNA transfer from D. scabra to D. ciliolata there- fore appears a more likely explanation for this pattern than the reverse. In short, introgression may account for cpDNA identity between D. sca- bra and D. ciliolata at Kilauea, but would not account for п = 13 Dubautia cpDNA in D. scabra. The evolutionary history of these species is being further analyzed, in part, by examination of cpDNA from D. scabra on the older island Maui, where the Hawaiian л = 13 taxa D. ciliolata, D. arborea, and D. linearis subsp. hillebrandei do not occur. IV. Dubautia genome | The distribution of D. knudsenii, D. planta- ginea, D. laxa, and D. microcephala on the cpDNA tree conflicts with their common possession of ge- nome | (Carr & Kyhos, 1986). Ancient and recent hybridization together could account for these dis- crepancies, i.e., cpDNA may not reflect phylogeny. It may be significant, however, that (1) cytogenetic variation within genome l species is imperfectly known, with sampling limited to single collections from one or two populations, and (2) these taxa are all found on the oldest жола island, Kauai, concordant with genomic antiqui The existence within eee эрешен of two or more genome types differing by chromosomal rearrangements, a situation well doc umented in North American Madiinae (Clausen, 195 1975; Carr & Carr, 1983), could explain the cpDNA results. For example, Dubautia planta- chro- arr, ginea and D. laxa are a minimum of two mosomal interchanges removed from the group with which they showed cpDNA alliance and with allozymes (Witter & Carr, 1988): D. scabra/n = 13 Dubautia. A structurally intermediate genome in one of the first two species could explain evo- lutionary transition from genome | to the D. sca- bra genome. Although no intraspecific chromo- somal variation is presently known from Hawaiian Madiinae, such undetected heterogeneity may ac- count for unexpected cytogenetic results in Ar- gyroxiphium (Carr & Kyhos, 1986). Expanded cpDNA investigations and molecular analyses of nuclear DNA are needed to clarify the relative extent to which hybridization, intraspecific differ- entiation, and evolutionary parallelism have influ- enced phylogenetic patterns in this remarkable group of plants. LITERATURE CITED BarpwiN, B. G. 1989. Chloroplast DNA Phylogenetics and ese Studies in Madiinae (Asteraceae). ssertation, Univ. of California, Davis, Cali- fornia. C. W., Ja. 1988. Evolution and variation in plant ea and E genomes. Pp. 23-53 in L. D. Gottlieb & S. K. Jain (editors), Plant Evolutionary юу. Chapman and ee Dui ZARLQUIST, 57. Leaf anatom togeny in Argyroxiphium and Wilkesia оо Amer. а, 96-705. BiRK Y, —. 1958a. Wood ато of Heliantheae (Com- posie), Trop. Woods 108: 1-30. = 58b. Structure and ontogeny of glandular vichames a Maa (Compositae). Amer. J. Bot. 45 {з а. - Veuebitive anatomy of Dubautia, Ar- gyroxiphium, and Wilkesia (Compositae). Pacific Sei. 13: 195-210. 959b. Studies on Madinae: anatomy, су- tology. and evolutionary relationships. Aliso 4: 171- 236. ——. 1965. Island Life: A Natural History of the Islands of be Natural History Press, Garden City, New 974. New York. Carr, G. D. 1975. Chromosome evolution and aneu- = reduction in Calycadenia pauciflora (Astera- tion 29: 681-699. . 1985. Monograph of the Hawaiian Madiinae (Asteraceae): Argyroxiphium, Dubautia, and Wil- kesia. Alle srtonia 4: eN 123. 87. Be s ticks and tarweeds: masters | adaptive Pc Trends Ecol. & Evol. 2: 192- um Biology. Columbia Univ. Press, D. W. Kvuos. 1981. Adaptive сеш іп he Hawaiian discal alliance (Composi dinae). ILC oo of spontaneous ЫИ. he lution 35: 543-556. o a ———- E Adaptive radiation of the Hawaiian silversword alliance (Compositae- Madi- Cytogenetics of ur ial and natural hy- Evolution 40: 959-9 inae). brids. Volume 77, Number 1 1990 Baldwin et al. 109 Chloroplast Evolution in the Hawaiian Silversword Alliance , E. A. PowELL & D. W. KyHos. 1986. Self- incompatibility i in the Hawaiian Madiinae (Composi- tae): an exception to Baker’s Rule. Evolution 40: 430-434. ———, R. RonicH AUX, M. S. WirrER & D. W. KvH W. Anderson (editors), Genetics, j nn and the Founder Principle. Oxford Univ. Press, New York. Carr, R. L. & С. D. Carr. 1983. Сое ume races and structural rg a in Calycadenia ciliosa Greene (Asteraceae). Amer. J. Bot. 70: 744-755 CARSON, H. L. 1983. nal sequences and in- terisland colonizations in Hawaiian Drosophila. Ge- netics 103: 465-482 CLAUSEN, J. 1951. Stages in the Evolution of Plant Species. Hafner, New York. CLEGG, M. T., J. Y. Rawson & К. Тном А К. 1984. Chloroplast DNA тт in pearl millet and related species. Genetics 106: 449-461. CRAWFORD, D. J., R. Wurrkus & T. F. STUESSEY. Plant evolution and p- 183-199 in K. M. Urbañska (editor), Differentiation Patterns in Higher Plants. Academic Press, New York. 1987. neriati Crins, W. . A. Boum & С. D. Carr. 1988. Fla- vonoids | as да сайа: of hybridization in a mixed population of lava-colonizing Hawaiian tarweeds (As- teraceae: Heliantheae: Madiinae). Syst. Bot. 13: 567- 7 5 DEBry, R. W. & N. A. SLADE. 1985. Cladistic analysis of restriction endonuclease cleavage maps within a maximum-likelihood framework. Syst. Zool. 34: 21- 34 то W. М. & E. MancoLiAsH. 1967. Construction of phylogenetic trees. Science 155: 279-284, JANSEN, В. К. & J. D. PALMER. 1987. Chloroplast DNA from lettuce and Barnedesia (Asteraceae): structure, gene ешн. = тоа of a lar ersion. Curr. ; M59 Kim, 1 1987. ани апа Жыз of some parents and dio of vg Hawaiian Madiinae (Asteraceae). Amer. J. Bot. mo KLUGE, А. С.& лан . Quantitative phy- Zool. J.S гет and the tii. x anurans. Syst. = NEI & W. Lr 1979. Pipinus р» models for esl genetic variation in s of o cer = donucleases. Proc. Natl. I USA 76: 5269-5 PALMER, J. D. . Comparative organization of d roplast genomes. Arenal Rev. Genet. 19: 325-354. — . Isolation and structural analysis of chlo- E de DNAs. Molec t. 190: 13-19. — —— & D. Zamir. 1982. Chloroplast DNA evolution and phylogenetic relationships in Lycopersicon. Proc. Natl. Acad. USA 79: 5006-5010 ‚ G. P. SINcH & D. T. N PiLaY. 1983a. Struc- ture and sequence evolution of three wor chlo- roplast papa Бена ‚ Сеп. "as 190: . ¿CR , D. B. Сонех & T. D ORTON. 1983b. rinde DNA eroi and the origin of amphiploid Brassica. Theor. Appl. Genet. 65: -189. . JANSEN, H. J. MicHaELs, М. W. CHASE & J. R. ia 1988. Chloroplast DNA vari- = m ip phylogeny. Ann. Missouri Bot. Gard. ТЫ: 206. Е L н. D. E. Sorris & J. D. PALMER. 1988. A molecular reexamination of introgression between Helianthus annuus an olanderi (Compositae). Evolution 42: 227 ROBICHAUX, R. H. 1984. Variation in the tissue water relations of two sympatric Hawaiian Dubautia species and their natural hybrid. Oecologia (Berlin) 65: 75- 81. 1 Tissue elastic properties of а mesic forest Hawaiian Dubautia species with 13 pairs of chromosomes. Pacific Sci. 39: 191-194. & J. E. CANFIELD. 1985. Tissue elastic prop- erties of eight Dubautia species that differ in habitat and diploid chromosome number. Oecologia (Berlin) SYTSMA, К. J. & L. D. GOTTLIEB. 1986. Chloroplast DNA evolution and phylogenetic relationships in Clarkia sect. Peripetasma (Onagraceae). Evolution 40: 1248-1261. B. A. ScHaaL. 1985. Phylogenetics of the di dam. skinneri (Gentianaceae) species complex n Panama utilizing restriction fragment analysis. Paus 39: 594-608. & J. F. SmrrH. 1988. DNA and morphology: comparisons in the Onagraceae. Ann. Missouri Bot. 37. 986. Genetic differentiation in the Hawaiian дел alliance (Compositae: Madi- . Ph. D. Dissertation, Univ. of Hawaii, Honolulu, & С. D. Carr. 1988. Adaptive radiation and genetic differentiation in the Hawaiian silversword alliance (Compositae: Madiinae). Evolution 42: 1278- 287. EVOLUTION IN THE MADIINAE: EVIDENCE FROM ENZYME ELECTROPHORESIS Martha S. Witter! ABSTRACT The genetic differences observed among species contain historical information that is a function of their ancestry, node of origin, and population structure over time Dubautia. Wilkesia, and Argyroxiphium, and among the mainland species m ма аге ге enitor-derivative species pairs as well a provide examples of rapidly evolved proge . Studies of isozyme differentiation among the isla nd du of viewed. groups ecies showing gradual Haba as a function of their time of separation. À comparison of island and mainland taxa a illustrates the effect of population Madinae suggest an ancient ‘tetraploid origin for the Hawaiian specie The Madiinae offer several advantages for the study of plant evolution. They are a cohesive, monophyletic group with a distinct biogeographic center of origin, the Pacific Coast of North Amer- ica. The mainland genera inhabit most of the major community and habitat types within California. The Hawaiian genera have some of the most remark- ably distinct morphologies found within the Com- positae. Early work with the Madiinae included the classic studies of plant evolution done at the Car- negie Institution of Washington at Stanford (Clau- sen, 1962). Layia and Madia were used because they included species at various stages of evolu- tionary differentiation (Clausen, 1962). The Car- negie group worked on the assumption that the evolutionary record of a group of living species is reflected in the existing genetic relationships among species (Clausen, 1962). One of their primary means of assessing genetic relatedness was through ex- perimental hybridization studies. Biochemical ge- netic techniques have provided new means of ana- lyzing the number of genes shared by related species and have reduced the need to resort to time-con- suming hybridization programs e electro- phoresis was the first technique that allowed the number of gene differences among species to be measured directly over a wide taxonomic range (Avise, 1974; Gottlieb, 1977). Modern molecular techniques promise to allow an even greater res- olution of phylogenetic patterns and genetic di- vergence among species (Palmer, 1985). This pa- per will review the isozyme data available in the Madiinae to illustrate how the pattern and degree of genetic differentiation can be used to make in- ferences on the mode and tempo of genetic diver- gence among species. GENETIC VARIATION AND GENETIC DIFFERENTIATION AMONG SPECIES: MAINLAND TAXA In a study of the genus Layia, Clausen et al. (1941) used extensive analyses of morphological characters, cytogenetics, and fertility relationships from interspecific hybridizations. They found that one group of six species formed a complex con- sisting of three morphologically discontinuous santhemoides; and L. jonesii, L. leucop L. munzii (Clausen et al., 1941; Clausen, 1962). They proposed that these species had arisen allo- patrically, through the gradual accumulation of many small genetic changes as a result of differ- ential selective pressures in different environments. The morphologically similar species represented more recent stages in the divergence and accu- mulation of genetic differences among taxa (Clau- sen, 1962). The degree of genetic relatedness and the pat- tern of genetic divergence predicted by the model of allopatric evolution was tested using allozymes. Warwick & Gottlieb (1985) examined the species ' Laboratory of Biomedical and Environmental Sciences, University of California, Los Angeles, California 90024, Э.А, ANN. MISSOURI Вот. GARD. 77: 110-117. 1990. Volume 77, Number 1 1990 Witter 111 Evolution in Madiinae L. platyglossa -880 L.jonesii L. munzii L. leucopappa -901 L. fremonti L. chrysanthemoides 824 FIGURE 1. Summary of interpopulation and interspe- cific genetic identities (Nei, 1972) among the species of the Layia platyglossa complex. (Redrawn from Warwick & Gottlieb, 1985). of the complex and found the genetic data to be fully concordant with an allopatric model of spe- ciation (Fig. 1). Genetic distances were substan- tially higher among morphologically discontinuous species than among morphologically similar taxa. The pattern of allozymic divergence supports the concept of gradual genetic divergence through shifts in gene frequency. Figure 2 shows the distribution of alleles shared among taxa of the L. platyglossa complex. The vast majority of alleles are shared by either two or all three of the major species groups, while very few alleles are restricted to a single taxon or species group. All species of the complex are extremely polymorphic: an average of 67% of all loci are polymorphic per species, and the average number of alleles per locus is 3.4 (Warwick & Gottlieb, 1985). These data support the notion that allozymic divergence and genetic differentiation have occurred primarily as the result of changes in gene frequency from a rich, common pool of polymorphic gene loci. This pattern con- trasts sharply with that observed in island popu- lations. ISLAND TAXA The Hawaiian species of the Madiinae are re- nowned for their striking range of morphological and ecological diversification (Carr et al., 1988) The pattern of morphological and cytological dif- ferentiation is strongly correlated with the biogeo- graphic distribution of species within the archipel- ago (Carr, 1978, 1985): many of the most mor- phologically specialized and differentiated species are endemic to the oldest island of Kauai and have the ancestral chromosome complement of 14 pairs of chromosomes. The cytologically most derived species, with 13 pairs of chromosomes, occur large- ly on the younger islands of Maui and Hawaii and are much less morphologically and cytologically L. platyglossa (55) i fremontii (37) es chrysanthemoides (56) L. jonesii (39)? n parentheses. Circles represent overlapping sets of in- and all three species subgroups of the complex. Numer x not in pa- rentheses show the numbers of alleles that occur uniquely Mise individual € within individual species groups, among any two subgroups, and the numbers of alleles that occur in all three subgroups (Gottlieb, un- published data). differentiated than the Kauai endemics. Estimates of allozymic divergence agreed with the hypothesis that the Kauai species were ancestral to the 13- paired species and that the 13-paired species were of relatively recent origin (Fig. 3; Witter & Carr, 8). As in previous studies of island endemics (Crawford et al., 1985; Helenurm & Ganders, 1985; Lowrey & Crawford, 1985), the 13-paired species had extremely high interspecific genetic identities, I > 0.90. The 14-paired taxa, however, were significantly differentiated, with values more WILKESIA 2SPECIES I = 0.971 T= 0.698 (0.59 - 0.77) DUBAUTIA n=13 6 SPECIES 9 SPECIES HE 0.745 т .48 - 0.99) 9 T = 0.686 i Т = 0.950 (0.43 - 0.93) (0.73 - 1.00) T= 0.795 T= 0.894 (0.67 - 0.95) (0.77 - 0.99) D. SCABRA n=14 2 SUBSPECIES I = 0.975 Summary of interspecific genetic wn dm FIGURE 3 (Nei, 1978). among rin species of Wilkesia and Dubaut 112 Annals of th Missouri E DH Garden LOCUS: Pgic 3 b D w gymnoxiphium O W. hobdyi e R laxa e D. paleata o D. raillardioides Q D. microcephala o D. pauciflorula O:.o latifolia Q, D. knudsenii ө D. laxa as plantaginea ! ! D. plantaginea 2 ө D. laxa di O D. plantaginea — Q A. sandwicense O A sandwicense | О A. kauense null O D. scabra © D scabra KAUAI Q) |W.OAHU SON E. OAHU "d w. мдш > E. MAUI A Y @ D. sherffiana с 'D herbstobatae O D. linearis ssp. /inearis O D.reticulata О D waianapan- O о ssp. pum Q D. ciliolata Q D platyphylia O D. arborea n=13 O D. menziesii FIGURE 4. Distribution of allelic variants at the Pgic 3 locus among the species of the Hawaiian Madiinae. The frequencies of particular alleles are illustrated in pie diagrams for each spec ies. Open circles represent the frequency of a single, common allele that was found in a number of species. Closed circles represent the frequency of all other alleles found, most of which were restricted to one or two species. typical of interspecific comparisons in other plant groups (I ~ 0.67, Gottlieb, 1981; Crawford, 1983). The results obtained for the 14-paired species of the Hawaiian Madiinae showed that insular plant populations do diverge genetically, but depend on sufficient lengths of time to accumulate genetic differences (Witter & Carr, 1988) The pattern of interspecific genetic differentia- tion observed in island species appears to be the direct result of population structure. Low genetic variability in the Hawaiian Madiinae was attributed to founder events and genetic drift within small populations (Witter € Carr, 1988). Insular groups are frequently composed of small, localized popu- lations with markedly decreased levels of genetic variation. For example, the Hawaiian Madiinae are approximately one-third as variable (Р = 0.24, A, = 1.29) as the species of Layia discussed previ- ously. Figure 4 illustrates the typical distribution of alleles at a single locus among the species of the Hawaiian Madiinae and can be used to make se eral points about the effect of genetic variability on genetic divergence within the Hawaiian Madi- inae. A single, common allele usually predominates at each locus among most species, suggesting a genetically depauperate common ancestor. The low frequency of polymorphisms indicates that allo- zymic divergence will occur only as new mutants arise, rather than through gradual divergence in frequencies of multiple alleles at a given locus, as was observed in continental species of Layia. The distribution of alleles at Pgic 3 (Fig. 4), for ex- ample, illustrates that the n = 13 species of Du- bautia are all fixed for the same allele, which they presumably inherited from a single founder. The results of such a genetic bottleneck limit allozymic divergence among these taxa to the production and incorporation of new mutants. High genetic iden- tities among the Hawaiian Madiinae are therefore not only a reflection of the length of time since divergence, but are also a function of the historical demography of species with consequent effects on genetic variability and structure of populations. PROGENITOR-DERIVATIVE SPECIES PAIRS Harlan Lewis was the first to propose the rev- olutionary notion of saltational speciation in an- nuals, whereby a new species could arise rapidly and be reproductively isolated from its progenitor Volume 77, Number 1 1990 Witter 113 Evolution in Madiinae through the fixation of structural chromosome rear- rangements (Lewis, 1973). Gottlieb (1981), in one of the early and striking systematic applications of enzyme electrophoresis in plants, showed that a number of species pairs believed to be related as progenitor and derivative had very high genetic identities and that the derivative species possessed a less variable subset of the alleles present in the progenitor. Recently derived species have histori- cally been of interest to evolutionary biologists for several reasons (Ayala et al., 1975 ottlieb, 1973, 1974; Carson, 1976). Better inferences can often be made about the recent past than about the distant past. Where polarity between species pairs can be identified, the direction of evolution in Character change can be studied. Recent species pairs are also often partially or fully compatible, and interspecific hybrids can be analyzed for the genetic basis of species differences. Layia discoidea is a highly localized, morpho- logically unusual species. Its most obvious mor- phological characteristics are an absence of ray florets with their enclosing bracts and a reduction of the pappus. Originally believed to be an ancient relictual form because of its highly distinctive mor- phology, experimental hybridization showed it to be fully interfertile with Layia glandulosa (Clau- sen, 1962). Subsequent allozyme analysis indicated that L. discoidea was a recent derivative of L. glandulosa subsp. lutea (Gottlieb et al., 1985). Unlike other annual progenitor-derivative species pairs, L. glandulosa and L. discoidea are fully interfertile and freely recombining. With their dis- tinct morphological characteristics, they are ideal for the study of the genetics of characters often used to separate taxa at higher taxonomic levels (Gottlieb et al., 1985; Ford & Gottlieb, 1989). The Hawaiian taxa also have a number of species pairs that fit the criteria of recent progenitor and derivative. Both species of Wilkesia are endemic to the island of Kauai. Wilkesia gymnoxiphium is a monocarpic rosette plant on an elongated woody stalk up to 3 m at reproductive maturity. It is widespread along the upper margins and outer slopes of Waimea Canyon. Wilkesia hobdyi is much more restricted, known only from two pop- ulations on the sea cliffs of the Na Pali coast. Unlike W. gymnoxiphium, it is a multiply branched ro- sette plant which flowers repeatedly. The genetic identity between these two Wilkesia species is I = 0.971. At 8 out of 11 loci the two species share either identical monomorphic alleles or common polymorphisms, while at the remaining three loci W. gymnoxiphium is polymorphic and W. hobdyi possesses only one of the alleles of the polymor- phism (Witter, 1986). Wilkesia hobdyi appears to be a recent derivative of the more widespread W. gymnoxiphium by the criterion of a reduced subset of genetic variability present in a derivative taxon. However, another possibility is that drift due to restricted population size could produce the de- creased genetic variability in W. hobdyi. Even if evolutionary polarity for the changes remains un- settled, the two species of Wilkesia offer an ideal opportunity to study the origin, ecological dynam- ics, and genetics of the complex life history trait of monocarpy. Although all members of the group of 13-paired species of Dubautia have high genetic identities (Fig. 3), D. dolosa and D. reticulata appear to be a good case of a progenitor-derivative species pair. Dubautia reticulata is a large tree in the rain- forests of east Maui, whereas D. dolosa is a rainforest shrub. The two species have a Nei’s I = 1.00, sharing the same alleles at all loci. Several unique alleles are found only in these two taxa, supporting their common divergence from the oth- er 13-paired taxa (Fig. 5). As with the two species of Wilkesia, the polarity of the relationship cannot be established on the basis of these allozyme data. Again, however, these species offer unique exper- imental opportunities. The distinct differentiation of life form in two species with a similar genetic background can be used to examine the genetic basis of the complex character trait of life form. Ecological hypotheses to explain the adaptive na- ture of life form differences suggest that trees in a tropical wet forest have evolved in response to shading and competition for light, while shrub forms are often associated with nutritional or other stress- es (Rundel, pers. comm.). Dubautia reticulata and D. dolosa can be used to test the functional basis of the adaptational differences between trees and shrubs in this habitat. DUPLICATE GENE ExPRESSION Duplicate gene expression can result either from duplication of a single locus or through polyploid- ization. Unlinked single gene duplications may have originated in plants heterozygous for overlapping reciprocal translocations and have a high proba- bility of being unique phylogenetic markers of shared ancestry (Gottlieb, 1986). Gottlieb (1986) docu- mented two such unlinked duplications in the Ma- diinae: the first is a PGM duplication found throughout the diploid mainland genera; the second is an IDH duplication found throughout Layia (Fig. 6). A second IDH duplication occurs in two species of Layia with a cytotype of n = 8, and may prove 114 Annals of the Missouri Botanical Garden #56 9 u G&G x ra 3 x à { ыбаа є Шш 3 > т со FED 2 x x l2 = єє cC о > c c Z c = c a J J TIO J a X dJ d a =: A а з э с o с о O с A A с 4 2 12 8 6 9 18 20 C.I.: 0.98 21 22 Taxa 11 Ch 12 W ISO OC FIGURE 5. геи a ctual distribution of character states in the Distribution of unique alleles among the species of Wilkesia and Dubautia. The tree most consistent manner possible. The tree is arbitrarily rooted D SCABRM D SCABRH * D PLATYP On a 26 Tpi2 27,28 PerA format was used ae polarity assumed and does not represent an actual phylogen to be a phylogenetic marker for that section of Layia (Gottlieb et al., 1985; Fig. 6). Within Layia several other isozymic patterns are consistent with gene duplication, or could have arisen through other mechanisms, and must therefore be con- firmed by genetic analysis. The LAP isozymes ob- served in the Layia munzii complex may be useful in resolving the trichotomy among the three taxa Fig. 6). awaiian Madiinae expressed three dupli- , PGIc, and TPIc, and possibly a fourth juge: PCI (Fig. 6). Because the base chromosome number in E Hawaiian Madi- in this group ( Th cations in all taxa inae is n = 14, in a group where the ancestral condition is n = 7, it is assumed that these dupli- cations result from polyploidization and are not multiple single gene duplications (Witter, 1988). This interpretation is supported additionally by the number of PGM loci expressed in the Hawaiian axa. All Hawaiian species appear to express four PGM loci (Witter, 1988). Th of expressed loci expected from an ancestral du- e maximum number plication in the Madiinae followed by polyploidiza- PGM loci were due to single gene duplications, it would tion is four loci, the number observed. If the be necessary to hypothesize two additional single gene duplications above the ancestral Madiinae PGM duplication. The majority of 13-paired Dubautia species ex- hibited clear evidence of duplicated ADH activity, whereas there was no evidence of expressed du- plications among the 14-paired species (Witter, 19 The complex genetic model necessary to explain the pattern of phenotypic variation of ADH isozymes of the Hawaiian Madiinae needs further genetic analysis; however, the 13-paired species are clearly differentiated in their pattern of dupli- cated ADHs (Fig. 7), and these may be a useful phylogenetic marker among these species which are otherwise difficult to differentiate cytologically or isozymically. Most interesting is the differentia- tion of the three subspecies of D. linearis, the most widespread and variable of the 13-paired taxa. PHYLOGENETIC AND BIOGEOGRAPHIC INFERENCES FROM THE SYSTEMATIC DISTRIBUTION OF UNIQUE ALLELES IN THE HAWAIIAN ТАХА Argyroxiphium and Wilkesia each have a num- ber of alleles that are unique to the genus and common to all taxa within the genus (Fig. 8). Du- bautia, in contrast, has a large number of alleles unique to the genus, but these are largely restricted to single species or closely related species groups Fig. 5). With the exception of a single PER notype common to D. scabra and the 13-paired — species of Dubautia, and a single polymorphism, Tpi2*/Tpi2*, common to all lineages except D latifolia, virtually no alleles define large clades within Dubautia (Figs. 5, 8). This suggests that some of the modern species of Argyroxiphium and h more recent than some of the lineages within Dubautia. In partic- Wilkesia are presumably muc ular, the high genetic identity between the two 115 Evolution їп Madiinae Witter Volume 77, Number 1 1990 "e[qe[^? езер Neues ou ‘иоцеоцапр jo изәцва ord£joueuq; 1901 pexuqu Geog '[ :uro1j әле елед 'әвипрер{ ayi jo e1eued uenemep pue pue[ureur ayi ur uorssa1dxe əuə3 ojeoidnp SuLMoys səuə3 jo uonnquisiq 99111105) “3 (9861) e «9WD4 П ж “sIoquiás (9861) JONI `S (8861) INIM ^ (S861) FINED 29 Pm A ^e (S861) Te 19 "9 ячпо1у] - ASUMEN "y H *ii*udáeid q £129 mneqnq *Iz9 tineqnq "1-0 ЯУМПаУИ NYIIYAAVH PRO ^M unrnqdroum3 ewoxna зочта #140002 Py 501071410 ОМУЛМІУИ "I п п munu ~] взәаог 71 maowasj -1 sopromaqyawsésy> -< g=u =ч МАУП VIAV'I 116 Annals of the Missouri Botanical Garden D. reticulata D. sherffian D. dolo e D. herbstobatae D. linearis D. ciliolata cabra D. linearis ssp. opposita . arborea | D. latifolia ssp. hillebrandii D. platyphylla D. menziesii DUPL. DUPL DUPL. y ADH1/ADH2 КОИ | ADHI/ADH2 POLYMORPH ADH ADHI ADH3 FiGURE 7. Differential expression of ADH isozymes among the species of Dubautia according to the genetic model of Witter (1988). All species are n = =. M 2 Е " и LL 2 = c c c < e 3 £ § Š Н * 5 $ 4 5 © x 3 E 3 Ы = E OS 7 2 © Q Q Q < аРЕК Dy aPER Оз ТР12® ТР128-“ ТР11 81 aPER E ТР15© ¡GOT3“ cae aPER A, arid GOT2*? PGI4^ id MN Ш Monomorphic locus а icut E Polymorphic locus сотта PGI39 FicunE 8. е: ibution of E alleles among the genera of the iian Madiinae. The tree format is used only to illustrate ө distribution of alleles and does not represent actual phylogenetic relationships. 13, except D. latifolia and D. scabra species of Wilkesia (Fig. 3) is significant to un- derstanding the diversity within the Hawaiian Ma- diinae. Biogeographically and genetically (Fig. 5) the genus Wilkesia appears to be relatively ancient within Hawaii. However, speciation within the ge- nus itself has been a much more recent phenom- enon, and evolution even on the oldest major Hawaiian island, where Wilkesia is endemic, has not been static. The modern diversity of the Hawai- ian Madiinae exists not only as the result of the accumulated products of ancient evolutionary events, but also because even the most ancient groups have continued to diversify and speciate. In considering the distribution of unique alleles among the species of Hawaiian Madiinae, the shared alleles are generally taken to indicate common an- cestry. However, accurately reconstructing phy- logenetic relationships by the distribution of unique alleles is confounded by the fact that alleles orig- inate in a polymorphic state and may persist for long periods as polymorphisms. Under these con- ditions the random loss of a synapomorphic allele may be fairly frequent, and true members of a clade may be excluded due to loss of alleles. Figure 5 illustrates just one of the possible tree topologies that can be generated using unique alleles with the Volume 77, Number 1 1990 Witter 117 Evolution in Madiinae MacClade program (Maddison & Maddison, 1987). It shows that a single isozyme character is consis- tent with the cytological, morphological, and dis- tributional data supporting D. scabra as the 14- paired species most closely related to the 13-paired taxa of Dubautia. Several species pairs that mor- phologically appear to be closely related are sup- ported by the enzyme data: D. reticulata and D. dolosa; D. sheriffiana and D. herbstobatae; D. paleata and D. raillardioides; D. pauciflorula and D. knudsenii (Fig. 5 LITERATURE CITED AVISE, J.C. 197 Pi le value of electrophoretic data. Syst. Zool. 23: 465-481. AYALA, К. J. 1975. Genetic differentiation during the speciation process. Evol. Biol. 8: 1-78. Carr, С. D. 1978. Chromosome numbers of Hawaiian flowering plants and the significance of у in selected taxa. Amer. J. Bot. 65: 236-2 Monograph of the Ehe. Madiinae (Asteraceae): сего Dubautia, and Wil- 3. — viden. M. 5. WITTER € D. W. Күн 1 aptive radiation of the Hawaiian silversword alliance (Compositae-Madiinae): a com- with the Hawaiian picture-winged Drosoph . W. Anderson pre Genetics, Speciation, and the Founder Principle. Oxford Univ. Press, New Yor CARSON, H. L. 1976. t between newly formed species. BioScience 26: 700-701. CLAUSEN, J. 1962. Stages in the Evolution of Plant ig od Publ., New York. . D. k & W. M. cor 1941. Ex- perim mental зооту, Carnegie Inst. Washington Year Book 40: 16 CRAWFORD, D. J. 1983. ү A EE systematic infe erences from electrophoretic studies. Pp. 257- 287 in S. D. Tankesley & T. J. Ort rton y „бын in Plant Genetics and Breeding, Par Elsevier, New York. . STUESSEY & M. SiLvA O. 1985. Allozyme divergence and evolution of Dendroseris (Composi- tae) on the Juan Fernandez Islands. Amer. J. Bot. -948. 72: 947 1973. Genetic pic sym- a diploid species 5 GOTTLIEB, L. D. patric speciation and the origin of E pe mer. J. ; 4. Genetic confirmation of the origin of inre ue Evolution 28: 244-250. —— Electrophoretic evidence and plant sys- tematics. Ann. Missouri Bot. Gard. 64: 161-180. 1 a ped evidence and plant pop- ulations, Progress. Phytochem -45. 1986. Phosphoglucomutase and isocitrate de- hydrogenase gene ic da in Layia (Compositae). Amer. J. Bot. 74: 9-1 , S. 1. WARWICK & V.S. Forp. 1985. Mor- phological and electrophoretic divergence between аз discoidea and L. glandulosa. gia Bot. 10: фай © ec HELENURM, K. & F. К. GANDERS. 1985. Adaptive ra- diation and genetic differentiation in Hawaiian Bi- dens. Evolution 39: 753-765. The origin of diploid neospecies in Clarkia. Amer. arto 107: 161-170, D. J. CRAwronp. 1985. Allozyme divergence and evolution in Tetramolopium (Com- positae: Astereae) on the Hawaiian Islands. Syst. Bot. 10: 64-72 MADDISON, W. P. & D. R. MappisoN. 1987. MacClade, à Har. e netic distance between populations. Amer. таби 106: 283-292. Estimates of average heterozygosity "m genetic pr for a small number of individ- uals. Genetics 89: 583-590. Pu J. D. 1985. Chloroplast DNA and molecular phylogeny. Bioessays 2(6): 263-267. Warwick, S. I. & L. D. GOTTLIEB. 1985. Genetic divergence and geographic speciation in Layia (Com- positae). dip 39: 1236-1241. WITTER, M. S. Genetic differentiation in the . D. Dissertation, Univ. of Hawaii, Honolulu, i. 1988. age expression of biochemical gene markers in the Hawaiian silversword w pos ol. Ee Beds Biochem. Syst. Ec -39 £^ C. D. ee 1988. Genetic ү Жөн с in the Hawaiian silversword alliance жа omposit Madiinae). чя 42: 1278-1287 NEW THELYPTERIS Alan R. Smith' (THELYPTERIDACEAE) FROM CENTRAL AMERICA ABSTRACT Studies on Thelypteris ч for Flora Mesoamericana have revealed the following eight new species, which are herein describe ypteris aureola, T. barvae, T. chiriquiana, T. croatii, T. grayumii, T. longisora, T. redunca, and T. MESSER me (subg. Goniopteris) aureola A. R. lamina of T. aureola. A few other species of The- nov. TYPE: Honduras. Yoro: Rio /ypteris apparently have entirely dunes vari- Pelo, Cordillera de Mico Quemado, 60-380 ants in an otherwise hairy species, e.g., . (Me- m, 19 Mar. 1923, Ames 136 (holotype, US). niscium) longifolia (Desv.) R. Tryon, pur it is Figure 10-Е. possible that this transformation is under simple ncn control. Nevertheless, it is a very striking Ex affinitate T. praetermissae (Maxon) А. R. Smith, nj) 4 and unusual feature and instantly makes the species | nicaraguensis (Four orton, et specierum affinium glandibus stipitatis aureis vel auriantiacis abax- recognizable. ialiter abundantibus in axibus laminis sorisque distinguen- i Thelypteris (subg. s barvae А. R. Smith, sp. nov. TYPE: Costa Rica. Heredia: along Rio San n Atlantic slope of Volcán Barva, 10?13'N, 84?05'W, 12 Apr. 1986, Grayum 7012 (holotype, UC; isotype, MO). Figure 2А-С. Rhizome not known, probably creeping; fronds 65 cm; petiole 32 cm х З mm; lamina 32 cm, with a conform apical pinna; buds lacking; pinnae 11 pairs, to 15 X 2.5 cm, incised ca. % their width, proximal ones abruptly narrowed but still lobed at their base; segments 4-5 mm wide, con- tiguous, suboblique and subfalcate; veins ca. 15 Thelypteris jimenezii affinis, a qua imprimis differt pairs per segment, proximal 2 pairs from adjacent pinnis reductis basi laminae paucioribus (ca. 4-jugatis), segmentis latioribus sinubus angustioribus, textura crassi- segments connivent at sinus; indument abaxially ore, lamina abaxialiter parce pilosa, soris suboblongis. on rachis, costae, veins, and laminar tissue of abun- dant golden or orangish, short-stipitate glands 0.1 Rhizome erect; fronds ca. 60 cm; stipe 13 cm mm, also with acicular hairs to 1 mm and stellate х 2-3 mm, at base with numerous, ovate, persis- or furcate hairs less than 0.1 mm on costae and tent, shiny scales, these glabrous or with adpressed rachis, adaxially the laminar tissue with scattered hairs 0.2-0.3 mm on surface; lamina ca. 45 cm, sessile or short-stipitate glands; sori subcostular, proximally with 4 pairs of subabruptly reduced exindusiate; sporangia and receptacle with numer- pinnae, lowermost ca. l cm; pinnae opposite or ous short-stipitate golden or orangish glands. subopposite, ca. 17 pairs, to 8 x 1.8 cm, incised o ca. 1.5-2 mm from costa; aerophores lacking; Known only from the type. : A P 5 ин : Led "^ segments 5 mm wide; veins 6-7 pairs per segment; The lamina color in T. aureola is lighter, yel- . : ; indument abaxially on rachis, costae, and costules lower green than in closely related species. Pu- ; . . . B ү pet of adpressed hairs 0.2-0.3 mm, laminar tissue bescence on the costae abaxially, a mixture of f | E | . abaxially glabrous or with a very few similar ad- short, simple and furcate hairs less than 0.1 mm . . - pressed hairs; sori oblong or elongate along distal and stout simple hairs greater than 0.5 mm, sug- third of veins, exindusiate. gests affinity with T. praetermissa (Maxon) A. R. Smith. Other close relatives include T. minor (C. Known only from the type, in primary forest in Chr.) A. R. Smith and T. nicaraguensis (Fourn.) low, damp places. C. Morton. However, all of these are eglandular Perhaps most closely related to T. jimenezü (as are most other members of subg. Goniopteris), | (Maxon € C. Chr.) C. Reed but differing by having in stark contrast to the copiously stipitate-glandular broader segments with narrower sinuses, thicker ' University Herbarium, University of California, Berkeley, California 94720, U.S.A ANN. Missouni Bor. GARD. 77: 118-124. 1990. Volume 77, Number 1 Smith 119 1990 New Thelypteris from Central America ES Y W 2 ee, KF ae = EET eom FIGURE 1. New species of Thelypteris. A-C. Thelypteris croatii (Croat 21951 UC) Lamina apex. — C. Ultimate segments, with detail of hairs. D-E. pinna. — E. Ultimate segments, with detail of hairs and laminar glands. F-H. Thelypteris redunca (Croat & Hannon 64239 UC). —F. Proximal pinna. —G. Ultimate segments, with detail of hairs. — H. . — A.Proximal pinna.— B. Thelypteris aureola (Ames 136 US). — D. Proximal 120 Annals of the Missouri Botanical Garden Men. “we di SS 7 BLE Prone \ T Ñ FIGURE 2. pinna. MO). — D. Habit. — E. Medial pinna. lamina, laminar tissue glabrous abaxially and only sparingly hairy adaxially, elongate sori, and fewer reduced pinna pairs. In blade cutting, it greatly resembles certain species of sect. Uncinella, e.g., T. atrovirens, but hamate hairs are entirely lack- ing. Thelypteris (subg. Amauropelta) chiriquiana A. В. Smith, sp. nov. TYPE: Panama. Chiriqui: Distrito Bugaba, Cerro Punta, from STRI house New species of Thelypteris. A-C. Thelypteris barvae (Grayum 7012 UC). —A. Habit — В. Medial C. Ultimate segments, with detail of hairs. D-E. Thelypteris grayumii (Haber ex Bello & Lierheimer 4532 to edge of mountain across river, 8%52'N, 82%33'E, 2,200 m, 24 Jan. 1985, van der Werff & Herrera 6319 (holotype, UC; iso- type, MO not seen). Figure 3A-C. Species indusiis amplis persistentibus fuscatis, pinnis subinaequilateris, stipitibus nitentibus, atropurpureis vel nigrescentibus, lamina e im pi i zomate longo erecto gracili a congeneribus diversa. Rhizome erect, caudex to 10 x 1 cm; fronds to 80 cm; petiole to ca. 15 cm X 2 mm, purple- Volume 77, Number 1 121 1990 Smith New Thelypteris from Central America Ys Y YS Y N le AN XN N S 8 }, N Ns i E NW NN 4 ДИЙ nn FIGURE З. New Species of Thelypteris. A-C. Thelypteris chiriquiana (van der Werff £ Herrera 6319 UC). — . Habit. — B. Medial pinna. —C. Ultimate segments, with detail of hairs. D-E. Thelypteris longisora (van der Werff & Herrera 6375 UC). — D. Proximal pinna. —E. Ultimate segments, with detail of hairs. black, shining, glabrous or with sparse hairs 0.1— slightly inequilateral with segments on basiscopic 0.2 mm, at base with castaneous, shining, glabrous side shorter and more oblique; aerophores absent scales to 5 X 1.5 mm; lamina to ca. cm, or very weakly developed; segments ca. 3 mm wide; proximally with ca. 6 pairs of subabruptly reduced veins to 6 pairs per segment; indument abaxially pinnae, lowermost vestigial; pinnae to 6 X 1.5 cm, lacking or of sparse hairs 0.1-0.2 mm on costae, 122 Annals of the Missouri Botanical Garden laminar tissue glabrous on both sides or sparsely short-hairy adaxially; sori medial to supramedial; indusium tan to dark brown, persistent, glabrous. Known only from the type, from cloud forest. Affinities of this species are uncertain. In the relatively large, darkened indusium and slightly inequilateral pinnae, it resembles 7. cinerea (So- diro) A. R. Smith but differs by having shining, purplish black stipe, eglandular lamina, and long, slender, erect rhizome specimen van der Werff & Herrera 6455 (UC), from the same area as the type, is similar but differs in the more numerous hairs on rachis and costae abaxially; scattered castaneous, linear scales to 0.5 mm on costae abaxially; tuberculiform aerophores; ciliolulate indusia; and creeping to ul- timately suberect rhizome. These differences may be sufficient to recognize it as a different species, but more material is needed. The two are more closely related to each other than to other 7he- lypteris. Thelypteris (subg. Goniopteris) croatii A. R. Smith, sp. nov. TYPE: Panama. Chiriquí: Buri- ca Peninsula, Rabo de Puerco, 8 km W of Puerto Armuelles, 50-150 m, 18 Feb. 1973, Croat 21951 (holotype, UC; isotype, MO). Figure 1A—C. Ex affinitate T. biolleyi (Christ) Proctor et specierum affinium pilis ашшы carentibus, venis infimis ай 1m conn 1s cuneata distincta; a 7. nephrodioides (Klotzse ah Proctor soris exindusiatis vel indusiis minutis dignoscenda Rhizome erect; fronds ca. 1 m; stipe ca. 45- 60 cm; lamina 50 cm, with a confluent, pinnatifid apex; buds lacking; pinnae ca. 20-25 pairs, to ca. 15 x 2.5 ст, incised ca. Y,—Y, their width, proximal ones abruptly narrowed toward their long-cuneate base; segments ca. 4-6 mm wide; veins ca. 10- 13 pairs per segment, proximal pair from adjacent segments connivent at sinus, rarely united below sinus; indument abaxially on rachis, costae, and veins entirely of sessile stellate hairs 0.1 mm, lam- inar tissue verrucose, with appressed stellate hairs on both sides, adaxially the veins with curved acic- ular hairs on costules; sori exindusiate or with an indusial fragment; sporangia with sessile stellate and simple hairs. Paratypes. Costa RICA. PUNTARENAS: between Las Cruces Botanical Garden and Rio Jaba, ca. 3 km SE of San Vito de Coto Brus, 8?47'N, 82%58' W. "E 050-1,200 m, Grayum 5606 (МО); San José: 13 km SW of San Isidro on road to Dominical, 800 m, 10 Aug. 1967, Lellinger 878 (MO, US ant seen). PANAMA. CHIRIQUI: along road from Puerto Armuelles to San Bartolo Limite, 7 mi. W of Puerto Armuelles, ca. 120 m, 19 May 1976, Croat 35064 (MO This belongs to the group of T. biolleyi (Christ) Proctor but differs in the complete absence of an- chor-shaped hairs and in the lowermost pair of veins connivent at the sinus, occasionally united below the sinus. From T. nephrodioides (Klotzsch) Proctor, T. croatii differs in being exindusiate or with only a minute fragmentary indusium. From both it differs in the narrowly cuneate pinna bases. Thelypteris (subg. Amauropelta) grayumii A. mith, sp. nov. TYPE: Costa Rica. Heredia: along Rio San Rafael, Atlantic slope of Volcan Barva, 10?13'N, 84%05'W, 1,500 m, 12 Apr. 1986, Grayum 7042 (holotype, UC; isotype, MO not seen). Figure 2D-E Thelypteris villanae affinis, a qua ани differt pin- nis infimis 1-2 paribus pinnarum aliquantum reductarum, 3-5 mi n longis (non valde reductis), asi stipitum exsqua- matis, pinnis integris vel crenulatis, ascendentibus, basin versus cuneatis. Rhizome suberect; fronds 5-15 em, fasciculate; stipe up to 2 cm X 0.7 mm, tan to brownish, glabrous; lamina to ca. 13 cm, proximally with 1 or 2 pairs of slightly shortened pinnae, lowermost ca. 3-5 mm long, distally the lamina pinnatifid and somewhat attenuate; pinnae alternate, often strongly ascending, up to ca. 15 pairs, up to 10 x mm, narrowed at the base, entire to crenulate, sometimes with a single acroscopic, nearly free basal lobe; aerophores lacking; veins simple from the costae, up to ca. 10 pairs per pinna; indument completely lacking on both sides of lamina; sori oblong or elongate along the proximal /,-% of t veins, exindusiate. Paratype. Costa RICA. PUNTARENAS: Monteverde Cloud Forest Reserve, road to TV towers, 10%20'N, 84°50'W, 1,600 m, 23 Apr. 1986, Haber ex Bello & Lierheimer 4542 (MO). This is probably most closely related to 7. vil- lana L. D. Gómez, from which it differs in fronds lacking strongly reduced proximal pinnae (2-4 pairs reduced in 7. villana, the lowest ca. 1 mm long), in the completely glabrous stipe (scaly stipe bases in T. villana) in the more ascending pinnae that are narrowed at the base, and in growing at higher elevations (200-1,100 m for T. villana). The ho- lotype was noted as growing on mossy boulders in a river. The species is named for its collector, Dr. Mi- chael Grayum, who brought it to my attention as possibly new, and who has made many important collections of and observations on Costa Rican pte- ridophytes. Volume 77, Number 1 1990 Smith 123 New Thelypteris from Central America Thelypteris (subg. Amauropelta) longisora A. R. Smith, sp. nov. TYPE: Panama. Chiriqui: Distrito Bugaba, Cerro Punta, from STRI house to nearby ridge, 2,200 m, 8*52'N, 82°33'%, 25 Jan. 1984, van der Werff & Herrera 6375 pe fl UC; isotype, MO not seen). Figure D-E. da soris elongatis secus venas, frondibus ad 2 m longis, , sinubus inter segmen nta latis, pilis hamatis carentibus, abaxialiter squamis appressis amorphis secus costas a congeneribus diversa. Rhizome erect; fronds to 2 m; stipe length un- known, 5+ mm diam.; lamina to ca. 1.5 m, proxi- mally with 5+ pairs of abruptly reduced pinnae, lowermost vestigial; pinnae 30 pairs or more, to 20 X 3 cm, incised within 1 mm of costa; aero- phores lacking or tuberculiform; segments to 5 mm wide, sinuses very broad; veins to 20 pairs per segment; indument abaxially of dense, short hairs 0.1 mm long on rachis, costae, and costules, also with a few appressed amorphous costal scales (ir- regular in shape and lacking obvious cell definition) to 1 mm, lamina adaxially with spreading slightly reddish hairs 0.2-0.3 mm; sori linear, running along half to nearly the entire length of veins, confluent and completely obscuring the lamina at maturity; exindusiate. Known with certainty only from the type, from cloud forest. In the flora area, only T. linkiana (C. Presl) R. Tryon and 7. atrovirens (C. Chr.) C. Reed have the sori so elongate along the veins, but these species both bear hamate hairs on the axes abax- ially. Other characters distinguishing T. longisora are the very broad sinuses between segments, deep- ly incised pinnae, fronds to 2 m with widely spaced pinnae, and appressed, amorphous scales on the costae abaxially. e specimen Smith et al. 2446 (UC), from the same general area as the type, is nearly iden- tical in lamina dissection and sorus shape but differs in the costae and rachis abaxially lacking hairs and having more numerous costal scales. Thelypteris (subg. Goniopteris) redunca A. R. Smith, sp. nov. TYPE: Honduras. Olancho: along Rio Olancho, on road between San Francisco 86°07'W, 1,300 m, 6 Feb. 1987, Croat & Hannon 64239 (holotype, UC; isotype, MO not seen). Figure 1F-H. Thelypteris biolleyi affinis, a qua imprimis differt venis infimis ad s sinu m ma incisis, indusiis magnis persistentibus rufo-brunneis, pilis abaxialiter cines et densioribus ad laminam ve- nasque. Rhizome apparently short-creeping; fronds ca. 140 cm; petiole ca. 70 cm X 5 mm; lamina ca. 70 cm, gradually narrowed distally to a confluent, pinnatifid apex; buds lacking; pinnae ca. 20 pairs, X 3 cm, deeply incised to within 2 mm of costa, basal segments of proximal pinnae free or nearly so, reduced; segments 4—5 mm wide, acutish at apex, subfalcate; veins to ca. 15 pairs per seg- ment, proximal pair from adjacent segments meet- ing margin at or just above sinus; indument abax- ially of dense anchor-shaped hairs 0.2-0.3 mm and stellate hairs on costae, veins, and laminar tissue, adaxially with appressed stellate hairs; sori with a large, reddish brown indusium, this stellate- hairy; sporangia glabrous. Known only from the t This is most closely аА to Т. biolleyi (Christ) Proctor but differs in having the lowermost pair of veins meeting the margin at the sinus; deeply in- cised pinnae; large, persistent, reddish brown in- dusia; and denser covering of anchor-shaped hairs on the lamina and veins abaxially. Another close relative, 7. nephrodioides (Klotzsch) Proctor, dif- fers in lacking anchor-shaped hairs and in being less deeply incised. Thelypteris croatii differs in lacking anchor-shaped hairs and in being nearly or quite exindusiate. Thelypteris ы Amauropelta) subscandens A , Sp. nov. TYPE: Costa Rica. Her- edia: N se of Cerros Las Marias, N slope of Volcán Barva, 10?10.5'N, 84°06.5'W, 2,100- 2,380 m, 19 Apr. 1986, Grayum 7273 (ho- lotype, UC; isotype, MO not seen). Figure 4. Species rhizomate scandenti, ad 1 m x 4 mm, rhachidi costis costulisque abax e о hamatis 0.2-0.3 mm praeditis, lamina adaxialiter pilis 2 T T 3 ' g. [a E L | E astaneis vel Bam e owi epilosis a congeneribus iem Rhizome prostrate to subscandent (erect on smaller plants), to 1 m X 4 mm, bearing ovate to ovate-lanceolate, castaneous, glabrous, shining scales 2-3 x э mm; fronds ca. 75 cm; stipe to ca. 15 cm х 2-3 mm, dark brown at base, with numerous hairs 0.1-0.3 mm, these slightly reddish, often + appressed; lamina to ca. 60 cm proximally with ca. 4 pairs of abruptly vaksal pinnae, lowermost ca. 1 mm; pinnae alternate, ca. 30 pairs, to 10 x 2.2 cm, deeply incised to within 0.5 mm of costa; aerophores lacking; segments ca. 3 mm wide, suboblique, margin entire to crenulate; veins to ca. l0 pairs per segment; indument abax- ially on rachis, costae, and costules of spreading, hamate hairs 0.2-0.3 mm, laminar tissue glabrous 124 Annals of the Missouri Botanical Garden ГИР бае С ham m ЗИЛИ qur S NV 3335572074 5220144 eed SS е d" S i 1 А SS RSEN y ПАДА 1, уму е EZ 77 А LLL LA SN MSN n, pain, ONTO OO ШҮҮ Эн Е? ТИЛИНИ {и LLL 1212222 2 1a 2429 у М N И ae UN КЫЫ лд as a [ hn S MN МАМЫ SAL "tete d X ' »» y ҮШ, dl 1 SIMI Sn no» NICE C 2022/2245 4.477 SS EZ Sa NN AA Sx) RED SAN A М FIGURE 4. with detail of hairs. —D. or nearly so on both sides; sori round, supramedial, with a relatively large, persistent indusium ca. 0.7- 1.0 mm diam., this dark-castaneous to blackish with a very narrow tan margin, minutely glandular at the margin, epilose. Known only from the type, from cloud forest on ridge. The hamate hairs suggest that this species be- Thelypteris subscandens (Grayum 7273 UC). — Sorus. — А. Habit. — B. Medial pinna. — C. Ultimate segments, longs to sect. Uncinella. However, T. subscandens differs remarkably from all known members of the section by the scandent rhizome. Also, it lacks adpressed hairs on the lamina adaxially, a character found in most other members of the section. In addition, it bears small castaneous scales on the costae abaxially. In sum, it is one of the more distinctive species in the section, perhaps in all of the subgenus, and close relatives are not apparent. THE GRASSES OF CHIQUITANIA, SANTA CRUZ, BOLIVIA! Timothy J. Killeen? ABSTRACT A checklist is provided for the Gramineae of the Brazilian Shield region in eastern Santa Cruz, Bolivia. Annotations include keys to genera and species, a regional synonymy, notes on habitat preference, geographic distribution, cytology, Der. and palatability for 275 species of tropical grasses; six new taxa are described: Andropogon sanlorenzanus, Eragrostis chiquitaniensis, Eriochrysis X concepcionensis, Paspalum kempffii, Schizachyrium beckii, and Thrasya crucensis. Chiquitania is a region in eastern lowland Bolivia which includes the Provinces of Nuflo de Chávez, Velasco, and Chiquitos in the Department of Santa Cruz. Recognized as a distinct cultural and geo- graphic region, it has a unique natural history when compared with other parts of eastern Bolivia. Sit- uated on the extreme western edge of the Brazilian (Precambrian) Shield, it is characterized by a vari- able topography (altitude 300-1,240 m) that sup- ports a high diversity of forest, savanna, and sa- vanna wetland communities. Encompassing икенен 170,000 km’, Chiquitania is bor- on the south by the Serranias Chiquitanas, to a west by the alluvial plains of central Santa Cruz, to the east by the Gran Pantanal of Brazil, and to the north by the Department of the Beni (Fig. 1). Although it was colonized over 250 years ago by Jesuit missionaries, the vegetation of the region remains largely unaltered as the local in- habitants have utilized the native savanna for cattle production while exploiting the forest for timber, rubber, or wildlife. Shifting agriculture, while com- mon, does not currently pose a threat to the existing forest vegetation. However, the recent introduction of exotic forage grasses adapted to forest soils threatens to disrupt traditional patterns of agri- culture. Due to increased productivity and better nutritional quality, these grasses allow for higher stocking rates and the easier management of cattle. Consequently, forest destruction for pasture estab- lishment is occurring at accelerated levels. It is hoped that a better understanding of the native grasses will promote range management techniques which improve the productivity of savanna vege- tation, while at the same time helping to preserve some natural communities. This checklist is the partial result of three years of work in the region studying the autecology of the native grasses. Analytic results of the autecological investigations, as well as general descriptions of the vegetation, will be reported in future publications. CLIMATE The region has a typical savanna climate or "Aw" in the Koppen classification. The regional airport at Concepcion (16%03'S, 62°10'W; altitude 500 m), has meteorological data spanning 35 years (Guaman & Valverde, 1982). Mean annual pre- cipitation is about 1,200 mm with a marked dry season of five months during the austral winter. Annual fluctuations vary greatly, ranging from 700 mm to 1,500 mm. Mean daily temperature varies only slightly por the dini reaching a max- imum in Novem 6°C) and a minimum in June (21°С). Strong cold E sweep through the region during the dry season, causing the temperature to drop to 10°С for short periods. Maximum temper- atures of about 33°С are common in November at the beginning of the rainy season. ' This research was financed by a fellowship from the Organization of American States (OAS) and grants from the World Food Institute of Iowa State University, the pa Museum of Natural ue ii the Amer rican Society of Plant Epl and the Smithsonian Institution. I am in Richard Pohl provided wise counsel and moral support throughout my investigations. James Solomon, Stephan Beck, and Monica Moraes provided logistic support during my residence in Bolivia. ? Department of Botany, Iowa State University, Ames, Iowa 50011, U.S.A. ANN. Missouni Вот. Garb. 77: 125-201. 1990. 126 Annals of the Missouri Botanical Garden | pr Ail | | | | C 63 61° 59 „И ut Y ° a: | . А 14 — «ЫЎ % л Parque Nacional pu \ * ^ Noel Kempff Mercado * E e Prva. Мино _ De | deChavez | Т8 d i | N his | Б, 18° — | Pra. ш i n San Jose а — Santiago o, 4 = s ^ кыз, FIGURE 1.—A. Ma capitals; dots are other a villages and cara border with Brazil, dot lice de ema 5 a a dash VEGETATION The vegetation in Chiquitania is similar to that described for the adjacent regions of central Brazil and, although Bolivians have developed their own vernacular terms for describing the vegetation in Chiquitania, I use the standard terminology de- veloped by Brazilian ecologists (Eiten, 1972, 1978; Ratter et al., 1988). Cerrado refers to a complex of intergrading communities that range from low ap of South America showing Bolivia C. En in detail: stippled area a Some pe to the Brazilian со Shield; ates the Department of San provincial boundaries, and a dotted line a i boundary of B Nacional “Noel Kempff Merc a. — B. Geo ographic position of Chiquitania within Bolivia. — ; stars represent provincial solid line marks the international Cras rom the Beni, е m serrania s or hill c forest to open grassland and that are characterized by the presence of tortuous, woody species with thick stems and coriaceous leaves. Cerrado is the most common savanna vegetation in Chiquitania and is usually restricted to eroded Tertiary plan- ation surfaces or the upper slopes of low hills (500- 700 m). Easily recognized due to its distinctive physiognomy, these well-drained savanna com- munities are western disjuncts separated from the cerrado province of central Brazil by the more Volume 77, Number 1 Killeen 127 Grasses of Chiquitanía, 1990 Santa Cruz, Bolivia TABLE 1. A comparison of the floristic similarity of the grasses of Chiquitanía with other neotropical floras. Species Index ota i of simi- Geographic region Source species: common larity? Chiquitania This study 250 — Andean Piedmont of Santa Cruz, Bolivia Foster (1966) and others? 120 100 53 Llanos de Mojos, Beni and northeastern Beck (1984) 112 78 43 La Paz, Bolivia Haase & Beck (1989) Paraná, Brazil Renvoize (1988) 329 120 40 Bahia, Brazil Renvoize (1984) 331 114 39 Santa LU Brazil Smith et al. (1982) 343 108 36 Costa Pohl (1980) 392 108 33 Entre Ríos, Argentina Burkhart (1969) 319 79 28 Central Andes Hitchcock (1927) 642 123 27 Tucumán, Argentina Türpe (1975) 298 57 21 ' Excludes cultivated grasses and those species Vid by the authors that occur only in adjacent regions. ? Sørensen (1948): es in common to А & index — (# o doe in А + х 1 of species in BX0.5 ы * Consisting mainly of the collections of José Steinbach as cited by Foster (1966) and additional collections known to the author, particularly those of Michael Nee (NY). * Excludes species recorded only from the Andean Piedmont of Santa Cruz (i.e., the collections of José Steinbach). abundant, seasonal forest vegetation. The Bolivians refer to this vegetation as pampa arbolada or arbolada. Floristically and structurally distinct from cerrado is a xerophytic savanna complex known as campo rupestre (Eiten, 1978). This treeless savanna occurs as islands restricted to ridge tops 800- 1,200 m above sea level. There are a number of campo rupestre localities in eastern Santa Cruz, of which the western escarpment of the Serranía de Caparuch (i.e., Parque Nacional “Noel Kempff Mercado”) is probably the most extensive. The Serrania de San Lorenzo, situated near the town of San Javier, is only 40 km from the western edge of the Brazilian Shield (16?15'S, 62?40'W; altitude 900 m) and is one of the most western campo rupestre localities in South America. This site has yielded two new grass species and several unique intraspecific variants; undoubtedly, more taxa will be described as the flora of this serrania is docu- mented. The locals refer to this vegetation type as campo or pampa, distinguishing it from cerrado (pampa arbolada) by the lack of woody vegeta- tion. Similarly, the wetland communities common to central Brazil occur throughout Chiquitanía. Val- ley-side campos are treeless savannas that occur along erosional surfaces, wherever there is a fluc- tuating, perched, water table that seeps out on gently sloping valley sides (Eiten, 1978; Goldsmith, 197 e overflow of rivers on alluvial plains leads to the formation of a pantanal complex of seasonally humid or seasonally inundated sa- vannas (Cole, 1986; Eiten, 1978). Scattered across these open grasslands are raised earth platforms built up by termites; these support cerrado species or low forest. At the base of valley-side campos and scattered across the pantanal complexes one finds savanna marsh, a savanna wetland com- munity with saturated soils throughout the year. Beck (1984) and Haase & Beck (1989) reported on similar extensive savanna wetlands in the De- partment of Beni. The residents of Chiquitania а=. refer to all savanna wetlands as curiches. uiz (1982) conducted a preliminary inventory of the seasonal forest communities near Concep- ción in the Province of Nuflo de Chávez. The forest TABLE 2. The geographic distribution of the grasses recorded to occur in Chiquitanía, Santa Cruz, Bolivia. Sout Southern Central Central Bolivian Pantropical Neotropical American Cone' Brazil? Andes endemics Other? 33 103 38 27 38 2 T 25 ' Includes central and southern (extra-amazonian) Brazil, Bolivia, Paraguay, Uruguay, and subtropical regions of Argentina. 2 Includes central Brazil and adjacent cerrado regions in Bolivia and Paraguay. * Cultivated grasses and species only collected in adjacent regions of Bolivia. 128 Annals of the Missouri Botanical Garden ABLE 3. Synopsis of the tribes and genera (number of species occurring in Chiquitania). Bambuseae (8) Actinocladum, Bambusa, Chusquea, Guadua, Rhip- idocladum Olyreae (4) Olyra Streptochaeteae (1) Streptochaeta Phareae (1) Pharus Oryzeae (7) Luziola, Centotheceae (1) Orthoclada Leersia, Oryza Arundineae (2) Arundo, Gynerium Aristideae (13) Aristida Pappophoreae (2) Pappophorum Eragrostideae (24) actyloctenium, Eleusine, Eragrostis, Gouinia, Lep- tochloa, Tripogon ies e (9) Chloris, Cynodon, Eustachys, Gymnopogon, Mi- сгос p loa Sporoboleae (6) Sporobolus Paniceae (144) Acroceras, Anthaenantiopsis, Arthropogon, Axono- chrus, Digita laena, Echinochloa, Erioc ie чаш Hy- pus, Brachiaria, Cen ria, Echino- menachne, Ichnanthus, Lasia Leptocorypheum, Melinis, Mea Oplis- menus, Otachyrium, Panicum, жул Penni- setum, Rhynchelytrum, Sacciolepis, Seta Thras Arundinelleae (3) Arundinella, Loudetia, Loudetiopsis Andropogoneae (51 Agenium, Andropogon, Bothriochloa, Coelorhachis, Elionurus, ysis, Hackelochloa, Hemar- thria, Hyparrhenia, Imperata, Rhytachne, Sac- Eriochr charum, Schizachyrium, Sorghastrum, Sor- ghum, Trachypogon, Tripsacum, Zea types described in his study are semideciduous, and the density of the understory is inversely correlated with the development of the canopy. High forest was described as having canopy trees 20-25 m tall and formed one end of a continuum of forest or forest scrub communities that eventually inter- grade with cerrado. In contrast, gallery forest, which occurs as a narrow strip 50 to 100 m wide on valley floors, is an evergreen community with trees 15-30 m tall. Locally, the various gradations of forest are recognized and referred to as monte alto, monte bajo, and monte humedo. In addition, granitic outcrops, locally known as lajas, exist as islands within cerrado, campo rupestre, savanna wetlands, and seasonal forest. These range in size from small outcrops scattered across the landscape to large domes (inselbergs) that can attain a height of 200 m and diameter of 400 m or more. FLORISTIC RELATIONSHIPS OF THE GRASSES OF CHIQUITANIA This is the first relatively complete inventory of native grasses from a region in central South Amer- ica. Table 1 compares the number of species shared with grass floras of Costa Rica, Brazil, Argentina, the Central Andes, and two adjacent regions in eastern Bolivia. The greatest similarity is with the grass flora of the Bolivian savanna regions, followed by the floras of the coastal states of Brazil. Chi- quitania shows little floristic affinity with the high altitude grasslands of the Andes or with the sub- tropical grasslands of Argentina. The similarity to the Brazilian floras is even more apparent when the species are classified according to their geo- graphic distributions (Table 2). The Bolivian plant collector José Steinbach made extensive collections of grasses near the town of Buenavista on the Andean Piedmont in western Santa Cruz (Foster, 1966). If one assumes that his collections are an essentially complete inventory of the grasses for that region, then the richness of the grass flora of the Brazilian Shield becomes apparent. One hundred thirty more species have been recorded for Chiquitania than for the Andean Piedmont even though the two areas have a high index of similarity. Only 20 species collected on the Andean Piedmont have not also been recorded for Chiquitania. The greater species richness of Chiquitania is most easily explained by the greater diversity of savanna communities in the region, and in part possibly by richness within the savan- nas. A similar argument can be made to explain the greater number of species reported for the extensive savanna wetlands of the Beni; however, this remote part of Bolivia has yet to be inventoried adequately. This study documents 80 new records for Bolivia, including two genera Actinocladum and Streptochaeta, and 112 new records for the Department of Santa Cruz. A total of 75 genera and 276 species are reported here (Table 3). In addition to the key, annotations for each Volume 77, Number 1 1990 Killeen Grasses of Chiquitanía, Santa Cruz, Bolivia 129 taxon include opimis. of recent monographic treatments rgentina, Uruguay, Paraguay, pur the Central Andes), dis- cussion of selected taxonomic problems, notes on habitat preferences, geographic distributions (in- formation collated from existing floras and mono- graphs), palatability, local flowering behavior (peak months italicized), and cytology. Fire is crucial in initiating flowering for many species and these can be found in bloom between July and November at the end of the dry season when ranchers system- atically burn native savannas to provide forage for cattle. Palatability estimates (0-4) are based on paired comparisons along fences dissecting over- grazed and lightly grazed savanna (unpublished data), as well as observations made of selective grazing by cattle. Species ranked as (0) either in- creased in relative abundance or showed no effect in response to overgrazing; a ranking of (1) is given to species that showed no evidence of grazing but were found to be less abundant in overgrazed sa- KEY TO GENERA KEY TO GROUPS la. Giant grasses; woody or semiwoody; culms 2- m tall .. lb. Not giant grasses; culms herbaceous or rarely 2a. Leaf 2b. аш cp lacking pseudopetiole, the bla Qo = Spikelets disarticulating withou vannas; rankings of (2-4) are for species that de- crease in relative abundance in overgrazed savan- nas and/or were observed to be grazed selectively by cattle. Chromosome counts are based on meiotic squashes made in the field and are being indepen- dently verified via mitotic preparations (Norman, Quarin & Killeen, in prep.). Specimens cited are those of the author unless otherwise indicated; bold- face numbers are voucher specimens for cytolog- ical preparations and usually have a matched set of seeds, which has been sent to the Instituto de Botánica del Nordeste in Corrientes, Argentina (CTES). The distribution of replicate sets of spec- imens are as follows: lowa State University (ISC), Herbario Nacional de Bolivia (LPB), Field Museum of Natural History (F), Missouri Botanical Garden (MO), United States National Herbarium (US), New York Botanical Garden (NY), Instituto de Botanica Darwinión, San Isidro, Argentina (SI), and Instituto de Botanica del Noreste, Corrientes, Argentina Key I iwoody; plants prostrate or up to 2 m ta blades а at the base into ренк. dde the ini lanceolate or ovate; forest grasses .. Key II des ied lanceolat tats. ast some spikelets ае with a ae or (rarely) spikelets рк оп ies e inflorescen t any attached bristles, rachis internodes, or pedicels. 4a. Spikelets disarticulating above the persistent glumes (glumes tached cbe. involucres, rachis internodes, and and lemmas falling in some species of Eragrostis but then the palea and rachilla ri spikelets 1—many- . 4b. Spikelets disarticulating below the glumes (glumes vestigial in some genera); spikelets 1-2 flowered „К y V KEY I Giant grasses; woody or semiwoody; culms 2-30 m tall. la. Foliage usually dimorphic: culms producing leaves with reduced, often deciduous blades, the branches with 2а. of culm with a dominant bud (or branch branch having numerous secondary br . Primary buds (or branches) at each midculm at the basal nodes of the primary bra base and hollow towards apex; nch; plants ar br abd н p. at зава at the base of the plant . Primary buds (or branches) at each midculm erous overlapping leaves with well-developed, pseudopetiolate еа plants blooming infrequently. ) and 1- maller branches, or with a single large anches inserted at its а nodes; branches thorny ог not thorny. node solitary; secondary branches (if present) —— boreal; culms erect, stout, hollow, or solid a e numerous, with one large bud and numerous smaller buds (or branches), frequently the major aes rem n dormant and obscured by branches oe from the numerous minor buds; plants vining or scandent; culms solid; branches lacking husquea ramosissima thor 2b. Nodes of cu producing several to numerous branches of equal size; WD not thorny. 4a. Branches originating from a triangular shield above the node; forest lia R 4b. Branches originating in a straight line; growing in open savanna or rarely in = hipi sie racemiflorum | мшш. du um vertic E a every da. Spikelets bisexual. Act oda dea not dimorphic, the leaves of culm and branches similar, lacking a а plants bloomin 130 Annals of the qa Botanical Garden 6a. Inflorescence a spicate panicle; spikelets glabrous or pubescent but not villous, subtended by an involucre of bristles Pennisetum 6b. oo a panicle or panicle of racemes; spikelets villous, not subtended by an involucre of Ta. Spikelets with 3-6 bisexual florets, the callus of the spikelet and glumes glabrous, the sarae villou undo donax ТЬ. Spikelets with 2 florets, the lower reduced to a single hyaline lemma, the callus Шош the lum es om or pubescent, the lemmas glabrous to sparsely ciliate „u Saccharum 5b. Spikelets unis 8a. Plants dioeci ious; spikelets villous; inflorescence paniculate ШШ... и sagittatum 8b. Plants monoecious; spikelets not villous; inflorescence racemose or highly m 9а. Foliage basal and caulescent; staminate ав apical and pistillate cee basal on the same inflorescence; cultivated and wil Tripsacum 9b. Foliage caulescent; e spikelets in көр racemes and pistillate spikelets on sillas cobs; cultivated plants ....... Zea mays KEY Il Grasses rarely approaching 2 m in height; culms herbaceous; leaf blades constricted at the base, forming a pseudopetiole, the blades lanceolate to ovate; forest grasses. la. Spikelets unisexual; plants monoecio 2a. Pseudopetiole twisted, the ibid al side of the blade facing upward; staminate and pistillate spikelets paired with each other; lemma of pistillate spikelet coriaceous, with uncinate (hooked) D or besc ent piri Jacke uncinate Бабы OR уга Spikelets bisexual. 3a. р о 3-40 mm long; nae with cross veins between the longitudinal veins clearly visible; spikelets terete or laterally compre 4а. ксн ар a spi ike; oi pue aristate, the awns becoming entwined when mature an the Spies to fall ав а group aue oe A treptochaeta spicata lr e rescence a diffuse open panicle; spikelets laterally compressed, awnless е Orthoclada laxa 3b. Pseudopetiole dee than 3 mm; blades lacking evident cross veins; ids dorsally compressed (glumes keeled in /chnanthus). 5a. Spikelets subtended by sterile panicle branches; inflorescence a spicate panicle „u Setaria 5b. Spikelets not subtended by sterile panicle branches; inflorescence not a spicate panicle. A Both glumes equal nn Homolepis aturensis 6b. Lower glume 2-25 the length of upper g glum 7a. Lemma of upper floret with a pair of а appendages or scars at base ........... Ichnanthus 7b. Lemma of upper floret lacking winged appendages or scars at base 8a. Spikelets subglobose, black at maturity, obliquely placed on pedicels; glumes a lower lemma with apical tufts of hairs „uinnsinn siacis 8b. Spikelets elliptic or ovate, variously green or purple at maturity, not obliquely placed on pedicels; glumes and lower lemma variously glabrous to pubescent but apices lacking tufts of hairs nnn Panicum = KEY Ш Spikelets a hate with attached bristles, involucres, rachis internodes, and pedicels, or rarely not disarticulating from the inflorese — w Spikelets in fascicles of 1-4, subtended by numerous bristles or enclosed by a spiny bur; inflorescence a spicate anicle. 2a. Bristles united in at least the basal 44, forming an indurate spiny bur „u enchrus 2b. Bristles separated to the base, forming an involucre „u .. Pennisetum Spikelets usually ipe ice not subtended by bristles nor enclosed by a spiny bur; inflorescence a raceme, O or panicle of racemes. Spikelets all sessile, or pa iis internodes and pedicels connate - c 4a. All spikelets unisexual; staminate spikelets terminal on racemes, subsessile, chartaceous, paired; рае spikelets basal on racemes, solitary at eac ch node, sunken in a hollow, cartilaginous, rachis тйегподе.... A нне ЫМАНЫ НЕ НЫ aM рн ннн ыйдын Ттїрзасит 4b Spikelets эш bisexual or staminate, never O staminate and ae spikelets paired with is hollow тарын interr 5a. Sessile spikelet «жы “the lower glume cartilaginous, strongly pitted; pedicellate spikelet chartaceous, lanceolate e Hackelochloa granularis Volume 77, Number 1 Killeen 131 1990 Grasses of Chiquitanía, Santa Cruz, Bolivia 5b. iie spikelet lanceolate, the lower glume leathery or chartaceous, not pitted; pedicellate Hemarthria altissima t similar to sessile spikele 3b. ётер pe ps pedicellate (ре айе spikelet vestigial in some species but pedicel always present); rachis internodes and pedicels not connate. ба. de seb an open or contracted panicle. 7a. Pedicellate spikelets nearly cd to the sessile spikelet in form, either bisexual or жек. : Glumes ciliate along the margins; callus hairs stiffly spreading, 14-34 times the len ngt the spikelet, Hosen or golden (rarely tan); panicle contracte riokhrysie 8b. ry et рни or pubescent but not ciliate along E ena callus hairs filiform, 34- e length of the spikelet, white; panicle O Saccharum 21 7b. Police vil unlike the sessile spikelet in form, itr staminate, vestigial, or lacking. node and pedicel be. a thin translucent line running lengthwise between r less villous Bothriochloa g nerves; inflorescence more 9b. Rachis ne and pedicel terete or flattened but lacking a thin translucent line running lengthwise between two strong nerves; inflorescence glabrous to pubescent. 10a. Pedicellate spikelets ld pres equal to the sessile spikelet in length ....... orghu lOb. Pedicellate spikelets lacking although ens present Sorghastrum 6b. Inflorescence of l-several racemes (sometimes the culm freely branched at upper nodes and forming a compound panicle, but then each raceme or 5 of several digitate racemes subtended by a spatheate sheath). lla. Raceme 1 per peduncle or spatheate — 12а. Lower glume of sessile spikelet ru : ies els and quse internodes ANN margins of the lower glume of the Coelorhachis essile spikelet 13b. Pedicels and des pm not auriculate; margins of the lower glume of e sessile spikelet not winged ..... Rhytachne 12b. Lower glume of sessile spikelet not r Lower glume of the sessile dE en sulcate between two lateral keels .. Andropogon 14b. Lower glume of the sessile spikelet weakly to strongly concave (rarely with a medial groove in Schizachyrium). 15a. Upper lemma of pei spikelet awned (awn vestigial in some forms of $ tenerum) from a deeply lobed apex; lower glume of sessile b e to acuminate izachyrium 15b. Upper lemma of sessile spikelet awnless; lower glume of ds spikelet bidentate (teeth sometimes minute in E. muticus) Elionurus llb. Racemes 2-several per peduncle or spatheate sheath. l6a. Sessile се with a sharp, needlelike callus and а stout twisted а 1 er glume of pedicellate spikelet foliaceous, 6-9 mm long, к ке 10-15- al the margins membranous; sessile spikelet cartilaginous, with a E tudinal groove 17b. "ida Ley of p spikelet not foliaceous, 4-6 mm long, arei: 0. ved, the margins firm; sessile spikelet concave, sulcate, or grooved. 18а. "Upper glume of the sessile spikelet with an awn 1-5 mm long ... di do 18b. Upper glume of sessile spikelet awnless or with an awn less than 1 m H урыса ong 16b. Sessile spikelet with a blunt or rounded callus, awned or awnles 9a chis internodes and pedicels with a thin, translucent line running lengthwise between 2 strong nerves; lower glume of the sessile spikelet concave or flat ER othriochloa 19b. Rachis internodes and pedicels terete or flat, lacking a thin ree a line running lengthwise between 2 strong nerves; lower glume of sessile spikelet a between 2 strong lateral keels ndropogon KEY IV Disarticulation above the persistent glumes nri and lemmas falling in some species of Eragrostis but then the palea and rachilla persistent); spikelets 1 -many-flowered. la. d" bearing 3-26 awns, the awns sometimes united below to form a column but then distinctly tripartite abov 2a. po 10-26, not united above to form a column, stout, unequal in length o Pappophorum 2b. Awns 3, distinct or united below to form a column, slender, equal or unequal in length (rarely the Aristida lateral awns reduced and inconspicuou v . Lemmas awnless or awned, but never with 3 or more awns arising from the same lemma. 3a. Spikelets unisexual, the plants monoecious, stamens 3-6; stoloniferous aquatics or grasses of marshy soils LT c . Luziola 132 Annals of the Missouri Botanical Garden 3b. Fags bisexual; stamens 1-3; plants of various habits. Spik d. lets 1-flowere in lets lacking a sterile rachilla internode prolonged past the base of the palea; lemma 5 5а. Spike -nerved orobolus 5b ни with а minute sterile rachilla internode prolonged past the base of the palea; lemm ] -3-nerve 6a. Iniorescurcé a digitate or subdigitate whorl of l-sided spikes 3-10 ст long; lemma 3-nerved occ ¿ynodon b. Inflorescence a large, open, ovoid panicle 20-40 cm long; lemma 1-nerved WW... Eragrostis atroides an 4b. a k : А -flowered, with reduced florets above or below fertile floret S ts dorsally compressed; florets 2, the lower floret reduced (staminate or neuter), the ~ jet ag dis bisexual. 8a. Lower glume with stout, tuberculate-based, golden hairs; inflorescence a false raceme; spikelets in triads at tips of flexuous branches arranged on a central axis cocino Loudetiopsis chrysothrix 8b. Lower glume glabrous; inflorescence an open, branched pani Ipper floret dues , 4-5 mm long, the lemma mies an awn 10-20 mm long; i ; Lattin fammida r 9b. e floret glabrou 2 mm long, the lemma with an awn l-13 r caespitose or with iaa rhizomes, but not a bunch grass; foliage cauline .................... Arundinella hispida 7b. Spikelets laterally adita florets 2- many, the lowest floret bisexual, the upper floret(s e staminate, or neu Bisexual florets 3- 15. all alike although the uppermost sometimes reduced or sterile. „. Eragrostis lla. Inflorescence paniculate llb. Inflorescence a spike or a panicle of spikes 12a. Disarticulation at "m. base of b pom floret, the florets falling together as a group Dactyloctenium aegyptium 12b. Disarticulation between the florets, the florets falling separately. 13a. Inflorescence a solitary, erect, or slightly arcuate spike ooo Von d spicatus 13b. Inflorescence composed of 2-50 spikes or spikelike raceme 14 Inflorescence a single (occasionally 2) whorl(s) of 3- 5 spikes "leusine indica 14b. We ce a panicle of ae borne on an elongate axis. Glumes distinctly 3-6-nerved Gouinia Glumes 1 -пегуе do hloa 10b. Pee florets solitary (rarely 2 in Gymnopogon) with 1-3 reduced gies a. croc hlod indica like Inflorescence a solitary, erect spi Y Inflorescence of l-several whorls “of digitate or subdigitate puta or spikeli racemes. 17а. Foliage cauline, the эсин lacking a midne Gymnopogon 17b. Foliage basal, or both basal and cauline ie Ыы: with a midnerve. Plants ee perennials; upper floret absent or up to 0.5 m 8a. long С ж 18b. Plants caespitose, perennial or annual; upper florets sterile or rarely fertile, well developed, 1-2 mm long 19a. Upper glume bilobed, a dark chestnut brown; upper боле oblong or club-shaped, the lower lemma ovate or broadly elliptic, usu E awnless; foliage strictly equitant ............... Eustachys 19b. Upper glume a ute to acuminate, der аз hight pues to green; lower floret, the pin: jesus usually pss foliage not strictly Chloris equitant KEY V Spikelets disarticulating below the glumes (or the glumes vestigial in Leersia and Oryza); falling singly, without attached rachis internodes or pedicels; rachis and/or panicle branches generally persistent; spikelets with 1-2 florets. la. оркен “ae ally compressed. 2a. Spikelets | -flowered; lemma and palea strongly scabrous; stamens ~ 6; plants usually aquatic. 3a. Spikelets wiih only a res and a palea; plants stoloniferou Leersia ЗЬ. Sy of 4 bracts, the fertile floret subtended dn two glumelike bracts; plants caespitose or dec ieri and ne. at the nodes Oryza 2b. Spikelets 2-flowered, the lower floret staminate or neuter, the upper floret bisexual; lemma and palea | 1 glabrous or pubescent but not strongly scabrous; stamens 3; plants terrestrial. Volume 77, Number 1 Killeen 133 1990 Grasses of Chiquitanía, Santa Cruz, Bolivia lb. 16a. 4a. id жерй glabrous; foliage with abundant glandular hairs giving the plants a molasseslike odor pio Шен T NUUS AUTRE ON SOIRS elinis minutiflora 4b. Spikelets pe cur foliage glabrous or pubescent but lacking glandular, aromatic hairs. за. Glumes gibbous, densely pubescent, the callus densely pubescent with hairs 10-20 m R hurt al repens Glumes elliptic, glabrous, the callus pubescent with hairs 1-10 mm long mm Ж Arthropogon 5b. I dorsally compressed (glumes keeled in Ichnanthus, Mesosetum, and Oplismenus but then the flo oret dorsally compressed), generally with 2 florets, the lower floret staminate or reduced, the upper floret ual. isex 6a. Spikelets paired, unequally pedicellate, the lower staminate and persistent, the upper perfect, disartic- ulating at the base of a pungent callus and provided with a long, stout, geniculate awn; inflorescence втв raceme or а conjugate pair of racemes rachypo gon plumosus 6b. Spikelets solitary, paired, or in groups of 3 or more, but not an unequally ко pair of spikelets with the lower staminate and the upper орн. the callus never ср awned ог awnless but the awn never geniculate; inflorescence an open panicle or of 1-many ra 7a. ое villous, the callus hairs а. than 2 times the length ce be ghumes, the upper lemma hyalin mperata 7b. Spikelets glabrous to pubescent but the hairs never more than 1.5 times the length of the spikelet, the upper lemma coriaceous to cartilaginous 8a. Spikelets subtended by 1-10 bristles (at least apically on panicle branches); inflorescence a spicate panicle Setaria 8b. Spikelets not subtended by bristles; inflorescence vario 9a. Spikelets with an indurate knob at base, this Nod of either a swollen pedicel or the first rachilla node and the sie adnate lower glume 10a. Rachis foliaceous, winged, 2-8 mm wide; spikelets paired but arranged in a single row, with the backs of the upper lemmas of adjacent spikelets facing one another rasya petrosa 10b. Rachis ii ig ie 0.5-1 mm wide; spikelets paired, arranged in 2-4 regular or gul ws, the d secund or with the back of the upper lemmas facing he midr ib, but not back to back Eriochloa 9b. Spikelets lacking an indurate kn dba t base. lla. “ш lemma with а laterally ia beak or truncate scar at a 12a. Glumes and lower lemma glabrous; upper lemma smooth; мен glabrous Acroceras 12b. Glumes and lower lemma coarsely hispid; upper lemma strongly r rugose; pedicels with a ring of stout hairs at apex o... rachiaria paucispicata 11b. Upper ae blunt or acute but not with a laterally flattened beak or truncate scar at apex. 13a. Lemma of upper floret with a pair of appendages or scars at base. 14a. Appendages composed x Ii tufts of thick hairs; upper floret sessile; foliage аа 1, the idi Panicum olyroides 14b. Appendages membra win Á or reduced to scars; ee floret on a short stipe, foliage canline, the blades lanceolate to ov 15а. Inflorescence a panicle Ichnanthus 15b. ries of 1-5 racemes arranged on a central axis; fre- Фе spikelets is e a dinis first glume surpassing the ikelet Echinolaena 13b. Lemma of fertile floret lacking appendages or scars at base 16 Inflorescence an open or contracted panicle. l7a. Both glumes equal in length Eg Spend the floret; leaf blades with a short dido lene molepis aturensis 17b. Lower glume lacking or from %-% the length of the spikelet; leaf blades mh. or without pseudopetiole. 18a. Lower glume absent; spikelets densely pubesce Leptocoryphium lanatum 18b. Lower glume present; spikelets glabrous or E ui 19a. Palea of the lower floret splitting when mature, the two halves becoming reflexed and the spikelets appearing winged; fertile floret brown and shining when mature ........ achyrium dias Ый 19b. Palea of the lower floret not splitting, the spikelets not appearing ML spikelet variously colored. 20a. Spikelets placed obliquely on pedicels, black at maturity, the glumes and lower emma with tufts of hairs at apices = 20b. Spikelets not placed obliquely on pedicels, more or less erect, variously color at maturity, bs glumes and lower lemma glabrous or irt and! t but not with ses of hairs at a 21a. bbc a narrow, contracted, spicate panicle. 22a. Culms pithy; upper lemma leathery, the margins flat; plants stolon- 134 Annals of the Missouri Botanical Garden e floating aquatics or (in the dry season) rooted in the mud the margins of streams and ponds ymenachne 22b. C im. hollow; upper lemma cartilaginous, the margins inrolled or flat; plants caespitose or decumbent, of hus habitats but never floating atics acciolepis dee 21b. CHE open or contracted, if somewhat spicate then with fewer than spikelets. TR Spikelets inflated, subglobose; upper lemma strongly rugose; n hed racemes with the back of the lower glum Brac hiaria iented towards the rachis 23b. rine not inflated, ovate to narrowly elliptic; upper lemma smooth (rugose in P. maximum); if the inflorescence somewhat racemose then spikelets neither inflated nor the lemmas rugose „u anicum 16b. ugs of 1 -many racemes. S pikelet 25a. 2 > c 26b. ts subaristate to awned. Blades ovate to lanceolate, less than 3 times as long as wide; awns arising from the bidentate a ismenus as wide; lower glume mucronate but not awned o d Poh ойра aristate, the pe glume and/or the lower lemma awne . Spikelets not awne 26 Inflorescence a solitary raceme Raceme with a single row wot paired spikelets oriented back to back, so that the lower glumes of each pair face each other and the upper glumes face apart; rachis foliaceous, rasya winged 27b. ee with solitary or paired spikelets not oriented back to back; rachis winged or 27a. not. 28a. ros Lini lacking; spikelets oriented with the back of the fertile lemma towards the aspalum 28b. poer ae well developed; spikelets oriented with the back of the upper lemma away from the rachis. 29a. Spikelets distichous, arranged in 2 rows spreading at 180° glumes and lower lemma pubescent with stiff, golden, or reddish hairs „u setum cayennensis the aes Mes 29b. е secund, arranged in 2 ог 4 rows but on the те side of s; glumes and lower lemma glabrous or pubescent with soft, white hairs Brachiaria Echinochloa Inflorescence of 2 -many conjugate, digitate, or paniculate racemes. 30a. Ligules absent; spikelets subglobose 30b. Tn present, membranous or ciliate; spikelets elliptic to subglobose 1 largins of the upper lemma indurate, inrolled, clasping the back of the palea, or (rarely) the floret open. 32a. Lower aes well developed, 1-3-nerved, 14-34 the length of the spikelet. 33a as narrowly elliptic, secund, irregularly arranged on short Panicum a 33b. Spikelets subglobose, appressed-ascending, regularly iris in 2 4 rows with the back of the lower glume facing the rachis ............ rachiaria 32b. ae шө lacking or a minute nerveless scale less than /; the oh of the 34a. Lowe glume a minute, nerveless scale; lower floret staminate, with well-developed palea; lemma of the upper floret not clasping the Ee alea, the floret open „u Anthaenantiopsis trachystachyum 34b. с glume absent; lower floret neuter, usually lacking а palea; of the e r floret Pm the palea, the floret open only ксы anthes 35a. Back of is upper lemma facing away from the rachis; ME solita 35b. Back “of the upper lemma facing toward the rachis; sakes solitary or paire aspalum 31b. Margins of the upper lemma hyaline, lying flat against the back of the palea. Plants a aquatics or (in the dry season) rooted in the mud along the ds, the culms solid with aerenchymous tissue; menachne onopus margins of streams and pon lades clasping culm; lower glume well developed „n 36b. Plants not floating aquatics, the culms hollow; blades not clasping culm; er glume either lacking or reduced to a small scale less than / the lowe Digitaria length of the spikelet Volume 77, Number 1 1990 Killeen Grasses of Chiquitanía, Santa Cruz, Bolivia 135 ENUMERATION OF GENERA AND SPECIES Acroceras Stapf in Prain Zuloaga, F. O., O. Morrone & A. A. Saénz. 1987. Estudio exomorfológico e histofoliar de las es- pecies americanas del género Acroceras (Po- aceae: Paniceae). Darwiniana 28(1-4): 191- KEY TO SPECIES la. Apex of upper lemma with a stout, ey compressed beak; pedicels ascending to ap- pressed, borne on racemose panicle branches . . zizanioides lb. Apex of upper lemma with a truncate, greenish scar, not beaked; pedicels preme the panicle open, its branches not racemo A. excavatum A. excavatum (Henrard) Zuloaga & Morrone, Darwiniana 28(1-4): 195. 1987. Panicum excavatum Henrard, Feddes Repert. Spec. Nov. Regni Veg. 23: 179. 1926. Lasiacis excavatum (Henrard) L. Parodi, Notas Pre- lim. Mus. La Plata 8: 92. 1943 Occasional in semideciduous forest, abundant in forest openings and logging roads; palatable (4); flowering January (April) to July; local name: ta- quarilla. Distribution: Venezuela, Brazil, Para- guay, and Argentina. (694, 830, 932, 967, 1001, 1740, 1891, 1966) A. zizanioides (H.B.K.) Dandy, J. Bot. 69: 54. 931. P. zizanioides H.B.K., Nov. Gen. Sp. 1: 100. 1816. P. grandiflorum Trin. ex Nees, Agrost. Bras. 143. 1829 (nomen nudum). Occasional, seasonally inundated savannas and roadside ditches; flowering December to May; 2n = 36; local name: cañuela. Distribution: Mex- Andropogon L. KEY TO SPECIES ico and the West Indies to northern Argentina; Bolivia: the Beni. (689, 998, 2414) Actinocladum McClure Soderstrom, T. R. 1981. Observations on a fire adapted bamboo of the Brazilian cerrado, Ac- tinocladum verticillatum (Poaceae: Bambu- soideae). Amer. J. Bot. 68(9): 1200-1211. A. verticillatum (Nees) McClure ex Soderstrom, Bot. 68(9): 1201. 1981. Arundi- naria verticillata Nees, Agrost. Bras. 2(1): 523. 1829. Ludolphia verticillata (Nees) A. Dietrich, Sp. Pl. 25. 1833. Rhipidocladum verticillatum (Nees) McClure, Smithson. Contr. Bot. 9: 106. 1973 Usually reported as a cerrado species but in Chiquitania it is restricted to campo rupestre and adjacent transitional scrub; both rhizomes and culms appear to be extremely resistant to fire; locally abundant and forming extensive colonies in Parque Nacional “Noel Kempff Mercado.” Distribution: central Brazil. (1378, 2759) Agenium Nees in Lindl. A. villosum (Nees) Pilger, Feddes Repert. Spec. Nov. Regni Veg. 43: 82. 1938. Heteropogon villosus Nees, Agrost. Bras. 362. 1829. An- dropogon neesii Kunth, Révis. Gramin. 2(39). 1832, non Trinius, 1832. Agenium nutans Nees in Lindl., Intr. Nat. Syst. Bot. 2: 447. 1836. Andropogon rw Steudel, Syn. Pl. Glumac. 1: 395. Rare, cerrado; flowering in February and March; 2n — 20. Distribution: Brazil, Paraguay, Uruguay, and Argentina; Bolivia: Andean Piedmont of Santa Cruz. (854, 2419) la. Upper glume of the sessile and pedicellate spikelets aristate. 2a. Racemes 1/peduncle; upper glume of pedicellate spikelet enlarged and winglike; caespitose pu е А. fastigiatus 2b. Касетеѕ 2/peduncle; upper glume of pedicellate — not nee perennial bunch grasses. 3 nder, 1-2 mm Callus of sessile spikelet pungent; culms sle pedicels 0.5-0.8 mm wide at the apex n diameter; rachis internodes and 3b. Callus of sessile spikelet Es culms 3-5 mm in diameter; rachis internodes and pedicels l- wide at the apex жоюш 1.5 anus — B э жён spikelet of A. sanlorenzanus). ulm abundantly desi at upper 4. gay per glume of sessile and sare spikelets acuminate, acute, or minutely bidentate (subaristate in the nodes to form a corymbose or elongate compound panicle of numerous (more than 30) spatheate inflorescences, each bearing 1-2(3) included or scarcely exserted racemes 136 Annals of the Missouri Botanical Garden 5a. Racemes 1 /spatheate inflorescence; pedicellate spikelet well developed and staminate at all the nodes of the raceme 6a. Rachis internodes and pedicels ciliate; sessile spikelet bisexual A. insolitus 6b. Rachis Уи ве and pedicels glabrous; sessile spikelet pistillate . virgatus 5b. Racemes 2-3/spatheate inflorescence; pedicellate spikelet staminate at terminal rachis nodes but vestigial below. Та. Sessile spikelets awnless; racemes Mini panicles e A. bicornis Tb. Sessile spikelets awned; racemes flexuous; pan nicles elongate glaziovii 4b. Culms branched or unbranched but not orig a compound ы nicle; racemes (2)3-6 per dn digitate; if inflorescences numerous (10-25 in s ome variants of 4. lateralis) then clearly exserted. 8a. Pedicellate spikelets well developed, н чөк ог surpassing the sessile spikelet in length. ere freely branched at the middle and upper nodes, 150-200 cm tall, 5-30 inflorescences/ . lateralis 9b. Cups unbranched or o branched, 30-40 em tall, 1-3 inflorescences/culm 10 ele du Sessile spikelet awnless, 5-5.5 inflorescences borne at die upper nodes; pedicellate spikelets staminate, 6 mm long A. : mm long; racemes 4-6/peduncle; culms with 2 -3 A l c c . Sessile spikelet awned, 4-4.5 mm long; racemes 3 per peduncle; culms iis anched, ries a single terminal inflorescence; pedicellate sois neuter (rarely i e 3-4 mm lon arinatus ng 8b. e ice roe vestigial, up to М the length of the sessile spikelet. ets awned; rachis hairs equaling the sessile spikelet; mature culms with a single, к. spikele nal inflorescence macrothrix llb. Sessile spikelets awnless; rachis hairs twice the length of the sessile spikelet; mature e culms with 2-3 terminal and axillary inflorescences. 12а. “Rac emes 3-5; leaf blades pne = ‘caespitose, the culms somewhat genic 12b. Racemes 2-3; leaf blades 1-3 Spe tema caespitose, the culms erec A. angustatus (C. Presl) Steudel, Syn. Pl. Glu- mac. 1: 370. 1854. Diectomis laxa Nees, ipe Bras. 340. 1829, non А. laxus Willd., 806. E MM Presl, Rel. Haenk. 1: с 18 Occasional, seasonally humid savanna and in white sandy soils on valley-side campos (upslope); flowering February to (April) May; meiosis ab- normal, plants probably apomictic. Distribution: Mexico and the West Indies to Brazil. (886, 960, 2078, 2459) A. bicornis L., Sp. Pl. 1046. 1753. Occurring in a wide range of natural and dis- turbed habitats; occasional in seasonally inundated savannas, cerrado, and forest margins; abundant in old fields and along most roadsides in savanna and forest soils; unpalatable (2); flowering November to (January to March) May. Distri- bution: Mexico to Argentina; Bolivia: Andean Pied- mont of Santa Cruz, Cochabamba, Beni, and the Yungas. (626, 672, 682, 1555) A. carinatus Nees, Agrost. Bras. 2: 330. 1829. . carinatus var. i Hack. in A. di Monogr. Phan. 6: 434. 1889, 4. cari- us var. exserens Hack. in | Mart., Fl. Bras. 29 288. 1883. 3-5 mm wide, carinate at ен spikelets 3-5 € long lat A. M mm wide, acuminate; peces 2.5-3.5 mm long; t A. leucostachyus Poorly represented in herbaria but locally abun- dant in campo rupestre on Serrania de Santiago; flowering is dependent upon fire. Distribution: Cos- ta Rica and Brazil. (2790) A. fastigiatus Sw., Prodr. 26. 1786. Diectomis MEA o P. Beauv., Ess. Agrostogr. 160. 1812 Occasional, in superficial soils on lateritic crests and granitic outcrops, locally abundant as a weed along road embankments in savanna soils; flowering November to (April) May; 2n = 20. Distribu- tion: throughout the tropics. (800, 1511, 1853, 2019, 2395) A. gayanus Kunth, Révis. Gramin. 1: 163. 1829. Cultivated forage grass of recent introduction by the British Mission in Tropical Agriculture (Pat- terson, 1984), usage not widespread; two different populations observed to flower in November and April. (1409A, 1964) A. glaziovii Hack. in Mart., Fl. Bras. 2(3): 286. 1883 Poorly represented in herbaria but abundant in Chiquitania, codominant with Saccharum trinii and Arundinella hispida in savanna marsh; coarse and Volume 77, Number 1 1990 Killeen 137 Grasses of Chiquitanía, Santa Cruz, Bolivia unpalatable (0); flowering in May and June. Dis- tribution: central Brazil. (983, 2031, 2082) A. insolitus Sohns, Mem. New York Bot. Gard. 9(3): 271. 1957 Closely related to 4. virgatus from which it is distinguished by ciliate pedicels and rachis inter- nodes, more elongate racemes, and a bisexual ses- sile spikelet; reminiscent of A. bicornis and Schi- zachyrium microstachyum. Rare, savanna marsh; flowering in April. Distribution: Venezuela and Bra- zil; only the fifth known collection. (Killeen 2484; additional specimens examined: VENEZUELA: Ma- guire & Wurdack 3577 US, NY; Davidse & González 12710 MO. Brazil: Harley et al. 15771 MO; Irwin 8734 US) A. lateralis Nees, Agrost. Bras. 329. 1829. А. brevis Trin., Mém. Acad. Imp. Sci. St.-Pé- tersbourg, Ser. 6, Sci. Math. 2: 268. 1832. A. glaucescens var. lateralis (Nees) Hack. in Mart., Fl. Bras. 2(3): 289. 1883. A. incanus Hack. in A. DC., Monogr. Phan. 6: 432. 1889. Populations vary in stature, size of the sessile spikelet (3-5 mm), pedicellate spikelet (4-8 mm), and the length of the awn of the sessile spikelet. Closely related taxa of doubtful status are 4. hy- pogynus Hack. (racemes 4-10), A. coloratus Hack. (awnless), A. herzogii Hack. (rachis and pedicels glabrous), and A. lindmanii Hack. ex Lindman (rachis and pedicels densely villous). This species is common in the sandy soils of seasonally humid savannas around San Ignacio de Velasco but is completely lacking from cerrado communi- ties near Concepcion, Lomerio, and San Javier, which are generally characterized by clay or sandy clay soils. Plants flower from August to January and blooming is possibly stimulated by fire. Dis- tribution: Brazil, Paraguay, Argentina and Uru- guay; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (754, 1114, 1246, 1550, 1662, 2295, 2819) A. leucostachyus H.B.K., Nov. Gen. Sp. 1: 187. 1816. A. lanuginosa H.B.K., Nov. Gen. Sp. 1: 187. 1816. Rare in Chiquitania, as a weed of pastures, dis- turbed savannas, and roadsides; flowering is stim- ulated by fire. Distribution: Mexico and the West Indies to Argentina; Bolivia: Andean Piedmont of Santa Cruz, Beni, and the Yungas. (1255, 1275, 1359, 1502, 2292, 2816; Thomas 5651 NY, LPB) A. macrothrix Trin., Mem. Acad. Imp. Sci. St.- Petersbourg, Sér. 6, Sci. Math. 2: 270. 1832. A. ternatus subsp. macrothrix (Trin.) Hack. in Mart., Fl. Bras. 2(3): 287. 1883. Similar to A. ternatus (Sprengel) Nees, which has fewer racemes (3 vs. 4-6), larger spikelets (5— 6 mm vs. 4-5 mm), longer awns (2-2.5 cm vs. 1.5-2 mm), and more densely villous rachis in- ternodes and pedicels. Rare, valley-side campo (midslope); flowering is stimulated by fire. Distri- bution: Brazil, Uruguay, and Argentina. (1869, 2195) A. sanlorenzanus Killeen, sp. nov. TYPE: Ser- гаша de San Lorenzo, 10 km W of San Javier, Prva. Nuflo de Chávez, Dpto. Santa Cruz, Bolivia (16?15'S, 62°40'Ұ), 800-900 m, 30 Oct. 1987, Killeen 2832 (holotype, ISC; iso- types, LPB, F, MO, US, SI). Figure 2. A. carinatus Nees, affinis sed culmis 3- 4 nodis, ramosis ongis, nonexcurrentibus, spiculis muticis Ere spiculis pedicellatis 5.5-6.0 mm lon gis nota Caespitose, perennial bunch grass; culm gla- brous, 30 cm tall, with 2-5 nodes visible above the basal foliage. Sheaths keeled, glabrous, strongly nerved, the margins membranous; ligules mem- branous, 0.2 mm long; blades broadest at base, equaling the sheath in width, folded, revolute when stressed, the adaxial surface glabrous with a thick- ened midrib, the abaxial surface pilose with an indistinct midnerve, 1-13 cm long, 2.4 cm wide (folded), reduced at the upper nodes. Inflorescences 2-3, terminal and axillary, inserted at the upper nodes, each composed of 4-6 digitate racemes 3- 6 cm long; spikelets paired, sessile and pedicellate; rachis internodes 3 mm long, flattened, with cilia 2.5-3 mm long; pedicels similar, united with the sessile spikelet and the rachis internode at the base to form a bearded callus, the hairs 2 mm long; disarticulation below the glumes, the sessile spike- lets falling attached to the rachis internodes and pedicels, the pedicellate spikelets falling separately. Sessile spikelets bisexual, 5 mm long; lower glume 5 mm long, bicarinate, the midnerve suppressed, each of the 2 keels composed of a single ridgelike nerve, strongly sulcate with the 2 keels reflexed and touching in the middle; upper glume 5 mm long, laterally compressed; lower lemma hyaline, 5-5 mm long, similar to the lower glume in form, ciliate on the margins, the lower palea lacking; 138 Annals of th e Missouri Botanical Garden zx NOR E AN n Y Lt ’ growth; innovations intravaginal; culms ме ект) 0-1 Cultivated forage grass, not widely planted in Chiquitania but common in the Gran Chaco; flow- ering November bs May; local name: estrella af- ricana. (978, 13 Dactyloctenium Willd. D. aegyptium (L.) Willd., Enum. Pl. 1029. 1809. ynosurus aegyptius L., Sp. FL. 72. 1753. D. mucronatum (Michx.) Willd., Enum. Pl. 1029. 1809 are, roadside weed. Distribution: throughout the tropics. (1700 Digitaria Haller Henrard, J. T. 1950. Monograph of the genus Digitaria. Univ. Pers. Leiden. Veldkampf, J. F. 1973. A revision of Digitaria Haller (Gra- mineae) in Malesia. Blumea 21: 1-80. Rugolo de Agrasar, Z. 1974. Las especies del género Digitaria (Gramineae) en la Argentina. Dar- winiana 19(1): 65-171. Webster, R. D. 1987 Taxonomy of Digitaria section Digitaria in North America (Poaceae, Paniceae). Sida 12(1): 209-222. -10 mm long (at least at base of non lant invested D. neesiana 2b. Spikelet hairs 5-10 mı n long; base of plant not shrouded by senescent sheaths; foliage cauline (at least at maturity); innovations extravaginal; culms freely branched, 3-6 nodes visible above the basal foliage D. insularis lb. Ке glabrous, short-pubescent or ciliate, the hairs, when present, less than 0.5 mm long. 3a. > m f£ Spike 2-2.5 mm army ы short, glandular-tipped ha Spikelets 1.8 mn me not б ке Ка‹ 4b. emes 3-8(10); spikelets 1.5-1.8 mm long b. "oo (8)10-30; spikelets 1-1.3 mm long ... or tan; vind borne in pairs; racemes conjugate or pees р. 5 3b. Upper floret lead- colored, green, 6a. Racemes 2, conjugate; plants stoloniferou pps er ce е reddish brown; spikelets borne in triads; racemes inserted on a 1 long, obovate; oe of glumes and lower са бй уот i D. n elongate ax and ee еы sensis 1 long, lanceolate; не eee of glumes and lower lemma in lines betwee agilis “lehmanniana uscescens 6b. Racemes 2-many, digitate at 1 or 2 iru plants of indefinite duration, decumbent and rooting at the nodes. Ta. ies glume lackin 8a. Upper glume а lower lemma glabrous; upper glume equaling upper floret . pper glume and lower lemma ciliate; upper glume about 14 the length of upper flor | р. 8b. l . D. n M rum ~ = Lower glume present. 9a. Spikelets маш ды. the pedicellate spikelet with stiffly spreading marginal hairs, the sile spikelet ... D. 9b. Spikelets all alike, pos pedicellate and sessile spikelet both with ascending marginal hairs D hairs lacking on s bicornis . ciliaris D. bicornis (Lam.) Roemer & Schultes, Syst. Veg. 2: 470. 1817. Paspalum bicorne Lam., Tabl. Encycl. 1: 176. 1791. sanguinalis sensu Hitchc., Contr. U.S. Natl. Syntherisma Herb. 24(8): 425. 1927 (in part). D. diver- siflora Swallen, Rhodora 65: 356. 1963. Common, as a weed of pastures and roadsides in forest and cerrado soils; flowering throughout Volume 77, Number 1 1990 Killeen 149 Grasses of Chiquitanía, Santa Cruz, Bolivia the year. Distribution: pantropical. (1262, 1280, 1406, 1539, 1696) D. ciliaris (Retz.) Koel., Descr. Gram. 27. 1802. Panicum ciliare Retz., Observ. Bot. 4: 16. 1786. Panicum adscendens H.B.K., Nov. Gen. Sp. 1: 80. 1816. D. adscendens (H.B.K.) Henrard, Blumea 1: 92. 1934. Syntherisma sanguinalis sensu Hitchc., Contr. U.S. Natl. Herb. 24(8): 425. 1927 (in part); see Veld- kampf (1973) for extensive synonymy. Common, as a weed of pastures and roadsides in forest and cerrado soils; rarely naturalized in superficial soils over granitic outcrops; flowering throughout the year. Distribution: throughout tem- perate and tropical regions of the New World. (603, 608, 678, 743, 1126, 1280, 1541, 1672, 1701, 2303) D. fragilis (Steudel) Luces, J. Wash. Acad. Sci. 32: 160. 1942. oo ane Steudel, Syn. Pl. Glumac. 1: 19. Common along roads in cerrado; occasional on superficial gravel soils of granitic outcrops and lat- eritic crests; flowering October to December. Dis- tribution: Venezuela and Colombia. (638, 798, 837, 1233, 1329, 1538, 1786) D. fuscescens ч Presl) Henrard, Meded. Rijks- Herb. 61: 8. 1930. Panicum fuscescens С. Presl, Rel. a 213. 1830. Syntherisma fuscescens (С. Presl) Scribner, Ann. Rep. Mis- souri Bot. Gard. 10: 49, t. 10. 1899. Occasional as a weed in compacted gravel soils of road embankments; flowering January to Feb- ruary. Distribution: throughout the tropics. (786, 1661, 1734) D. insularis (L.) Mez ex Ekman, Ark. Bot. 11: 17 12. Andropogon insularis L., Syst. Nat. ed. 10. 2: 1304. 1759. Panicum leu- cophaeum H.B.K., Nov. Gen. Sp. 1: 97. 1816. Acicarpa sacchariflora Raddi, Agrost. Bras. 31, pl. I, f. 4. 1823. Trichachne insularis (L.) Nees, Agrost. Bras. 87. 1829. Tri- chachne sacchariflorum (Raddi) Nees, Agrost. Bras. 87. 1829. Leptocoryphium penicilli- gerum Speg., Anales Soc. Ci. Argent. 16: 102. 1883. Valota penicilligera (Speg.) Chase ex L. Parodi, Revista Fac. Agron. Veterin. 4: 46. 1922. Digitaria sacchariflora (Raddi) Hen- гага, Blumea 1: 99. 1934. Abundant, an aggressive weed along roads in savanna and forest; flowering in November, De- cember, and irregularly throughout the year. Dis- tribution: southern United States to Argentina. (595, 1388, 1405, 1571, 1695, 1730) D. lanuginosa (Nees) Henrard, Meded. Rijks- Herb. 61: 5. 1930. Paspalum lanuginosum Nees, Agrost. Bras. 63. 1829. D. laetevirens Mez, Bot. Jahrb. Syst. 56(125): 8. 1921. Pan- icum cuyabense Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 3: 206. 1835. Syntherisma cuy- abensis (Trin.) Hitchc., Contr. U.S. Natl. Herb. 22: 468. 1922. D. cuyabensis (Trin.) L. Par- odi, Physis 8: 378. 1926 Rare, in seasonally humid sandy soils; flowering throughout the year. Distribution: the Guianas, Brazil, Paraguay, Argentina, and Uruguay; Bolivia: Andean Piedmont of Santa Cruz. (772, 1252, 1578, 1583, 2109, 2293) D. lehmanniana Henrard, Blumea 1: 107. 1934. Rare, cerrado; flowering in November. Distri- bution: Colombia, Peru, and Argentina; Bolivia: the Yungas. (1488) D. mattogrossensis (Pilger) Henrard, Meded. Rijks-Herb. 61: 1. 1930. Panicum adustum Nees var. mattogrossensis Pilger, Bot. Jahrb. Syst. 30: 131. 1901. Rare, cerrado/forest margin; flowering in March and April. Distribution: central Brazil. (1922, 2400) D. neesiana Henrard, Blumea 1: 99. 1934. Tri- chachne velutina Nees, Agrost. Bras. 90. 1829, non D. velutina Forssk. ex P. Beauv., 1812, nec D. velutina (DC.) Hitchc., 1927. Panicum vestitum Kunth, Révis. Gramin. 1: 39. 1829, non D. vestita Figari & DeNotaris, 1854. Valota vestita (Kunth) Kuhlm., Co- miss. Linhas Telegr. Estrateg. Mato Grosso Amazonas 67, Annexo 5, Bot. 11: 40. 1922. Local populations of this species are composed of functionally dioecious plants: pistillate plants have spikelets with vestigial anthers and nonfunc- tional pollen, while staminate plants have spikelets lacking gynoecia. Common in ungrazed cerrado; palatable (3); flowering is dependent upon fire; 2n — about 60. Distribution: Colombia and Brazil. (1140, 1192, 1457, 2189, 2787; Thomas 5694 NY, LPB) D. setigera Roth ex Roemer & Schultes, Syst. Veg. 2: 474. 1817. Syntherisma digitata 150 Annals of the o Botanical Garden (Sw.) Hitche., Contr. 142. 1908 (in part). Similar to D. horizontalis Willd., which has a vestigial lower glume. Rare but probably over- looked, a roadside weed. Distribution: throughout the tropics. (1427) U.S. Natl. Herb. 12: Echinochloa P. Beauv. Gould, F. W., M. A. Ali & D. E. Fairbrothers. 1972. A revision of Echinochloa in the United States, Amer. Midl. Naturalist 87: 36—59. KEY TO SPECIES la. — about 2 mm long, the lower lemma to mucronate; racemes an the pibes crowded and ы 4-rowed 222... Е " E. colona 1b. Spikelets 2.5 ET mm n long, the: lower lemma mucronate to aristate; lower racemes кашта d, irregularly 4-rowed ccoo E. cruz-pavonis E. colona (L.) Link, Hort. Berol. 2: 209. 1833. Panicum colonum L., Syst. Nat. 10(2): 870. 1759. Oplismenus colonus (L.) H.B.K., Nov. Gen. Sp. 1: 108 Occasional, roadside ditches. Distribution: throughout the tropics. (2299B, 2304) crus-pavonis (H.B.K.) Schultes, Mant. 2: 269. 1824. Oplismenus crus-pavonis H.B.K., Nov. Gen. Sp. 1: 108. 1816. Panicum sabulicola Nees, Agrost. Bras. 258. 1829. E. sabulicola (Nees) Hitchc., Contr. U.S. Natl. Herb. 17: 257. 1913. E. crusgalli var. crus-pavonis (H.B.K.) Hitchc., Contr. U.S. Natl. Herb. 22: 148. 1920. Occasional, roadside ditches. Distribution: throughout the tropics. (2299, 1718) Echinolaena Desv. KEY TO SPECIES la. Lower glume cuspidate, about 5 mm long, sur- paring the spikelet but not foliaceous, glabrous; 3-6; айба caespitose perenni ial 7. minarum = c 2 чаы 2 © = o "4 delicate, decumbent and rooting at the ed E. gracilis Swallen, J. Wash. Acad. Sci. 23: 457. 1933 Rare but inconspicuous and probably over- looked, seasonally inundated savanna; flowering December to May. Distribution: Central America, Colombia, and Venezuela; Bolivia: the Beni. (2083, 2261, 2415; Bruderreck 289 ICS, LPB) E. minarum (Nees) Pilger, Notizbl. Bot. Gart. Berlin-Dahlem 11: 246. 1931. Oplismenus minarum Nees, Agrost. Bras. 268. 1829. Ich- nanthus minarum (Nees) Doell in Mart., Fl. Bras. 2(2): 294. 1877. 1. lilloi Hack. ex Stuckert, Ann. Conserv. Jard. Bot. Genéve 17: 288. 1914. I. sandiense Mez, Bot. Jahrb. Syst. 56(124): 5. 1921. І. riparia Swallen, Phytologia 11: 150. 1964. Rare, scrub thickets on margins of granitic out- crops; flowering in February; 2n — 20. Distribu- tion: Peru, Brazil, and Argentina; Bolivia: Cocha- bamba and the Yungas. (1820, 2333) Eleusine Gaertn. E. indica (L.) Gaertn., Fruct. € Sem. 1: 8. 1788. Cynosurus indicus L., Sp. Pl. 72. 1753. Common, weed of roadsides and gardens; flow- ering throughout the year. Distribution: throughout the tropics. (597, 742) Elionurus Willd. Renvoize, S. A. 1978. Studies in Elionurus (Gra- mineae). Kew Bull. 32(3): 665-672. KEY TO SPECIES la. Culms slender, at most ] mm in diameter, un- 1 — С . Culms stouter che to produce 3- 8 dicus and axillary racemes; cilia of iid glume of sessile spikelet lashlike, . gradually reduced towards apex 2a. Lower glume of sessile spikelet 2 mm wide, the nerves 10-13, distinct .............. . rostratus 2b. Lower glume of sessile sid 1-1.5 m wide, the nerves 5- 7, indistir 3a. Ba ok of на glume of sessile spikelet yilose; cilia of low- er a of sessile spikelet 0.2 mm CA ces s tripsacoides E Back of lower glume of sessile spikelet ubescent, cilia of lower glume of ses- sile spikelet 0.8 mm long .......... E. ciliaris o c E. ciliaris H.B.K., Nov. Gen. Sp. 1: 193. 1816. E. tripsacoides var. ciliaris (H.B.K.) Hack. in A. DC., Monogr. Phan. 6: 333. 1889. Intergrading with E. tripsacoides and E. mu- ticus. Uncommon, as a weed of roadsides, pastures, Volume 77, Number 1 1990 Killeen 151 Grasses of Chiquitanía, Santa Cruz, Bolivia and disturbed cerrado near villages; unpalatable (0); flowering July to January, possibly stimulated by fire. Distribution: Mexico to Bolivia. (1408, 1682, 1683, 2118, 2774, 2820) tr) . muticus (Spreng.) Kuntze, Révis. Gen. Pl. 3(2): 350. 1898. Lycurus muticus Sprengel, Syst. Veg. 4(2): 32. 1827. р сап- didus Trin., Mem. Acad. Imp. Sci. St.-Pé- tersbourg, Sér. 6, Sci. Math. 2: 266: 1832. A. adustus Trin., Mém. Acad. Imp. Sci. St.- Pétersbourg, Ser. 6, Sci. Math. 2: 259. 1832. E. latiflorus Nees in Steudel, Syn. Pl. Glumac. 1: 364. 1854. E. candidus (Trin.) Hack. in Mart., Fl. Bras. 2(3): 306. 1883. E. adustus (Trin.) Ekman, Ark. Bot. 13(10): 5. 1913. E. viridulus Hack. in Stuck., Anales Mus. Nac. Hist. Nat. Buenos Aires 13: 414. 1906 An extremely variable taxon with numerous lo- cal races (allopatric species?). The most striking divergence in Chiquitania can be observed in cam- po rupestre populations: on Serrania de San Lo- renzo plants have villous racemes and the teeth of the lower glume of the sessile spikelet are 2-3 mm long (similar to E. candidus var. bisetosus Hack. & Lindman); in contrast, plants of the Serranía de Santiago have moderately pubescent racemes, and the teeth of the lower glume of the sessile spikelet are virtually lacking (E. adustus); cerrado popu- lations are generally intermediate between these two extremes. Abundant, the dominant grass in cerrado com- munities, providing more than 70% of the her- baceous cover in badly overgrazed savannas; very unpalatable (0), usually (but not always) with an KEY TO SPECIES odor of citrus. Cattle will only graze this species for a few weeks at the end of the dry season after it has been burned. Common but not abundant in campo rupestre, which is usually ungrazed; local name: paja carona, paja bruta, or pasto amargo; flowering is dependent upon fire; 2n — 20. Distri- bution: northern South America to Argentina, trop- ical Africa; Bolivia: Andean Piedmont of Santa Cruz, Beni, and the Yungas. (517, 654, 734, 877, 1105, 1196, 1239, 1276, 1454, 1733, 2187, 2777, 2785) E. rostratus Nees, Agrost. Bras. 357. 1829. An- dropogon rostratus (Nees) Trin., Mém. Acad. Imp. e 2 oe Ser. 6, Sci. Math. 2: 261. Rare roadside weed. Distribution: Argentina; Bo- livia: the Beni. (734, 770) E. pene Humb. & Bonpl. ex Willd., Sp. Pl. 4: 941. 1806. Restricted to the Andean Piedmont and inter- montane valleys, in well-drained sandy soils; flow- ering in August, possibly stimulated by fire. Dis- tribution: Mexico to Bolivia. (1113) Eragrostis Wolf Harvey, L. H. 1948. Eragrostis in North and iddle America. Ph.D. Dissertation. Univ. of Michigan, Ann Arbor, Michigan. Nicora, E. С. 1940. Nota taxonómica sobre Eragrostis neesii y Eragrostis articulata. Revista Ar- gent. Agron. 7(4): 257-273. la. 2 1(2- = lemma about 1 mm long, = sterile rachilla internode prolonged past the palea; panicle cm long and 20 cm wide; foliage glabrou airoides lb. Florets 2- q lemmas, panicles, and lag various; if only one floret developed, then the lemma greater than 1.5 mm long and the foliage pubescen 2а. Spikelets disarticulating from the apex ет the base, the glumes persistent, the rachilla internodes s falling. and palea 3a. Panico spicate to narrowly elliptic, the length 10 times the width; spikelets densely y de dn on short, erect, or ascending branches, the lowermost distantly placed; lemmas a. Paleas ong-ciliate on the keels; panicles spicate; plants 5-25 cm tall 4b. Paleas glabrous or minutely scabrous on the keels; panicles narrowly elliptic; plants 30- 100 tal m long. E. ciliaris japonica w = the cm ta Panicles neither spicate nor narrowly elliptic, pyramidal, or ovate, the length less than 5 | times width; spikelets loosely arranged on the bra 9a. Spikelets linear, lead-colored, the length about 10 times the widt m long. h; poss annuals nches; lemmas 2-2.5 m E. т 5b. Te elliptic or ovate, purple, tan, or lead-colored, the length about 3 times the width ual als or perenni i Esc Kuss 2 mm wide, the florets densely imbricate and obscuring the rachilla; uals solida 6b. Spikelets visible; perennials. lon 1-1.5 mm wide, the florets loosely imbricate and the rachilla more or less 7а. Spikelets tan; plants robust, the culms 1-1.5 m tall; panicles elliptic, 30-50 cm g. 152 Annals of the Missouri Botanical Garden 8a. Panicle branches distinctly verticillate at all nodes; lemmas acute; foliage cauline E . orthoclada 8b. olay па alternate (subverticillate at the lower nodes); lemmas acumi- e; folia . macrothyrsa -J c А Spikelets lead- isd the culms 30-70 em tall; panicles ovoid to pyramidal, 20- 9a. Spikelets (1)2-4-flowered; panicle disarticulating at the base of the peduncle; culms with no nodes visible above the basal folia . polytricha age 9b. Spikelets 3-6-flowered; panicle not disarticulating at the base of the peduncle; with 1-2 nodes visible above the basa | foliage ens 2b. RNA E from the base towards the apex, the glumes and lemmas deciduous, the rachilla and pale ersistent. 10a. Pedic els with a glandular ring (rarely lacking in E. neesii); foliage coarsely hirsute with papillose- airs; annuals; spikelets tan. Ma Foliage cauline; culms branched at lower and middle nodes, about 60 cm is spikelets ide, the lemmas 2.4 mm long l lb. Foliage qe culms unbranched, 10-20 cm tall; spikelets less than 2 mm ида, = lemm 2-3 mm wi 12a. 2b. Leaf blades lacking glandular pits; panicle contra . Pedicels not pe = foliage glabrous or н онй ыи or perennials; spikelets — c o” purple, tan, or lead-c 13a. Panicle chen o. 5-2 cm long, reflexed or spreading; foliage m 14a. Leaf blades 2-10 cm long, spikelets tan, ovate-elliptic; annuals ene Leaf ‘blades with glandular pits on the midnerve; cd OPEN зш ы ез E. articulata ted E . neesii esii var. lindmanii E. ne 14b. з blades 15-50 cm long, spikelets lead-colored, falcate at maturity; perennial y. perennis nch se 13b. Panicle pes n im longer than 2 em or ascending; foliage generally glabrous 5a. Low er glume narrowly lanceolate, acuminate, subequaling or surpassing the adjacent floret: annuals 16a. Axils of = branches and pedicels densely pilose; lemmas acuminate to Е E {ыле eis cuspi 16b. Axils of Mera branches and pedicels glabrous; lemmas acu E. Lower к ovate to lanceolate, only half as long as the жне. floret; “annuals 15b. or perennials. rufescens 17a. Spikelets 0.8 mm wide, the florets loosely imbricate and erect; margins of lemmas 17b. embranous; an c Un E ры FE [7] = un ы id | ho л muals ilosa wide, the florets pat imbricate and spreading (or ascending); margins of lemmas firm; per 18a. Axils of panicle пиз glabrous, "he NR laxly ascending, 5-15 cm 1-2 mm diria E acute to . Axils of panicle branches pilose, ^w branches strictly ascending, - : cm long; spikelets subsessile, crowded, tan with purple spots, 2-3 wide; lemmas acuminate E E. airoides Nees, Agrost. Bras. 509. 1829. Aira brasiliensis Raddi, Agrost. Bras. 36. 1823, non E. brasiliensis (Raddi) Nees, 1829. Spo- robolus brasiliensis (Raddi) Hack. ex Stuck., Anales Mus. Nac. Hist. Nat. Buenos Aires 21: 91. 1911. Agrosticula brasiliensis (Raddi) Herter, Revista Sudamer. Bot. 6: 145. 1940. Uncommon, forming distinct colonies in season- ally humid and shallowly inundated savanna; flow- ering in January and February. Distribution: north- ern South America to Argentina and Uruguay. (722, 1564, 1833, 2321) E. articulata (Schrank) Nees, Agrost. Bras. 502. 1829. Poa articulata Schrank, Sylloge Ra- tisb. 1: 194. 1827. E. glareosa Trin., Мет. A and distantly placed, lead-colored, E. bahiensis y. sec ш баш Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math. 1: 406. 1831. E. articulata var. gla- brescens Henrard, Meded. Rijks-Herb. 40: 69. 1921. Е. neesii var. laxa Jedwabn., Bot. Arch. 5: 206. 1924. Е. articulata var. eglan- dulosa Paca Revista Argent. Agron. 7: 259, 272, f. 4, 3. 1940. As treated by Nicora (1940), the E. articulata complex is composed of two species with several intergrading varieties, all of which are small caes- pitose annuals with basal, coarsely hirsute foliage. Some Bolivian specimens are intermediate between the varietal taxa (Table 4). Common, superficial soils over granitic outcrops and lateritic crests; occasional, as a weed along roads and as a pioneer of sand dunes in the Andean Piedmont; flowering Volume 77, Number 1 1990 Killeen 153 Grasses of Chiquitanía, Santa Cruz, Bolivia TABLE 4. Morphological variation for selected characters in the Eragrostis articulata complex. Glandular Contracted Glandular Sulcate Basa Branched midnerve' panicle pedicel caryopsis foliage culms Е. neesii = + + » Ра E var. expansiflora = — + + P var. lindmanii - + — - $ E Killeen 1338? — +/— — + + E E. articulata + — + + + ud Killeen 1247? + - + = + 2 var. eglandulosa = +/— 4 + 3 = E. chiquitaniensis + = + — n ! Refers to the midnerve of the leaf blade. * Intermediate specimens provisionally placed in the preceding taxon. August to January? Distribution: Brazil, Paraguay, and Argentina; Bolivia: Cochabamba, Andean Pied- mont of Santa Cruz, and the Gran Chaco. (1122, 1247, 1481, 1568) E. bahiensis Schrader ex Schultes, Mant. 2: 318. 1824. Poa microstachya Link, Hort. Berol. 1: 185. 1827. E. expansa Link, Hort. Berol. 1: 190. 1827. E. psammodes Trin., Мет. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math. 1: 400. 1831. E. firma Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 4(1): 74. 1838. E. blepharophylla Jedwabn., Bot. Arch. 5: 197. 1924. E. macra Jedwabn., Bot. Arch. 5: 200. 1924. A difficult taxon with variable inflorescence mor- phology and indeterminate spikelets (i.e., the num- ber of florets/spikelet ranges from 5 to 30 de- pending upon the robustness of the individual plant). Occasional, valley-side campo (upslope), seasonally humid savanna, and seasonal ponds; frequently in disturbed habitats; flowering throughout the year; п = 60. Distribution: Peru, Brazil, Paraguay, and Argentina: apparently introduced into the United States; Bolivia: Andean Piedmont of Santa Cruz and the Yungas. (779, 782, 785, 821, 1125, 1235, 1256, 1336, 1579, 1656, 1664, 1809, 2110, 2271; Bruderreck 108 ISC, LPB) E. chiquitaniensis Killeen, sp. nov. TYPE: Boliv- ia. Santa Cruz: Estancia San Ignacio, 25 km N of San José de Chiquitos, Prva. Chiquitos, 17%35'S, 60°45'W, 320 m, 1 Feb. 1986, Killeen 1728 (holotype, ISC; isotypes, LPB, F, MO, US, SI, NY, CTE). Figure 3. E. arti d erue Nees affinis sed robustis, 60 cm altis, culmis 4-6-nodis, ramificantibus c copiose ad no- dos medios et тайт. foliis caulinis, laminis 15-25 cm longis, 4-7 mm latis; paniculis 16-27 cm longis, 4-6 cm latis; spiculis 2-4 mm latis, lemmatibus 2.0-2.5 mm longis notabilis. Caespitose annuals; culms branched, 60 cm tall, glabrous. Foliage cauline, coarsely hirsute with pa- pillose-based hairs; sheaths longer than the inter- nodes below but shorter than internodes at upper nodes, glandular-pitted on the abaxial midrib; lig- ules ciliate, 1 mm long; blades cordate based, flat, lax, 15-25 cm long, 4-7 mm wide, broadest at the base, the apex long acuminate. Inflorescence an open, elliptic panicle, 16-27 cm long and 4- 6 cm wide; branches alternate to subverticillate at the lower nodes, ascending; axils of branches glan- dular, bearded with stiffly spreading hairs; axis, branches, and pedicels scabrous; pedicels 2-10 mm long, ascending, with a glandular ring 1-2 mm below the insertion of the spikelet. Spikelets pyramidal, 4—12-flowered, the lower bracts spread- ing, the upper ascending, 4-8 mm long, 2-4 mm wide, glabrous; disarticulation from the base of the spikelet towards the apex, the glumes and lemmas falling, the rachilla and paleas persistent; lower glume ovate, acute, 1.4—1.8 mm long, keeled, the lateral nerves indistinct; upper glume similar, 1.6- 2.2 mm long, covering Y of the adjacent lemma; lemmas ovate, acute, keeled at the apex, rounded at the base, the lateral nerves indistinct but visible, 2.4 mm long, 0.4 mm wide when folded; palea 1.6 mm long, minutely ciliate on the keels; caryopsis 0.6 х 0.6 mm, truncate, not sulcate. This species is closely related to E. articulata but differs by its more robust habit, branched culm, cauline foliage, and larger spikelets. Locally com- mon in an open deciduous savanna woodland with a canopy 10—15 m tall and about 60% grass cover. The land north of San José de Chiquitos has rel- atively fertile, alluvial soils and is rapidly being converted to mechanized agriculture; if narrowly Annals of the Missouri Botanical Garden FIGURE 3. Eragrostis chiquitaniensis (Killeen 1728). — A. Habit and inflorescence (bar = 5 cm). —B. Spikelet i ) ui and pedicel (bar = 1 mm; arrow pointing to glandular ring on pedicel). Volume 77, Number 1 1990 Killeen 155 Grasses of Chiquitanía, Santa Cruz, Bolivia restricted to this area, this species will soon be extinct. E. ciliaris (L.) R. Br. in Tuckey, Narr. Exp. Congo . 1818. Poa ciliaris L., Syst. Nat., ed. 10. 8: 875. 1759. Macroblepharus contrac- tus Philippi, Linnaea 19: 101. 1858. Common as a weed in streets, yards, and gardens of villages. Distribution: throughout the tropics. (795 E. japonica (Thunb.) Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math. 1: 405. 1831. Poa japonica Thunb., Fl. Japon. 51. 1784. P. glomerata Walter, Fl. Carol. 80. 1788. E. hapalantha Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math. 1: 409. 1831. E. interrupta (Lam.) Doell in Mart., Fl. Bras. 2(3): 157. 1878. E. glome- rata (Walter) L. Dewey, Contr. U.S. Natl. Herb. 2: 543. 1894. Not yet collected in Chiquitanía, as a weed in sandy soils along roads on the Andean Piedmont of Santa Cruz. Distribution: throughout the tropics. (2731) E. lugens Nees, Agrost. Bras. 505. 1829. E. flaccida Lindman, Kóngl. Svenska Vetensk. Acad. Handl. 34(6): 17. 1900. Similar to E. soratensis Jedwabn., which has 3- 6 florets but narrower panicles and no visible nodes above the foliage. Rare in Chiquitanía, more com- mon in the Gran Chaco and Andean Piedmont; flowering is stimulated by fire, October to January. Distribution: southern United States to Argentina; Bolivia: Tarija and the Andean Piedmont of Santa Cruz. (1237, 1277, 1565, 1573) E. macrothyrsa Hack., Feddes Repert. Spec. Nov. Regni Veg. 8: 47. 1910 Rare, a single population 10 km S of Concep- ción, cerrado / forest margin; flowering in March. Distribution: Argentina and Paraguay. (1832) E. ja (H.B.K.) Steudel, Syn. Pl. Glu- ac. 1: 276. 1854. Poa maypurensis H.B.K., Nov. Gen. Sp. 1: 161. 1816. P. racemosa Vahl, Ecolg. Amer. 1: 7. 1796, non Thunb., 1794. Р. vahlii Roemer & Schultes, Syst. Veg. 2: 563. 1817. E. vahlii (Roemer € Schultes) Nees, Agrost. Bras. 499. 1829. Е. acuminata Doell in Mart., Fl. Bras. 2(3): 153. 1878. Not yet collected in Chiquitanía but common on the Andean Piedmont of Santa Cruz and the Beni. Distribution: Mexico to Brazil. E. neesii Trin., Mém. Acad. Imp. Sci. St.-Pé- tersbourg, Sér. 6, Sci. Math. 1: 405. 1831. E. brasiliana Nees, Agrost. Bras. 510. 1829. E. villosa Steudel, Syn. Pl. Glumac. 276. 1854. E. lindmanii Hack., Kóngl. Svenska Vetensk. Acad. Handl. 34(6): 19. 1900. E. neesii var. lindmanii (Hack.) Ekman, Ark. Bot. 13(10): 51. 1913. Specimens without glandular pedicels are usually placed in E. neesii var. lindmanii (Table 4). Oc- casional, superficial soils over granitic outcrops and lateritic crests in cerrado; flowering October to February. Distribution: Brazil, Paraguay, Uru- guay, and Argentina. (pedicels glandular: 592, 805; lacking glands on pedicels: 747, 1338, 1800) E. orthoclada Hack., Bull. Herb. Boissier II. 4(3): 281. 1904. F. longipila Hack. in Stuck., Anales Mus. Nac. Hist. Nat. Buenos Aires 21: 132. 1911. Scrub forest, San José de Chiquitos; flowering in January. Distribution: Paraguay and Argentina. (1709) E. perennis Doell in Mart., Fl. Bras. 2(3): 144. Locally common in cerrado at Santiago de Chi- quitos; flowering is dependent upon fire. Distribu- tion: Brazil. (2793) E. pilosa (L.) P. Beauv., Ess. Agrostogr. 71, 162. 1812. Poa pilosa L., Sp. Pl. 68. 1753. Common, as a weed in streets and gardens of villages. Distribution: throughout the tropics. (1128) E. polytricha Nees, Agrost. Bras. 507. 1829. F. lugens var. glabrata Doell in Mart., Fl. Bras. 2(3): 140. 1878. E. lugens var. glabrescens Doell in Mart., Fl. Bras. 2(3): 141. 1878. Common and widespread but never abundant, cerrado. The panicle disarticulates at the base of the peduncle and forms a tumbleweed at maturity; flowering is dependent upon fire; 2n — 60. Distri- bution: Central America to Argentina and Chile; Bolivia: Andean Piedmont of Santa Cruz. (524, 615, 1104, 1195, 1371, 2188, 2241) E. p Schrader ex Schultes, Mant. 2: 319. 824. E. acicularis Trin., Mém. Acad. Imp. 156 Annals of th Missouri ía Garden Sci. St.-Pétersbourg, Ser. 6, Sci. Math. 1: 406. 1831. Е. multipes S. L. Moore, Trans. Linn. Soc. Bot. II. 4: 511. 1895. F. poor Bot. Arch. 5: 205. 1924 neura Jedwabn., Common, as a weed along roads in sandy or clay soils; naturalized in cerrado and seasonally humid savannas; flowering December to May. Dis- tribution: Brazil. (620, 633, 780, 913, 997, 1638, 1678, 1879) E. secundiflora C. Presl, Rel. Haenk. 1: 276. 1830. E. compacta Salzm. ex Steudel, Syn. Pl. Glumac. 1: 275. 1854. Common, as a weed along paths in seasonally humid savannas and valley-side campos; occasion- al, in cerrado; flowering October to June. Distri- bution: Brazil. (613, 1337, 1342, 1358, 1482, 1542, 1665, 2457) E. solida Nees, Agrost. Bras. 501. 1829. E. ај finis Salzm. ex Steudel, Syn. Pl. Glumac. 1: 277. 1854 Occasional, as a weed in cerrado and well-drained soils; flowering August to April. Distribution: Brazil and Paraguay. (1127, 1560, 2005; Bruderreck 298 ISC, LPB) E. tenuifolia (A. Rich.) Hochst. ex Steudel, Syn. Pl. Glumac. 1: 268. 1854. Poa tenuifolia A. Rich., Tent. Fl. Abyss. 2: 425. 1851. Common weed along streets and yards of towns. Distribution: throughout the tropics. (59 Eriochloa H.B.K. Shaw, R. B. & F. E. Smiens. 1981. Anatomical and morphological characteristics of the Er- iochloa (Poaceae) of North America. Bot. Gaz. (Crawfordsville) 142: 534-544. KEY TO SPECIES la. Upper lemma with a mucro + mm long (hidden by the lower lemma and upper glume); panicle us 4-15 racemes; MEE eds arranged; achis pubesc E. punctata lb. а кнн (eom a mucro; panicle c of 1-4 racemes; spikelets regularly 2-rowed, secund; rachis densely hispid. 2 s 3- 4 mm long, culms 25- 50c A distachya Spikelets 5-7 mm long; culms 50-100 c tall E. lo ко = E. distachya H.B.K., Nov. Gen. Sp. 1: 95, t. 30. 1816. Helopus brachystachys Trin., Sp. Gram. 324: 277. 1831. Common to locally abundant in savanna wetland communities; rarely in cerrado and superficial soils of granitic outcrops; palatable (2); flowering Jan- uary to (March) June; 2n = 18. Distribution: Cen- tral America to Brazil and Paraguay; Bolivia: An- dean Piedmont of Santa Cruz and the Beni. (700, 895, 946, 1458, 1859, 1999, 244.1, 2592) E. grandiflora (Trin.) Benth., J. Linn. Soc. Bot. 19: 39. 1881. Helopus grandiflora Trin., Sp. Gram. 3(24): 278. 1831. Uncommon, cerrado and cerrado/forest tran- sition; palatable (3); flowering in March and April. Distribution: Brazil and Paraguay. (1844, 1955, 006) E. зн (L.) Desv. in Hamilt., Prodr. Pl. Ind. Occ. 5. 1825. Milium punctata L., Amoent. Acad. 5: ч 1759. Helopus cognatus Steu- del, Syn Pl. Glumac. 1: 101. 1854. Common, a weed of pastures and roadside ditch- es; locally abundant in some seasonally inundated savannas; flowering throughout the year. Distri- bution: southern United States to Argentina; Bo- livia: Andean Piedmont of Santa Cruz, Beni, and the Yungas. (921, 971, 996, 1187, 1316, 1506, 714) Eriochrysis P. Beauv. Swallen, J. 1966. Notes on grasses. Phytologia 14(2): 88-91. KEY TO SPECIES la. Leaf blades velutinous; lower glume long-ciliate, the hairs 1.5-4 mm long, very dark reddish ikelets. brown, a uring the s 2a. Hairs borne on the upper '4 th of the lower glume and obscuring the tridentate or blunt apex . 7. cayanensis 2b. Hairs marginal only, not on the upper 4% th of the lower glume, the apex rounded ..... %. ео Leaf blades glabrous, scabrous or 3C zd to 3 mm), light p brown to whitish, not obscuring the spikele Ji ower glume bo rounded at apex ... E. laxa 3b. Lower glume lanceolate or ee falcate, acuminate or bidentate at the a 4а. о 6-10(15) em long; plants О. pa tall. E. holcoides 4b. Pale 20-30 cm long; plants 1-2 tall E . 10 arminge ana Volume 77, Number 1 1990 Killeen 157 Grasses of Chiquitanía, Santa Cruz, Bolivia FIGURE 4. = 5 ст). — Species of Eriochrysis. A-C. E. x сае ois (Killeen аа — А. Habit. let.—D. E. с Lower glume of the sessile spike —B. Inflorescence (A nsis (Killeen 2265). Lower glume of 1 mm). and B: bar С do sessile spikelet. — E. E. laxa (Killeen 2264). Lower glume of the sessile prr. (C, D, and E: bar = E. cayanensis P. Beauv., Ess. Agrostogr. 8, pl. 4, f. 11. 1812. Saccharum cayennense (P. Beauv.) Benth., J. Linn. Soc., Bot. 19: 66. 1881 Common, valley-side campo (midslope) and sa- vanna marsh; occasional, seasonally humid or in- undated savanna; rarely in seasonal ponds and laguna margins; unpalatable (0); flowering Novem- ber to January and irregularly throughout the year; 2n — 20. Distribution: Mexico and the West Indies to Argentina; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (1131, 1346, 1438, 1510, 1594, 1860, 2265) E. xconcepcionensis Killeen, nothosp. nov. TYPE: Bolivia. Santa Cruz: Estancia Santa Maria, 10 km S of Concepción, Prva. Nuflo de Chavez, 16?13'S, 62%0'W, 500 m, 16 Mar. 1987, Killeen 2384 (holotype, ISC; iso- types, LPB, F, MO, US, SI, CTE, NY). Fig- e 4. xa d affinis sed pilis nodis culmorum 2 mm longis; zit i cuprinis, cum pilis callorum spiculas padirenta notililli. Formula: E. cayanensis P. Beauv. x E. laxa Swallen. Caespitose perennial; culms unbranched, the nodes densely bearded with hairs about 2 mm long, the internodes glabrous. Foliage mainly basal; sheaths glabrous below, pubescent towards the apex; 158 Annals of the Missouri Botanical Garden ligule a ciliate membrane 0.8 mm long; blades 6- 25 cm long and 3-6 mm wide, long-attenuate at the base, reduced at upper nodes, velutinous on both surfaces, flat or involute when stressed. In- florescence exserted, a lobed, spicate panicle 14 cm long an .5 сш wide, the lower branches distantly placed. Spikelets paired, sessile and ped- icellate, the two spikelets bisexual and similar, oc- casionally only pistillate; both subtended by a bearded callus of reddish hairs 2—3 mm long; sessile spikelets 3-3.5 mm long, the pedicellate spikelets 2-2.5 mm long, largest at the base of the branches, gradually reduced in size at upper rachis nodes; disarticulation below the glumes, the sessile spikelet falling with the attached rachis internode and ped- icel, the pedicellate spikelet falling separately. Low- er glume narrowly elliptic to obovate, dorsally com- pressed, indurate and nerveless, rounded at the tip, glabrous on back, ciliate on the margins, the stiffly spreading, reddish brown hairs 1.5-3.0 mm long; upper glume laterally compressed, ciliate on the margins; lower lemma hyaline, similar to lower glume in form and length; upper lemma 1.2-1.7 mm long, hyaline, ciliate on the margins; palea minute or lacking; stamens 3, the anthers 1.5 mm long, purple-orange; lodicules membranous, strong- ly nerved. A single elliptic population composed of about 300 plants growing in the same habitat with both parental species. The hybrid has the foliage ves- titure and spikelet pubescence of the local forms of E. cayanensis, and the inflorescence mor- phology and the shape of the lower glume is like E. laxa. The three taxa were distinct and no in- termediates were observed. Pollen development and seed set in the hybrid was abnormal (nonstained pollen: 82%; seed set: 0%) compared with both parents (nonstained pollen: 8% in E. laxa and 8% in E. cayanensis; seed set: 34% in E. laxa and 3796 in E. cayanensis). Phenology of the hybrid is unique— the plants bloomed in March and April, two to three months after the peak flowering season of both parents; locally common, midslope on a valley-side campo. (Paratype: Killeen 1867) E. holcoides (Nees) Kuhlm., Comiss. Linhas Te- leg. Estrateg. dass Grosso Amazonas 67, An- nexo 5, Bot. 11: 89. 1922. Anatherum hol- coides Nees., ы. Bras. 324. 1829. Andropogon holcoides (Nees) Kunth, Revis. Gramin. 2: 4 29. Saccharum holcoides (Nees) Hack. in Mart., Fl. Bras. 2(3): 254. 1883. S. holcoides var. brevipilum Hack. in A. DC., Monogr. Phan. 126. 1889. 5. hol- coides var. penicillare Hack. in A. DC., Monogr. Phan. 126. 1889 A morphologically diverse taxon; plants with burned foliage generally have asymmetrically fal- cate spikelets with strongly nerved, bidentate, lower glumes (E. holcoides s.s.) specimens with un- burned foliage have elliptic to lanceolate spikelets with weakly nerved, acuminate lower glumes (S. holcoides var. brevipilum Hack.). Spikelet hairs range from almost white to dark brown (S. hol- coides var. penicillae Hack.), and glumes range from glabrous to sparsely ciliate. Bolivian material collected to date is like the type of the species. Rare, valley-side campos (midslope); locally abun- dant in a boglike seep in a campo rupestre complex (Serrania de Santiago); flowering is dependent upon fire. Distribution: Colombia, Brazil and Paraguay. (1190, 2170, 2792) E. laxa Swallen, Phytologia 14: 89. 1966. Occasional, valley-side campos with E. caya- nensis and Paspalum malmeanum; unpalatable (0); flowering October (December) to April; 2n — 20. Distribution: central Brazil. (644, 737, 1365, 1437, 1595, 1870, 2264) E. warmingeana (Hack.) Kuhlm., Comiss. Lin- has Teleg. Estratég. Mato Grosso Amazonas 67, Annexo 5, 11: 29. 1922. Saccharum warmingiana Hack. in Mart., Fl. Bras. 2(3): 254. 1883. Not yet collected in Chiquitania but probably existing in the extensive pantanal region of Alto Paraguá (Prva. Velasco). Distribution: Brazil; Bo- livia: Andean Piedmont of Santa Cruz and the Beni where it is a zonal dominant of seasonally inundated savannas. (2596; Steinbach 7032 US) Eustachys Desv. KEY TO SPECIES la. MUR о with lashlike сша about 10 mm lor eel lacking or weakly developed, te racemes many Е; ee lb. Lemmas ovate, with ecd cilia 0.2-0.5 n long, the keel strongly e pubescent (rarely glabrous); racemes 2-ma . E. caribaea E. caribaea (Spreng.) Herter, Rev. Sudamer. Bot. 6: 147. 1940. Chloris caribaea Spreng., Syst. Veg. 1: 295.1825. C. bahiensis Steudel, Syn. Pl. Glumac. 1: 208. 1854. Е. bahiensis (Steu- del) Herter, Fl. Ilustr. Uruguay 1: 85, f. 339. 1941. C. capensis var. bahiensis (Steudel) L. Parodi, Revista Argent. Agron. 20: 26. 1953. Volume 77, Number 1 1990 Killeen 159 Grasses of Chiquitanía, Santa Cruz, Bolivia Rare, roadside weed; flowering November to February. Distribution: Brazil, е Uru- guay, and Argentina. (1407, 1713) E. distichophylla (Lag.) Nees, Agrost. Bras. 418. 1829. Chloris distichophylla Lag., Gen. Sp. Pl. 4. 1816. С. d Trin. in Sprengel, Neue Entd. 74. 1821. C. fasciculata Schrader ex TEA Mant. 2: 339. 1824. Paspalum superbum Sprengel, Syst. Veg. 1: 248. 1825. C. acuminata Trin., Sp. Gram. 3: t. 305. 1831. Occasional, roadside weed; flowering January to June. Distribution: Peru, Brazil, Paraguay, Argen- tina, and Chile. (695, 977) Gouinia Fourn. Swallen, J. R. 1935. The grass genus Gouinia. Amer. J. Bot. 22: 31-41. KEY TO SPECIES la. Panicle branches lacking pr on basal 1⁄4- inal nerves for А wee culms (2)3-5 mm in diameter „u. G. latifolia lb. Panicle branches bearing spikelets to the base; lemma ciliate on the marginal nerves for 14-14 e its length, the awns 8-15 mm long; culms mm in diameter G. virgata G. latifolia (Griseb.) Vasey, Contr. U.S. Natl. Herb. 1: 365. 1895. Tricuspis latifolia Gri- seb., pee Konig]. Ges. Wiss. Gottingen 19: 259. 1874. Rare, roadside weed in forest; flowering in May. Distribution: Mexico to Argentina. (948) С. virgata (С. Presl) Scribn., Bull. U.S.D.A. Div. Agrost. 4: 10. 1897. Bromus virgatus C. Presl, Rel. Haenk. 1: 263. 1830. Common, roadside weed in forest soils; flowering February to June. Distribution: Mexico to Brazil. (833, 969, 1885; Puerto Suarez, Chase 11151 Guadua Kunth Young, S. M. 1985. The taxonomy and natural history of the Bambusa guadua complex (Po- aceae: Bambusoideae). M.S. Thesis, Univ. of Florida, Gainesville, Florida. McClure, F. A. 1973. Genera of the bamboos native to the New World (Gramineae; Bambusoideae). Smithson. Contr. Bot. 9: 1-148. 1973. KEY TO SPECIES . Culms 1-5 cm in diameter, 2-15 m tall; sheaths of culm leaves 10-30 cm lon 2a. Culms thin-walled, ideal when young green; pseudopetioles 10 mm long; foliage leaf blades about 7 times as long as w m- weberbaueri 2b. Culms solid or thick-walled, the intei hispid, bearing эше whitish hairs when young, becoming glabrous with maturity, yellow-green; ре кос i З mm long; (oigo leaf blades 10-15 times as long as Кы С. paniculata lb. Culms 5-20 cm in diameter, 20-30 m tall; sheaths of ef leaves 30-40 cm long 3a. Abaxial surface of sheath deny pa as long as wide; pre "I the culms hi di packed clumps 3-8 m in diame superba 3b. Abaxial surface of sheath uisum foliage leak ies aut 10 times as long as wide; igate, not forming dense, well- part de 2 . paraguayana G. paniculata Munro, Monogr. Bamb. 85. 1868. Bambusa paniculata (Munro) Hack., Oes- terr. Bot. Z. 53: 195. 1903. B. munroi Hack., Feddes Repert. Spec. Nov. Regni Veg. 7: 374. 1909. Abundant, forming colonies along cerrado/for- est margins and dominating tens of thousands of hectares of low forest known locally as *guapa- sales." Scattered populations began to flower in October 1987 after a severe drought and wide- spread forest floor fires; only unburned individuals were blooming initially. Populations also flowered in May 1977. Culms provide the matrix for the traditional mud and tile roof of local construction and are used to make pig-proof fences for vegetable gardens; palatable (3); local name: guapa. Distri- bution: Costa Rica to Brazil. (752, 2305, 2329, 2762, 2807; Krapovickas & Schinini 32436 US; Thomas 5659 NY, LPB) G. paraguayana Doell in Mart., Fl. Bras. 2(3): 179. 1880. Bambusa paraguayana (Doell) Bertoni, Anales Ci. Parag. 2(2): 159. 1918. Rare, Parque Nacional “Noel Kempff Merca- do”; forming large diffuse colonies in mesic forest; local name: taquarembó. Distribution: Brazil, Par- aguay, and northeastern Argentina. (2763, 2764) G. superba Huber, Bol. Mus. Goeldi 4: 479. 1904. Bambusa superba (Huber) McClure, Smithson. Contr. Bot. 9: 68. 1973. 160 Annals of the Missouri Botanical Garden Locally abundant along the Rio Quizer near San Ramón and 10 km northwest of San Javier; local name: taquarembó. Distribution: Brazil. (2302) G. weberbaueri Pilger, Feddes Repert. Spec. Nov. Regni Veg. 1: 257. 1830. Bambusa weberbaueri (Pilger) McClure, Smithson. Contr. Bot. 9: 68. 1973. A single known colony in Chiquitania, on the grounds of the municipal swimming pool at San José de Chiquitos; local name: taquara. Distribu- tion: Peru; Bolivia: the Yungas. (2818, 2627) Gymnopogon P. Beauv. Smith, Jr., genus Gymnopogon (Gramineae). lowa State Univ. J. Sci. 45: 319-385 J. P. 1971. Taxonomic revision of the KEY TO SPECIES la. Leaf blades 5-20 mm wide; awns straight, not hygroscopic; spikelets distantly placed on the of the spicate branches, crowde d to- rds apex 7. кн Lb. Leaf blades 3-5 mm wide; awns flexuc groscopic; spikelets uniformly distrae ns the whole BUD of the spicate brane b байн С. fastigiatus Nees, Agrost. Bras. 430. 1829. Monochaete fastigiata (Nees) Doell in Mart., Fl. Bras. 2(3): 79. 1878. Doellochloa dn giata (Nees) Kuntze, Rev. Gen. Pl. 2: 773. 1891. G. jubiflorus Hitchc., Contr. U.S. us Herb. 24: 412. 1927. G. fastigiatus subsp. jubiflorus (Hitchc.) J. P. Smith, Iowa State Univ. J. Sci. 45: 361. 1971. Common in valley-side campos (midslope), oc- casionally in seasonally inundated savannas and savanna marsh; flowering April to (May) July. Dis- tribution: Colombia, Venezuela, and Brazil; Bolivia: Beni and the Yungas. (1029, 1144, 1350, 2039, 2077, 2588 С. spicatus (Sprengel) Kuntze, Revis. Gen. Pl. 3(3): 354. 1891. Polypogon spicatus Spren- gel, Syst. Veg. 1: 243. 1825. G. biflorus Pilger, Bot Jahrb. Syst. 30: 139. 1901. Plants are variable within populations for stat- ure, size of inflorescence, and the number of flo- rets/spikelet. This species is a common constituent of cerrado, occupying open spaces between large bunch grasses; unpalatable (0); flowering February to (May) July; 2n — 20. Distribution: Mexico to Uruguay and Argentina; Bolivia: Andean Piedmont of Santa Cruz. (890, 924, 1011, 1813, 1990, 2017, 2092, 2481) 1248, 1567, Gynerium Willd. ex P. Beauv. Conert, H. J. 1961. Die Systematik und Anatomie der Arundinoideae. Cramer Verlag, Wein- heim. С. sagittatum (Aubl.) P. Beauv., Ess. Agrostogr. 138, 153, pl. 24, f. 6. 1812. Saccharum sagittatum Aublet, Hist. Pl. Guiane 1: 50. 1775. G. procerum P. Beauv., Ess. Agrostogr. 164, pl. 24, f. 6. 1812. G. saccharoides Humb. & Bonpl., Pl. Aequin. 2: 105, pl. 115. 1813. Arundo saccharoides (Humb. & Bonpl.) Poir. in Lam., Encycl. Suppl. 4: 703. 1816. G. parviflorum Nees, Agrost. Bras. 463. 1829. Widespread along the banks of the Rio Grande; local name: caria brava. Distribution: Central America to Argentina; Bolivia: the Yungas and the eni. Hackelochloa Kuntze H. granularis (L.) Kuntze, Revis. Gen. Pl. 2: 776. 1891. Cenchrus granularis L., Mant. Pl. 2: 575. 1771. Manisuris granularis (L.) Sw., Prodr. 25. 1788. R ytilix granularis (L.) Skeels, U.S.D.A. Bull. Foreign Pl. Intr. 282: 20. 1913. Mnesithea granularis (L.) Koning & Sosef, Blumea 31(2): 303. 1986. Uncommon, on gravel soils of lateritic crests in cerrado; flowering November to May. Distribution: throughout the tropics. (634, 892, 1487, 1615) Hemarthria R. Brown H. altissima (Poir.) Stapf & C. E. Hubb., Bull. Misc. Inform. 1934: 109. 1934. Rottboellia altissima Poir., Voy. Barbarie 2: 105. 1789. R. fasciculata Lam., Tabl. Encycl. 1: 204. 1791. Manisuris altissima (Poir.) Hitchc., J. Wash. Acad. Sci. 24: 282. 1934. M. fascicu- lata (Lam.) Hitchc., Am. J. Bot. 2: 299. 1915. Rare, in disturbed soils of seasonally inundated savanna; palatable (3). Distribution: throughout the tropics. (1508, 1553) Homolepis Chase Zuloaga, F. O. & T. R. Soderstrom. 1985. Clas- sification of the outlying species of New World Panicum de Paniceae). Smithson. -63 Contr. Bot. 59: Volume 77, Number 1 Killeen 161 1990 Grasses of Chiquitanía, Santa Cruz, Bolivia Н. aturensis (H.B.K.) Chase, Proc. Biol. Soc. KEY TO SPECIES Wash. 24: 146. 1911. Panicum aturense H.B.K., Nov. Gen. Sp. 1: 103. 1816. P. viridiflorum Nees, Agrost. Bras. 135. 1829. Р, een Presl, Rel. Haenk. 1: 312. 1830 Rare, streambank in forest; flowering in Janu- ary. Distribution: Mexico to Brazil; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (1676) Hymenachne P. Beauv. KEY TO SPECIES la. Panicle branches appressed to the axis, forming a spicate panicle; leaf blades 0.5-2(3) cm 1. de lb. Panicle branches laxly ascending to o the panicle not spicate; leaf blades (1.5)2-4 с wide H. елда Н. amplexicaulis (Rudge) Nees, Agrost. Bras. 276. 29. Panicum amplexicaule Rudge, Pl. Guian. 1: 21, t. 27. 1805. Agrostis mono- stachya Poir., Encycl. Suppl. 1: 256. 1810. P. perdensum Steudel, Syn. Pl. Glumac. 1: 65. 1854. Common, roadside ponds and along banks of streams in full sunlight, especially disturbed cattle crossings; palatable (4); flowering October to April; local name: cañuela. Distribution: Mexico to Uru- guay and Argentina; Bolivia: the Beni. (790, 1362B, 1691, 1739, 1778, 2380) H. donacifolia (Raddi) Chase, J. Wash. Acad. Sci. 13: 177. 1923. Panicum donacifolium Raddi, Agrost. Bras. 44. 1823. P. auricula- tum Willd. ex Sprengel, Syst. Veg. 1: 322. 1825. P. cordatum Doell in Mart., Fl. Bras. 2(3): 239. 1880. H. auriculata (Willd.) Chase, Proc. Biol. Soc. Wash. 21: 5. 1908. Hymen- achne cordata (Doell) Kulhm., Comiss. Linhas Telegr. Estratég. Mato Grosso Amazonas 67, Annexo 5, Bot. 11: 45. 1922. Similar to Н. amplexicaulis in habit, distribu- tion, and phenology, the two species frequently occurring in mixed populations; palatable (4); local name: cañuela. Distribution: Costa Rica to Para- guay and northern Argentina; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (683, 1362A, 1743, 2450) Hyparrhenia Andersson in Stapf. Clayton, W. D. 1969. A revision of the genus Hyparrhenia. Royal Botanic Gardens, Kew. Additional Series II. London. la. Spatheate sheaths аан racemes hirsute; spikelet pairs 2-3 per e; lower glume of sessile spikelet with 2 medial pi grooves; racemes reflexed at maturity, rachis hairs white A A, —À H. bracteata lb. еы sheaths subtending racemes gla- brous; spikelet pairs 4-8 per raceme; lower zia me ki sessile spikelet lacking grooves; ra- cemes ascending, rachis hairs reddish H H. bracteata (Willd.) Stapf in Prain, Fl. Trop. Africa 9: 360. 1919. Andropogon bracteatus Humb. & Bonpl. ex Willd., Sp. Pl. ed. 4. 914. 1806. Anthistiria foliosa H.B.K., Nov. Gen. Sp. 1: 191. 1816. Anthistiria reflexa H.B.K., Nov. Gen. Sp. 1: 191. 1816. Anthistiria pi- losa J. S. Presl & С. Presl, Rel. Haenk. 1: 348. 1830. Andropogon trachypus Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math. 2: 280. 1832. Anthistiria an- dropogonoides Steudel, Syn. Pl. Glumac. 1: 402. 1855 Common, seasonally inundated savanna and sa- vanna marsh; somewhat palatable (2), foliage coarse, but less abundant in overgrazed savanna; flowerin April to (May) June; 2n = about 40. Distribution: Mexico to Brazil and tropical Africa; Bolivia: the Beni. (888, 897, 912, 2020, 2108, 2480, 2595) Н. rufa (Nees) Stapf in Prain, Fl. Trop. Africa 9: 304. 1919. Trachypogon rufus Nees, Agrost. Bras. 345: 1829. Andropogon rufus (Nees) Kunth, Révis. Gramin. 2(39). 1832. An important cultivated forage grass that has become naturalized throughout the region in cer- rado com and as a weed along roads. When seeded into degraded cerrado, this species will re- place Elionurus muticus as the dominant grass if grazed lightly for two to three years; flowering May to July, a secondary peak of flowering in Novem- ber; local name: yaragua. (516, 1008, 2108A) Ichnanthus P. Beauv. Stieber, М. Т. 1982. Revision of /chnanthus sect. Ichnanthus (Gramineae, Panicoideae). Syst. Bot. 7(1): 85-115. Stieber, M. T. 1987. Re- vision of Ichnanthus sect. Foveolatus (Gra- mineae: Panicoideae). Syst. Bot. 12: 187- KEY TO SPECIES la. Leaf blades 3-5 mm wide; panicles 5-10 cm long; branches with 3-8 spikelets clustered on 162 Annals of the Missouri Botanical Garden the lower side of the midportion of the branches, and a single long-pedicellate spikelet at apex . . procurrens lb. Leaf blades 5-20 mm wide; panicles cm long, spikelets irregularly arranged along the branches 2a. Paired vinglike appendages at base of up- per floret present; blades stiffly spreading, usually fons culms stout ... /. inconstans 2b. Paired appendages lacking (scars only); blades lax, icd EAS culms not stout E A co Ё pallens I. inconstans (Trin. ex Nees) Doell in Mart., Fl. Bras. 2(2): 287. 1877. Panicum inconstans Trin. ex Nees, Agrost. Bras. 132. 1829. І. velutinus Ekman, Ark. Bot. 13: 31, p. 2, f. 2. 1913. I. peruvianus Mez, Feddes Repert. Spec. Nov. Regni Veg. 15: 129. 1918. /. polycladus Mez, Feddes Repert. Spec. Nov. Regni Veg. 15: 129. 1918; see Stieber (1987) for extensive synonymy. Common, rooting in organic soils of cracks on granitic or sandstone outcrops, rarely in cerrado or scrub; palatable (3); flowering January (Feb- ruary) to April and irregularly throughout the year; n = 20. Distribution: Peru, Brazil, Paraguay, and Argentina; Bolivia: the Yungas. (1449, 1731, 1818, 1975, 2331, 2782B) I. pallens (Sw.) Munro ex Benth., Fl. Hongk. 414. 1861. P. pallens Sw., Prod. 23. 1877. 1. axillare (Nees) Нисһс. & Chase, Contr. U.S. Natl. Herb. 18: 334. 1917. I. tipu- aniensis Rogers, Phytologia 26: 59-64. 1973; see Stieber (1982) for extensive synonymy. Occasional, forming colonies in humid soils of forest and as a weed following shifting agriculture; palatable (4); flowering November to May. Distri- bution: throughout the tropics. (1376, 2104, 2448) I. procurrens (Nees ex Trin.) Swallen, Phytologia 11: 149. 1964. P. procurrens Nees ex Trin., Gram. Panic. 183. 1826. Echinolaena pro- currens (Nees ex Trin.) Kunth, Revis. Gramin. 1: 54. 1829 Common, valley-side campo (upslope to down- slope), seasonally humid to inundated savanna and savanna marsh, rarely in cerrado; unpalatable (0); flowering September (January) to July; 2n = 20. Distribution: Venezuela, Brazil, Paraguay, and Ar- gentina; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (591, 612, 631, 712, 776, 1424, 1642, 2285, 2590; Thomas 5652, 5671 NY, LPB) Imperata Cyrillo Gabel, M. L. 1982. A systematic study of the genus Imperata (Gramineae: Andropogoneae). Ph.D. Dissertation. Iowa State Univ., Ames, Iowa. KEY TO SPECIES la. Inflorescence 7-12 cm long; base of blades as wide as the apex of the sheath, not narrow to a thickened, ea midrib ........... ds Inflorescence 15-30 cm long; base of blades narrower than the apex xof the sheath, reduced to a thickened, indurate midrib. 2a. Lower panicle branches 3-10 cm long; the narrowed base of the blade 5-10 cm long pov Shep MON contracta 2b. Lower panicle branches 1-2 cm longi the narrowed base of the blade about 1 cm = I. brasiliensis Trin., Мет. Acad. Imp. Sci. St.- Pétersbourg, Sér. 6, Sci. Math. 2: 331. 1832. Saccharum sape St. Hil., Voy. Distr. Diam. 1: 386. 1833. I. brasiliensis var. mexicana Rupr., Mém. Acad. Roy. Sci. Bruxelles Bel. 9: 245. 1842. I. sape (St. Hil.) Anderss., Ofvers. Fórh. Kóngl. Svenska Vetensk.-Akad. 12: 160. 1855. Г. arundinacea var. ameri- cana Anderss., Ofvers. Fórh. Kóngl. Svenska Vetensk.-Akad. 12: 160. 1855. / caudata Chapman, Fl. South. U.S., 2nd edition. 1883. Abundant, forming colonies in densely wooded cerrado and on cerrado / forest margins, occasion- ally in campo cerrado; a common weed of pastures and roadsides, becoming a serious pest in banana and coffee plantations. Unpalatable (0); flowering is dependent upon fire; 2n — 20; local name: sujo. Distribution: southern United States to Argentina; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (1171, 2192, 2217) I. contracta (H.B.K.) Hitchc., Ann. Rep. Missouri Bot. Garden 4: 146. 1893. Saccharum con- tractum H.B.K., Nov. Gen. Sp. 1: 182. 1816. S. dubium H.B.K., Nov. Gen. Sp. 1: 183. 1816. S. caudata G. Meyer, Prim. Fl. Esseq. 68. 1818. I. caudata (С. Meyer) Trin., Mem. Acad. Imp. Sci. St.-Pétersbourg, Ser. 6, Sci. Math. 2: 331. 1832. Anatherum berteria- num Sprengel ex Schultes, Mant. 2: 443. 1824. I. exaltata var. caudata Hack. in DC., Monogr. Phan. 6: 99. 1889. /. longifolium Pilger, Bot. Jahrb. Syst. 30: 136. 1901. Not yet collected in Chiquitania but a zonal dominant of shallowly inundated savannas in the Volume 77, Number 1 1990 Killeen Grasses of Chiquitanía, Santa Cruz, Bolivia 163 Beni; probably unpalatable; flowering in July. Dis- tribution: Mexico to Argentina. (2607) I. tenuis Hack. in A. DC., Monogr. Phan. 6: 689. 1889. I. exaltata var. angustifolius Hack. in A. DC., Monogr. Phan. 6: 99. 1889. Common, valley-side campo (midslope to down- slope), seasonally humid and inundated savanna and savanna marsh; unpalatable (0); flowering De- cember to (February) June. Distribution: central Brazil and northeastern Argentina; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (655, 719, 768, 818, 961, 1129, 1580, 1808, 1872, 2117, 2482, 2605) Lasiacis (Griseb.) Hitchc. Davidse, G. 1978. A systematic study of the genus Lasiacis (Gramineae: Paniceae). Ann. Mis- souri Bot. Gard. 65: 1133-1254. KEY TO SPECIES ‚ ligulata Ib. Panicle branches ascending or dpi У sheaths L. papillose-hispid sorghoidea la. Panicle branches reflexed; sheaths glabrous L. sorghoidea (Desv.) Hitchc., Contr. U.S. Natl. Herb. 18: 338. 1917. Panicum lanatum Sw., Prodr. 24. 1788, non Rottb., 1778. P. sor- ghoidea Desv. in Ham., Prodr. Pl. Ind. Occ. 10. 1825. L. guaraniticum (Speg.) L. Parodi, Notas Mus. La Plata, Bot. 8: 95. 1943; an extensive synonymy is given in Davidse (1978). Common, seasonal forest and forest/savanna margins, forming robust colonies along new roads in forest; very palatable (4), an important source of forage in the dry season when most savanna grasses are senescent; flowering January to July; n = 36; local name: taquarilla. Distribution: Mexico to Argentina; Bolivia: Andean Piedmont of Santa Cruz, Cochabamba, and the Yungas. (553, 675, 883, 906, 1888, 2014, 2099, 2332) L. ligulata Hitchc. & Chase, Contr. U.S. Natl. Herb. 18: 337. 1917 Occasional, forest openings and margins. Dis- tribution: West Indies, northern South America and Brazil; Bolivia: the Yungas. (965, 1000) Leersia Sw. Pyrah, G. L. 1969. Taxonomic and distributional studies in Leersia (Gramineae). Iowa State J. Sci. 44: 215-270 L. hexandra Sw., Prodr. 21. 1788. L. mexicana H.B.K., Nov. Gen. Sp. 1: 195. 1816. L. con- tracta Nees, Agrost. Bras. 516. 1829. L. glaberrima Trin., Mém. Acad. Imp. Sci. St.- Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 5: 172. 1840. Oryza hexandra (Sw.) Doell in Mart., Fl. Bras. 2(2): 10. 1877. Common, forming colonies in shallow lakes, sea- sonal ponds, and seasonally inundated savannas; very palatable (4), savanna wetlands with a high proportion of this species are prized by ranchers; flowering in March, April, and irregularly through- out the year; populations are largely infertile (seed set <5%); 2n = 48; local name: arrocilla. Dis- tribution: throughout the tropics. (701, 762, 815, 943, 1635, 1777, 2023, 2416) Leptochloa P. Beauv. KEY TO SPECIES la. Lemma papillose, strongly laterally ~re glabrous or sparsely pilose on bac e margins “уй on upper half, awned or aviles glumes lanceolate; spikes distinctly 2-ra mucronate, acute, or irregularly lobed; glumes ovate or oblong; spikes obscurely 2-ranked or secund. 2a. Apex of lemma irregularly toothed or lobed, the midnerve forming a mucro; spikelets obscurely 2-ranked; racemes 10-20 per panicle . uninervia . Apex of lemma entire, acute or mucronate but not irregularly toothed or lobed; spike- lets distinctly secund; racemes 20-40 per le . scabra N c L. uninervia (C. Presl) Hitchc. & Chase, Contr. U.S. Natl. Herb. 18: 383. 1917. Megastach- ya uninervia C. Presl, Rel. Haenk. 1: 283. 1830. Diplachne uninervia (C. Presl) Parodi, Univ. Nac. Buenos Aires Revista Centr. Estud. 18: 147. 1925. D. tarapacana Phil., Anales Mus. Nac. Chile Bot. 8: 88. 1891. D. cari- nata Hack., Bol. Acad. Ci. (Córdoba) 16: 253. 1900. An extremely variable taxon similar to and pos- sibly not distinct from L. fascicularis (Lam.) A. Gray. Occasional, as a weed in seasonal ponds and roadside ditches. Distribution: United States to Ar- gentina; Bolivia: Andean Piedmont of Santa Cruz, Beni, and Cochabamba. (825, 1712) L. scabra Nees, Agrost. Bras. 435. 1829. Occasional, roadside ditches in standing water; flowering November to February. Distribution: southern United States to Brazil. (1425, 2300) 164 Annals of th Missouri Botanical Garden L. virgata (L.) P. Beauv., Ess. Agrostogr. 71, 1812. Cynosurus virgatus L., Syst. Nat. ed. 10. 2: 876. 1759. Chloris digitaria H.B.K., Nov. Gen. Sp. 1: 168. 1816. £. dig- itaria (H.B.K.) Nees, Agrost. Bras. 433. 1829. A d. (H.B.K.) Nees, Syll. Pl. Ratisb. 1: 824. L. domingensis (Jacq.) Trin., Fund. с. 133. 1820. L. barbata Desv., Opusc. Sci. Phys. Nat. 104. 1831. L. mutica Steudel, Syn. Pl. Glumac. 1: 208. 1854. L. villosa Ekman, Ark. Bot. 10(7): 31, p. 3, 6. 1911. Common, roadside weed in forest soils; flowering throughout the year. Distribution: southern United States to northern Argentina; Bolivia: Andean Pied- mont of Santa Cruz and the Yungas. (602, 607, 639, 827, 828, 840, 973, 1261, 1312, 1567A, 1681, 1708, 2298) Leptocoryphium Nees L. lanatum (H.B.K.) Nees, Agrost. Bras. 84. 1829. Paspalum lanatum H.B.K., Nov. Gen. Sp. 1: 94, t. 29. 1816. L. molle Nees, Agrost. Bras. 85. 1829. Anthaenantia lanata (H.B.K.) Benth., J. Linn. Soc. Bot. 19: 39. 1881. Milium juncoides Speg., An. Soc. Cient. Argent. 16: 105. 1883. Occasional in cerrado, zonally abundant in sandy soils in valley-side campos (upslope); flowering is dependent upon fire; 2n = 20. Distribution: Central America to Argentina; Bolivia: Andean Piedmont of Santa Cruz, Beni, and the Yungas. (781, 1111, 1163, 1368, 1430, 1455, 1660, 2190, 2768, 2784; Thomas 5630 NY, LPB) Loudetia Hochst. ex A. Braun L. flammida (Trin.) С. E. Hubb., Bull. Misc. Inform. 1936: 361. 1936. Arundinella flam- mida Trin., Sp. Gram. 3: 267. 1831. Tri- chopteryx flammida (Trin.) Benth., J. Linn. Soc., Bot. 19: 59. 1882. Common, in savanna marsh, seasonally humid, and seasonally inundated savanna; characteristic of valley-side campos where it displays a bimodal distribution at the top and bottom of the soil-mois- ture gradient (but absent midslope); moderately palatable (2); flowering January (February) to May; 2n = 20. Distribution: Brazil and Paraguay; Bo- livia: Andean Piedmont of Santa Cruz and the Yungas. (718, 757, 905, 1747, 1861, 2324) Loudetiopsis Conert L. chrysothrix (Nees) Conert, Bot. Jahrb. Syst. 77: 285. 1957. Tristachya chrysothrix Nees, Agrost. Bras. 460. 1829 Eiten (1972) reported this species to be char- acteristic of cerrado communities in central Brazil, and it is common in the dry savannas of the Andean foothills near Samaipata, Santa Cruz (1,500 m). However, it is rare in Chiquitania and restricted to seasonally humid savannas with sandy soils; flow- ering in January and February. Distribution: Brazil and Paraguay. (791, 1666, 1738, 1834, 2488) Luziola Juss. ex Gmel. Swallen, J. R. 1965. The grass genus Luziola. Ann. Missouri Bot. Gard. 52: 472-475. KEY TO SPECIES la. Achene striate; e pe Lagen when immature; pistillate panicle racemose branches, ds ORE оа. to ascending - 2. bahiensis . Achene smooth; pistillate glumes not pleated; pistillate par nicle open, freely branched, the ped- | . peruvianum — c icel spreading .... L. bahiensis (Steudel) Hitchc., Contr. U.S. Natl. Herb. 12: 234. 1909. Caryochloa bahiensis Steudel, Syn. Pl. Glumac. 5. 1854. L. lon- givalvula Doell in Mart., Fl. Bras. 2(2): 17. 1871. L. striata Bal. & Poit., Bull. Soc. Hist. Nat. Toulouse 12: 231, t. 4, f. 2. 1878. L. pusilla S. Moore, Trans. Linn. Soc. Bot. II. 4: 507, t. 37, f. 1-8. 1895. L. contracta Hack., Oesterr. Bot. Z. 52: 8. 1902 Common, in savanna marsh growing underneath the robust dominant Saccharum trinii; flowering November (December) to July, possibly stimulated by fire. Distribution: southern United States to Ar- gentina. (622, 1026, 1422, 1647, 1780, 2250) L. peruvianum Juss. ex Gmel., Syst. Nat. 2: 637. 1791. L. mexicana H.B.K., Nov. Gen. Sp. 1: 199. 1816. Milium natans Spreng., Syst. Veg. 1: 250. 1825. L. brasiliana Moric., Pl. Nouv. Amer. 94, t. 60, 1840. L. leiocarpa Lindman, Kongl. Svenska Vetensk. Akad. Handl. 34(6): 12. 1900. L. doelliana Pro- doehl, Bot. Arch. 1: 240. 1922. ommon, in seasonal ponds and occasionally in seasonally inundated savanna; palatable (4); flow- ering October to (January) May. Distribution: southern United States to Argentina; Bolivia: Tari- ja, Andean Piedmont of Santa Cruz, and the Beni. (850, 1244, 1648, 2085, 2283, 2288) Melinis P. Beauv. M. minutiflora P. Beauv., Ess. Agrostogr. 54. 1812. Panicum melinis var. inerme Doell in Mart., Fl. Bras. 2(2): 242. 1877. Volume 77, Number 1 1990 Killeen 165 Grasses of Chiquitanía, Santa Cruz, Bolivia Although occurring as a roadside weed in humid montane regions of Bolivia, in Chiquitania it is a naturalized species characteristic of bromeliad thickets on granitic outcrops; its highly aromatic and glandular foliage makes it very palatable to cattle (4); flowering in May; local name: capim ceroso. Distribution: throughout the tropics. (933, 2102) Mesosetum Steudel Filgueiras, T. de S. 1986. O género Mesosetum Steudel (Gramineae: Paniceae). Thesis. Uni- versidad Estadual de Campinas, S.P., Brasil. Swallen, J. R. 1937. The grass genus Meso- setum. Brittonia 2: 363-392 M. cayennense Steudel, Syn. Pl. Glumac. 118. 1854. Occasional, superficial soils on granitic outcrops, rarely in gravel soils of cerrado (campo limpo) or campo rupestre; flowering April to September. Dis- tribution: West Indies to Brazil. (908, 1217, 1491, 1981) Microchloa R. Brown M. indica (L.f.) P. Beauv., Ess. Agrostogr. 115, t. 20, f. 8. 1812. Nardus indica L. f., Suppl. Pl. 105. 1781. Common, superficial soils over granitic outcrops; rarely on lateritic outcrops in cerrado; flowering November to May. Distribution: throughout the tropics; Bolivia: Cochabamba and the Yungas. (799, 1443, 1484, 1547, 1782, 1826, 2091) Olyra L. Zuloaga, F. O. & T. R. Soderstrom. 1989. A revision of the genus Olyra and the new seg- regate genus Parodiolyra (Poaceae: Bambu- soideae: Olyreae). Smithson. Contr. Bot. 69: 1-79 KEY TO SPECIES == p . Pistillate floret к lower glume long- anicle branches lacking pistil- О. ciliatifolia Pistillate floret glabrous; lower glume short-aris- tate to acuminate; all panicle branches with cup spikelets. anicle more or less branched, the branch- alternate; pistillate spikelets 1-2 = моз. branch, the florets ovate and smooth O. latifolia Panicle of 4-8 verticillate racemes inserted at 1-2 nodes; pistillate spikelets and florets = N = various. 3a. Pistillate spikelets (2)3-4 per raceme, the cai narrowly elliptic and dis- tinctly pitted . fasciculata А Pistillate ре 1-2 рег raceme, the florets ovate and smooth О. caudata w = О. caudata Trin., Linnaea 10: 292. 1836. O. dimidiata Hochstetter ex Steudel, Syn. Pl. Glum. 1: 36. 1853. O. рене) Hack., Oesterr. Bot. Z. 51: 461. 190 Rare, forest at Parque Nacional “Noel Kempff Mercado.” (Thomas 5734 NY) O. ciliatifolia Raddi, Agrost. Bras. 19. 1823. O, cuneatifolia Desv., Opusc. Sci. Phys. Nat. 106. 1831 Occasional, cerrado/forest margins and tran- sitional scrub; palatable (2); flowering December to April; 2n = 44; local name: taquarilla. Distri- bution: northern South America to Brazil and Par- aguay; Bolivia: the Beni. (649, 728, 869, 1614, 1889, 1920, 2386) O. fasciculata Trin., Mém. Acad. Imp. Sci. St.- Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 3: 113. 1835. O. heliconia Lind- man, RB Svenska Vetensk. Acad. Handl. 34(6): 11. 1900 Uncommon, in deep shade of seasonal forest, rarely as a weed along roads in forest soils; flow- ering October to December; local name: taqua- rilla. Distribution: Peru and Brazil; Bolivia: Andean Piedmont of Santa Cruz and the Yungas. (671, 1377, 2385, 2828) O. latifolia L., Syst. Nat. ed. 10. 2: 1261. 1759. O. arundinacea H.B.K., Nov. Gen. Sp. 1: 197. 1816. O. cordifolia H.B.K., Nov. Gen. Sp. 1: 198. 1816. O. pubescens Raddi, Agrost. Bras. 18. 1823. O. scabra Nees, Agrost. Bras. 306. 1829. O. brasiliensis Desv., Opusc. 106. 1831. O. media Desv., Opsuc. 106. 1831. Occasional, seasonal forest and forest openings, locally abundant in gallery forest and along streams in seasonal forest; flowering November to July; local name: taquarilla. Distribution: Mexico to Ar- gentina; Bolivia: Beni and the Yungas. (554, 677, 931, 1047, 1379, 1496, 1774, 1969, 2013) Oplismenus P. Beauv. Scholz, U. 1981. Monographie der Gattung Oplis- menus. Phanerogamarum Monographiae To- mus XII. J. Cramer, Vaduz. Davey, J. C. & W. D. Clayton. 1978. Some multiple discrim- 166 Annals of the Missouri Botanical Garden inant function studies on Oplismenus (Gra- mineae). Kew Bull. 33(1): 147-157. KEY TO SPECIES la. Awn antrorsely scabrous; racemes hispid .......... O. burmannii O. hirtellus lb. Awn smooth; racemes not hispid ............ O. burmanni (Retz.) Р. Beauv., Ess. Agrostogr. 54. 1812. Panicum burmanni Retz., Observ. Bot. 3: 10. 1783 Rare, granitic outcrop; flowering in May. Dis- tribution: throughout the tropics. (939) O. е (L.) Р. Beauv., Ess. Agrostogr. 54. 12. Panicum hirtellum L., Syst. Nat. e "i 870. 1759. P. setarium Lam., Tabl. Еп- сус]. 1: 170. 1791. O. setarius (Lam.) Roe- mer & Schultes, Syst. Уер. 2: 481. 1817. O. brasiliensis Raddi, Agrost. Bras. 40. 1823. O. velutinum (G. Meyer) Schultes, Mant. 2. 271. 1824. Sometimes treated as two species (O. setarius), but the two forms intergrade and share the same habitat. Common in forest, especially where dis- turbed by cattle; palatable (3); flowering January to (April) July. Distribution: United States to Ar- gentina; Bolivia: Andean Piedmont of Santa Cruz, Chapare, the Yungas, and the Beni. (693, 750, 1030, 1881, 1886, 1967, 2002, 2009, 2030, 2074 Orthoclada P. Beauv. O. laxa (L. Rich.) P. Beauv., Ess. Agrostogr. 70, 149, 168. 1812. Aira laxa L. Rich., Actes Soc. Hist. Nat. Paris 1: 106. 1792. Not yet collected in Chiquitania but found in forest islands in the Beni. Distribution: Mexico to Brazil. Oryza L. Tateoka, Т. 1962. Taxonomic studies of Oryza 1. O. latifolia complex. Bot. Mag. Tokyo 75(893): 418-427. Tateoka, Т. 1962. Taxo- nomic studies of Oryza II. Several species complexes. Bot. Mag. Tokyo 75(894): 455- 461. Tateoka, T. 1963. Taxonomic studies of Oryza III. Key to species and their enu- meration. Bot. Mag. Tokyo 76(899): 165- 173. KEY TO SPECIES la. Glumelike bracts about as long as the spikelet . grandiglumis lb. Glumelike bracts less than !4 the length of the spikelet. 2a. Spikelets na green at maturity (rarely reddish), 5-6(7) mm long, erect on ei pedicels; lemma muticate or aristate, th awns up to 2 cm long cc О. latifolia 2b. Spikelets reddish or tan at maturity, 7-10 mm long, obliquely inserted on pedicels. Awn 4-10 cm long; spikelets oblong, 9-10 mm long, spikelets ЛҮҮ з ). pulipogen 3b. Awn absent or up to 2 cm nt spike lets elliptic to oblong, 7-9 mm long, spikelets persistent O. sativa 2 c O. grandiglumis (Doell) Prodoehl, Bot. Arch. 1: 233. 1922. O. sativa var. grandiglumis Doell in Mart., Fl. Bras. 2(2): 8. 1871. O. latifolia var. grandiglumis (Doell) A. Chev., Rev. Int. Bot. Appl. Agric. Trop. 12: 1207. 1932. Occasional, along streams in forest, especially where light intensity is high; flowering throughout the year; 2n = 48. Distribution: northern South America, Peru, and Brazil. (922, 1052, 1677, 2232) O. latifolia Desv., J. Bot. 1: 77. 1813. O. sativa L. var. latifolia (Desv.) Doell in Mart., Fl. Bras. 2(2): 7. 1871. O. platyphylla J. A. Schultes & J. H. Schultes, Syst. Veg. 7: 1364. 1830. Distribu- tion: Mexico to northern Argentina; Bolivia: An- dean Piedmont of Santa Cruz and the Beni. (691) Rare, roadside ditches in forest soils. O. rufipogon Griffiths, Ic. Pl. Asiat. 3: 5, pl. 144, f. 2. 1851. O. glumipatula Steudel, Syn. Pl. Glumac. 1: 3. 1854. O. perennis Moench sensu Hitchcock, Grasses W. Ind. 145. 1936. O. paraguayensis Wedd. ex Fourn., Compt. Rend. Cong. Int. Bot. & Hort. Paris 1878: 233. 1880. O. sativa var. paraguayensis L. Parodi, Physis 11: 244. 1933. Commonly collected in Amazonia along river banks, where it is a floating aquatic of indefinite duration with caulescent foliage; however, in Chi- quitanía it occurs as a caespitose perennial with basal foliage. Common, seasonal ponds of pantanal complex, rarely in seasonally inundated savanna and small pools on granitic outcrops; palatable (4); flowering April (May) to August. Distribution: throughout the tropics. (941, 1894, 2086) O. sativa L., Sp. Pl. 333. 1753. Cultivated, usually interplanted with maize; a staple of the local diet. Volume 77, Number 1 Killeen 167 1990 Grasses of Chiquitanía, Santa Cruz, Bolivia Otachyrium Nees (2-3); flowering is stimulated by (but not dependent Eoduliby, Т. 8 T.H. Soderetrom. 1984, Revision upon) fire, August to May. Distribution: Colombia, O. versicolor (Doell Henrard, Blumea 4: 511. 1941. Panicum truncatum Nees, Agrost. Bras. Plants variable between populations for spikelet size and morphology. Locally abundant in shallowly inundated or seasonally humid savanna, rarely in low forest scrub (San José de Chiquitos); palatable of the genus Otachyrium (Poaceae: Panicoi- Venezuela, the Guianas, Brazil, and Paraguay; Bo- дене}, Simiihson. Conie: Bot. ST: 1924. mo Andean Piedmont of Santa Cruz and the Beni. 864, 1197, 1522, 1623, 1724, 2769) Panicum L. 215. 1829, non Trin., 1826. P. versicolor Zuloaga, К. O. 1986. Systematics of new world Doell in Mart., Fl. Bras. 2(2): 254. 1877. species of Panicum (Poaceae: Paniceae). Pp. 287-309 in T. R. Soderstrom, K. W. Hilu, C. S. Campbell & M. E. Barkworth (editors), Grass Systematics and Evolution. Smithsonian Institution, Washington, D.C. KEY TO SPECIES 1а. омы 5-7 mm long; panicles open, я at the base of the peduncle at maturity; the base of e upper florets with paired tufts of thick hai , lb. Spikelets 1-3 mm long; panicles open, iei elliptic, spicate, or racemose; the base of the upper florets olyroides lacking tufts of thick hairs 2a. Lowermost panicle branches verticillate. ко = За. Upper floret rugose, only the lowermost branches verticillate; caespitose perennial .......... P. maximum 3b. Upper floret deba) he panicle branches verticillate at all the nodes; culms decumbent an ting at the n . mertensii Panicle branches poda or subopposite, not verticillate, if several inserted at a single node, then borne on one side and of unequal size. 4a. Glumes and lower lemma vesciculate-pubescent; upper lemma к гир за. Primary panicle branches bearing spikelets to the a bearing l- 3 pem S panicle branches; pedicels spreading to ascending, 1-5 mm lon millegrana 9b. Primary panicle branches lacking spikelets on lower vil %, unbranched; pedicels uniformly short, 0.5-2 mm long and appressed to branches F sellowii 4b. Glumes and lower lemma glabrous or puberulent but not vesciculate-pubescent; upper lemma smoot 6a. Paid 2 from each of the upper nodes, A npe, the bases included in the sheath; a usually coarsely hispid; decumbent an a. Spikelets about 3 mm long, the glumes pe De lemma acuminate, the floret ovate to elliptic rudgei Tb. reed oh 2 mm long, the glumes and lower lemma acute, the floret elliptic to . cayennense Ób. Panicle 1 pem sen node, “а inflorescences (if present) on short leafy branches; sheaths Pc glabrous to pubescent but not coarsely hispid; annuals or perennials. a. Low e %-% n sin of spikelet, the apex blunt or rounded; foliage cauline and uniformly distributed on the culms; plants stoloniferous or decumbent and rooting at the nodes 9a. Spikelets 1.8-2.2 m ong. sed ee velutinous (at least abaxially), lax; panicle of 2-5 racemose br s, pyram . pantrichum 10b. Leaf Viet spi strictly ascending to erect; panicle freely branched, Р. s chwackeanum globos 9b. Skelet 1.2-1.5 mm lon la. Panicles about 7 x 7 cm; leaf blades > 5 cm long „u P. cyanescens llb. Panicles about З х 3 cm; leaf blades 2-4 cm long ooo... parviflorum 8b. Lower glume up to A the > length of the spikelet, the apes acute (somewhat blun t in P. rhizomatous, or stoloniferous. P. pulchellum 12a. Back of lower lemma with paired, glandular pits 12b. Back of lower lemma not glandular-pitted. 13a. Panicles spicate, the spikelets 6-25, borne on short, erect, appressed branch- es; plants caespitose, sedgelike, bunch grasses 14а. ed s 10-25, 1.5-1.8 mm long; rachilla internode elongate; foliage ually glabrous . stenodes 14b. iw. 5-10, 2.2- d i mm long; rachilla internode not elongate; foliage usually pubesce P. caricoides 13b. Panicles жы spicate; La numerous or if few, then the branches spread ing; plant habit various 168 Annals of the Missouri Botanical Garden 15a. Spikelets 1-1.5 mm long. 16a. Lower glume reduced to a membranous, nerveless scale l6b. Lower glume well developed, 1-5-nerve P. trichanthum 18a. Lower ee gibbous, loosely clasping the spikelet; panicles narrowly elliptic, the нна. ending; nodes glab racemose, the spike 17b. Lower lemma with a well-developed 19a. Spikelets 1-1.2 mm long; leaf emose. sparsely bran 20a. vate, rous 18b. Lower ina not шы tightly clasping the spikelet; panicles pyramidal, the es elets lanceolate, secu bearded P. ched, rac Spikelets uniformly distributed on a pilose rachis 20b. Spikelets irregularly distributed on a glabrous rachis. ridum nd; nodes bearded aaa polygonatum alea. blades ovate; panicles open, diffuse P. 19b. Spikelets about 1.5 mm long; leaf blades linear to lanceolate; primary panicle branches trichoides P. pilosum в. s stou e 5 mm in diameter; leaf blades cordate and clasping the culm a i Culms ns 21 mm in diameter; leaf blades not cordate-clasping at the base . Р. . hylaeicum laxum — л c . erus 2-3 mm lon bright orange when ng. 2a. Lower glume 44-14 the length of spikelet, 1-nerved, the apex broadly acute to truncate; anther re P. di tom anm 22b. c colored, usually yellow or purple. 23a. P a panicle of secund raceme Inflorescence an open or narrowly elliptic pa anicle. 23b. sn Lower glume about !4 the length of the spikelet, acute to acuminate, 3-5-nerved; anthers variously P. stoloniferum eaf blades lanceolate, acuminate; plants of shaded habitats... Р. haenkeanum 24b. Leaf blades linear; heliophytes. 25а. U pper florets dark chestnut brown, rotated 90° in the spikelet; foliage basal. Bracts subtending the upper floret 5: l 2 glumes, 2 lower lemmas, and a quadriglume рата f 26b. Bracts subtending the upper floret 4: 2 glumes, 1 lower lemma, and a palea P. peladoense 25b. Upper qasi stramineous to dull brown, not rotated 90°; foliage cauline 2 Plants robust la ы de: 27b. Plants 30-5 P. caricoides Nees ex Trin., Gram. Panic. 149. 1 Р. коө a Proc. Amer. Acad. Arts. : 497 ‚ Р. junciforme Steudel, Syn. ES тан q B 1854, nomen nudum Easily confused with P. stenodes, which grows in more humid or seasonally inundated habitats; both species have been observed growing in adja- cent savanna E without evidence of introgres- sion (2824 vs. 2825 and 2263 vs. 2268, 2269). Common, паат humid ѕауаппа апа valley- side campos (upslope); flowering September to Jan- uary, flowering is stimulated by (but not dependent upon) fire. Distribution: Central America, the West Indies to Brazil. (1373, 2263, 2824) Р. cayennense Lam., Tabl. Encycl. 1: 173. 1791. Р. pedunculare Willd. ex Steudel, Nomencl. Bot. 2(10): 260. 1841. Rare, cerrado and superficial sandy soil over granitic outcrops; flowering November to January. iduous: pedi cm tall, caespitose or decumbent annuals; та not brittle; leaves not ae eae pedicels divergent Р m tall, rhizomatous perennials; culms brite leaf icels ascending richolaenoides hirticaule Distribution: Mexico to Argentina; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (797, 1486) P. cyanescens Nees in Trin., Gram. Panic. 202. 1826. P. carannense Mez, Notizbl. Bot. Gart. Berlin-Dahlem 7: 73. 1917 Uncommon, seasonally humid savannas; flow- ering January to July. Distribution: northern South America to Argentina; Bolivia: Andean Piedmont of Santa Cruz. (1096, 1658; Bruderreck 122 ISC, LPB) P. dichotomiflorum Michx., Fl. Bor. Amer. 1: 48. 1803. P. chloroticum Nees in Trin., Gram. Panic. 236. 1826. P. proliferum var. ri- с Doell in Mart. Fl. Bras. 2(2): 200. 877. P. proliferum var. xanthochlorum io ex Bertoni, Anales Ci. Parag., Ser. 2. 150. 1918; see Zuloaga (1986) for extensive synonymy. Common in seasonal ponds on granitic outcrops and shallow roadside ditches, occasional in season- Volume 77, Number 1 1990 Killeen 169 Grasses of Chiquitanía, Santa Cruz, Bolivia ally inundated savanna, and rarely as a weed along roadsides in dry gravel soils; flowering Novem- ber to February. Distribution: United States to Argentina; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (648, 756, 1504, 1626, 1735) P. haenkeanum C. Presl, Rel. Haenk. 1: 304. 1830. P. costaricensis Hack., Oesterr. Bot. Z. 51: 428. 1901. Central American specimens are much more robust and have larger panicles than specimens collected in Bolivia and Brazil; however, spikelet morphology and the shape of the leaf blade are characteristic. Rare, in diffuse light of Orbignya forest; flowering in August. Distribution: Central America and Brazil. (1094) P. hirticaule C. Presl, Rel. Haenk. 1: 308. 1830. Р. flabellatum Fourn., Bull. Soc. Bot. France II. 27: 293. 1880, non Steudel, 1854. P. caatingense Renv., Kew Bull. 37: 325. 1982. Rare, as a weed in pastures and roadsides in humid soils near San José de Chiquitos; flowering in February. Distribution: United States to Argen- tina; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (1702, 1711) P. hylaeicum Mez, Notizbl. Bot. Gart. Berlin- Dahlem 7: 75. 1917. P. minutiflorum Doell in Mart., Fl. Bras. 2(2): 253. 1877, non Rasp., 1825. P. laxum var. pubescens Doell in Mart., Fl. Bras. 2(2): 213. 1877, in part. P. laxum var. amplissimum Hack., Feddes Repert. Spec. Nov. Regni Veg. 6: 343. 1909. P. doelli Mez, Bot. Jahrb. Syst. 56(125): 27. 1934. P. boliviense Hack. sensu Smith, Wasshausen & Klein, Fl. Ilustr. Catarinense (Gramineas). 671. 82 Rare, roadside ditch in forest soils; flowering in November. Distribution: northern South America to Paraguay. (1426) P. laxum Sw., Podr. 23. 1788. P. agrostidiforme Lam., Tabl. Encycl. 1: 172. 1791; see Zu- loaga (1987) for extensive synonymy. Plants are variable within and between popu- lations for stature and inflorescence morphology. This grass is abundantly represented as a constit- uent of seasonally inundated savannas, seasonal ponds, and as a weed in roadside ditches; it is rarely found in savanna marsh and valley-side campos. An important source of forage, especially in abused savanna wetlands, as this species colonizes openings in the herbaceous canopy caused by cattle tram- pling; palatable (4); flowering throughout the year but more abundantly during the rainy season. Dis- tribution: Mexico and the West Indies to Argentina; Bolivia: Andean Piedmont of Santa Cruz, Chapare, Beni and the Yungas. (609, 628, 632, 641, 647, 657, 727, 755, 774, 775, 813, 851, 1063, 1124, 1554, 1575, 1625, 1686, 1687, 1716, 1717, 2286, 2294, 2316, 2587) P. maximum Jacq., Coll. Bot. 1: 76. 1786. P. jumentorum Pers., Syn. Pl. 1: 83. 1805. P. praticola Salzm. ex Doell in Mart., Fl. Bras. 2(2): 203. 1877. A productive, high quality, cultivated, forage grass adapted to forest soils. The traditional variety (hierba guinea) is a robust plant reaching 2.5 m in height, newly introduced dwarf cultivars (panico verde, panico petrie, etc.) can be planted in mixed swards with the legume Neonotonia wightii (W.&A.) Verdc. and offer improved pasture man- agement capabilities (Patterson, 1984). Most plants flower March to June, but the dwarf cultivars also flower irregularly throughout the year. (1315) P. mertensii Roth ex Roemer & Schultes, Syst. Veg. 2: 458. 1817. P. altissimum G. Meyer, Prim. Fl. Esseq. 63. 1818. P. elatior Kunth, Révis. Gramin. 1: 38. 1829, non L., 1781. P. megiston Schultes, Mant. 2: 248. 1824. P. equisetum Nees ex Doell in Mart., Fl. Bras. 2(2): 206. 1877, as synonym. Occasional, an emergent aquatic forming colo- nies in roadside ditches and along disturbed stream- sides under bridges; flowering in February. Distri- bution: Mexico and the West Indies to Argentina; Bolivia: the Beni. (1745) P. millegrana Poir. in Lam., Encycl. Suppl. 4: 278. 1816. P. rugulosum Trin., Gram. Panic. 195. 1826. P. lasianthum Trin., Sp. Gram. 3, pl. 245. 1830. P. dispersum Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 3: 282. 1835. P. subglobosum Hack., Bull. Herb. Boissier, ser. 2, 4(3): 274. 1904. Similar to and possibly not distinct from P. sel- lowii. Common, seasonal forest, forest/savanna margins, and transitional scrub, forming large col- onies in forest openings; palatable (4); flowering November to (January, February) June. Distri- bution: Mexico to Argentina. (835, 966, 1489, 1675, 2320) 170 Annals of the Missouri Botanical Garden Р. olyroides H.B.K., Nov. Gen. Sp. 1: 102. P. densiflorum Willd. ex Spreng., Syst. Veg. 1816. P. probiscidium Trin., Gram. Panic. 184. 1826. P. = Steudel, Syn. Pl. Glumac. 1: 77. 1854. Common, cerrado and in the well-drained, sandy soils of the Andean Piedmont; somewhat palatable (2), coarse but plants lacking in overgrazed savan- na; flowering January (March) to May; the panicle forms a tumbleweed at maturity; 2n = 36. Dis- tribution: Venezuela to Argentina; Bolivia: Andean Piedmont of Santa Cruz. (868, 1257, 1559, 1646, 1769, 1847, 2398; Thomas 5567 NY, LPB) P. — Hack., Verh. Zool.-Bot. Ges. Wien 195: 72. 1915. P. protractum Mez, Notizbl. Bot. e Berlin-Dahlem 7: 77. 1917. P. warmingii Mez, Bot. Jahrb. Syst. 56(125): 1. 1921 Locally abundant in low forest scrub on Serrania San Lorenzo, spreading upslope into margins of campo rupestre; rarely in Orbignya forests near Santa Rosa de la Roca; flowering November to April. Distribution: Central America to Argentina; Bolivia: the Yungas. (1391, 1972) P. parviflorum Lam., Tabl. Encycl. 1: 173. 1791. P. une Sprengel, Syst. Veg. 1: 321. 182; Common, in seasonally inundated savanna, sea- sonal ponds, and lake margins; flowering November to (January) May. Distribution: throughout the tropics; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (735, 801, 1251, 1254, 1523, 1779, 2278) P. peladoense Henrard, Blumea 4: 504. 1941. Р. bergii var. leiophyllum Hack. & Lindman, Kongl. Svenska Vetensk. Acad. Handl. 34(6): 10, pl. 4B. 1900. Р. campestre Nees, Agrost. Bras. 197. 1829, non Nees in Trin., 1826. Р. cayennense var. campestris (Nees) Pilger, Bot. Jahrb. Syst. 30: 132. 1901. Rare, disturbed ground; more common on mar- gins of savanna/scrub of the Andean Piedmont of Santa Cruz; flowering October to January. Distri- bution: Brazil to Argentina. (823, 1236) P. pilosum Sw., Prodr. Veg. Ind. Occ. 22. 1788. P. distichum Lam., Encycl. 4: 731. 1798. Setaria disticha (Lam.) H.B.K., Nov. Gen. Sp. 1: 112. 1816. P. pilisparsum С. Meyer, Prim. Fl. Esseq. 57. 1818. P. trichophorum Schrader ex Schultes, Mant. 2: 247. 1824. 1: 320. 1825. P. distans Willd. ex Spreng., Syst. Veg. 1: 305. 1825. Uncommon, gallery or swamp forest; common as a weed along roads in forest and savanna soils; palatable (5); flowering throughout the year. Dis- tribution: Mexico and the West Indies to Argentina; Bolivia: Beni and the Yungas. (605, 730, 748, 792, 1054, 1515, 1689, 1741, 1880, 1961, 2319) P. polygonatum Schrader ex Schultes, Mant. 2: 256. 1824. P. potamium Trin., Gram. Panic. 239. 1826. Setaria polygonata (Schrader) Kunth, Révis. Gramin. 1: 47. 1829. P. hy- drophyllum Trin. ex Nees, Agrost. Bras. 208. 1829, non Schultes, 1824. P. pilosum var. polygonatum (Schrader) Doell in Mart., Fl. Bras. 2(2): 211. 1877. P. ecuadorense Mez, Bot. Jarhb. Syst. 56(125): 3. 1921. Rare, humid forest in deep shade; flowering in October. Distribution: Mexico to Paraguay; Bolivia: Andean Piedmont of Santa Cruz and the Yungas. (1380) P. pulchellum Raddi, Agrost. Bras. 42. 1823. Eriochloa pulchellum (Raddi) Kunth, Révis. Gramin. 1: 30. 1829. P. bipustulatum Schlect., Linnaea 26: 135. 1853. Rare, seasonal forest and forest margins; flow- ering December to May. Distribution: Brazil; Bo- livia: the Yungas. (653, 2073) P. quadriglume (Doell) Hitchc., Contr. U.S. Natl. Herb. 24(8): 460. 1927. P. cayennense var. quadriglume Doell in Mart., Fl. Bras. 2(2): 220. 1877. P. bergii var. quadriglume Hen- rard, Meded. Rijks-Herb. 40: 52. 1921. P. eccentricos Hitchc. & Chase ex Rojas, Revista Jard. Bot. Mus. Hist. Nat. Paraguay 2: 164. 1930, nomen nudum. Common but never abundant, cerrado; some- what palatable (3); flowering January (March) to May; 2n — 18. Distribution: Brazil and Paraguay; Bolivia: Andean Piedmont of Santa Cruz and the Yungas. (736, 758, 856, 891, 1563, 1616, 1770, 1828, 1848, 2397) P. rudgei Roemer & Schultes, Syst. Veg. 2: 444. 1817. P. scoparium Rudge, Pl. Guian. 1: 21, pl. 29, 1805, non Lam., 1798. P. pilosissima Roth ex Roemer & Schultes, Syst. Veg. 2: 458. 1817. P. dasytrichum Spreng., Syst. Volume 77, Number 1 1990 Killeen 171 Grasses of Chiquitanía, Santa Cruz, Bolivia Veg. 1: 317. 1825. P. rhigiophyllum Steu- del, Syn. Pl. Glumac. 1: 76. 1854. P. cay- ennensis var. divaricatum Doell in Mart., Fl. Bras. 2(2): 220. 1877. Occasional, weed along roads and as a natural constituent in campo rupestre /forest scrub margin on Serranía San Lorenzo; coarse and unpalatable (0); flowering November to March. Distribution: Costa Rica to Brazil; Bolivia: the Yungas. (1397, 1962) P. scabridum Doell in Mart., Fl. Bras. 2(2): 201. 1877. Rare, seasonally inundated savanna and forest scrub; flowering November to January. Distribu- tion: the Guianas and Brazil; Bolivia: the Beni. (1514, 1688) P. schwackeanum Mez, Bot. Jahrb. Syst. 56(125): 1. 1921. P. helobium Mez ex Ek- man, Ark. Bot. 11: 23. 1921. P. cyanescens var. latifolium Doell in Mart., Fl. Bras. 2(2): 263. 1877. Occasional, seasonally inundated savanna; flow- ering January to March; 2n = 60. Distribution: Brazil, Paraguay, Uruguay, Argentina. (1634, 2413) P. sellowii Nees, Agrost. Bras. 153. 1829. P. beyrichii Kunth, Révis. Gramin. 2: 231, pl. 27. 1830. P. puberulum Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 3: 277. 1835, non Kunth, 1829. P. probandum Steudel, Syn. Pl. Glumac. 1: 76. 1854. P. rugulosum var. condensatum Hack., Feddes Repert. Spec. Nov. Regni Veg. 6: 343. 1909. Occasional, forest margin and low forest scrub; palatable (5); flowering January to February. Dis- tribution: Mexico and the West Indies to Argentina; Bolivia: Chapare and the Yungas. (831, 1732, 2105 P. stenodes Griseb., Fl. Brit. W. Ind. 547. 1864. P. caricoides var. glabriusculum Doell in Mart. Fl. Bras. 2(2): 239. 1877. Common, seasonally inundated savanna; coarse but somewhat palatable (2) and less common in overgrazed savannas. Flowering is stimulated by fire but plants also bloom irregularly throughout the year. Distribution: Central America and the West Indies to Brazil; Bolivia: the Beni. (610, 710, 1025, 1199, 2075, 2268, 2269, 2770, 2825) P. stoloniferum Poir. in Lam., Encycl. Suppl. 4: 274. 1816. P. frondescens G. Meyer, Prim. Fl. Esseq. 56. 1818. P. olyrifolium Raddi, Agrost. Bras. 43. 1823. P. ctenodes Trin., Sp. Gram. 2, pl. 171. 1829. Common, forming colonies in gallery forests or along streams in seasonal forest; palatable (5); flow- ering November to May. Distribution: Central America and the West Indies to Argentina; Bolivia: Andean Piedmont of Santa Cruz, Beni, and the Yungas. (606, 682, 1386, 1519, 2029) P. trichanthum Nees, Agrost. Bras. 210. 1829. P. guayaquilense Steudel, Syn. Pl. Glumac. 1: 85. 1854. Common weedy species of forest openings, pas- tures, and along roads, from deep shade to full sunlight, in well-drained, seasonally inundated or marshy soils; palatable (4); flowering throughout the year. Distribution: Mexico and the West Indies to Paraguay; Bolivia: Chapare and the Yungas. (899, 1053, 1091, 1259, 1556, 1970, 2773) P. trichoides Sw., Prodr. Veg. Ind. Occ. 24. 1788. P. capillaceum Lam., Tabl. Encycl. 1: 173. 1791. Common weed in forest openings and along well- shaded roads in forest; palatable (4); flowering No- vember to April. Distribution: throughout the trop- ics. (552, 604, 836, 1892, 1971) P. tricholaenoides Steudel, Syn. Pl. Glumac. 1: 68. 1854. P. junceum Nees, Agrost. Bras. 159. 1829, non Trin., 1826. P. bambusoides Speg. ex Arechav., Anales Mus. Nac. Mon- tevideo 1: 128, pl. 9, 10. 1894, non Desv. ex Ham., 1825. P. pilgeri Herter, Herb. Corn. Osten. Com. 1: 2. 1925, non Mez, 1904. Occasional, seasonally inundated savanna; flow- ering November to January. Distribution: Colom- bia to Argentina; Bolivia: Andean Piedmont of San- ta Cruz and the Beni where it is a zonal dominant in seasonally inundated savannas (Beck, 1984). (720, 778, 1505) Pappophorum Schreb. Pensiero, J. F. 1986. Revisión de las especies Ar- gentinas del género Pappophorum (Grami- neae, Eragrostoideae, Pappophoreae). Dar- winiana 27: 65-87. KEY TO SPECIES la. Basal lemma of each spikelet inflated, 2-2.5 mm X 1.8-2.5 mm wide (when viewed later- 172 Annals of the Missouri Botanical Garden ally), with 19- " awns, the second lemma b P. elongatum Sprengel, Syst. Veg. 4(2): 34. sexual (rarely sterile) ...................... : п 1827. Р. polystachyum Kunth, Révis. Gra- lb. Basal lemma at [ each ele thin, 1-1.2 2- 435. 183 31. P. h des Oreh, x 0.3-1 mm wide, with 11-15 awns, the sec- min. SACCNATOLAES rise ond lemma sterile |... P. pappiferum Symb. Fl. Arg. 301. 1879. А " Rare, as a weed along roads near San Ignacio P. krapovickasii Roseng., Comun. Bot. Mus. ge Velasco; flowering October to January. Distri- Hist. Nat. Montevideo 1V(58): 1-5. 1975. bution: Mexico to Argentina; Bolivia: Cochabamba, Rare, as a weed in disturbed places; flowering Tarija, and the Gran Chaco. (1308, 1684) November to February. Distribution: Argentina and Paraguay. (1520, 1729) Paspalum L. Chase, A. 1929. The North American species of P. pappiferum (Lam.) Kuntze, Revis. Gen. Pl. Paspalum. Contr. U.S. Natl. Herb. 28(1): 1- 3(2: 365. 1898. Saccharum pappiferum 310. Chase, A. The South American species Lam., Tabl. Encycl. 1: 155. 1791. P. alo- of Paspalum, unpublished manuscript on file pecurotaeum. Valk Symb. Bot. Upsal 3: 10. at the Hitchcock-Chase Library, U.S. National 794. P. laguroideum Schrader ex Schultes, Herbarium, Smithsonian Institution, Wash- Mant. 2: 342. 1824. P. macrostachyum . ington, Schrader ex Schultes, Mant. 2: 342. 1824. KEY TO SPECIES la. ig rad densely pubesce 2a. Upper glume elabrous, оа winged and cordate-based; lower lemma with stiff papillose-based hairs; spikelets 5.5-7 mm lon to a . pectinatum Upper glume pubescent, ciliate, or glabrous, neither winged nor cordate-based; lower lemma glabrous or pubescent; spikelets 0.8-5 mm long Racemes 10-20; spikelets pa! Pilose P. urvillei 3b. Racemes 1-6; spikelets variously pubescent. 4a. Upper Eus with stiff, erect or spreading, d uk or cilia. 5a. Rachis foliaceous, 5-9 mm wide; pedicels with a ring of hairs at apex; racemes 1, subtended by a ste un bract a at the apex of the к үкен үе bract d into a second conjugate raceme) . stellatum 5b. Rachis not foliaceous or only dish so, den than 3 mm wide; pedicels various; racemes -many 6a. Spikelets narrowly elliptic, 4-5 mm long; submarginal cilia of the p well de- veloped on the lower half (2- 4 mm long) but lacking or short-ciliate abov [e] c P carinatum . Spikelets elliptic to ovate, 2-3.5 mm long; submarginal cilia of the glume well r half (2-4 mm long) or along hr entire length. 7a. Racemes 1-5, quss on an elongate axis; spikelets 2-3.5 mm long; upper glume irregularly long-ciliate (1-4 mm) cul рш its length, the stouter cilia tuberculate-based, the margins c rky P. polyphyllum 7b. Racemes 2, on ugate; spikelets 2 mm long; upper glume long-ciliate towards the apex mm), 2 ciliate on the lower half (0-0.2 mm), the cilia not berculate-based, the margins not cork Р 4b. ^ m glume unifor O lacking stiff submarginal hairs or cili Raceme . malmeanum mly а spike elets 1.8-2 mm long, the hairs on upper dine and lower lemma P Бе аы . ekmanianum 8b. Racemes 3- ^n scum 3-5 mm long, the hairs on upper glume and lower lemma not tuberculate-based. 9a. Leaf blades 8-20 mm wide, flat; spikelets 3.5-5 mm long; lower lemma acute ..... P. erianthum 9b. Leaf blades 2-4 mm wide, involute; spikelets 2.8-3.5 mm long; lower lemma nA mmodes lb. Spikelets not densely pubescent Upper glume Ge the upper lemma expose lla. Upper lemma prominently 3-5-nerved llb. U pee lemma nerveless either glabrous, puberulent, or minutely ciliate the lower glume) lacking or reduced to a membrany scale, the back of = с ы Р. malacophyllum г the nerves indistinct. ex of pedicel т ч, upper glume la rdnerianum cking Р. gar 12b. de of pedicel glabrous; upper glume a membranous scale covering 4 of upper floret inaequivalve Volume 77, Number 1 Killeen 173 1990 Grasses of Chiquitanía, Santa Cruz, Bolivia 10b. Upper glume well Es equal or subequal to spikelet. За. т ikelets 1-2 m 4а. Dex uel dark chestnut brown section Plicatula, go to lead 28b P Upper floret stramineous. . The margins of the glume ciliate, the spikelets suborbicular, acute ....... P. conjugatum 15b. The glume glabrous or puberulent but not ciliate, the pes рег | elliptic, or obovat 1ба. Racens 6-15(20); caespitose perennials. 17a. Racemes stiffly ascen wi the rachis ciliate, 1-5 cm long; foliage . hu succulent, the blades 2-4 mm wide, e, stiffly ascending ........... umigenum 17. Racemes lax, the rachis glabrous, 3-10 cm long; foliage not succulent, the blades 5-15 mm wide, lax P. paniculatum lób. Racemes 1-3(4); caespitose annuals or perennials. pper glume and lower lemma with glandular hairs or globular, glassy papillae; annuals. 19a. cop 0.8 mm p pedicels flattened, erect; racemes 2-3, n a short ax P. parviflorum 19b. Spikelets 1-1.8 iuda ong; pedicels terete or triquetrous; racemes emma provided with glandular-tipped hairs ... P. clavuliferum 18b. Upper glume and lower lemma glabrous; annuals or perennials. Upper floret rugose; upper glume and lower lemma not covering the sides of the upper floret; inea — annual ... P. pictum 21b. Upper floret smooth; upper and lower lemma covering the sides of the upper floret; сас elliptic; perenni... aus P. pumilum 13b. е 2-4 mm lon a. Racemes 2, con njugate or nearly so (rarely a third raceme below). 23a. Spikelets mottled with purple (brown in old or alcohol specimens), 2.2-2.5 т long, broadly elliptic, blunt P. maculosum 23b. ЕК not mottled, 2.2-5 mm long, narrowly elliptic їо orbicular, acute to blunt. acemes reflexed at maturity; plants stoloniferous perennials of sandy soils P. vaginatum 24b. Racemes ascending at maturity; plants caespitose, stoloniferous or rhizo- matous perennials 25a. эшен narrowly elliptic, 3 times as long as wide; plants caespitose, ra not a ss. 26a. Spikelets 1.5-2 mm wide; racemes subconjugate or conjugate, ilose at base P. lineare 26b. ce 1 mm wide; racemes strictly conjugate, glabrous at P. pallens 25b. Spiele ‹ ovate to suborbicular; rhizomatous turf grasses. 5-2.8 mm long, orbicular, blunt 0. P. minus 27b. Spikelets 3-4 mm long, broadly elliptic, aute сь P. notatum 22b. Racemes 1 many, borne on a short to elongate axis, neither conjugate nor subconjugate. 28a. Rachis 1-3 mm wide, A or t ne upper florets slightly convex, stra- mineous, tan, or dark bro 29a. Rachis 2-3 mm ar plants decumbent and rooting at the n 30a. Rachis 3 mm wide; our. acuminate; upper floret a te culms prostrate or scan P. acuminatum 30b. т E mm ase spikelets blunt; upper floret stramineous, tawny or dark n, ide brown then the margins of the upper lemma = dime Р. wrightii white е; 29Ь. Rachis about 1 mm ae flattened; plants caespitose perennials. 3la. Racemes more than 40; basal sheaths strongly equitant, succulent, and with reticulate venation. 32a. Spikelets elliptic; axis of inflorescence (20)30-50 cm lon | intermedium 32b. Spikelets suborbicular; axis of inflorescence 15-20(25) cm long P. de ensum 31b. Racemes 4-15; basal sheaths Um to rounded but not strongly equitant, lacking reticulate ven a. Racemes 1-3 cm long, "i ascending; upper florets stra- mineous; leaf blades a serrated margins „u P. lividum 174 Annals of the Missouri Botanical Garden 33b. Racemes 5-15 cm long, laxly ascending; upper florets cream- colored to dark brown; leaf blades with serrated (cutting) margins. 34a. Spikelets 2.5-3.2 mm long, strongly obovate; upper glume d lower lemma green or brown, not purple spotted, the upper glume pubescent near the apex; upper floret very dark Spikelets 2.2-2.5 mm long, elliptic to obovate; upper 34b. > brown . virgatum glume and lower lemma purple шак glabrous; upper Р. с е 28b. Rachis 0.5 mm wide, aaa or terete; upper florets pers plano-convex; ark brown and shiny sia eam-colored to dark brown o... conspersum Plicatula). 35a. Spikelets 1.8-3 mm lon 36a. Lower Md indurate e, dark brown г mn similar to the upper floret; spikelets biconvex (ie., both upper and ds conv ex 36b. Lower lemma not indurate, bicolored (the margins light tan and thick- ened, surrounding a dark brown, papery center); spikelets plano-con- vex. 37a. 38a. 38b. Spikelets 2-3 mm long; lower lemma plicate, with transverse wrinkles spreading inward from the thicken Spik compressed; plants of savanna wetlands Spikelets 2.5-3.0 mm long, broadly elliptic to obovate; d margins elets 2-2.5 mm long, elliptic; blades folded, the sheaths P. lenticulare nie flat, the sheaths somewhat compressed or rounded on 37b. Spikelets transverse wrinkles 35b. e 3-4 m U erally e back; plants of well-drained soils ........... P . plicatulum na 2 mm long; lower lemma not plicate, lacking P. limbatum pper glume a чадан lemma with irregular reticulate wrinkles, gen- ous between the wrinkles . gemniflorum 39b. Upper я and pe lemma either lacking reticulate wrinkles, or ly. with transverse wrinkles (plicate) on the lowe 40a Я pu stout, r lemma on about 5 mm in diameter; racemes 10-20; spikelets P. atratum 40b. o iptic ' Culms l- Palin mm in diameter; racemes 2-6; spikelets obovate or oe Lower le bicolored, with transverse wrinkles spreading inward from the thickened margins (plicate); axils of ra- ceme кы ose ?, guenoarum 41b. lower lemma not bicolored, the margins similar to the P. acuminatum Raddi, Agrost. Bras. 25. 1823. Rare, as a floating aquatic in artificial pond. Distribution: United States to Argentina; Bolivia: 5) the Beni. (232 P. ammodes Trin., Gram. Panic. 120. 1826. P. sordidum Hack., Oesterr. Bot. Z. 51: 197. 1901 Locally common in cerrado and campo rupestre near Santiago de Chiquitos; flowering is dependen on fire. Distribution: northern South America to Brazil and Paraguay. (2781, 2786) king transverse wrinkles; axils of racemes glabrous or sparsely pilose. 4 Spikelets 3-3.4 mm long, glabrous at base of bracts; lower glume absent; axils of racemes glabrous . Spikelets 3.7-3.9 mm long, minutely pubescent а at the base along the line of the insertion of the glum an un the low emma; lower glume present, vestigial narrow ngular bract М the length of t subequaling га del axils of racemes нер ilos pi P. кетр P. carinatum Humb. & Bonpl. ex Flügge, Gram. Monogr., Paspalum 65. 1810. P. kappleri Hochst. in Steudel, Syn. Pl. Glumac. 1: 21. 1854. P. stellatum sensu Trin., Sp. Gram. 1: 119. 1828. Occasional, seasonally humid savannas with sandy soils near Santa Rosa de la Roca (Prva. Velasco); flowering October to January, appar- ently stimulated by fire. Distribution: northern South America to Brazil. (1436, 1659, 2823) P. clavuliferum Wright, Anales Acad. Cienc. Habana 8: 203. 1871. P. falcula Doell in Volume 77, Number 1 1990 Killeen 175 Grasses of Chiquitanía, Santa Cruz, Bolivia Mart., Fl. Bras. 2(2): 61. 1877. P. horticola Salzm. ex Doell in Mart., Fl. Bras. 2(2): 61. 1877, as synonym. P. pittieri Hack. ex Beal, Gr. of N. Amer. 2: 88. 1896. Uncommon, in seasonal pools or humid sandy soils on granitic outcrops; flowering November to January. Distribution: Mexico and the West Indies to Brazil. (811, 1534) P. conjugatum Bergius, Act. Helv. Phys. Math. 7: 129, t. 8. 1762. P. renggeri Steudel, Syn. Pl. Glumac. 17. 1854 Common, as a weed in pastures and along paths and roadsides in forest; palatable (4); flowering December to (January) April. Distribution: throughout the tropics. (650, 673, 749, 814, 1679, 1752) P. conspersum Schrader ex Schultes, Mant. 2: 174. 1824. Common, seasonally inundated savanna; palat- able (2); flowering October to (January) February; 2n — about 40, abnormal meiosis indicates the species is probably apomictic. Distribution: Mexico to Argentina; Bolivia: the Yungas. (845, 1632, 2317) P. densum Poir. in Lam., Encycl. 5: 32. 1804. P. paniceum Smith in Rees, Cycl. 26: 14. 1813. Rare, roadside ditch in seasonally inundated soils; flowering in January. Distribution: Central America and the West Indies to Brazil; Bolivia: Andean Piedmont of Santa Cruz and the Beni. (2301) P. ekmanianum Henrard, Meded. Rijks-Herb. 40: 49. 1921. This species is similar to P. verrucosum Henrard of Paraguay and southern Brazil, which has more distantly placed spikelets 2.6-3.2 mm long and a zigzag rachis (vs. spikelets 1.8-2.2 mm long and a straight rachis in P. ekmanianum). Rare in Chi- quitania where it is restricted to sandy, superficial soils on Serranía de San Lorenzo (campo rupestre), more common on the Andean Piedmont and the montane tropical savannas near Samaipata; flow- ering April to June. Distribution: apparently a Bo- livian endemic. (1980, 2491) P. erianthum Nees ex Trin., Gram. Panic. 121. 1826 Common, in cerrado and campo rupestre. Somewhat palatable (2), although the foliage is coarse, the relative abundance of this species is markedly decreased in overgrazed savannas. In- dividual plants are variable within populations for foliage vestiture, inflorescence morphology, and spikelet size; all populations sampled near Concep- ción exhibited an abnormal meiosis indicating that the species is probably apomictic; flowering is de- pendent upon fire. Distribution: Central America, Brazil, and Paraguay; Bolivia: Andean Piedmont of Santa Cruz. (1139, 1194, 1369, 2191, 2779) P. gardnerianum Nees, Hooker's J. Bot. Kew Gard. Misc. 2: 103. 1850. P. gardnerianum var. oligostachyum Doell in Mart., Fl. Bras. 2(2): 42. 1877. P. gardnerianum var. ves- titum Kuhlm., Comiss. Linhas Telegr. Estra- teg. Mato Grosso Amazonas 67, Annexo 5, 11: 49. 1922 Locally abundant on Serrania de San Lorenzo (campo rupestre), possibly present on other ser- ranias but overlooked due to lack of flowering pop- ulations at time of visits; flowering in April. Dis- tribution: northern South America to Argentina. (1396, 1984) P. humigenum Swallen, Phytologia 14: 362. 1966. P. denticulatum var. ciliatum Trin., Sp. Gram. 123. 1829. Common to abundant in seasonal ponds and vember to (December) January; 2n = 20. Distri- bution: Venezuela and Brazil. (702, 708, 1655, 2246; Bruderreck 113 ISC, LPB) P. inaequivalve Raddi, Agrost. Bras. 28. 1823. P. inaequivalve var. glabriflorum Hack. ex Stuck., Anales Mus. Nac. Hist. Nat. Buenos Aires 21: 23. 1911. P. glabriflorum (Hack.) Herter, Revista Sudamer. Bot. 6: 138. 1940. Uncommon, forest floor in deep shade and for- est/savanna margins; flowering February to May. Distribution: Brazil, Uruguay, and Argentina; Bo- livia: Beni and the Yungas. (870, 1753, 2028) P. intermedium Munro ex Morong, Ann. N.Y. Асад. Sci. 7: 258. 1893. Very similar and possibly not distinct from P. plenum Chase and P. turriforme R. W. Pohl of Central and northern South America. Abundant in savanna marsh and seasonally inundated savanna; flowering November (December) to February; ab- normal meiosis indicates the species is probably apomictic. Distribution: Brazil, Paraguay, Uru- guay, and Argentina. (1631, 1673, 2258) 176 Annals of the Missouri Botanical Garden P. lineare Trin., Gram. Panic. 99. 1826. P. angustifolium Nees, Agrost. Bras. 64. 1829, non Leconte, 1820 nec Nees ex Trin., 1826. Р. neesii Kunth, Révis. Gramin. 1: 25. 1829. P. tropicum Doell in Mart., Fl. Bras. 2(2): 83. 1877. P. furcellatum S. Moore, Trans. Linn. Soc. London, Bot. 2(4): 505, pl. 34. 1895. Plants are variable among populations for spike- let size and the presence of a third subdigitate raceme. This species is zonally abundant in sandy soils of valley -side campos (upslope) and seasonally humid savanna; it is less common in seasonally inundated savanna and savanna marsh; coarse and unpalatable (0). Flowering is dependent upon fire; abnormal meiosis indicates the species is probably apomictic. Distribution: Mexico and the West In- dies to northern Argentina; Bolivia: the Beni. (783, 1160, 1324, 1366, 1803, 2218, 2772) P. lividum Trin. in Schlect., Linnaea 26: 383. 1854. P. proliferum Arechav., Anales Mus. Nac. Montevideo 1: 52. 1894. P. hieronymii Hack., Oesterr. Bot. Z. 51: 198. 1901. Rare, seasonal pond. Distribution: Mexico to Argentina. (1719) P. maculosum Trin., Gram. Panic. 98. 1826. Paspalum notatum var. maculatum Nees, Hooker’s J. Bot. Kew Gard. Misc. 2: 104. 1850. Occasional, seasonally humid savanna and val- ley-side campos; palatable (4); flowering from Jan- uary to April; 2n = 40. Distribution: northern South America to Uruguay and Argentina. (1640, 1645, 1959, 2282) P. malacophyllum Trin., Sp. Gram. 3: pl. 271. 1831. Anachyris paspaloides Nees, Hook- er’s J. Bot. Kew Gard. Misc. 2: 103. 1850. This species is part of the Anachyris group, an extremely variable taxon with several intergrading forms. Burkhart (1969) recognized three species Р. malacophyllum with leaf blades 1-3 cm wide, Р. simplex Morong with leaf blades 2—5 mm wide, and P. elongatum Griseb. with pilose rachis mar- gins. In Chiquitania the morphology ranges from plants with stout culms, leaf blades 3 cm wide, and large panicles with more than 50 racemes each 10-13 cm long to delicate plants (rarely scandent with leaf blades about 5 mm wide and panicles with 4-5 racemes, each 2-6 cm long; spikelet size М varies independently, ranging from 1.5 to 2.5 mm long. Significantly, life form is not strictly corre- lated with habitat and wide-bladed plants can occur in relatively full sunlight (Killeen 1855), while narrow-bladed plants can occur underneath a closed canopy (Killeen 1845). The entire group is in need of a careful systematic revision. Common, restrict- ed to forest/cerrado margins and low scrub; pal- atable (3); flowering January to (March) April; 2n = 40. Distribution: northern South America to Paraguay; Bolivia: Cochabamba and the Yungas. 1095, 1722, 1727, 1845, 1855, 1918, 1974, 2330, 2449) — P. malmeanum Ekman, Ark. Bot. 10(17): 12, pl. 4. 1911. Rare in herbaria but zonally abundant on valley- side campos (midslope); foliage coarse and unpal- atable (0); flowering April to May; 2n — 20. Dis- tribution: central Brazil. (898, 2024, 2076, 2478) P. minus Fourn., Mex. Pl. 2: 6. 1886. Rare, forming colonies in disturbed, seasonally humid soils; palatable (4); flowering November to January. Distribution: Mexico to Paraguay; Boliv- ia: the Yungas. (1521, 1584) P. multicaule Poir. in Lam., Encycl. Suppl. 4: 309. 1816. P. papillosum Sprengel, Novi Provent. 47. 1819. Common, in seasonally humid sandy soils at top of valley-side campos and superficial soils (or small ephemeral ponds) on granitic outcrops; flowering September to May. Distribution: Mexico and the West Indies to Brazil; Bolivia: the Yungas. (796, 903, 1218, 1250, 1535, 1540, 1644, 2025) P. notatum Flügge, Gram. Monogr., Paspalum 06. 1810. P. distachyon Willd. ex Doell in Mart., Fl. Bras. 2(2): 73. 1877. P. uruguay- ense Arechav., Anales Mus. Nac. Montevideo 1: 53. 1894. The traditional cultivated forage grass of lowland Bolivia; adapted to forest and savanna soils, this species is highly resistant to overgrazing and soil compaction. Although newly introduced exotic species are more productive and offer better-quality forage, this species is still favored for pastures near the ranch house and as a lawn in parks, yards, and football (soccer) fields; occasionally naturalized in humid sandy soils of valley-side campos. Popula- tions flower November to (January) February; lo- Volume 77, Number 1 1990 Killeen 177 Grasses of Chiquitanía, Santa Cruz, Bolivia cal name: grama negra. Distribution: Mexico to Argentina. (594, 1637) P. pallens Swallen, Phytologia 14: 365. 1966. Common, forming dispersed colonies in seasonal ponds and seasonally inundated savanna; palatable (4); flowering November (January) to May; 2n = 20. Distribution: central Brazil. (703, 802, 826, 944, 1421, 1464, 1500, 1532, 1590, 2284) P. paniculatum L., Syst. Nat. 10(2): 855. 1759. Р. hemisphericum Poir., Encycl. 5: 31. 1804. P. compressicaulis Raddi, Agrost. Bras. 29. 1823. P. supinum Rupr. ex Galeotti, Bull. Acad. Roy. Sci. Bruxelles. 9: 237. 1842. P affine Bello, Anales Soc. Esp. Hist. Nat. 12: 125. 1883. Abundant, as a weed of pastures and roadsides in forest soils; flowering October to June. Distri- bution: throughout the tropics. (599, 601, 658, 659, 660, 696, 744, 972) P. parviflorum Rhodé ex Flúgge, Gram. Mono- gr., Paspalum 98. 1810. P. vestitum Steudel, Syn. Pl. Glumac. 1: 17. 1854, nomen nudum. P. parviflorum var. humilis Nees, Doell in Mart. 2(2): 45. 1877. Rare, in seasonal pool on granitic outcrop; flow- ering in November. Distribution: Central America and the West Indies to Brazil. (1537) P. pectinatum Nees ex Trin., Sp. Gram. 1: 117. 1828. Anastrophus pectinatus (Nees ex Trin.) Schlecht. ex B. D. Jacks., Index Kew. 1: 1, 118. 1893 Abundant in campo rupestre (Serrania de San Lorenzo and Serrania de Santiago); flowering is dependent upon fire. Distribution: Mexico to south- ern Brazil; Bolivia: the Beni. (2782A, 2833) P. pictum Ekman, Ark. Bot. 10(17): 11. 1911. Р. maculatum Nash, М. Amer. Fl. 17: 186. 1912 Common, seasonally inundated savanna; flow- ering November to May. Distribution: Central America to Brazil; Bolivia: Andean Piedmont of Santa Cruz. (810, 945, 1490, 1945, 2087, 2454; Bruderreck 261 ISC, LPB) P. polyphyllum Nees ex Trin., Gram. Panic. 114. 1826. P. blepharophorum sensu Trin., Sp. Gram. 2(11): pl. 134. 1829. P. disticho- phyllum sensu Doell in Mart., Fl. Bras. 2(2): 65. 1871. Р. macroblepharum Hack., Оез- terr. Bot. Z. 51: 196. 1901. P. biciliatum Mez, Feddes ee Spec. Nov. Regni Veg. 15: 27. 1917 This species has two variants: sparsely branched, large-leaved plants that grow in relatively well- developed soils and freely branched, small-leaved plants that are rooted in the cracks of granitic outcrops and boulders. The Andean species P. humboldtianum Fligge exhibits a similar phenom- enon but is distinguished from P. polyphyllum by the presence of uniform, lashlike cilia on the upper glume rather than having cilia of irregular length interspersed with stout trichomes. Uncommon, in campo rupestre (Serrania de San Lorenzo and Serrania de Santiago); only senescent specimens have been collected (October), probably flowering late in the rainy season (May). Distribution: Brazil, Paraguay, Uruguay, and Argentina; Bolivia: An- dean Piedmont and montane tropical grasslands near Samaipata, Santa Cruz. (1398, 2795; Cutler 7019 US) P. pumilum Nees, Agrost. Bras. 52. 1829. P. campestre Trin., Мет. Acad. Imp. Sci. St.- Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 3: 144. 1835. P. bicrurulum Salzm. ex Steudel, Syn. Pl. Glumac. 1: 21. 1854. Rare, ephemeral ponds of granitic outcrops; flowering in November. Distribution: the West In- dies to Uruguay and Argentina. (2335) P. stellatum Humb. & Bonpl. ex Flügge, Gram. onogr., Paspalum 62. 1810. P. stellatum var. monostachyum Nees, Agrost. Bras. 78. 1829. P. stellatum var. distachyum Nees, Agrost. Bras. 78. 1829. P. cujabense Trin., Sp. Gram. 3: pl. 284. 1831. P. wagnerianum Schlect., Linnaea 26: 133. 1853. P. splen- dens var. sphacelatum Hack., Oesterr. Bot. Z. 51: 239. 1901. P. stellatum f. hirsuta Hack. in Stuck., Anales Mus. Nac. Hist. Nat. Buenos Aires 21: 28. 1911 Two variants of this species exist in Chiquitanía with distinctive morphologies, chromosome num- bers, and habitat preferences. The more common genotype (P. cujabense) has deep purple rachis wings, spikelets 2 mm long, and a chromosome number of 2n — 20 (Killeen 2487); it is found as a dominant in a zonal sequence on valley-side cam- pos (upslope) and in seasonally humid savannas. Less common in herbaria but locally abundant 178 Annals of the Missouri Botanical Garden around Concepción is a more robust genotype (P. stellatum var. monostachyum) with golden brown rachis wings, spikelets 3 mm long, and a chro- mosome number of 2n = 32 with normal meiosis (Killeen 2477); this is restricted to well-drained cerrado and frequently forms distinct colonies in association with lateritic crests. The US type frag- ment (Humboldt & Bonpland s.n.) of this species (sensu lato) has spikelets intermediate to the two genotypes and the traditional circumscription of the species is maintained. Both variants appear to be somewhat palatable (2). Different populations flower een in March, April, and May; A savaral escences of both genotypes were scored for seed set, but none produced seed. tribution: Mexico and the West Indies to ш Bolivia: Andean Piedmont of Santa Cruz. (687, 904, 1952, 1998, 2011B, 2458, 2473, 2474, 2477, 2487) P. urvillei Steudel, Syn. Pl. Glumac. 1: 24. 1854. Р. ovatum var. parviflorum Nees, Agrost. Bras. 43. 1829. P. larranagae Arechav., An- ales Mus. Nac. Montevideo 1: 60. 1894. P. dilatatum var. parviflorum Doell in Mart., Fl. Bras. 2(2): 64. 1877. Not yet collected in Chiquitania. Rare, as a weed in roadside ditches of the Andean Piedmont but more common in the irrigation ditches of the in- termontane valleys of the Andes. Distribution: United States to Argentina. (2121, 2508) P. vaginatum Sw., Prodr. 21. 1788. P. vagina- tum var. longipes Lange, Vidensk. Meddel. Dansk. Naturhist. Foren. Kjobenhavn 1854: 44. 1855 Seasonal pond with sandy soils. Distribution: throughout the tropics. (1574) Р. virgatum L., Syst. Nat. ed. 10. 2: 855. 1759. Common, as a weed in disturbed marshes, streamsides, and roadside ditches; flowering Oc- tober to February; local name: cortadera. Distri- bution: southern United States to Argentina; Bo- livia: Andean Piedmont of Santa Cruz, Tarija, Beni, and the Yungas. (585, 716, 1317, 1503; putative hybrids with P. plicatulum s.l.: 1720, 1721) P. wrightii Hitchc. & Chase, Contr. U.S. Natl. Herb. 18: 310. September 1, 1917. P. vir- pe var. platyaxis Doell in Mart., Fl. Bras. 2(2): 89. 1877. P. virgatum var. glabrius- Fl. Bras. 2(2): 89. 1877. culum Doell in Mart., P. plicatulum var. multinode Hack., Feddes Repert. Spec. Nov. Regni Veg. 6: 342. 1909. P. platyaxis (Doell) Mez, Feddes Repert. Spec. Nov. Regni Veg. 15: 73. December 31, 1917. P. texanum Swallen, Proc. Biol. Soc. Wash. 55: 94. 1942. P. luticolum Swallen, Phyto- logia 14(6): 373. 1967. Specimens differ in foliage morphology (blades 2 mm wide and involute to 10 mm wide and pla- nate) and the degree of pigmentation of the upper floret (tawny to dark chestnut brown); apparently allied to sect. Plicatula, P. wrightii is closely related to P. hydrophyllum Henrard, which has larger spikelets (2.5-3 mm). Rare, in seasonally inundated savanna and along streambanks; flow- ering January to March. Distribution: United States to Argentina. (764, 1883 southern Paspalum sect. Plicatula An extremely variable species complex with nu- merous intergrading taxa and a center of diversity in south-central Brazil. Populations with identical morphology and phenology are usually restricted to a narrowly defined habitat and/or local geo- graphic area. Most plants sampled in Chiquitania had abnormal meiosis and disfunctional pollen de- velopment (10-70% not staining in analine blue); nonetheless, plants consistently produced viable caryopses at low levels of fertility (10-30% of total florets). Infrequent hybridization followed by apo- mixis is probably responsible for the origination and maintenance of the genetic diversity in this difficult group. Populations differ in stature, foliage morphology (vestiture, length / width ratios, and the compression of the sheath and blade), inflorescence morphology (number, length, and arrangement of racemes), spikelet size (1.5-4.0 mm long), spikelet shape (narrowly elliptic, obovate, orbicular), and the form of the lower lemma (margins thickened or not, plicate wrinkles present or not, bicolored or concolorous). Numerous species and varieties have been described; most modern floristic treat- ments have recognized certain distinctive and/or well-collected genotypes, while uniting other forms under P. plicatulum. In Chiquitania, there is a clear correlation between morphology and habitat: populations with conduplicate foliage and small (1.8-2.5 mm long) elliptic spikelets are common in savanna wetland habitats, while populations with planate foliage and larger (2.5-4.0 mm long) widely elliptic to obovate spikelets are common in well- drained cerrado. There a 5, particularly in central Brazil (e.g., Р. ramosum аны which Volume 77, Number 1 1990 Killeen 179 Grasses of Chiquitanía, Santa Cruz, Bolivia has conduplicate foliage and spikelets 3.2 mm long, is known to occur in humid soils). The following treatment is provisional. P. atratum Swallen, Phytologia 14(6): 378. 1966. P. plicatulum var. robustum Hack., Bull. Herb. Boissier, sér. 2 4(3): 269. 1904. Rare, as a weed in pasture and disturbed places; flowering February to March. Distribution: central Brazil. (1846) P. gemniflorum Steudel, Syn. Pl. Glumac. 1: 25. 1854. P. plicatulum var. oligostachyum Doell in Mart., Fl. Bras. 2(2): 77. 1877. P. reticulatum Hack., Oester. Bot. Z. 51: 199, 1901 As in other taxa within sect. Plicatula, popu- lations are variable for a number of characters, notably inflorescence morphology and spikelet size; a similar taxon, P. plicatulum var. leptogluma Pilger, has suborbicular spikelets 2-2.2 mm long. Occasional to locally abundant in cerrado; palat- able (4); flowering December (January) to May. Distribution: northern South America to southern Brazil. (656, 706, 715, 733, 739, 867A, 876, 893, 1618, 1866, 1875, 1877, 1956, 2015, 2443, 2451) P. guenoarum Arechav., Anales Mus. Nat. Mon- tevideo 1: 50. 1894. P. guenoarum var. ves- titum Henrard, Feddes Repert. Spec. Nov. Regni Veg. 18: 240. 1922. P. plicatulum var. guenoarum (Arechav.) Roseng., Arrill. & Izag., Gram. Urug. 373. 1970. Occasional in cerrado, particularly on lateritic crests; palatable (4); flowering December to (Feb- ruary) March. Distribution: Brazil, Paraguay, Uru- guay, and Argentina. (878, 2394) P. kempffii, Killeen, sp. nov. TYPE: Bolivia. Santa Cruz: Estancia La Pachanga, 5 km S of Con- cepción, Prva. Ñuflo de Chávez, 16?08'S, 62°05'W, 500 m, Killeen 2272 (holotype, ISC; isotypes, LPB, F, MO, US, SI, CTE, SP, NY). Figure 5. P. macedoi Swallen, affinis sed 2 longis, 2- 0. 4 nite longis; spiculis 4, 6-3.9 тт longi is, 2.4-2.7 m latis, ovatis vel late ellipticis; glumis infernis vesti- баш vel bracteatis anguste triangulariter, 2.5 mm lon- gis, evolutis in dimidiis spicularum; flosculis superis rin- gentibus, marginalibus lemmatum albis notabilis. “a racemis; ES cm b М Caespitose perennial from extravaginal inno- vations; culms erect, 1.2- m tall; internodes 3-4 mm in diameter, glabrous, ridged, hollow; nodes 3—5 per culm, bearded, the hairs 3 mm long, stifly ascending. Foliage mainly basal, with the blades reduced above and lacking at the ultimate node; sheaths glabrous, longer than the internodes below, shorter above, keeled, the margins mem- branous, auriculate; ligules 2.5 mm long, mem- branous, adnate to the auricle of the sheath; with a wedge-shaped thickening on the abaxial surface between the sheath and the blade; blades 30 cm long at base, 4-6 mm wide, flat or folded, glabrous on the adaxial surface, glaucous on the abaxial surface, the margins smooth, somewhat narrowed at the base, the apex long-acuminate. Inflorescence a panicle of 2-3 racemes placed 5-6 cm apart on a sulcate, glabrous axis; racemes 7-13 cm long, sparsely pilose in the axils, the hairs 5 mm long, stiff; rachis flattened, 1 mm wide, zigzag, bearing spikelets to the tip; pedicels 0.5-1 mm long. Spike- lets paired, 3.6-3.8 mm long, 2.5 mm wide, broad- ly elliptic, strongly plano-convex, disarticulating below the glumes, the back of the upper lemma oriented towards the rachis; a pair of spikelets overlapping the proximal pair by 44-14 of their length; a delicate ring of hairs 0.2-0.4 mm long inserted at the base of the bracts. Lower glume lacking, vestigial, or a well-developed, triangular bract 2.5 mm long, distinctly 3-nerved, more reg ularly developed on the interior spikelet of each pair; upper glume convex, 5-7-nerved, charta- ceous, the lateral nerves submarginal, subequaling the upper floret but not covering the apex nor the sides of the upper lemma; lower lemma flat, T-nerved, the lateral nerves submarginal, charta- ceous, green, not bicolored nor with plicate, trans- verse wrinkles along the unthickened margins; low- er palea lacking. Upper floret gaping prior to and after anthesis; upper lemma 3.3-3.6 mm long, broadly elliptic to subglobose, convex, 1.5 mm wide, the apex rounded, dark chestnut brown, car- tilaginous, longitudinally striate, the margins white, inrolled, clasping the palea below but not at the apex; upper palea 3-3.3 mm long, subglobose, similar in texture to the upper lemma, with paired lobes at the base enclosing the floral parts. Lodi- cules 2, truncate, fleshy, 0.8 mm long; anthers 3, 2 mm long, orange; style branches 2, naked and translucent, 0.5 mm long; stigmas plumose, purple, 1.5 mm long, exserted; spikelets chasmogamous. Locally abundant in campo cerrado, particu- larly over lateritic crests; palatable (3). In addition to several morphological characters, notably the 180 Annals of the Missouri Botanical Garden I Р aspalum kempffui (Killeen 2272). — A. Habi view of Е pair. —D. Lateral view of la (C and D: bar = distinctly larger spikelets, this segregate species of the Plicatula group differs from similar taxa found in well-drained cerrado soils (i.e., P. plicatulum s.s., P. guenoarum, and P. macedoi) by flowering earlier in the rainy season, from December to Jan- uary. An abnormal meiosis (2n = about 28, З plants sampled), a high proportion of nonstained pollen grains (57%, 8 plants sampled), and low but consistent levels of seed set (14% of total florets, 13 plants sampled) indicates the species is apo- mictic. Р. o H.B.K., Nov. Gen. Sp. 1: 92. 816. P. ana Steudel, Syn. Pl. Glu- mac. 1: 25. 1854. P. humile Steudel, Syn. it. — B. Inflorescence (A and B: bar = 5 ст). — C. Ventral ] mm). Pl. Glumac. 1: 25. 1854. P. compressifolium Swallen, Phytologia 14(6): 381. 1967. P. paludosum Swallen, Phytologia 14(6): 379. 1967. P. pontanalis Swallen, Phytologia 14(6): 376. 1967. ма Swallen, Phy- tologia 14(6): 379. Common to locally abundant in seasonally in- undated and humid savannas, less common in val- ley-side campos and very rarely in cerrado; flow- ering September to (January, February, March) June, individual populations with distinct phenol- ogies. Distribution: Central America to Argentina. (698B, 699, 707, 765, 771, 789, 822, 842, 858, 865, 1231, 1232, 1492, 1498, 1527, 1557, Volume 77, Number 1 1990 Killeen 181 Grasses of Chiquitanía, Santa Cruz, Bolivia 1561, 1601, 1765, 1838, 1862, 1874, 1882, 2001, 2111, 2112, 2270, 2279, 2291, 2322, 2396, 2417) P. lenticulare f. intumescens (Doell) Killeen, comb. & stat. nov. BASIONYM: Р. plicatulum var. intumescens Doell in Mart., Fl. Bras. 2(2): 78. 1877. Although forming distinct populations, the fo- liage, inflorescence morphology, and spikelet size are similar to that of P. lenticulare, and a similar phenomenon of an indurated lower lemma occurs in the related plicatuloid species of P. convexum Humb. $ Bonpl. ex Willd. and P. foveolatum Steudel. In Chiquitania, indurated lower lemmas rarely occur in large-spikelet genotypes as well. Locally abundant at Santa Rosa de la Roca in seasonally humid and shallowly inundated savanna; flowering November to (January) February. (761, 763, 803, 844, 1526, 1667, 1670, 1671, 1723) P. limbatum Henrard, Blumea 4: 511. 1941. Common to locally abundant in seasonally hu- mid/inundated savanna; palatable (3); flowering December (January) to February; 2n = 20, mei- osis normal; possibly the diploid progenitor of the apomictic species of sect. Plicatula. Distribution: Paraguay and Brazil. (698, 809, 816, 862, 863, 1622, 1669, 1812, 2003, 2276, 2453) P. macedoi Swallen, Phytologia 1 4(6): 377. 1967. The type of the species has 3-4 racemes 12— 16 cm long; material from Chiquitania is similar in spikelet morphology, but inflorescences tend to be somewhat pyramidal (racemes 2—10 cm long). Occasional in cerrado, particularly on lateritic crests; flowering January (February) to March. Distribution: central Brazil. (617,714, 1643, 1781, 1796, 1797, 1827, 2323) Р. plicatulum Michx., Fl. Bor. Amer. 1: 45. 1803. P. multiflorum Desv., Opusc. Sci. Phys. Nat. 58. 1831. P. montevidense Sprengel, Syst. Veg. 1: 246. 1825. P. saxatile Salzm. ex Doell in Mart., Fl. Bras. 2(2): 76. 1877, nomen nudum. P. plicatulum var. longipilum Hack., Feddes Repert. Spec. Nov. Regni Veg. 6: 342. 1909. This variant most closely resembles the type fragment of the species (sensu lato), as well as more recent collections from the southern United States where the type collection was made. Different pop- ulations (genotypes) possibly originated separately via polyploidization and apomixis from diploid ge- notype(s) or from existing but distinct apomictic populations. Although it is common in most cerrado localities, it is locally abundant in ungrazed savan- nas. The more pubescent genotypes provide the best forage of any native grass in cerrado com- munities; palatable (3-4). Populations flower from January to (March, April) May. Local names: gamalote, camalote, gramalote. Distribution: southern United States to Argentina. (769, 867B, 1657, 1840, 1854, 1865, 1876, 1878, 1917, 1993, 2390, 2444, 2455) Pennisetum Rich. Brunken, J. 1977. A systematic study of Penni- setum sect. Pennisetum (Gramineae). Amer. J. Bot. 64: 161-176. Parodi, L. В. 1925. Las Gramineae del género Pennisetum. Anales Mus. Nac. Hist. Nat. Buenos Aires 32: 501- 526. KEY TO SPECIES la. Inner whorl of bristles ciliate; culms herbaceous, de in diameter. 2a. Inner whorl of bristles flattened; spikelets 2-3 per involucre; panicles yellow or green P. ciliare 2b. Inner whorl of bristles terete; spikelets 1 er involucre; panicles purple ...... P. setosum lb. Inner whorl of bristles not ciliate; culms stout, semiwoody, 5-10 mm in diameter. 3a. Spikelet stipitate within the involucre; one bristle distinctly longer than the rest; pan- icles purple . purpureum Jb. Spikelet sessile; bristles of various lengths but one not distinctly longer than the rest; panicles yellow nervosum P. ciliare (L.) Link., Hort. Berol. 1: 213. 1827. Cenchrus ciliaris L., Mant. 302. 1771; see DeLisle (1963) for extensive synonymy. Cultivated forage grass adapted to the sandy soils of the Gran Chaco; flowering in October; local names: bufel, bufelo. (1258) P. nervosum (Nees) Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 3: 177. 1835. Gymnotrix ner- vosa Nees, Agrost. Bras. 277. 1829. Cen. chrus nervosus (Nees) Kuntze, Rev. Gen. Pl. 3(3): 347. 1898. Rare, roadside embankment; flowering in Jan- uary. Distribution: Ecuador, Argentina, and Brazil. (1690) 182 Annals of the Missouri Botanical Garden P. purpureum Schum., Beskr. Guin. Pl. 44. 1827. Cultivated forage grass of African origin; due to its robust stature and stout woody culms, the forage must be harvested by hand and fed to cattle; utilized to a limited extent on almost all estancias; local names: mercurón, elefante. (979) P. setosum (Sw.) L. Rich. in Pers., Syn. Pl. 1: 72. 1805. Cenchrus setosus Sw., Prodr. 26. 1788. Panicum cenchroides L. Rich., Actes Soc. Hist. Nat. Paris 1: 106. 1792, non Lam., 1798 nec Elliott, 1816. Pennisetum pallidum Nees, Agrost. Bras. 285. 1829. Similar to P. polystachyon (L.) Sw., an Old World annual species; however, material from Chi- quitanía is clearly perennial with short, stout rhi- zomes. Occasional in scrub on margins of granitic outcrops, rarely in cerrado; flowering December to July. Distribution: Mexico and the West Indies to Brazil; Bolivia: the Yungas. (635, 839, 1821, 2007) Pharus R. Browne Judziewicz, E. J. 1987. Taxonomy and morphology of the tribe Phareae. Ph.D. Dissertation. Univ. of Wisconsin, Madison, Wisconsin. P. lappulaceus Aubl., Hist. Pl. Guiane 2: 859. 1775. Р. glaber H.B.K., Nov. Gen. Sp. 1: 196. 1816. P. pubescens Sprengel, Neue Ent- deck. Pflanzenk. 1: 241, p. 1, f. 1-4. 1820. P. brasiliensis Raddi, Agrost. Bras. 21. 1823. P. micranthus Schrader ex Nees, Agrost. Bras. 302. 1829. Р. angustifolius Doell in Mart., Fl. Bras. 2(2): 23. 1871. P. latifolius L. var. parviflorus (Doell) Prodoehl, Bot. Arch. 1: 250. 1922 Occasional, in deep shade of seasonal forest. Distribution: Mexico and the West Indies to Ar- gentina; Bolivia: the Yungas. (547, 676, 832, 1773, 1890, 1965) Rhipidocladum McClure R. racemiflorum (Steudel) McClure, Smithson. Contr. Bot. 9: 106. 1973. Arthrostylidium racemiflorum Steudel, Syn. Pl. 336. 1854 Glumac. I: Occasional, a liana in seasonal forest; palatable (3); local name: taquara. Distribution: Central America to Peru; Bolivia: the Yungas. (729, 751, 2447, 2734) Rhynchelytrum Nees В. repens (Willd.) С. E. Hubb., Bull. Misc. In- form. (1934): 110. 1934. Saccharum repens Willd., Sp. Pl. 1: 322. 1791. Tricholaena rosea Nees, Linnaea 11: 129. 1837. R. ro- seum (Nees) Stapf & С. E. Hubb. 9(5): 880. 30. Rare, roadside weed; flowering in October. Dis- tribution: throughout the tropics. (2827) Rhytachne Desv. R. subgibbosa (Winkl. ex Hack.) W. D. Clayton, Kew Bull. 20: 261. 1966. Rottboellia lori- cata Trin. subsp. subgibbosa Winkl. ex Hack. in Mart., Fl. 2(3): 311. 1883. Rott- boellia loricata Trin. subsp. glaberrima Hack. in Mart., Fl. Bras. 2(3): 311. 1883. Bras. Rare, San Ignacio de Velasco. Distribution: Bra- zil, Paraguay, and Argentina. (Bruderreck 149 ISC, LPB) Saccharum L. Molina, A. M. 1981. El genero Erianthus (Gra- mineae) en la Argentina y paises limitrofes. Darwiniana 23: 559-585. Swallen, J. 1966. Notes on grasses. Phytologia 14(2): 91-95. z т < TO SPECIES Spikelets awned, 6-8 mm long; rachis inter- nodes 12-2 the length of the spikelet; foliage basal, the sheaths strictly equitant and covered with a thick, chalky wax S. trinii . Spikelets awnless, 3.5-4.5 mm long; rachis in- ternode as long as or twice the length of the spikelet; foliage cauline, the ш. ee on the back, not covered with chalky wax . — o — c ©. ‚осп inarum S. officinarum L., Sp. Pl. 54. 1753. Sugar cane, cultivated on a small scale in Chi- quitania for sugar production and fodder for live- stock, particularly pigs; cultivated commercially on the alluvial plains near Santa Cruz de la Sierra; flowering April to July. (2120, 2445) S. trinii (Hack.) Renv., Kew Bull. 39: 184. 1984. S. saccharoides var. trinii Hack. in Mart., Fl. Bras. 2(3): 258. 1883. Erianthus trinii (Hack.) Hack. in A. DC., Monogr. Phan. 6: 135. 1889. E. balansae Hack. in A. DC., Monogr. Phan. 6: 133. 1889. E. purpureus Swallen, Phytologia 14: 92. 1966. E. gla- Volume 77, Number 1 Killeen 183 1990 Grasses of Chiquitanía, Santa Cruz, Bolivia brinodis (Hack.) Swallen, Phytologia 14: 93. or = nerves of glumes and lower lemma 1966. not indur S. angustissima Individuals are variable within and between pop- ulations in their foliage vestiture (velutinous to sca- brous) and spikelet pubescence (glumes villous, cil- iate, or glabrous; callus hairs less than or surpassing the glumes). When burned, populations flower en masse within four weeks producing purple, chas- mogamous spikelets on racemose, exserted pani- cles. Unburned populations bloom irregularly from December to May, producing pale tan, cleistoga- mous spikelets in globose panicles, which are in- cluded or only slightly exserted from the sheaths. These variants intergrade, and intermediate indi- viduals have been observed in wet savannas which were only superficially burned and along the in- terface of burned and unburned savanna patches. Zonally abundant on valley-side campos (down- slope) and in savanna marsh (providing more than 25% of the herbaceous cover), occasional in sea- sonally inundated savanna, seasonal ponds, and laguna margins; unpalatable (0). Distribution: Mex- ico to Argentina; Bolivia: the Beni and the Yungas. (cleistogamous phenotypes: 646, 767, 1787, 1806, 2281, 2589; chasmogamous phenotypes: 590, 1340, 2219; intermediate phenotypes: 1135, 1460, 2776) Sacciolepis Nash KEY TO SPECIES la. —€— d puis d reddish, the blades flat; pan- icles m long; spikele ets glabrous, nerves of ies mes d lower lemma indurate 5 lb. Foliage not шн аш. green, the lute; panicles 6-8 cm long; spikelets glabrous KEY TO SPECIES la. Racemes flexuous, the rachis internodes and pedicels recurved 1b. Racemes straight, the rachis internodes an a S. angustissima (Hochst.) Kuhlm., Comiss. Lin- as Telegr. Estrateg. Mato Grosso Amazonas 67, Annexo 5, Bot. 11: 92. 1922. Panicum angustissimum Hochst. ex Steudel, Syn. Pl. Glumac. 1: 66. 1854. 5. karsteniana Мет, Feddes Repert. Spec. Nov. Regni Veg. 15: 123. 1918. Common to locally abundant, valley-side cam- pos (midslope), seasonally inundated savannas, sa- vanna marshes and seasonal ponds; flowering March to (May) June, and irregularly throughout the year; 2n = 36. Distribution: Venezuela and the Guianas to central Brazil. (593, 688, 902, 1136, 1198, 1325, 1429, 1641, 1871, 1946, 2027, 2452) S. myuros (Lam.) Chase, Proc. Biol. Soc. Wash. : 7. 1908. Panicum myuros Lam., Tabl. Encycl. 1: 172. 1791. Common, seasonally inundated or humid savan- nas, valley-side campos, and seasonal ponds; usu- ally along cow paths; flowering November to (May) July. Distribution: Mexico to Brazil and Paraguay; Bolivia: Andean Piedmont of Santa Cruz, Chapare, Beni, and the Yungas. (889, 1117, 1253, 2022, 2119, 2603) Schizachyrium Nees Túrpe, A. М. 1983. Revision of the South Amer- ican species of Schizachyrium (Gramineae). Kew Bull. 39: 169-178 S. microstachyum nd pedicels not recurved, erect and appressed to the spikelet. Blades 2-7 cm long; rachis internodes delicate, 0.2-0.6 mm wide at apex; annuals or of indefinite duration, decumbent or scandent; foliage strictly cauline. und panicle of numerous exserted racemes; compou 2 mm wide at apex; sessile spikelet 3 mm long, the lower glume with sulcatum 3b. Culms sparsely bran the culm; rachis internodes 0.4-0. rounded on the bac ine : ched; racemes few to several, not exserted and borne singly at the nodes of 6 mm wide at apex; sessile spikelet 4 mm long, the lower glume S . maclaudii N Е" asal and cauline . Blades 10-70 cm long; rachis internodes 0.8-1.0 mm wide at apex; caespitose perennials; foliage b : 4a. ideis solitary (rarely 2), terminal, exserted; pedicellate spikelet 3-6 mm long, awnless, "ET strictly basal d; foliage EN = m3 ‚ 2 б Ф =] в” А nerum orne at the upper nodes, the base aly included in the sheath; "рейсе baristate, not strongly nerve uli 5a. Lower glume of sessile spikelet strongly concave, coriaceous ; foliage basal and cau . sanguineum c cave, cartilaginous or chartaceous. glume of sessile spikelet scabrous, cartilaginous; mar юны up to 1 m А sobtada 6b. уен glume of sessile spikelet smooth, chartaceous; pedicellate spikelet 2-2.5 mm S. beckii 184 Annals of the Missouri Botanical Garden FIGURE 6. dide gc Aland beckii (Killeen 1987).— Spikelet pair (bar = 1.5 mm).—C. Le S. beckii Killeen, sp. nov. TYPE: Bolivia. Santa ruz: Serranía de San Lorenzo 10 km W o San Javier, Prva. Ñuflo de Chavez, 16°15'S, 62°40'W m, 17 Apr. 1986, Killeen 1987 ом ISC; isotypes, LPB, F, MO, US, ISC). Figure 6. 5. sanguineum (Retz.) Alston affinis sed culmis rami- ficantibus copiose ad nodos superos formantibus paniculas compositas elongatas, 40-60 racemis spathiformibus; spathis 1.5-2.0 cm longis; racemis 1 -3 cm longis, spiculis A. Habit and dl See inflorescence (bar — ower glume of the sessile spikelet (bar — 1 mm). 5 cm).— 4-6 binatim; e rachidium 3.0-3.5 mm longis, callis minutis 0.3 1 infernis lanceo rey Bs chartaceis, nec convexis nec coriaceis, 0.7 mm latis notabilis. Caespitose perennial bunch grass; innovations intravaginal; culms solid, 4.5 m pressed, glabrous, smooth, branched at the upper nodes. Foliage basal and cauline; sheaths glabrous keeled, strongly equitant, longer than the inter- nodes; ligule membranous, 2 mm long; blades flat Volume 77, Number 1 1990 Killeen Grasses of Chiquitanía, Santa Cruz, Bolivia 185 TABLE 5. Morphological variation for selected characters in the Schizachyrium microstachyum complex. Spike let Speci- Rachis size abitat Flowering men! Panicle internodes (mm) preference phenology Latitude Region 2594 elongate strongly recurved 4 seasonally humid July 14°35'S Beni 1572 elongate weakly recurved 5-6 well drained January 17°40'S Andean Piedmont of Santa Cruz 1566 согутроѕе strongly recurved 4 seasonally humid January 17°40'S Andean Piedmont of Santa Cruz 724 corymbose strongly recurved 4 seasonally humid January 16°05'S Chiquitanía 2475 elongate weakly recurved 4-5 well drained May 16%05'S Chiquitania ' All specimens those of the author. or folded, glaucous, up to 23 cm long, reduced at the upper nodes, abruptly narrowed at the base, the junction between the blade and the sheath indistinct, the apex carinate. Inflorescence an elon- gate, compound panicle of spatheate racemes; spathes 1.5-2 cm long; racemes of 4-6 pairs of spikelets, 1-3 cm long, not exserted; rachis inter- nodes ciliate, 3-3.5 mm long, broadest at the apex, narrowed at the base, coriaceous, not inflated; ped- icels slender, flattened, 2.5-3 mm long, 1 mm wide; the rachis internode and pedicel united at base to form a short callus 0.3 mm long, the callus hairs 1 mm long. Sessile spikelet 5 mm long, inclusive of the callus; lower glume as long as the spikelet, lanceolate, chartaceous, dorsally compressed, flat, 0.7 mm wide, with 2 submarginal, minutely sca- brous keels, the apex minutely bidentate, the mid- nerve suppressed; upper glume laterally com- pressed, keeled, 4.5 mm long, 0.5 mm wide when folded; lower lemma hyaline, similar to the lower glume, the palea lacking; upper lemma hyaline, 3.5 mm long, awned from between two lobes 2.5 mm long, the awn 17 mm long, 1-geniculate, the palea vestigial, 0.2 mm long; lodicules 2, truncate, evidently nerved; stamens 3, the anthers 2 mm long, reddish; style branches 2, naked at base, the stigmas plumose, 1.5 mm long. Pedicellate spikelet vestigial, 2-2.5 mm long, 0.2 mm wide, a single awned bract, the awn 2.5 mm long, inserted be- tween 2 minute teeth. This species is allied to the S. sanguineum com- plex but has smaller spikelets 5 mm (vs. 6.5-8 mm) long, the flat, chartaceous lower glume of the sessile spikelet (vs. strongly convex and coria- ceous), a callus 0.3 (vs. 0.5-1) mm long, and the highly branched compound panicle with 40—60 (vs. 5-25) racemes. Similar in some respects to S. riedelii var. multirameus Hack., with which it shares the large compound panicle; however, the latter species has the strongly convex, coriaceous lower glume typical of S. sanguineum s.l. Locally abundant in campo rupestre and occurring with typical variants of S. sanguineum (i.e., Killeen ). 1985 S. maclaudii (Jacques-Félix) S. T. Blake, Proc. Roy. Soc. Queensland 80(6): 78. 1969. S. brevifolium var. maclaudii Jacques-Félix, Rev. Inst. Bot. Appl. Agric. Trop. 32: 432, f. 5b. 1953. Similar to S. brevifolium (Sw.) Nees, which has smaller spikelets (3 vs. 4 mm), slender rachis in- ternodes (0.2 vs. 0.6 mm wide at apex), and shorter (2-4 cm vs. 3-8 cm), more blunt leaf blades; in addition, some specimens from Chiquitanía have a glabrous callus similar to local forms of S. sulca- tum. Inconspicuous, in rank vegetation of season- ally inundated savanna or shallowly rooted in su- perficial soils of granitic outcrops (plants then caespitose and annual); flowering May to July. Distribution: northern South America and Brazil. (915, 2084, 2606) S. microstachyum (Desv. ex Ham.) Roseng., Arrill. & Izag., Bol. Fac. Agron. Montevideo. 103: 35. 1968. Andropogon microstachyus Desv. ex Ham., Prodr. Pl. Ind. Occid. 8. 1825. A. scoparius С. Presl, Rel. Haenk. 1: 338. 1830, non Michx., 1803. 4. paniculatus Kunth, Enum. Pl. 1: 494. 1833, non Lam., 1778. A. lhotzkyi Steudel, Syn. Pl. Glumac. 384. 1855. Andropogon condensatus subsp. elongatus Hack. in Mart., Fl. Bras. 2(3): 297. 1883. S. paniculatum (Kunth) Herter, Re- vista Sudamer. Bot. VI (5-6): 135. 1940. S. neoscoparium Herter, Revista Sudamer. Bot. VI (6-8): 193. 1943. S. microstachyum subsp. elongatum (Hack.) Roseng., Arrill. & Izag., Bol. Fac. Agron. Montevideo 103: 37. 1968. 186 Annals of the Missouri Botanical Garden A polymorphic species complex with numerous intergrading geographic races and ecotypes. Їп eastern Bolivia four phenotypes with distinct mor- phology, phenology, and habitat distribution have been documented (Table 5). Common in cerrado throughout Chiquitanía, as well as in well-drained and seasonally humid savannas with sandy soils of the Andean Piedmont and the seasonally humid savannas with heavy clay soils in the Beni; palatable (4); flowering April to (May) July in Chiquitania; 2n = 20. Distribution: Central America to Argen- tina and Uruguay. (Chiquitania: 724, 859, 875, 896, 990, 1093, 1121, 1273, 1509, 1552, 1572, 1816, 1842, 2012, 2058, 2289, 2475, 2483; Beni: 2594; Andean Piedmont of Santa Cruz: 1566, 1572, 2289, 2296; Steinbach 6809, 6950, 6951, 6952, 6953 US) S. sanguineum (Retz.) Alston, Suppl. Fl. Ceylon 334. 1931. Rottboellia sanguinea Retz., Ob- serv. Bot. 3: 25. 1783. 5. hirtiflorum Nees, Agrost. Bras. 334. 1829. 5. semiberbe Nees, Agrost. Bras. 336. 1829. Andropogon hir- tiflorus (Nees) Kunth, Révis. Gramin. 2(39). 1832. A. semiberbis (Nees) Kunth, Révis. Gramin. 2(39). 1832. A. riedelii Trin., Mem. Acad. Imp. Sci. St.-Pétersbourg, Ser. 6, Sci. Math. 2: 263. 1832. S. riedelii (Trin.) A. Camus, Ann. Soc. Linn. Lyon 70: 88. 1923. S. weberbaueri Pilger, Notizbl. Bot. Gart. Ber- lin-Dahlem 8: 452. 1923. Plants vary in foliage vestiture, the length/ width ratio of leaf blades, pubescence of racemes, and the form of the lower glume of the sessile spikelet. Monomorphic populations with glabrous racemes are zonally abundant upslope in valley-side campos (S. semiberbe sens. str.), while cerrado populations are polymorphic and are composed of a mixture of intergrading genotypes each with a unique phe- nology. Typically, a colony with some distinctive trait (such as glaucous foliage or densely pubescent racemes) is associated with a restricted area of gravel soils developed over a laterite crest, while another genotype is randomly distributed in the well-drained, red clay soils common to cerrado palatability between genotypes is variable (3-4); flowering January (February to communities; May) to July; 2n = 50-60, an abnormal meiosis Setaria P. Beauv. KEY TO SPECIES la. Leaf blades pleated, palmlike, 4-15 cm wide; spikelets about 3 mm long s 1.5-2.5 mm long lb. Leaf blades not pleated, 0.3-3 cm wide; spi ikelet indicates that the species is apomictic. Distribution: throughout the tropics but the greatest morpho- logical diversity is found in the New World; Bolivia: Beni and the Yungas, absent from the savannas of the Andean Piedmont of Santa Cruz. (Pubescent variants: S. hirtiflorum (sens. str.): 1010, 1551, 2010, 2093, 2476, 2479, 2486, 2629; glabrous variants: S. semiberbe (sens. str.): 732,855, 1636, 1783, 1835, 1851, 2018, 2026, 2095, 2421, 2442; intermediate variants: 909, 1390, 1960, 1985, 1997, 2094, 2485, 2629) S. scabriflorum (Кирг. ex Hack.) A. Camus, Ann. Soc. Linn. Lyon 70: 89. 1923. Andro- pogon scabriflorus Rupr. ex Hack. in Mar- tius, Fl. Bras. 2(3): 299. 1883. Occasional in cerrado; palatable (4); flowering January to (February) May; 2n = 40. Distribution: central Brazil. (857, 879, 1619, 1772, 1829, 1852, 1996, 2328) S. suleatum (Ekman) S. Т. Blake, Proc. Roy. Soc. Queensland 80(6): 78. 1969. Andro- pogon sulcatus Ekman, Ark. Bot. 10(7): 4, p. 1, f. 3 & p. 6, f. 3. 1911. A. brevifolius var. leptanthus Hack. in A. DC., Monogr. Phan. 6: 364. 1889. Rare, seasonally humid savanna near San Ig- nacio de Velasco; flowering in February. Distri- bution: Colombia and Brazil; Bolivia: the Beni. (Bruderreck 195 ISC, LPB) S. tenerum Nees, Agrost. Bras. 336. 1829. 5. filiforme Nees, Agrost. Bras. 338. 1829. An- dropogon gracilis C. Presl, Rel. Haenk. 1(4/ 5): 336. 1830, non Sprengel, 1825. 4. preslii Kunth, Révis. Gramin. 2(32): 489. 1831. 4. tener (Nees) Kunth., Révis. Gramin. 2(39): 565. 1832. A. neesii Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, бег. 6, Sci. Math. 2: 264. 1832, non Kunth, 1832. 4. campestris Kunth, Révis. Gramin. 3(41-44): 617. 1833. Rare, seasonally humid savanna and valley-side campo; flowering February to March. Distribution: Central America to Argentina; Bolivia: Andean Piedmont of Santa Cruz and the Yungas. (1775, 1802, 1) 5. poiretiana 2a. Bristles with retrorse scabrosities at tips, as well as antrorse nies below, the panicles gne to clothing . scandens Volume 77, Number 1 1990 Killeen 187 Grasses of Chiquitanía, Santa Cruz, Bolivia 2b. Bristles ingen жез чин only, the panicles not adhering to clothing. cels adnate to 5 or va os tles S. parviflora ww b. Pedicels ei to a single bris 4a. Leaves with an external ligule or ridge of thickened tissue on the abaxial surface at the junction of the blade and sheath, the bla S. v des 1-3 cm wide . vulpiseta 4b. Leaves lacking an external ligule, the gapa 0. us hs 5 cm wide. . Upper floret strongly rugose, broadly uals 9b. Upper floret smooth or weakly rugose, price pmi. S. fiebrigii R. Herrm., Beitr. Biol. Pflanzen 10: 30. 56. 1910. Occasional weed in sandy soils near San Jose de Chiquitos; flowering January to February? Distri- bution: Brazil, Paraguay, and Argentina; Bolivia: the Gran Chaco. (1286, 1697, 1698, 1703) S. leiantha Hack., Anales Mus. Nac. Hist. Nat. Buenos Aires 4: 78. 1904. S. argentina R. Herrm., Beitr. Biol. Pflanzen 10: 30, 56. 1910. Chaetochloa argentina (Herrm.) Hitchc., Contr. U.S. Natl. Herb. 24(8): 480. 1927. Occasional, as a weed in San José de Chiquitos. Distribution: Paraguay, Argentina, and Uruguay; Bolivia: Cochabamba and the Gran Chaco. (1260, 1706) un . parviflora (Poir.) Kerguélen, Lejeunia 120: 1987. Cenchrus parviflorus Poir. in Lam., Tabl. Encycl. 6: 52. 1804. S. genicu- lata (Lam.) P. Beauv., Ess. Agrostogr. 51, 178. 1812. S. gracilis H.B.K., Nov. Gen. Sp. 1: 109-110. 1816. S. purpurascens H.B.K., Nov. Gen. Sp. 110. 1816. P. im- berbis Poir. in Lam., Encycl. Suppl. 4: 272. 1816. S. imberbis (Poir.) Roemer & Schultes, Syst. Veg. 2: 891. 1817. P. flavum Nees, Agrost. Bras. 238. 1829. P. dasyurum Nees, Agrost. Bras. 238. 1829. P. penicillatum Willd. ex Nees, Agrost. Bras. 238. 1829. Chaetochloa geniculata (Lam.) Millsp. & Chase, Field. Mus. Nat. Hist., Bot. Ser. 3: 37. 1903 А pantropical species which varies in a number of characters when studied over its entire geo- graphic range. In eastern Bolivia, the typical (more variable) variant occurs in well-drained habitats as a common weed. However, a well-defined genotype characterized by elongate internodes, glaucous fo- liage, and reduced bristles is restricted to seasonally inundated savannas. The short-bristled variant has a chromosome number of 2n — 72; previous re- ports for the species are 2n — 18 and 36 (Gould, 1960). Distribution: throughout the tropics. (com- mon variant: 525, 588, 679, 711, 1507, 1710, 1766, 1814, 1627, 2227, 2273, 2389; short- S. fiebrigii S. leiantha bristled variant: 645, 1245, 1499, 1628, 2116, 2259, 2274, 2327) S. poiretiana (Schrader) Kunth, Révis. Gramin. 1: 47. 1829. Panicum elongatum Poir. in Lam., Encycl. Suppl. 4: 278. 1816, non Sal- isb., 1824 nec Pursh, 1814. P. poiretianum Schultes, Mant. 2: 229. 1824. P. flabellatum Steudel, Syn. Pl. Gram. 53. 1854. Chaeto- chloa poiretiana (Schultes) Hitchc., Contr. U.S. Natl. Herb. 22: 159. 1920. Rare, logging roads in seasonal forests; flowering December to March. Distribution: Central America to Brazil; Bolivia: the Yungas, Chapare, and Cerro Amboró in western Santa Cruz. (680, 1856) S. scandens Schultes, Mant. 2: 279. 1824. S. trinii Kunth, Enum. Pl. 1: 151. 1833. Pan- icum scandens var. vulgare Doell in Mart., Fl. Bras. 2(2): 171. 1877. Chaetochloa scan- dens (Schultes) Scribn. in J. D. Smith, Pl. Guatem. 5: 91. 1899. Occasional, as a weed in savanna and forest soils. Distribution: Mexico to Paraguay; Bolivia: the Yungas. (804, 807, 838, 866) S. vulpiseta (Lam.) Roemer & Schultes, Syst. Veg. 2: 495. 1817. Panicum vulpisetum Lam., Encycl. 4: 735. 1798. S. composita H.B.K., Nov. Gen. Sp. 1: 111. 1816. Chae- tochloa vulpiseta (Lam.) Hitchc. & Chase, Contr. U.S. Natl. Herb. 18: 350. 1917. Individuals of this species vary in stature, the length/ width ratio of leaf blades, and size of inflo- rescence; however, the external ligule is charac- teristic. Common, in forest scrub, on forest/sa- vanna margins, and as a weed along roads in forest soils; rarely occurring in the deep shade of seasonal forest; flowering January to (March) July. Distri- bution: Mexico to Paraguay; Bolivia: Andean Pied- mont of Santa Cruz and the Beni. (829, 834, 847, 999, 1079, 1674, 1744, 1884, 2106, 2318) Sorghastrum Nash Dávila, P. A. 1988. Systematic revision of the genus Sorghastrum (Poaceae, Andropogo- 188 Annals of the Missouri Botanical Garden neae). Ph.D. Dissertation. lowa State Univ., Ames, lowa. KEY TO SPECIES la. Awns stout, 5-10 cm long, pubescent, 2-ge- niculate; spikelets 7-9 mm long, the callus pun- gent 5. minarum lb. бее delicate, 1-8 mm long, glabrous, twiste 1 -geniculate; spikelets 3.5-4.5 mm long, the ile blunt _ 5. setosum S. minarum (Nees) Hitchc., Contr. U.S. Natl. Herb. 24: 501. 1927. Trachypogon mina- rum Nees, Agrost. Bras. 349. 1829. Andro- ogon minarum (Nees) Kunth, Révis. Gra- min. 2(33): 507. 1831. Stipa penniglumis Trin., Mem. Acad. Imp. Sci. St.-Petersbourg, Sér. 6, Sci. Math. 1: 77. 1831. Sorghum minarum (Nees) Hack. in Mart., Fl. Bras. 2: 276. 1883. Chrysopogon minarum (Nees) Benth., J. Linn. Soc., Bot. 9: 73. 1881. Similar to S. balansae Hack., which has an open panicle, a smaller callus (1 mm vs. 2.5 mm) and shorter awn (1.5-2.5 mm vs. 6-8.5 cm). Uncom- mon, cerrado and along cerrado/forest margins; flowering February to (April) May; 2n = 20. Dis- tribution: Brazil, Paraguay, and Argentina; Bolivia: Andean Piedmont of Santa Cruz. (1768, 1843, 1994, 2387) S. setosum (Griseb.) Hitchc., Contr. U.S. Natl. Herb. 12(6): 195. 1909. Sorghum parviflo- rum Desv. ex Ham., Prodr. Pl. Ind. Occid. 12. 1825, non P. Beauv., 1812. Trachypo- gon stipoides (H.B.K.) Nees var. beta Nees, Agrost. Bras. 351. 1829. Andropogon seto- sus Griseb., Cat. Pl. Cub. 235. 1866. 4. fran- cavillanus Fournier, Mex. Pl. 2: 56. 1881. A. agrostoides Speg., Anales Soc. Ci. Argent. 16: 136. 1883. S. nutans subsp. micranthum var. submuticus (Hack.) Hack. in Mart., Fl. Bras. 2(3): 275. 1883. Sorghastrum parvi- florum (Desv.) Hitchc. & Chase, Contr. U.S. Natl. Herb. 18: 287. 1917. S. stipoides (H.B.K.) Nash. subsp. agrostoides (Speg.) Ro- seng., Arrill. & Izag., Gram. Urug. 201. 1970. Occasional, in valley-side campo, savanna marsh, and seasonally humid savanna but locally abundant in some seasonally inundated savannas; flowering January (February) to May; 2n — 20. Distribution: Central America to Argentina; Bolivia: Andean Piedmont of Santa Cruz. (846, 861, 885, 1630, 1764, 1811, 1958, 1991, 2114, 2306, 2382) Sorghum Moench KEY TO SPECIES la. Rhizomatous perennials; caryopsis conc Pane by the glumes 5. halapense lb. Caespitose оеш caryopsis usually bursting from the glume . bic olor S. bicolor (L.) Moench, Meth. Pl. 207. 1794. Holcus bicolor L. Mant. Pl. 2: 301. 1771. Newly introduced forage sorghums are being used by large estancias for fattening cattle; local name: sorgo, sorgo forrajero. (2446) S. halepense (L.) Pers., Syn. Pl. 1: 101. 1805. Holcus halepensis L., Sp. Pl. 1047. 1753. Common, as a weed along roadsides, particularly in the fertile soils of the alluvial plains SW of the Brazilian Shield; an important source of forage on cattle drives, palatable (3); flowering October to January. Distribution: throughout the tropics. (692, 1314 Sporobolus R. Br. Clayton, W. D. 1965. The Sporobolus indicus complex. Kew Bull. 19: 287-295. KEY TO SPECIES la. Inflorescence a pyramidal кщ at least the ower mont branches verticillat Spikelets about 1.5 mm ie glumes and puis white or translucent; caespit ose an S. они 2b. Spikelets 2 .5-3.5 mm long; glumes a lemma golden- ess -colored; caespitose perennial bunch grasses 3a. Panicle эю жа rlord or spread- ing at P leaf blades involute, 1-3 mm wid S. cubensis 3b. Panicle pea hes ascending or flex- VASA flat or folded, 5-15 S. sprengelit Inflorescence an um en or contracted panicle; branches not verticillate, if several per node, then к оп опе side. a. Spikelets 0.8-1 mm long; stamens 1; pan- ia open; delicate caespitose annuals ....... S. monandrus mm long; stamens 3; ; perennial 4b. Spikelets 1.5-2.5 anicles contracted or open bunch grasses. 5a. на branches strictly erect and a O axis ‚ 5. indicus var. e Ре 5b. Panicle branches ascending to sprea ing S. jac алай S. cubensis Hitchc., Contr. U.S. Natl. Herb. 12: 237. 1909 Volume 77, Number 1 1990 Killeen 189 Grasses of Chiquitanía, Santa Cruz, Bolivia Rare, sandy soils of transition zone between cer- rado and valley-side campo (Santa Rosa de la Roca); flowering is dependent upon fire. Distribu- tion: West Indies to Brazil. (2826) S. gg var. exilis (Trin.) Koyama, J. Jap. Bot. 37: 235. 1962. Agrostis tenacissima pu S Pl. Rar. 16. 1787, non L.f., 1781. Vilfa tenacissima sensu H.B.K., Nov. Cen. Sp. 1: 138. 1816. Axonopus poiretii Roemer & Schultes, Syst. Veg. 2: 318. 1817. Vilfa exilis Trin., Mem. Acad. Imp. Sci. St.-Pé- tersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 6(1): 89. 1840. Sporobolus berteroanus (Trin.) Hitchc. & Chase, Contr. U.S. Natl. Herb. 18: 370. 1917. S. poiretii (Roemer & Schultes) Hitchc., Bartonia 14: 32. 1932. Not yet collected in Chiquitania but common as a pioneer species on the sand dunes of the Andean Piedmont; flowering throughout the year. Distri- bution: southern United States to Argentina; Bo- livia: Andean Piedmont of Santa Cruz, Cochabam- ba, Beni, and the Yungas. (1116, 1582) S. jacquemontii Kunth, Révis. Gramin. 2: 427, t. 127. 1831. Vilfa jacquemontii (Kunth) Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 6 92. 1840. S. indicus sensu Hitchc., Contr. U.S. Natl. Herb. 24(8): 393. 1927. Common, as a weed of roadsides and pastures, in forest or savanna soils; occasional, in well-drained or seasonally humid savannas; locally abundant in cerrado SW of San Javier; coarse but somewhat palatable (2); flowering October to (/V ber) April Distribution: United States to Argentina; Bolivia: Andean Piedmont of Santa Cruz, Cochabamba, Beni, and the Yungas. (596, 721, 746,819, 1234, 1318, 1363, 1370, 1461, 1513, 2193, 2277) S. monandrus Roseng., Arrill. & Izag., Bol. Fac. Agron. Univ. Montevideo 103: 12. 1968. Common, rooted in superficial soils on granitic outcrops, occasionally in cerrado on lateritic crests; flowering September to May. Distribution: Brazil, Paraguay, Uruguay, and Argentina. (806, 914, 1230A, 1442, 1483, 1825) S. pyramidatus (Lam.) Hitchc., U.S. Dept. Agr. Misc. Publ. 243: 84, f. 48. 1936. Agrostis pyramidata Lam., Tabl. Encycl. 1: 161. 1791. Vilfa arguta Nees, Agrost. Bras. 395. 1829. Sporobolus argutus (Nees) Kunth, Enum. Pl. 1: 215. 1833. Rare, seasonal pond; flowering in February. Dis- tribution: United States to Argentina; Bolivia: Cochabamba, Beni, and the Yungas. (1715) S. sprengelii Kunth, Révis. Gramin. 1: 68. 1829. Agrostis sporobolus Sprengel, Nov. Prov. Hal. 46. 1 Part of an intergrading species complex that includes S. cubensis Hitchc., S. adustus (Trin.) Roseng., S. aeneus (Trin.) Kunth, S. acuminatus Trin.) Hack., and S. eximius (Nees) Ekman. The only collection from Chiquitania is intermediate to currently accepted concepts of S. aeneus and S. sprengelii and is provisionally placed in the latter. Locally abundant in sandy soils of campo rupestre (Serranía de Santiago); flowering is stimulated by (probably dependent upon) fire. Distribution: Brazil. (2791) ~ Streptochaeta Schrader ex Nees Judziewicz, E. J. & T. R. Soderstrom. 1989. Mor- phological, anatomical, and taxonomic studies in Anomochloa and Streptochaeta (Poaceae: Bambusoideae). Smithson. Contr. Bot. 68: 1- 52 S. spicata Schrader ex Nees, Agrost. Bras. 537. 1829. Lepideilema lancifolium Trin., Mém. Acad. Imp. Sci. St.-Pétersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 3: 172. 1840. Rare, in deep shade of seasonal forest; flowering April and May. Distribution: Mexico to Argentina. (970, 1968) Thrasya H.B.K. KEY TO SPECIES la. Racemes 10-30 cm long; rachis 2-4 mm wide when folded, the margins glabrous; j paeis 5 mm long; upper floret concealed by the upper lume T. petrosa lb. Racemes 3-10 cm long; rachis 1-1.5 m wide when folded, the margins ipu or glabrous; spikelets about 3 mm lon r floret not concealed by the upper ue 2a. Rachis ciliate, bus hairs stout, Lo based, 1-1.5 mm long, turning yellow maturity; bueh grass with е vaginal in in- novations; foliage basal ............... hrasyoides 2b. Rachis glabrous or ciliate, the hairs deli- cate, not tuberculate-based nor y lowing h age; caespitose with extravaginal in- novations; foliage cauline . crucensis 190 Annals of the Missouri Botanical Garden FIGURE 7. Кат crucensis (Killeen 2334). – (А, В, апа С: Ба = 5 сш). mm). T. crucensis Killeen, sp. nov. TYPE: Bolivia. Santa Cruz: granitic dome 1 km N of Rancho Puesto Nuevo, 35 km S of ens Prva. Nuflo de Chavez, 16%25'S, O'W O m, 26 Feb. 1987, Killeen A ieee ISC; is- otypes, LPB, F, SP, MO, US, SI). Figure 7. T. petrosa (Trin.) Chase, affinis sed innovationibus ex- travaginalibus, culmis ramificantibus copiose ad nodos A. Base of plant. — B. Mid-portion of the culm. — D. Spikelet pairs and rachis. — — C. Raceme E. Lower glume and lower lemma (D and E: bar = 1 medius faciens rami multi ame foliis caulinis, ligulis 2- 5 mm longis, i 2-4. 1 latis; rachidibus race- morum 2.8 mm latis, ийме ы аш pilosis molliter, pilis 0.2-0.5 mm ae spiculis 3.0 mm longis, lemmatibus inferior apex et basis pilosis glaber notabilis. Caespitose perennial from extravaginal inno- vations, the rhizomes short. Culms terete, the in- ternodes glabrous. Foliage cauline; sheaths longer than internodes, compressed, glabrous, auriculate; Volume 77, Number 1 1990 Killeen 191 Grasses of Chiquitanía, Santa Cruz, Bolivia auricles adnate to a membranous ligule 2-5 mm long; blades 4-25 cm long, 2-4 mm wide, linear, reduced at the upper nodes, revolute, glabrous. Inflorescence a single exserted raceme 5-15 cm long, the rachis foliaceous, 1.4 mm wide when folded, clasping the lower V5 of spikelets, the mar- gins softly pilose with white hairs 0.2-0.5 mm long; spikelets 3 mm long, paired, unequally pedicellate, the pedicels adnate to the midrib, each pair oriented in a single row with the backs of the lower lemmas facing each other; disarticulation below the glumes. Lower glume narrowly triangular, 2 mm long, ad- nate to the first rachilla internode at the base, or reduced to a cupulate bead; upper glume narrowly lanceolate, acute, sparsely pubescent along the margins, equaling the upper floret but not covering the sides of the upper lemma; lower floret stami- nate, the lemma chartaceous, splitting lengthwise to the base along a medial groove, hispid apically on the two lateral keels, the palea equaling the lemma, glabrous, acuminate; upper floret 2.2 mm long, narrowly elliptic, the lemma indurate, stra- mineous, smooth, bearded at the apex, the margins inrolled, the palea similar in texture, free at apex. The spikelet morphology of Т. crucensis is sim- ilar to the widespread species 7. petrosa; however, the plants of S. crucensis have branched culms with cauline foliage (vs. unbranched culms with basal foliage), racemes 5-9(12) cm long (vs. (10)15-30 cm long), rachis margins 1 mm wide (vs. 2-3 mm wide), and spikelets 3 mm long (vs. 4-5.5 mm long). Moreover, 7. crucensis has a markedly different habitat preference and phenol- ogy. It is restricted to cracks and superficial soils on bald granitic domes (inselbergs), and flowers January to (February) March; 2n = 20, the type population. (Paratypes: 808, 1819) T. petrosa (Trin.) Chase, Proc. Biol. Soc. Wash. 24: 115. 1911. Panicum petrosum Trin., Sp. Gram. 3: 280. 1836. Tylothrasya petrosa (Trin.) Doell in Mart., Fl. Bras. 2(2): 296. 1877. Individual plants differ in stature, foliage ves- titure, length of racemes, width and color of rachis wings, and the presence or absence of cilia on rachis margins. Different variants occur in mixed popu- lations; very palatable (4). Common, to locally abundant in ungrazed cerrado; flowering January to (March) June; abnormal meiosis indicates prob- able apomixis. Distribution: Central America to Paraguay; Bolivia: the Beni. (738, 871, 1736, 1784, 1850, 1953, 1988, 2391, 2393, 2418; Cutler 7002 US) T. thrasyoides (Trin.) Chase, Proc. Biol. Soc. Wash. 24: 114. 1911. Panicum thrasyoides Trin., Gram. Panic. 126. 1826. T. hirsuta Nees, Agrost. Bras. 94. 1829. Locally abundant in campo rupestre (Serranía de San Lorenzo); flowering is dependent upon fire. Distribution: central Brazil. (2830) Trachypogon Nees Т. plumosus (Humb. & Bonpl. ex Willd.) Nees, Agrost. Bras. 344. 1829. Andropogon plu- mosus Humb. € Bonpl. ex Willd., Sp. Pl. 918. 1806. A. montufari H.B.K., Nov. Gen. Sp. 1: 184. 1816. T. montufari (H.B.K.) Nees, Agrost. Bras. 342. 1829. Locally abundant in Lomerio on rocky hillsides and ridgetops (campo limpo) and over lateritic crests; occasional in campo cerrado with deep red clay soils, in the sandy soils of seasonally humid savannas, and on valley-side campos (upslope); somewhat palatable (2); flowering inhibited by fire, November to (January) April; 2n = 20. Distri- bution: Central America to Argentina; Bolivia: Tar- ija, Beni, Andean Piedmont of Santa Cruz, and the Yungas. (636, 637, 704, 777, 1448, 1617, 1651, 1758, 1776, 1817, 2310) Tripogon Roemer & Schultes T. spicatus (Nees) Ekman, Ark. Bot. 11: 36. 912. Bromus spicatus Nees, Agrost. Bras. 471. 1829. Diplachne simplex Doell in Mart., Fl. Bras. 2(3): 97. 1878. D. spicata (Nees) Doell in Mart., Fl. Bras. 2(3): 160. 1878. Triplasis setacea Griseb., Abh. Konig]. Ges. Wiss. Gott. 24: 304. 1879. Sieglingia spi- cata (Nees) Kuntze ex Stuck., Anales Mus. Nac. Hist. Nat. Buenos Aires 11: 128. 1904. Rabdochloa spicata (Nees) Kuntze ex Stuck., Anales Mus. Nac. Hist. Nat. Buenos Aires 11: 121. 1904. Common, superficial soils over granitic outcrops; flowering September to March. Distribution: United States to Argentina; Bolivia: Cochabamba. (812, 1224, 1463, 1543) Tripsacum L. de Wet, J. M. J., J. R. Harlan & D. E. Brink. 1982. Systematics of Tripsacum dactyloides (Gramineae). Amer. J. Bot. 69: 1251-1257. de Wet, J. M. J., D. H. Timothy, K. W. Hilu & G. B. Fletcher. 1981. Systematics of South American Tripsacum (Gramineae). Amer. J. 192 Annals of the Missouri Botanical Garden Bot. 68: 269-276. de ШЫ J. M. J., J. К. Gray & J. R. Harlan. remates of aa Phytologia Sen 203-227. KEY TO SPECIES la. Culm internodes with lanulose tomentum; sheaths glabrous, poes ent or tomentose; staminate spikelets sE bsessile M T E сс = T. australe lb. Culm s and sheaths ae stami- nate spikelets subsessile and pedicellate .............. ` andersonii T. andersonii J. R. Gray, Phytologia 33: 204. 1976 Introduced into the area in the 1970s as a forage grass by the Fundación Baviera-Boliviana but not widely adopted; a few estancias may still have remnant populations. Local name: pasto guate- ala. T. australe Cutler & E. S. Anderson, Ann. Mis- souri Bot. Gard. 28: 259. 1941 Occasional, forest margins and scrub thickets around granitic outcrops; palatable (3); flowering December to (February) April; 2n = 36. Distri- bution: Venezuela, Colombia, Brazil, and Paraguay; Bolivia: Beni and the Yungas. (793, 841, 1815, 2004, 2326; Cutler 6007 US; Cutler 6008 MO) Zea L. Z. mays L., Sp. Pl. 971. 1753. The local cultivar “Cubano Amarillo” is used as an animal feed and to make a sweet chica, which is a staple of the local diet. LITERATURE CITED Beck, S. G. Comunidades vegetales de las banas dadas de NE Bolivia. Phytocoenologia 12: 321-350. BLack, G. A. . Grasses of the genus Axonopus (a taxonomic treatment). /n: L. B. Smith (editor), Ad- vancing Frontiers of Plant Sciences 5: 1-186. BURKHART, Flora Ilustrada de on oa In (Argentina), Parte П p neas .T.A., Volume 6, Pt. 2. Buenos Ai Caro, |. А. & SÁNCHEZ Eres Pe es species de Cynodon ( Gramineae) de la República Argentina. Kurtziana 5: 191-252. . 1986. The Savannas, Biogeography and (в obotany. Academic Press, London DavipsE, С. 1978. A systematic dy of the genus Lasiacis cepe Paniceae). Ann. Missouri Bot. Gard. 65 1254 DELISLE, D. Е pe "Pasos and wire g the ganu Cenchrus. lowa State Coll. 1. 259-351. EITEN, E 1972. The cerrado vegetation of Brazil. Bot. . (Lancaster) 38: 201-341. 1978. ob^ of the cerrado concept. Vegetatio WE 169-178. FosrER, В. С. 1966. um us gu ie of Bolivia. IV. Gramineae. Rhodora 68: GARÓFALO-SPALDING, B. M AUN of the Genus Axonopus Section Cabrera (Gramineae- Paniceae). Masters Thesis. Iowa State Univ., Ames, Iowa. GOLDSMITH, F. B. 1974. Multivariate analysis of tropical Brazil. J. E communities in Mato Grosso, 11-122 ~ 1960. Chromosome numbers in south- Amer. J. Bot. 47: 873-877. A ALVERDE C. 1982. Bio GOULD, F S. western grasses. П. GU AMÁ .& № tudio de Suelos. CORDECRUZ, Santa Cruz, Bolivia. Р W. € S. Beck. 1989. Structure and compo- sition of savanna vegetation in northern Bolivia: a preliminary report. Brittonia 4: 80-100. HENRARD, J. T. 1927. critical tevion м! Ph genus Aristida. Meded Rijks-Herb. 54А: Hrrencock, А. S. 1927. The grasses of | е Реги, and Bolivia. Contr. U.S. Natl. Herb. 24(8): 291- A. J. GUNPLAND & C. KUNTH. Nova ponds et Specie ш жок Volume l, Pt. 1/2: 1-120. Lutatiae Parisi orum, Paris. NiCORA, E. С. pu: Bod taxonómica sobre Eragrostis neesii y Eragı vista Argent. Agron. Mirar DT, A. W., 181 PALISOT 1 DE Br. AUVOIS, A. М. 1812. Essai d'une nouvelle agrostographie. Imprimie de Fain, Paris. PATTERSON, R. Т. 1984. Investigación y Desarrollo de ea stos Tropicales, Santa Cruz. CIAT Missión Bri- Pom. E FRIETAS, G. ARGENT, P. E. GIBBS, 1 SEMI MIR, C Sine & J. TAMASHIRO. 1988. Floristic composition and гооо Е of a southern cerrado area in Brazil s Roy. Bot. Gard. E 45: 137-151. RENVOIZE, ©. 984. The Grasses of Bahia. Royal а си ————. 1988. Hatschbac hs Paraná Grasses. Royal e Gardens, Kew. Ruiz, R. 1982. Levantamiento Integrado de los Recursos Naturales de la Región de Concepción, Prva. Nuflo de Chávez, Estudio Forestal. CORDE- CRUZ, Santa Cruz, Bolivia. SMITH, L. B., . WASSHAUSEN & R. KLEIN. 1982. Gramineas. p" R. Reitz uu Flora Ilustrada Ca tarinense, Volumes 1, 2, 3. IOESC, Itajai, Santa Catarina, Brazil. SORENSEN, T. A. A method of establishing ate of equal amplitude і in plant sociology based on ilarity of species content, and its application to anal ysis of the isa on ier commons. K. da e Selsk Biol. : 1-34. M. T. 1982. Re vision of [с И sect. aaa i INR Panicoideae). Syst 7: 85-115 STIEB Revision of Ichnanthus sect. Foveo- 1987. Volume 77, Number 1 1990 Killeen Grasses of Chiquitanía, Santa Cruz, Bolivia 193 latus (Gramineae: Panicoideae). Syst. Bot. 12: 187- 216 Твімоѕ, C. В. 1828. Species ies ica Iconibus et oe Illustravit. Volume. 1, Pt. 9: 97- 1 nsis Academie Каре ‘Scanian. St. Petersbur ш TÜRPE, А. M. 1975. Los generos de Gramineas de la Provincia de Tucumán (Argentina). Opera Lilloana 4: боку. 1. T. 1973. A revision я pes Haller (Gramineae) in Malesia. Blumea -80. ZULOAGA, Е. О. 1986. Systematics of rm World "eg of Panicum (Poaceae: Paniceae). Pp. 287-309 in T. R. Soderstrom, K. W. Hilu, C. S. Campbell & Barkworth (editors), Grass Systematics and Evolution. Smithsonian Institution, Washington, D.C. Index to species names; synonyms are in italic. Ас и ли 149 Асгосе 135 excavatum 135 paucispicatum 146 1 5 Agrosticula ias 152 grostis monostachya 161 pyram midata 189 sporobolus 189 tenacissima 189 ira m 152 laxa 166 К mphilepkis exaristatus 145 Anachyris paspaloides red Anastrophus pectinatus 17 Anatherum berterianum 162 1 var. leiophyllus 136 coloratus 137 condensatus subsp. elongatus 185 fastigiatus 136 francavillanus 188 gayanus 136 leucostachyus 137 subsp. selloanus 139 polydactylon 147 preslii 186 riedelii 186 rostratus 151 rufus saccharoides var. hassleri 145 139 fecundis 187 u trachypus anata Anthaenantiopsis 139 trachystachyum 139 Anthistiria n i 161 161 var. aculeolata 140 laucescens var. lateralis 137 6 194 Annals of th Missouri veneer Garden implexa 141 hirsutus 144 var. aequa 141 leptostachyus 144 inversa 141 longecilius 145 leptochaeta 140 macrostachyus 144 longifolia 140 marginatus 144 longiramea var. boliviana 140 minutus 144 macrophylla 14 paranaensis 145 gapotamica 141 pellitus mendo 14 perlongus 143 var. macrantha 141 pilosus neesi poiretii 184 pallens pulcher 145 uina forma breviaristata 141 siccus 143 var. tenuifolia 1: spr paraguayensis 141 stragalus 144 recurvata 141 к 145 riedeliana 141 типго 159 riparia 141 Шш: 159 гозасеа 140 рагадиауапа 159 sellowii 141 super is 50 spadicea 141 vulgari subarticulata 140 var. Q 145 succedeana 140, 141 var. vittata 145 tincta 141 weberbaueri 159 torta 141 Bothriochloa 145 venustula 141 exaristata 145 Arthropogon 141 hassleri 145 bolivianus 142 Brachiaria 145 scaber 142 brizantha 145 villosus 142 decumbens 145, 146 var. glabrescens 142 echinulata 1 Md eminii 146 cemiflorum 182 lorentziana 146 duds paucispicata 146 verticillata 135 plantaginea 146 Arundinella 142 Brom rasiliensis 142 spicatus 191 confinis virgatus 158 flammida 164 Cabrera chrysoblepharis 143 hispida 136, 142 Caryochloa bahiensis 164 mikani 142 Cenchrus 146 Arundo 142 ide uroides 147 donax 142 brownii saccharoides 160 ciliaris 181 Axonopus 142 crinitus 146 Axonopus section Cabrera 144 affinis aureus 14 da muricatus 146 barbigerus 143 myosuroides 147 bijugas vost isum, 143 parviflorus 187 cane un 145 pauciflorus 146 chry a es pungens 146 Us 143, 144 scabridum 147 chrysodactylon 143 setosus 182 compressus 3 viridis 146 cuatrecasasi 143 aetaria dissitiflorus 143 gibbosa 140 eminens 143 pallens var. tenuifolia 141 var. bolivianus 143 torta 141 exasperatus 144 Chaetochloa excavatum 143 argentina 187 fissifolius 144 geniculata 187 herzogii 144 poiretiana 187 Volume 77, Number 1 1990 Killeen 195 Grasses of Chiquitanía, Santa Cruz, Bolivia scandens 187 polydactyla 147 i donat 147 ramosissima 147 Coelorachis 147 dactylon 148 nlemfuensis 148 pascuus 148 plectostachyus 148 Cynosu e angustata 136 fastigiata 136 laxa 136 Digitaria 148 — 149 earindto 163 simplex 191 Doellochloa fastigiata 160 Echinochloa 150 colona 150 crus-pavonis 150 crusgalli var. crus-pavonis 150 sabulicola 150 Echinolaena 150 gracilis 150 minarum 150 procurrens 162 pya 151, 152 t н 152, 153 var. glabrescens 152 bahiensis 153 blepharophylla 153 ea 155 brasiliensis 152 chiquitaniensis 153, 154 155 ciliaris longipila 155 lugens 155 var. glabrata 155 var. Ee 155 153 ma macrothyroa 155 neesii 151, var. expansiflora 153 5 var. laxa Arara 153, 155 pilosa 15 polyneura 156 196 Annals of the Missouri Botanical Garden polytricha 155 Hymenachne 161 psammodes 153 pi ~ rufescens 155 ол secundiflora 156 cordata 161 solida 156 donacifolia 161 soratensis 155 Hyparrhenia 161 tenuifolia 156 i é vahlti 155 rufa a villosa 155 Hypogy Erianthus 182 spathiflorum 139 balansae 182 atus 139 glabrinodis 182 E. 161 purpureus 182 axillare 162 trinii 182 inconstans 162 Eriochloa 156 lilloi 150 brasiliensis 143 minarum 150 distachya 156 pallens 162 grandiflora 156 peruvianus 162 pulchellum 170 polycladus 162 punctata 156 procurrens 162 Eriochrysis 156 riparia 150 cayanensis 157, 158 sandiense 150 holcoides 158 tipuaniensis 162 var. ыт н 158 velutinus 162 var. penicillae 158 Imperata 162 laxa 157, 158 arundinacea var. americana 162 warmingeana 158 brasiliensis 162 concepcionensis 157 var. mexicana 162 Eustachys 15 caudata 162 bahiensis 158 contracta 162 caribaea 15 exaltata distichophylla 159 var. angustifolius 163 Goldbachia mikanii 142 var. caudata 162 Gouinia 159 longifolium 162 latifolia 159 аре 162 virgata 159 tenuis 163 Guadua 159 Lappagopsis bijuga 143 paniculata 159 Lasiacis 163 paraguayana 159 excavatum 135 superba 159 guaraniticum 163 weberbaueri 160 ligulata Gymnopogon 160 Moses 163 biflorus 160 Leersia 163 (адаш 160 contracta 163 subsp. клн 160 о "is йош 60 hexa 16 spicatus 160 mexicana Ye Gymnotrix nervosa 181 Lepideilema lancifolium 189 Gyneri m 160 Leptochloa 163 parviflorum 160 barbata 164 procerum 160 digitaria saccharoides 160 domingensis 164 sagittatum 160 fasc s 163 Hackelochloa 160 gracilis 164 granularis 160 mutica 164 Helopus : brachystachys 156 uninervia 163 cognatus 156 villosa A rrandiflora 156 virgat: Hemarthria 161 Lepocorspium 164 altissima 161 lanatu 64 Heteropogon molle | villosus 135 pe nich 149 I Loudetia 1 bicolor 188 flammida 164 alepensis 188 Loudetiopsis 164 Homolepis 160 chrvsothrix aturensis 161 Ludolphia verticillata 135 Volume 77, Number 1 1990 Killeen Grasses of Chiquitanía, Santa Cruz, Bolivia 197 Luziola 164 y muticus 151 Macroblepharus contractus 155 cavennense 165 juncoides 164 Mon lane as 160 Nardus indica 165 Olyra 165 arundinacea 165 brasiliensis 165 i ciliatifolia 165 Orthoclada 166 laxa 166 Oryza 16 glumaepatula 166 grandiglumis 166 hexandra 163 latifolia 166 var. grandiglumis 166 paraguayensis perennis 1 platyphylla 166 rufipogon 166 sativa 166 var. grandiglumis 166 var. latifolia 166 var. paraguayensis 166 Otachyrium 1 adscendens 149 adustum var. mattogrossensis 149 69 см гш 161 angustissimum 183 6 ba mbusoides 171 b егей var. leiophyllum 170 70 bipustulatum 170 boliviense 169 var. quadriglume 170 chrysoblephare 144 chrysodactylon 143 ciliare 149 deis 6 ar. boliviense 146 198 Annals of the Missouri Botanical Garden ecuadorense 170 elatior 169 elongatum 187 minens 146 eminu 146 equisetum 169 flavum 187 9 hydrophyllum 170 каен 169 imberbis 187 inconstans 162 lasianthum 169 laxum 169 var. amplissimum 169 а 146 melinis var. inerme 164 mertensi 169 millegrana 169 m 169 myuro 3 olyraefolium 171 s 170 paucispicatum 146 rtlisparsum 170 pilosissima 170 pilosum 170 var. polygonatum 170 plantagineum procurrens 162 proliferum var. richardi 16 var. xanthoc om 168 protractum 170 puberulum 171 pulchellum 170 pulchrum 145 quadriglume 170 ee 171 rudgei ae 169 var. condensatum 171 sabulicola ) scabridum 171 scandens var. vulgare 187 trachystachyum 139 trichanthum 171 trichoides 171 tricholaenoides 171 trichophorum 170 sum forma cos 146 forma viride 146 versicolor 167 vestitum 149 viridiflorum 161 Pappophorum 171 pappiferum 17 € ammodes 174 a ieu. 176 atratum 179 o 5 R4 = > E ~ = - blepharophorum 177 brasiliense 14 campestre 177 canescens 143 Volume 77, Number 1 Killeen 1990 Grasses of Chiquitanía, Santa Cruz, Bolivia carinato vaginatum 144 montevidense 181 carinatum 174 multicaule 176 UR ут ш 143 multiflorum 181 chrysites 144 neesii clavuliferum 17 notatum 176 compressicaulis 177 var. maculatum 176 compressifolium 180 ovatum var. parviflorum 178 compressum var. arenarium 144 pallens 177 conjugatum 175 paludosum 180 conspersum 175 paniceum 175 convexum 181 paniculatum 177 cujabense 177 papillosum 176 parviflorum 17 denticulatum var. ciliatum 175 var. humilis 177 dilatatum var. parviflorum 178 pectinatum 177 dissitiflorus 143 pictu 7 istachyon pittieri 175 distichophyllum 177 platyaxis 178 ekmanianum 175 platycaulon 143 elongatum 176 lenum 175 erianthum 175 plicatulum 178, 181 erythrochaetum 144 var. oligostachyum 179 exasperat var. guenoarum 17 alcula 17 var. intumescens 181 fissifolium 144 var. leptogluma 179 formosum 180 var. longipilum 181 foveolatum 181 var. multinode 178 fragile var. robustum 17 furcellatum 176 polyphyllum 177 gardnerianum 175 pontanalis 180 ar. oligostachyum 175 proliferum 176 var. vest pulchrum 145 gemniflorum 179 pumilum 177 glabriflorum 175 ramosum 178 guenoarum 179 renggeri 175 r. vestitum 179 reticulatum 179 hieronymi savannarum 143 emisphericum 177 saxatile 181 horticola simplex 176 humboldtianum 177 sordidum 174 humigenum 175 splendens var. а а 177 humile 180 stellatum 139, сан aae 178 forma Eds Ir inaequivalve 175 var. distachyum 177 var. glabriflorum 175 var. monostachyum 177 intermedium 17 superbum 159 kappleri 174 supinum 177 kempfhi 179, 180 texanum 178 lanatu tropi 1 lanuginosum 149 turriforme 175 larranagae 178 uruguayense 176 lenticulare 180, 181 urvillei 178 forma intumescens 181 vaginatum 178 leptostachyum 144 var. longipes 178 lim 181 verrucosum 17 lineare 139, 176 vestitum 1 lividum 176 virgatum luticolum 178 var. glabriusculum 178 macedoi 180, 181 var. platy macroblepharum 177 agnerianum 1 maculatu wri maculosum 1 Pennisetum 146, 181 malacophyllum 176 ciliare 181 malmeanum 158, 176 myosuroides 147 marginatum 144 nervosum 181 minus 176 pallidum 182 200 Annals of the Missouri Botanical Garden polystachion 182 hirtiflorum sl purpureum 182 maclaudi 18 setosum 182 e ee 136, 185 Pharus 182 subsp. elongatum 185 angustifolius 182 neoscoparium 185 brasiliensis 182 paniculatum 185 glaber 182 riedelii 186 lappulaceus 182 var. multirameus 185 laus и var. parviflorus 182 sanguineum 185 micranthus 182 scabriflorum 186 pubescens 182 semiberbe 18 Piptatheria sulcatum 186 confine 142 tenerum 186 weberbaueri 186 diuum 152 Setaria 186 ciliaris 155 argentina 187 glomerata 155 composita 187 japonica 155 disticha 17 maypurensis 155 fiebrigii 187 microstachya 153 geniculata 187 pilosa 155 gracilis 187 racemosa 155 imberbis 187 tenuifolia 156 leiantha 187 vahlii 1: parviflora 187 Polypogon poiret 187 spicatus 160 polvgonata 17 Rabdochloa spicata 191 purpurascens 187 Rhipidocladum 182 scandens 187 racemiflorum 182 trinii "d verticillata 135 8 Rhynchelytrum 182 Sieglingia spicata 191 repens 182 Sorghastrum roseum 182 balansae 188 Rhytachne 147, 182 minarum 188 subgibbosa 182 Rottboellia 147 subsp. micranthum var. submuticus 188 И pa parviflorum 1 aurita setosum 188 alió 160 stipoides subsp. agrostoides 188 loricata Sorghum 18 subsp. subgibbosa 182 bicolor 188 subsp. ы 182 halepense 188 sanguinea 1 minarum 1 Rytilix a 160 заго И 188 Saccharum Sporobolus 1 caudata cuminatus 189 ayennense 157 adustus 18 ontractum 162 aeneus 189 dubium 162 argutus 189 holcoides 158 berteroanus 189 officinarum 182 brasiliensis 152 pappiferum 172 cubensis 188 repens 182 eximius 189 saccharoides var. trinii 182 indicus 189 sagittatum 16 var. exilis 189 sape 162 jacquemontii 189 trinii 136, 164 monandrus 18 е 158 poiretii 189 Sacciole : pyramidatus 189 ia 183 sprengeli 189 karsteniana 183 Stipa myuros 183 жетекш 188 Schizachyrium 183 Streptoc beckii 184 coe 189 brevifolium 185 Syntherisma var. oe 185 cuyabensis 149 filiforme 18 digitata 149 Volume 77, Number 1 1990 Killeen Grasses of Chiquitanía, Santa Cruz, Bolivia 201 Trachypogon 191 8 flammid | a 164 Tricuspis latifolia 159 Triplasis setacea 191 64 Tylothrasya petrosa 191 Valota penicilligera 149 vestita Alfa arguta 189 exilis 189 Jacquemontii 189 tenacissima 189 Zea 192 mays 192 NOTES A NEW SPECIES AND COMBINATION IN THELYPTERIS FOR GUYANA Thelypteris (subg. Goniopteris) schomburgkii A T. nephrodioides (Klotzsch) Proctor pilis longioribus A. R. Smith, sp. nov. TYPE: Guyana: Esse- laril baxialiter, indusio parvo quibo, Schomburgk 135 (holotype, K; isotype, К). Figure 1. interdum inconspicuo, et venis late anastomosantibus infra sinum iffert. differ FIGURE 1. Thelypteris schomburgkii, based on the type and isotype. — A. Lamina and distal portion of stipe. — B. Pinna. — C. Segments, showing venation and indument, with detail of hairs. ANN. Missouni Bor. Garb. 77: 202-203. 1990. Volume 77, Number 1 1990 Notes 203 Rhizome unknown. Fronds 60-70 cm long; lam- ina with a gradually reduced, confluent, pinnatifid apex, with 17—20 pairs of lateral pinnae, these to 5 X 2.2 cm, incised ca. 14-294 their width, trun- cate at base and with basal segments of larger pinnae reduced half or more; rachis buds lacking; segments rounded to truncate at the tip, with up to 14 pairs of veins, proximal pair of veins from adjacent segments uniting at an acute or obtuse angle with an excurrent vein to sinus; rachis, cos- tae, and costules abaxially with dense, stellate, fur- cate, and acicular hairs intermixed, these 0.1-0.5 mm long, unbranched hairs the longest, branched hairs short-stalked; laminar tissue on both sides with sessile, appressed, stellate hairs, not verrucose. Sori inframedial, with a small, setose or furcate- hairy indusium. Known only from the type collection, at low elevation. This is probably most closely related to T. neph- rodioides (Klotzsch) Proctor, but differs by having lighter green lamina, small inconspicuous indusium (large and persistent in 7. nephrodioides), longer hairs along the axes abaxially (with many acicular unbranched hairs in addition to the stalked-stellate ones), and veins that are anastomosing at a broadly acute to obtuse angle below the sinus with an ex- current vein to the sinus. In this last character, T. schomburgkii comes closer to T. biolleyi (Christ) Proctor, which differs in having some long anchor- shaped hairs on the costae abaxially and in having obtusely united veins below the sinus. Thelypteris (subg. Goniopteris) gonophora (Weath.) A. R. Smith, comb. nov. Dryopteris gonophora Weath. in A. C. Smith et al., Lloy- dia 2: 164, fig. 1(1-3). 1939. TYPE: Guyana: western extremity of Kanuku Mts. in drainage of Takutu River, 4. C. Smith 3283 (holotype, GH not seen, photo — Alan R. Smith, University Herbarium, Univer- sity of California, Berkeley, California 94720, U.S.A. STEYERMARKOCHLOA ANGUSTIFOLIA (SPRENGEL) JUDZIEWICZ, A NEW COMBINATION (POACEAE- ARUNDINOIDEAE- STEYERMARKOCHLOEAE) The recent return to the U.S. National Her- barium of a loan of species of Pariana (Poaceae- Bambusoideae-Olyreae) included a fragment of the type of Pariana angustifolia Sprengel, which proved to be identical to the species described as Steyermarkochloa unifolia (Poaceae-Arundinoi- deae Aes e psa by Davidse & Ellis (Ann. Missouri Bot. Gard. 71: 994-1012. 1984). This was confirmed by examination of the IDC micro- fiche of B- WILLD sheet 18792. The following new combination is therefore required: Steyermarkochloa angustifolia (Sprengel) Judziewicz, comb. nov. Based on Pariana an- ANN. Missouni Bor. GARD. 77: 204. 1990. gustifolia Sprengel, Syst. Veg. 2: 609. 1825. TYPE: Venezuela. Amazonas: Atabapo, Hum- boldt s.n. (holotype, B not seen; fragment US). о hloa e и & Ellis, Ann. Mis- 1 Bot. Gard. 71: In addition to Venezuelan and Colombian pop- ulations cited by Davidse & Ellis (loc. cit.), the species is also known from nearby Amazonas, Brazil at Sao Felipe on the Rio Negro (Fróes 28788, US). —Emmet J. Judziewicz, Department of Botany, Washington, D.C. Smithsonian Institution, 20560, U.S.A. CESTRUM NEBLINENSE (SOLANACEAE), A NEW SPECIES FROM VENEZUELA During preparation of the family treatment of material of a distinctive and undescribed species the Solanaceae for the upcoming Flora of the of Cestrum was encountered. Guayana Highlands (Steyermark et al., in prep.), FiGURE 1. Cestrum neblinense үш 16661). — А. Branch with flower buds.—B. Young inflorescence. — C. Flower. —D. Opened corolla. — E. Fruit. —F, G. Anthers. — Н. Leaf, abaxial side. ANN. Missouni Bor. GARD. 77: 205-206. 1990. 206 Annals of the Missouri Botanical Garden Cestrum neblinense, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dept. Rio Ne- gro, Cerro de La Neblina, 0%51'N, 65%57'W, 700 m, Mar. 1984, Liesner 1666 1 (holotype, MO Frutex 1-2 m altus, ramis rufo- tomentosis, pilis curvis, шаш, multicellularibus. Folia solitaria, angue lanceolata, apice acuminato, parce pilifera. Flos calyce gla rato, 2 5 mm longo, bulboso; corolla 27 mm longa, extus et intus glabra praeter partem imum tomentosum; ие К rectis, edentatis, glabris, an- theris inclusis Shrub 1-2 m tall; twigs reddish, tomentulose with curved, ascending, multicellular hairs, older branches glabrate, longitudinally furrowed and mi- nutely tuberculate from the bases of the caducous trichomes; bumpy from the nodes (bases) of fallen ves. Leaves narrowly lanceolate, sometimes slightly oblique, subcoriaceous or stiffly papyra- ceous, mostly 8-10 cm long, - mm wide, apically narrowly acute or acuminate, basally acute or obtuse, drying slightly discolorous, shiny above, glabrescent above except sometimes near the mid- vein and sometimes glabrescent beneath, the mid- vein greatly elevated above and beneath, the lateral venation of 4-6 prominent and conspicuous, loop- ing, brochidodromous veins on each side of the midvein, the lateral veins lighter-colored beneath, the minor venation evident, the margins slightly revolute, both sides of margins puberulent with scattered curved whitish hairs; petioles 2-4 mm long, indistinct from the blade. Minor leaves want- ing. Inflorescences 1-3-flowered clusters on axil- lary short-shoots or terminal; rachis tomentulose, 2-5 mm long; bracts 4-5 mm long; bractlets 2 mm long; pedicels obsolete or up to 1.5 mm long. Flowers with the calyx glabrate, 2.5 mm long, sub- globose, indistinctly nerved, undulate-lobed, the lobes ciliolate, 0.3 mm long, the cup basally nar- rowed into the pedicel; corolla white(?), 27 mm long, the tube slender, slightly expanded in the upper 4, glabrous outside except for a few hairs in the sinuses near the apex, inside glabrous except for a small tomentose region at the very base, the lobes lanceolate, 2.5 mm long, ciliolate; stamens subequal, ca. 1 mm free, straight, edentate, gla- brous, the anthers included, ca. 0.8 mm long. Berry ca. 9 mm long; seeds 7, 5-7 mm long. Additional specimens examined. VENEZUELA. TERRITORIO FEDERAL AMAZONAS: Dept. Rio Negro, Cerro de La Neblina, Maquire et al. 42500 (MO, US) The willowlike leaves of this species resemble those of Cestrum salicifolium Jacq., also from Venezuela, and those of C. skutchii Morton from Guatemala, but these species have glabrate twigs, membranaceous or chartaceous leaves, and stami- nal filaments with dentate appendages. The twigs of C. neblinense, with their ascending, curved tri- chomes, resemble those in С. tubulosum Sendt. of southern Venezuela, which in contrast has thick, coriaceous, broadly ovate leaves and stamens with filaments that are free for 6 mm (vs. 1 mm in С. neblinense). This species is named for Cerro de La Neblina, the locality of the type collection. — William С. D'Arcy, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166, U.S.A.; and Carmen Benitez de Rojas, Facultad Agronomia, Universidad Central de Venezuela, Maracay, Venezuela. A NEW VITEX (VERBENACEAE) FROM MADAGASCAR The eastern rainforests remain the least known, and currently most threatened, of the diverse vege- tation types of Madagascar. Recent exploration of the forests around the Bay of Antongil, north of Maroantsetra, and on the Masoala Peninsula, the largest remaining tract of eastern rainforest, has yielded the following spectacular new species of Vitex L Vitex masoalensis G. E. Schatz, sp. nov. TYPE: Madagascar. Toamasina: Masoala Peninsula, 6 hr. walk inland from Antalavia along An- talavia River, 15%46'S, 50%03'E, 450 m, 17 Apr. 1987 (fl, fr), Schatz & Suzon 1353 (holotype, MO; isotypes, K, MO, P, TAN). Figure 1. Arbor 7-metralis non ramosa simulans V. hirsutissima J. С. Baker, sed foliis epetiolatis, folioliis aequalis, laminis glabrescentibus, foliis consociatis inflorescentis (= brac- teae) 3-foliolatis, et corolla alba. Unbranched tree 7 m tall, 8 cm diam., the bark smooth, gray, the vegetative bud densely ferrugi- nous-hirsute. Leaves restricted to several nodes at the apex, opposite, palmately compound, epetio- late, 5-foliolate; leaflets + equal, chartaceous, sub- coriaceous, somewhat bullate when fresh, obovate, 56-92 cm long, 13-24 cm broad, the apex cus- pidate, the base long-attenuate, the margins un- dulate along basal half, the venation eucampto- dromous to weakly brochidodromous with ca. 20 veins per side, the midrib slightly elevated adaxi- ally, prominently elevated abaxially and very stout toward the base, the lateral veins slightly impressed adaxially and elevated abaxially, the upper laminar surface glabrous, minutely foveolate, the midrib sparsely hirsute, the lower surface glabrous, very sparsely puberulous to hirsute along the midrib; petiolule 1-2 cm long, very stout and swollen at the point of attachment. Inflorescence a much- branched, condensed cyme borne in fascicles along the main trunk, i.e., trunciflorous, sometimes as- sociated with foliaceous bracts; bracts 3-foliolate, the leaflets + equal, chartaceous, subcoriaceous, sessile, narrowly elliptic to oblanceolate, 29-33 cm long, 3.5-6 cm broad, the apex long-acuminate to cuspidate, the base attenuate, the margins entire, the venation eucamptodromous to weakly brochi- dodromous with ca. 11 lateral veins per side, the midrib impressed adaxially, prominently elevated abaxially, the upper laminar surface glabrous, the midrib sparsely hirsute, the lower surface very sparsely appressed pubescent, essentially glabrous; pedicel slender, 0.5-0.8 cm long, densely ferru- ginous hirsute, bearing l-several densely ferrugi- nous-hirsute, linear to filiform bracteoles; calyx campanulate, 0.5 cm long and broad, the 5 lobes apiculate, 0.4 cm long, densely ferruginous-hir- sute; corolla zygomorphic, short-tubular, 1.5-1.8 cm long, 0.2-0.3 cm broad at the base, to 1.2 cm broad at the apex, white, fading to pinkish white, with longitudinal venation evident, the upper 2 and central 2 lobes triangular, 0.7 cm long, 0.6 cm broad, the apex acute, the margin entire, the lower lobe (lip) narrowly elliptic, 1.2 cm long, 0.5 cm broad, the apex rounded, the margin crenulate, all the lobes ferruginous-hirsute outside, glabrous in- side; stamens 4, adnate to the corolla 0.7 cm from the base, the filaments 2.5 cm long, white, basally hirsute for 0.6-0.7 cm, the trichomes to 0. long, the anthers 0.2 cm long, pink fading to gray- blue, 2-celled; style 2-3 cm long, white, the stigma bifid, the branches 0.3 cm long. Calyx accrescent in fruit, cotyliform, enclosing % of the fruit. Fruit a drupe, globose, 0.8 cm diam., pink turning to purple, the exocarp glossy, the mesocarp thin, somewhat fleshy, the endocarp hard. With over 40 endemic species, Madagascar rep- resents an important center of radiation for the genus Vitex (Moldenke, 1956). Vitex masoalensis is undoubtedly most closely related to V. hirsutis- sima J. С. Baker, also known from the Masoala Peninsula, with which the former shares a similar monocaulous habit and trunciflorous inflores- cences. From V. hirsutissima, the new species differs in having sessile leaves with the leaflets all more or less equal. In addition, the lamina is soon glabrescent, lacking the persistent hirsute pubes- сепсе characteristic of V. hirsutissima. The large, foliaceous bracts sometimes associated with inflo- rescences are three-foliolate in V. masoalensis ver- sus one-foliolate in V. hirsutissima. Special thanks are due to John Myers for the excellent illustration and to Gerrit Davidse for com- ments on the manuscript. ANN. Missouni Вот. GARD. 77: 207-208. 1990. 208 Annals of the Missouri Botanical Garden FIGURE l. Vitex masoalensis (Schatz & Suzon 1353).— A. Flower. — B. Fruit.—C. Adaxial surface of leaf. — axial surface of leaf showing swollen petiolules, and detail of venation. — E. as bract associated with an ea —F. Habit. LITERATURE CITED —George E. Schatz, Missouri Botanical Garden, MoLDENKE, Н. N. 1956. Verbenacées. б | Mad. Р.О. Box 299, St. Louis, Missouri 63166, U.S.A. agascar et des Comores. Famille 174: A NEW SPECIES OF MACROLOBIUM (FABACEAE: CAESALPINIOIDEAE) FROM MESOAMERICA In the course of botanical exploration in the region of Cerro Coronel in northeastern Costa Rica in 1986 and 1987, botanists from the Missouri Botanical Garden and the Museo Nacional de Costa Rica made many collections from this isolated group of low rolling hills. Two species of Macrolobium were collected, M. costaricense W. Burger and an undescribed and closely related one which is here named in honor of Gerardo Herrera, an astute and energetic Costa Rican botanist and co-collector of the type gathering. In the last revision of the genus Macrolobium (Caesalpinioideae-Amherstieae) Cowan (1953) recognized 48 species. Since then, 22 additional species have been added, 19 of them by Cowan, the majority from previously unexplored areas of Venezuelan Guayana (TROPICOS, 1989). Macro- lobium is considered to be wholly neotropical, the African taxa previously considered as part of the genus having been relegated mostly to the African genera Gilbertiodendron and Anthonotha by J. Léonard (see Cowan & Polhill, 1981). Of the 25 genera recognized for the tribe Amherstieae, only Macrolobium and Dicymbe are native to the New World. Although not a single Costa Rican specimen of Macrolobium was known to Cowan in 1953, two species from that country have since been de- scribed by Burger (1968) and Cowan (1985), M. costaricense and М. hartshornii, respectively. Both of these species, as well as the new taxon described herein, are members of sect. Stenosolen Harms and have leaves with few to many pairs of leaflets. Other closely related species are M. trinitense Urban from Trinidad, M. stenosiphon Harms of Pacific coastal Colombia and adjacent Ecuador, and М. colombianum (Britton & Killip) Killip, with six varieties, in northwestern South America and Panama. Macrolobium herrerae Zarucchi, sp. nov. TYPE: Costa Rica. Limon: hills 2 airline km SSE of Islas Buena Vista in the Rio Colorado, 14 of Barra del Colorado, pre- montane wet forest on low hills, 10%40'N, 83°40'W, 10-120 m, 13-14 Sep. 1986 (fl), airline km S Gerrit Davidse & Gerardo Herrera 31126 (holotype, MO; isotypes, AAU, CR n.v., F, K, MEXU, NY, US). Figure 1. et M. Species Macrolobio costaricensí W. Burger Small or medium-sized tree to 25 m. Branchlets dark purplish, essentially glabrous, sparingly len- ticellate. Stipules absent or possibly early caducous. Leaves 8.5-12(-15) cm long with (5-)6-7 pairs of leaflets, petiole (3-)5-12 mm long with the rachis not extending past the terminal pair of leaf- lets; rachis narrowly winged with the upper surface flat or slightly canaliculate, sparingly pubescent; leaflets sessile to subsessile, 5-7(-8) x 1.4-2 (-2.3) cm, elliptic and generally becoming falcate, moderately to strongly inequilateral at the base, apex acuminate and ultimately retuse, both sur- faces glabrous except for the sparingly tomentose midrib above; midvein of leaflets slightly impressed above, prominent below, secondary venation dis- cernible on both surfaces. Inflorescences axillary, generally on older branches at leafless nodes or where older leaves persist, racemose, dense; axis 5-10(-15) mm long, often several to many inflo- rescences densely clustered; pedicels 3-4 mm long, finely pubescent. Bracteoles ca. 4 mm long, con- nate for 2-9 of their length, splitting and persis- tent through early fruit development. Flowers with the hypanthium cylindric, 3-4 mm long, ca. 1 mm in diameter, glabrous, the stipe 1-2 mm long; sepals 4, elliptic to elliptic-oblong, 4-5 x 1.5-2 mm, unequal with the dorsal one wider than the rest, glabrous; single petal obovate, undulate, 13- 1 -8 mm, basally obtuse, apically rounded, glabrous, white with yellow lines; stamens 3, the filaments 13-15 mm long, villosulose along the basal third, purple; anthers versatile, ca. 1. long; gynoecium slightly sigmoid; ovary ca. 3 x 1 mm, very densely tomentose; gynophore 2-3 mm long; style 9-12 mm long, very thin, terete; stigma capitellate. Fruit 8-10 x 2.5-3.3 cm, nar- rowly elliptic-oblong or slightly obovate, glabrous, the margins slightly thickened, attached obliquely ANN. Missouni Вот. GARD. 77: 209-211. 1990. 210 Annals of the Missouri Botanical Garden FIGURE 1. Macrolobium herrerae. —A. H after Davidse & Herrera 31126, E after Alduvin 160). at the rounded base to a stipe 5-7 mm long, the acumen 1.5-3 mm long. Seeds 2-4 per fruit, oval, flattened, immature. Paratypes. HONDURAS. ATLÁNTIDA: Aldea El Pino a 10 km de La Ceiba, faldas del Cerro Pico Bonito, 2 Apr. abit. — B. Inflorescence. — C. Flower. — D. Gynoecia. — E. Fruit. (A-D 1977 (fr), Carolina Alduvín 160 (MO, ?TEFH n.v.). CosTa RICA. LIMÓN: Cerro Coronel, E of Rio Zapote, along and above new road within 1 km of Rio Colorado, tall evergreen forest and edge of Raphia swamp on gentl to moderate slopes, 10°40'N, 83°40'W, 10-40 m, 13- 14 Sep. 1986 (fl), W. D. Stevens & O. M. Montiel 24327 (CR n.v., К, MEXU, MO). Volume 77, Number 1 1990 Notes 211 This new species is apparently most closely re- lated to Macrolobium costaricense, also collected in the vicinity of Cerro Coronel, and additionally known from Panama and the Department of An- tioquia in Colombia. The single available fruiting collection of M. herrerae (Alduvín 160) is the northernmost record (ca. 16°N) for the genus. The new species is best distinguished from M. costari- cense by having a densely tomentose ovary and more numerous pairs ([5- J6-7) of leaflets with the distal pair more or less equal in size to those at- tached lower on the rachis. In M. costaricense the terminal pair of leaflets is larger. Also, the fruit of M. costaricense has valves markedly broader to- ward the apex, whereas the valves in M. herrerae may be only slightly wider distally. Macrolobium herrerae differs from all varieties of M. colombia- num by the densely tomentose ovary and generally smaller floral parts, especially the hypanthium. Since Macrolobium herrerae was collected along the Costa Rica-Nicaragua border in tropical wet, evergreen forest and is also known from Honduras, it probably occurs in Nicaragua. I thank John Myers for the excellent illustration, and Gerrit Davidse, John Dwyer, Rupert Barneby, and W. Douglas Stevens for comments on the manuscript. Botanical exploration in Costa Rica by Stevens, Montiel, and Davidse, all of the Missouri Botanical Garden, was supported by the National Geographic Society Grant #2785-84 LITERATURE CITED BurcEr, W. 1968. Notes on the flora of Costa Rica, 1. Fieldiana, Bot. 31: 273-275. Cowan, R. 53. A taxonomic revision of the genus Macrolobium (арлат "T onc Mem. w York Bot. Gard. 8: 257-342. 1985. Studies in tropical American Legumi- nosae-IX. Brittonia 37: 291-304. & R. . 1981. Tribe 5. Amherstieae. Pp. 135-142 in R. M. Polhill & P. H. Raven (ed itors Advances in Legume Systematics. Royal Bo- tanic Gardens, Kew. Tropicos. 1989. Computerized legume data base at the Missouri Botanical Garden, St. Louis, Missouri. —James L. Zarucchi, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166, U.S.A A NEW SPECIES AND NEW COMBINATION IN CAMPYLONEURUM С. PRESL (POLYPODIACEAE) During a taxonomic revision of the neotropical fern genus Campyloneurum C. Presl, I found that a new combination and new species name are need- ed for two South American taxa. Campyloneurum nitidissimum (Mett.) Ching var. abruptum (Lindman) B. León, comb. nov. Polypodium repens Aublet var. abruptum Lindman, Ark. Bot. 1: 245. 1903. TYPE: Bra- zil. Matto Grosso: Serra do ltapirapuam, ad arbores, 28 Apr. 1894, Lindman ( Regnell Exped. 1) 3345 (lectotype, here designated, S; isolectotype, K). Figure 1A, Polypodium nitidissimum Mett. var. latior Rosenstock, Repert. Spec sni Veg. 12: 474 3 TYPE: Bolivia. La Paz: Yungas septentrionalis, Polo Polo, prope Coroico, 900 m, Buchtien 3526 (ho- otype, S; isotypes, F, US). Campyloneurum nitidissimum var. abruptum occurs from Venezuela and Colombia to Bolivia and Brazil, usually at 100-1,500(2,000) m. It grows terrestrially on slopes of rock, sand, or clay, at the base of trees, and rarely as a hemiepiphyte in lower montane or lowland forests. It is usually misidentified in herbaria as C. coarctatum (Kunze) Fée, from which it differs by its narrow, long- creeping stem, widely spaced phyllopodia, and leaves less than 40 cm long. Campyloneurum nitidissimum is characterized by its dark linear scales, and stem 5-10 mm wide. Campyloneurum nitidissimum var. ni- as asymmetrically divided rown, tidissimum generally primary areoles and two to three sori between secondary veins, and always has subcoriaceous leaves. Campyloneurum nitidissimum var. abrup- tum differs from the typical variety by having un- divided primary areoles, two sori between second- ary veins, and herbaceous-chartaceous leaves. Selected specimens examined. VENEZUELA. MERIDA: ESE of Santa Bárbara, 7°41'N, 71°28'W, 9 Mar. 1980, Liesner & González 9236 (MO). TÁCHIRA: W of Pinal, W of bridge over Rio Frio, 27-30 Aug. 1966, Steyer- mark & Rabe 96710 (GH). s Pedraza, above El Algarrobo, 8°31'N 3 Aug. 1983, Werf < Ortiz 5810 (UC). ыр че BOLÍV ‘AR: Boca Verde, = Sint, 13-14 Feb. 1918, Pennell 4216 (NY). SUR SANTANDER: vicinity of Barranca Bermeja, Magdalena valley, between Sogamoso and Colorado rivers, 19 Feb. 1935, Haught 1567 (US). Boyaca: Casanare, Taura- mena, bosques del Rio Caja, 13 Apr. 1963, Uribe 4281 yee TOLIMA: Alto de Consuelo, Honda, July 1923, Ariste 4997 (F, GH). HUILA: 10 km SSW (by road) of La Plata, on road to Purace, 29 Apr. 1972, Duncan 1889 (UC). ECUADOR. PICHINCHA: between Nono and Nanegal, 14 km of Nono, 4 Sep. 1976, Croat 38838 (UC). МАРО: Rio Wai Si Aya, a northern ves i Rio Aguarico, 8 Aug. 1980, Brandbyge et al. 1 (AAU). PASTAZA: Río Bufeo, northern tributary of he Bobonaza, 2?20'S, 76%40'W, 19 July 1980, ae et al. 34779 (AAU). RU. SAN MARTIN: Chazut Klug 4080 (F, GH, 5, UC. US). HUANUCO: near confluence of Rio Cayumba with Rio ¿rota 10 Oct. 1936, Mexia 8272 (BM, F, GH, S, UC). Pasco: BE Quebrada Honda: camino à Tunqui, 17 жы 985, León 667 (GH, USM). JUNÍN: camino a Tarm antes de Carpapata, 1 Oct. 1982, León 340 (GH, USM), ee DE DIOS: Tambopata, lodge "Cuzco Amazónico,” 986, León 884 (USM). BOLIVIA. LA PAZ: SE been Caranavi and Guanay, 15?33'S, 67%45'W, 980, Croat 51658 (MO, UC); Nor Yungas, PR : tis Ps Yolosa, 16?12'S, 67°50'W, 19-20 Oct. 1982, Solomon 8549 (МО); Tumupasa, 22 Jan. 1912, R. S. Williams 1059 (GH, NY). TARIJA: Arce, vic. Comunidad Sidras, 22?14'S, 64°32'W, 6 May 1983, MEA 10535 (NY, UC). BRAZIL. Ronan Posto Mucaj io Mucajai, vic. a airstrip, 13 . 1971, ed al al. 10923 (K, NY, JS). AMAZONAS: Rio Curuquete, Cachoeira República, 95 July 1971, Prance et a 14584 (NY, US). s ank of Rio Maicuru, ca. 23 km from ms 955'S. 54?26'N, 29 July 1981, с. et al. 3702 nu RODÓNIA: basin of Rio Madeira, 2 km below con- fluence of Rio Abuna, 12 Nov. 1968, Prance et al. 8341 (S, US). MATO GROSSO: Mato do Curupira, 18 Feb. 1894, Lindman (лие Exped. I) 3075 (Sy gorge of Veu а dos Guimaraes, 17 Oct. 1973, Prance { ías: Yateri, Balsamo, 9 Feb. 1895, Macedo 5298 (S, US). MINAS GERAIS: Rio Branco, 13 Nov. 1930, Mexia 5298 (BM, NY, S). — Campyloneurum wurdackii B. Leon, sp. nov. TYPE: Venezuela. Bolivar: Cerro Pijiguao, Sier- ra Suapuré, 400-450 т, 19 Jan. 1956, Wur- dack & Monachino 41303 (holotype, MO; i dé US). Figure 1D, t haec speci ies a Compyloneunim repens 3-4 . Mediae venulae nitidissimum paleis lanceolatis, 3- 4 mm longis Terrestrial. Stem long-creeping, not — 2-3 mm diam.; scales light brown, lanceolate, 3- ANN. Missouni Bor. Garb. 77: 212-214. 1990. Volume 77, Number 1 1990 Notes 213 0.8mm ir 97, E ERA ee 89272927225 ў FIGURE 1. A-C. Campyloneurum nitidissimum var. abruptum.—A. Stem scale (100 х). — B. Cell structure from the middle part of the stem scale (100x) (A-B, Mexia 8272 F). — C. Detail of venation pattern (1 х) (Karsten Campyloneurum wurdackii (Wurdack & Monachino 41303 MO). — D. Stem scale. — E. Cell structure from middle part of the stem scale (100 x). — F. Detail of the venation pattern (1 х). (Hatched area on both scales represents the insertion zone. 4 mm long, 1-1.3 mm wide, with bases auriculate, apices obtuse or rarely acute; phyllopodia 5 mm apart. Leaves 19-41 cm long; petiole 4-8 cm long, stramineous. Lamina simple, lanceolate, with bases cuneate then long decurrent, apices acuminate to slightly caudate. Veins prominent or prominulous, secondary veins straight, 60—65° divergent from the costa, tertiary veins forming 7-8 primary ar- eoles between the costa and margin, excurrent veinlets 3-4 per areole, simple or furcate, central veinlet generally anastomosed with the transverse vein forming asymmetric secondary areoles. Sori subterminal or medial, rarely basal, on the free excurrent vein; paraphyses absent. arat pe VENEZUELA. TERRITORIO FEDERAL AMAZONAS: Atures, 23 km NE of Puerto Ayacucho, near Cachama, 5°51'N, 67°24'W, 17-19 Apr. 1978, Davidse & Huber 15306 (MO). Campyloneurum wurdackii is known only from Venezuela, where it has been collected twice from mossy places in lowland forests at 90-500 т. 214 Annals of the Missouri Botanical Garden Campyloneurum wurdackii and С. nitidissi- mum var. abruptum have similar cuneate leaf bas- es that are decurrent on the petiole. However, the former species has a stem 2-3 mm wide, lanceolate scales, and venation with divided primary areoles (Fig. 1D-F), while the latter has a stem 5-10 mm wide, linear scales, and venation with usually un- divided primary areoles (Fig. 1A-C). Campyloneurum wurdackii appears to be closely related to C. brevifolium, as both species have asymmetrically divided areoles, approximate leaves, and slightly clathrate stem scales. However, the 2-3 mm wide and lanceolate stem scales, whereas the latter has a stem 5-10 mm wide and ovate-lanceolate stem scales. Vareschi (Fl. Venez. 1(2): 950. 1969) previ- ously named Campyloneurum wurdackii as Poly- podium repens Aublet var. spathulatum Vareschi a nomen nudum because it lacked a Latin descrip- former has a stem tion. He cited Wurdack 41130 as the type of his new variety, but this number represents an angio- sperm (J. Kalluncki, pers. comm.). The number 41130 undoubtedly is a typographic error for Wur- dack 41303, a specimen of Campyloneurum that corresponds to the description of his new variety and from the same locality cited by Vareschi. The name honors Dr. J. J. Wurdack, who collected the type. I thank M. Lane, R. C. Moran, B. Ollgaard, A. R. Smith, R. C. Stolze, R. M. Tryon, and K. Young for commenting on the manuscript, and D. Lorenz for kindly helping with the illustrations. For assis- tance with the Latin description, I thank P. Klein and J. Dwyer. — Blanca Leon, Museo de Historia Natural, Av. Arenales 1256, Casilla 14-0434, Lima 14, Peru. A NEW SPECIES OF PALICOUREA (RUBIACEAE) FROM COSTA RICA Palicourea Aublet is a genus of about 200 species of shrubs and small trees found throughout the moist and wet Neotropics, with 24 species presently known from Costa Rica, including this new one. The recent treatment of Palicourea in Central America (Taylor, 1989) does not include this species. Palicourea gomezii С. М. Taylor, sp. nov. TYPE: osta Rica. Heredia: quebrada Sangrijuela, O m, flower & fruit, Gómez, Chacón Herrera 20896 (holotype, CR; isotype, MO). igure 1. Frutex 2-5 m altus, pilosulus. Folia petiolata; laminis ellipticis, 11-20.5 em long. x 5-9 cm lat., chartaceis, i ariarum munitis; stipu- larum foliacearum laminis sol aa 8- 1 2 mm lon- ovatis. Inflorescentia bulo. ex thyrso oliato : ramis inferioribus congue constans, viridis, 4-4.5 cm long. x -8 cm lat.; bracteis 0.5-7 mm longis; pedicellis l. 5- 9 mm longis; florum lobis calycinis 1-3 mm longis, in eodem flore inaequalibus; corolla infundibulari basi valde gibbosa, alba colore roseo suffusa, externe glabra, tubo 7 mm longo, lobis 3-5 mm longis. Fructus obovoideus, 6 mm ngus. Succulent shrubs or small trees to 5 m tall, sparsely to densely pilose to pilosulous, sometimes becoming glabrescent in older parts. Leaves with p “у at арех acute to acuminate with the „9—1 cm long, at base acute to cuneate, 1l. E 120 cm iis 9-9 cm wide, about 2.2-2.3 times as long as wide, chartaceous, secondary veins 11-15 on each side of costa, widely angled with the costa, widely curving, not looping to connect, with 1(-2) well-marked intersecondary veins pres- ent between each pair of secondary veins, margins not revolute; petioles 1-5 cm long; stipules form- ing ovate interpetiolar laminae, 8-12 mm long, at apex broadly rounded and emarginate, the sinus between the lobes about 1-3 mm deep. /nflores- .9-0.6 times as long as broad at base, peduncles 0-0.5 cm long (the panicle often appearing tripartite), not geniculate at base, bracts triangular to lanceolate, 0.5-7 mm long, those subtending primary branches about 5- 7 mm long and those subtending pedicels about 0.5 mm long; pedicels 1.5-5 mm long; peduncle, branches, bracts, and pedicels green, glabrescent. Flowers with calyx green, glabrescent, the free portion divided to base, lobes triangular to lingulate, acute, 1-3 mm long and strongly unequal in length on an individual flower; corolla carnose, funnel- form, swollen and strongly gibbous at base, the tube curved at ca. 90? just above the swelling, lilac to rose or white flushed with these colors, externally glabrous, internally glabrous except for a ring ca. 5 mm wide of trichomes just above the basal swell- ing and extending to the stamen insertion, tube 7 mm long, lobes triangular to lingulate, acute, 3-5 mm long; anthers in short-styled flowers 4 mm long; styles in short-styled flowers ca. 5 mm long, stigmas 1 mm long. /nfructescences similar in size and proportion to inflorescences; immature fruit obovoid, flattened, 6 mm long; well-developed pyr- enes not seen. This new species is distinguished by emarginate laminar stipules with rounded lobes; subsessile or sessile, relatively broad inflorescences; relatively long calyces with unequal lobes 1-3 mm long; and оү BÉ rose or lilac (or white flushed with with proportionately long lobes, half or more She m of the corolla tube. Among Central American plants the new species can be confused with Psychotria copensis Dwyer from Panama, which is probably better placed in Pali- courea (Taylor, in prep.). The Panamanian species differs in its usually more numerous secondary leaf veins, (13-)15-24 rather than 11-15; acute rath- er than rounded stipule lobes; and blue corollas with proportionately shorter lobes, a third or less the length of the tube. Palicourea gomezii also resembles an undescribed species from Panama (Taylor, in prep.), which differs in its more con- tracted inflorescences 2-2.5 cm long by 3-4 cm broad in contrast to 4—4.5 cm long by 6-8 cm broad in P. gomezii; pedicels 1-2 mm long in contrast to 1.5-5 mm long in P. gomezii; acute stipule lobes; three-veined calyx lobes in contrast to lobes with no evident venation in P. gomezii; and farinaceous corollas in contrast to the glabrous corollas of P. gomezii. The epithet honors Sr. Luis Diego Gómez P., ANN. Missouni Bor. GARD. 77: 215-216. 1990. 216 Annals of the Missouri Botanical Garden FIGURE 1. Palicourea gomezii (Gómez et al. 20896 MO). - partial longitudinal section Director of the Museo Nacional of Costa Rica dur- ing a long and productive period. All flowers examined resemble the short-styled Rela- tively few specimens have been seen, however, so y р morph of distylous species of Palicourea. this species may be distylous. All the collections seen were collected in January or March and are simultaneously flowering and fruiting. Paratypes. COSTA RICA. HEREDIA: Rios Sucio y Hor dura, Parque Braulio Carillo, 500 m, Gómez-Laurito 6401 (CR); 13 km en la carretera a Guapiles O m, Gómez- Laurito 6423 (CR); 11 km E of Сайан, 10°16'N, 84°05'W, 1,060 m, Loiselle 258 (MO). -a. Inflorescence and node with leaves. —b. Flower, I thank the curators of MO and CR who kindly provided specimens for examination. | also thank oy E. Gereau for assistance with Latin compo- sition, and William Burger, John D. Dwyer, and arry Hammel for advice. LITERATURE CITED TAYLOR, С. М. 1989. Revision of Palic CHIA a a Mexico and Central America. Syst. Bot 26 102. . Mon. — Charlotte M. Taylor, Departamento de Biol- ogía, Universidad de Puerto Rico, Río Piedras, I Puerto Rico 00931, U.S.A. NOTES ON THE FLORAL MORPHOLOGY AND ECOLOGY OF MARGARITARIA DISCOIDEA (EUPHORBIACEAE) AT MUFINDI, TANZANIA Margaritaria discoidea (Baillon) Webster is a common and widespread African tree found in de- ciduous woodland, fringing forest, dry evergreen forest, rainforest, and disturbed vegetation at al- titudes from near sea level to over 2,000 m (Rad- cliffe-Smith, 1987). At Ngwazi, Mufindi District, Iringa Region, Tanzania (08°31'S, 3510'E, alti- tude 1,850 m, rainfall 850 mm/year, mean monthly temperature from 17.5°C in January to 13.5°C in June with occasional frosts) М. discoidea var. nitida (Pax) R.-Sm. occurs as a tree 5 m tall in tree clumps associated with termite mounds in grassland. This variety also grows at the edges of planted wattle breaks (Acacia mearnsii De Wild., Leguminosae: Mimosoideae) and flowers in mid- October at the end of the dry season but before the start of the rains, which are initially short, heavy thunderstorms. The flowers are produced before and during a flush of leaves, and before the stipules fall. Old leaves can persist until just before the new leaf flush. The trees are generally dioe- cious, although one individual was seen with female flowers on a predominantly male tree (Lovett 3248, DSM, К, MO). Armstrong & Irvine (1989) ob- served a similar occurrence in the dioecious My- ristica insipida R. Br. (Myristicaceae) of Queens- land, Australia. The Flora of Tropical East Africa (Radcliffe- Smith, 1987), followed taxonomically here, and Webster (1979) described Margaritaria as having four sepals and four stamens. However, in a sample of 1,000 flowers from a female tree (Lovett 3251, DSM, K, MO), 657 flowers had four sepals, 326 had five sepals, and 17 had six sepals. In a sample of 1,000 flowers from a male tree (Lovett 3247, DSM, K, MO), 8 had two stamens, 37 had three stamens, 926 had four stamens, and 29 had five stamens. On another tree a flower with six stamens was also seen. In male flowers the sepals are re- flexed in a square, so are not as visible and easy to count as the sepals on female flowers. Female flowers were visited by ants and honey bees (Apis mellifera), which are kept at Ngwazi for honey production. Associated species in the tree clumps include the trees: * Albizia gummifera (J. Gmelin) C. A. Smith var. gummifera (Leguminosae: Mimosoi- deae), *Apodytes dimidiata Arn. var. dimidiata (Icacinaceae), Bequaertiodendron magalismon- tanum (Sonder) Heine & J. Hemsley (Sapotaceae), *Bersama abyssinica Fresen. subsp. abyssinica var. abyssinica (Melianthaceae), Buddleja salvi- ifolia (L.) Lam. (Loganiaceae), Canthium lactes- cens Hiern (Rubiaceae), Сай edulis Vahl (Apo- cynaceae), *Cassipou rea (B ) ) Alston (Rhizophoraceae), *Catha edulis (Vahl) F orsskal ex Endl. (Celastraceae), Croton macrostachyus Del. (Euphorbiaceae), Cussonia arborea Hochst. ex A. Rich. (Araliaceae), *Cussonia spicata Thunb. (Araliaceae), Dais cotinifolia L. (Thymelaeaceae), *Diospyros whyteana (Hiern) F. White (Ebena- ceae), Dombeya rotundifolia Harvey (Sterculi- aceae), *Ekebergia capensis Sparrman (Meli- aceae), Erythrina abyssinica Lam. ex DC. subsp. abyssinica (Leguminosae: Papilionoideae), Ery- thrina lysistemon Hutch. (Leguminosae: Papilio- noideae), *Euclea divinorum Hiern (Ebenaceae), Flacourtia indica (Burman f.) Merr. (Flacourti- aceae), *Garcinia kingaensis Engl. (Clusiaceae), Heteromorpha arborescens (Sprengel) Cham. & Schldl. (Apiaceae), Maytenus cf. heterophylla Ecklon 8: Zeyher) N. Robson (Celastraceae), *Olea capensis L. (Oleaceae), *Olinia rochetiana Adr. Juss. (Oliniaceae), Osyris abyssinica Hochst. (San- talaceae), *Peddiea fischeri Engl. (Thymelae- aceae), *Prunus africana (Hook.f.) Kalkman (Ro- saceae), *Psychotria mahonii С. Н. Wright var. puberula (Petit) Verdc. (Rubiaceae), *Rapanea melanophloeos (L.) Mez (Myrsinaceae), *Roth- mannia fischeri (Schumann) Bullock (Rubiaceae), *Schrebera alata (Hochst.) Welw. (Oleaceae), *Syzygium guineense (Willd.) DC. subsp. afro- montanum F. White (Myrtaceae), Tarenna neu- rophylla (S. Moore) Bremek. (Rubiaceae), Teco- maria capensis (Thunb.) Spach subsp. nyassae (Oliver) Brummitt (Bignoniaceae), and *Tricho- cladus ellipticus Ecklon & Zeyher subsp. malo- sana (Baker) Verdc. (Hamamelidaceae). Climbers — ANN. MISSOURI Bor. GARD. 77: 217-218. 1990. 218 Annals of the Missouri Botanical Garden include: Asparagus setaceus (Kunth) Jessop (Lil- iaceae), Byrsocarpus orientalis Baillon (Connara- ceae), Clematis hirsuta Perrier & Guillaumin (Ra- nunculaceae), *Dalbergia lactea Vatke (Leguminosae: Papilionoideae), *Dracaena laxis- sima Engl. (Agavaceae), Jasminum goetzeanum Gilg (Oleaceae), Keetia gueinzii (Sonder) Bridson (Rubiaceae), Rhus longipes Engl. var. longipes (Anacardiaceae), Rhoicissus tridentata (L.f.) Wild rumm. (Vitaceae), *Rubia cordifolia L. subsp. conotricha (Gand.) Verdc. (Rubiaceae), Smilax aspera L. (Smilacaceae), and *Toddalia asiatica (L.) Lam. (Rutaceae). Shrubs include: *Clausena anisata (Willd.) Hook.f. ex Benth. (Ru- taceae), Myrsine africana L. (Myrsinaceae), and *Psychotria zombamontana (Kuntze) Petit (Ru- biaceae). Flowering at the same time as Margaritaria discoidea were Albizia gummifera, Apodytes dimidiata, Byrsocarpus orientalis, Dombeya ro- tundifolia, Erythrina abyssinica, Erythrina ly- sistemon, and Rothmannia fischeri. One kilometer away in Brachystegia woodland Brachystegia spiciformis Harms (Leguminosae: Caesalpini- oideae), Cussonia arborea, and Parinari curatel- lifolia Planchon ex Benth. (Chrysobalanaceae) were also flowering. Flowering just before the onset of the rains evidently avoids damage to the flowers by the initial heavy thunderstorms and allows de- velopment of fruits through the whole rainy season. Jacaranda trees (Jacaranda mimosifolia D. Don, Bignoniaceae) planted at Ngwazi and that flower at the start of the rains failed to set seed in 1986 and 1987 following heavy rains in 1985 and 1986, but did set seed in 1988 following the late and poor rains of 1987. Following a thunderstorm, many flowers from these trees are found knocked to the ground. Of the associated species, those marked with an asterisk (*) in the above list also occur in the moist forests of the Mufindi escarpment 12 km east of Ngwazi, where the rainfall exceeds 1,600 mm/ year. All these species are widespread Afromontane trees. Moist forest species occurring on termite mounds outside their normal climatic range have been observed in Ethiopia (Friis et al., 1987), and it seems likely that termite mounds offer a route whereby moist forest species can disperse across otherwise climatically unsuitable areas. We gratefully acknowledge constructive com- ments on an earlier version of this manuscript from Dr. C. Taylor and an anonymous reviewer. The Tanzania Commission for Science and Technology kindly gave us permission to conduct scientific re- search in Tanzania. These observations were made during the course of collecting plant samples for the National Cancer Institute. LITERATURE CITED ARMSTRONG, J. E. & A. К. IRVINE. 1989. Flowering, sex ratios, pollen-ovule ratios, fruit set, and repro- uctive effort of a dioecious tree, Myristica insipida Myristicaceae) in two different rain forest commu- nities. Amer. J. Bot. 76: 74- x: Frus, L, M. С. GILBERT & К. VOLLESEN. 1987. ditions to the flora of Ethiopia, 2. Willdenowia 16: 564. ~ RADCLIFFE-SMITH, A. 1987. Euphorbiaceae (part 1). In: M. Polhill (editor), Flora of Tropical East Africa. Balkema, Rotterdam. WEBSTER, G. 1979. A revision of Margaritaria (Euphorbiaceae). J. Arnold Arbor. 60: 403-444. — Jon C. Lovett, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S.A. (presently: Department of Botany, University of Dar es Salaam, P.O. Box 35060, Dar es Salaam, Tanzania) and Roy E. Gereau, Missouri Botan- ical Garden, P.O. Box 299, St. Louis, Missouri, 63166-0299, U.S.A. SISYMBRIUM LLATASII AND S. MORRISONII (BRASSICACEAE), NEW SPECIES FROM COASTAL PERU The genus Sisymbrium L. includes more than 90 species and is most highly diversified in South America, where nearly half of the species grow (Al-Shehbaz, 1988). It was treated on a worldwide basis by Schulz (1924) and has been recently re- vised for Argentina by Romanczuk (1982). AI. though several South American species remain to be described, the major taxonomic difficulty in Si- symbrium involves generic boundaries. The genus is highly heterogeneous and consists of several lin- eages of poorly understood phylogenetic interre- lationships. Schulz (1924) recognized these lin- eages as independent genera on the basis of minor morphological differences. 1 have adopted in this paper and elsewhere (Al-Shehbaz, 1988, 1989) a somewhat broad generic concept of Sisymbrium, and I believe that most of Schulz's (1924) South American segregates might not merit recognition. Sisymbrium llatasii Al-Shehbaz, sp. nov. TYPE: Peru. Lambayeque: Cerro Reque, 580 m, 28 Sep. 1986, S. Llatas Quiroz 2102 (holotype, F; isotype, СН). Figure 1. Herba annua glabra, 0.2-0.8 m alta; folia caulina petiolata, oblonga vel bubus. ag sa vel subacuta, repanda vel dentata, 2.5-7 cm longa, 0.7-2.5 cm lata; sepala oblonga, patula, 3.5-4 mm ae petala ем obo- vata, alba vel lilacina, nonunguiculata, apice rotunda, basi cuneata, 6-7 mm longa, 3.5-4 mm lata; pedicelli fruc- tiferi recti, divaricati, 9-13 mm longi; siliquae teretes, poor lineares, 2.1-3.9 cm longae, 1.7-2 mm latae; styli 0.5-1 mm longi; semina biseriata, oblongo-ovata, 1— 1.2 mm longa, 0.8-0.9 mm lata. Glabrous annual herb. Stems erect, 0.2-0.8 m high, branched above and below. Cauline leaves petiolate, glabrous, oblong to lanceolate, obtuse to subacute at apex, repand to obscurely dentate, cuneate at base, 2.5-7 cm long, 0.7-2.5 cm wide, gradually reduced in size upward; petioles 1-2 cm long. Inflorescences ebracteate, corymbose ra- cemes, elongated and lax in fruit. Sepals oblong, spreading, caducous, glabrous, narrowly scarious at margin, 3.5-4 mm long, 1.7-2 mm wide. Petals white to lavender, dark-veined, broadly obovate, not clawed, rounded at apex, cuneate at base, 6- 7 mm long, 3.5-4 mm wide. Stamens slightly tetra- dynamous; filaments erect, white, 3-3.7 mm long; anthers oblong, sagittate at base, recurved at apex after dehiscence, 0.7-0.9 mm long. Nectar glands well developed, confluent, ringlike, surrounding bases of lateral stamens, subtending those of me- dian ones. Fruiting pedicels divaricate to somewhat ascending, straight, glabrous, slender, much nar- rower than fruits, 9-13 mm long. Fruits terete, narrowly linear, straight, obscurely torulose, sub- sessile, 2.1-3.9 cm long, 1.7-2 mm wide; valves glabrous, rounded at both ends, with a conspicuous midvein and obscure lateral veins; style 0.5-1 mm long; stigma slightly 2-lobed; septum hyaline, not veined. Seeds oblong-ovate, orange-brown, coarse- ly reticulate, biseriately arranged, 1-1.2 mm long, .8-0.9 mm wide; cotyledons incumbent. Additional specimen examined. PERU. LAMBAYEQUE: Chiclayo, Cerro Reque, 540 m, 24 Oct. 1978, S. llatas [Quiroz] 331 (NY). Sisymbrium llatasii, which is named after its collector, resembles and perhaps is related to S. litorale Philippi, a species endemic to central Chile near Concepción. Sisymbrium llatasii is an annual with obscurely dentate to repand leaves, broadly obovate petals, slender fruiting pedicels 9-13 mm long, slender fruits to 2 mm wide, and biseriate, orange-brown seeds 1-1.2 mm long. In contrast, S. litorale is a suffruticose perennial with serrulate- denticulate leaves, oblong-spatulate petals, stout fruiting pedicels 2-3 cm long, stout fruits, and uniseriate, dark reddish brown seeds ca. mm long. Both are coastal, narrowly endemic species that are separated from one another by some 3,400 air kilometers. Sisymbrium morrisonii Al-Shehbaz, sp. nov. TYPE: Peru. Arequipa: 8 km S of Mollendo, silty flat, near sea, base of uplifted conglom- erate bench, 28 Sep. 1938, C. R. Worth & . L. Morrison 15729 (holotype, UC; pho- tocopy, А). Figure 2. Herba perennis basi lignosa, pili densis minutis - 0.3 mm longis; folia caulina lanceolata vel ovata, e ANN. Missouri Вот. Garp. 77: 219-222. 1990. FIGURE l. Sisymbrium llatasii. —a. Plant. —b. Flower. — c. Sepal. —d. Petal. —e. Stamen. —f. Fruit. —g. Seed. Scales a, Ї = 1 cm; b-e, р = 1 mm. Drawn from the holotype by the author. Volume 77, Number 1 Notes 221 1990 FIGURE 2. Sisymbrium morrisonii. —a. Plant.—b. Leaf. — c. Sepal. —d. Petal.—e. Fruit. — f. Portion of replum and septum showing venation. Scales a, b, e = 1 cm; c, d, f = 1 mm. Drawn from the holotype by the author. 222 Annals of the Missouri Botanical Garden dentata vel repanda, cuneata, 2-4 cm longa, 0.5- 1.5 cm lata; racemi ebracteati; sepala oblonga, erecta, glabra, .5-4 mm longa; petala alba, spathulata, 5-5.5 mm longa; pedicelli fructiferi recti, divaricati, 8-13 mm oe siliquae teretes, anguste lineares, subfalca tae, glabra 6.8 cm longae, 1.4-1.6 mm latae; septum ло styli tenues, p E mm longi; semina oblonga, uniseriata, 1.4-1.6 mm longa, 0.9-1 mm lata. w Perennial herb. Stem woody at base, annually producing herbaceous stems 10-25 cm high, densely and minutely pubescent with short, stiff trichomes 0.1-0.3 mm long. Cauline leaves peti- olate, ovate to lanceolate, dentate to repand, cu- neate at base, subacute at apex, 2-4 cm long, 0.5- 1.5 cm wide, reduced in size upward, densely pu- bescent along the petioles, sparsely so along the midrib and margins, glabrescent on both surfaces. Inflorescences ebracteate, corymbose racemes, elongated considerably in fruit. Sepals oblong, erect, caducous, glabrous, nonsaccate, 3.5-4 mm long, ca. 1.5 mm wide. Petals white, spatulate, narrowed to a clawlike base, 5-5.5 mm long, ca. 1.8 mm wide. Stamens erect, slightly tetradynamous; fila- ments 3.5-4 mm long; anthers oblong, sagittate at base, ca. 0.8 mm long. Nectar glands confluent, mE Е the bases of filaments. Fruiting p ‚ straight, а slender and much narrower than fruits, 8-13 mm long. Fruits terete, narrowly linear, somewhat ae divari- cate to ascending, 5-6.8 cm long, 1.4-1.6 mm wide; valves glabrous, smooth, conspicuously 3-veined; septum complete, prominently 3-veined; style slender, 1 -2 mm long; stigma subentire. Seeds oblong, coarsely reticulate, на arranged, 1.4-1.6 mm long, 0.9- Sisymbrium morrisonii is named after John L. Morrison, a student of the Brassicaceae and one of the collectors of the holotype. It is closely related to the Peruvian endemic 5. gracile Wedd. (= macrorrhizum (Muschler) J. F. Macbr.), from which it differs in having undivided leaves, larger flowers, slender styles, and longer fruits. Sisymbrium gra- cile has pinnatifid to pinnatisect middle leaves, smaller flowers, obsolete to stout styles less than 1 mm long, and fruits 3-4(4.5) mm long. Schulz (1924, 1936) segregated the latter (as 5. mac- rorrhizum) to Phlebiophragmus O. E. Schulz, which he distinguished from Sisymbrium solely on the basis of having 2- to 4-veined instead of l-veined or veinless septa. However, this feature is unreliable and does not justify recognition of Phlebiophragmus as an independent genus. Sisymbrium morrisonii is also related to the Peruvian S. oleraceum O. E. Schulz. The latter has shorter fruits 4-4.5 mm long, glabrous stems and leaves, yellow petals to 6.5 mm long, and stout styles less than 1 mm lon I am most grateful to Michael Dillon for sending the material of Sisymbrium llatasii for study, to the curators and directors of F, NY, and UC (ab- breviations follow Holmgren et al., 1981) for send- ing the loans, and to Barbara Nimblett for typing the manuscript. LITERATURE CITED AL-SHEHBAZ, I. A. 1988. The genera of Sisymbrieae (Crucifera ae; oo ч g 1 the southeastern United Sta Arb. 213-237. Са аге p die (Brassica- ceae), a new species from Peru. Ann. Missouri Bot. Gard. 76: 1176-1178 HOLMGREN, P. K. N K. SCHOFIELD. "n 81. uus кнн 7th edition. Regnum . 106: usce к, M. о ҮТ El género Sisymbrium (Cru- ciferae) en la Argentina. Darwiniana 24 E ScHULZ, O. E. 1924. Cruciferae-Sisymbrieae. In: A. m (editor), Püsnsentech IV 105(Heft 86): 1- 1936. Cruciferae. /n: H. Harms Wed S Ман: Pflanzenfamilien, ed. 2. 17B: 2 —Ihsan A. Al-Shehbaz, Arnold Arboretum, Har- vard University, 22 Divinity Avenue, Cambridge, Massachusetts 02138, U.S. А NOTICES THE 1989 Jesse М. GREENMAN AWARD The 1989 Jesse M. Greenman Award has been won by Carol А. Todzia for her publication **Chlo- ranthaceae: Hedyosmum,” which appeared in Flora Neotropica Monographs, Volume 48. This mono- graph is derived from a Ph.D. dissertation sub- mitted to the University of Texas, under the di- rection of Dr. Beryl B. Simpson. The genus Hedyosmum is comprised of 40 species of pre- dominantly montane, neotropical shrubs and trees. The comprehensive monograph, which includes four newly described species, reexamines previous treat- ments of the genus and presents new data on anat- omy, morphology, ecology, and geography. Syn- opses of the taxonomic history, palynology, cytology, and uses are also provided. This award is named for Jesse More Greenman (1867-1951), who was Curator of the Missouri Botanical Garden Herbarium from 1919 until 1943. A cash prize of $500 is presented each year by the Garden, recognizing the paper judged best in vascular plant or bryophyte systematics based on a doctoral dissertation published during the pre- vious year. Nominations for papers published during 1989 are now being accepted for the 22nd annual award, which will be presented in the sum- mer of 1990. Reprints of such papers should be sent to the Research Division, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166- 0299, U.S.A. In order to be considered for the 1990 award, reprints must be received by 1 June 90. Volume 76, Number 4, pp. 945-1192 of the ANNALS OF Bs MissoURI BOTANICAL GARDEN was published on November 7, 9, BOOK REVIEW Godfrey, Robert K. 1988. Trees, Shrubs, and Woody Vines of Northern Florida and Adjacent ы gia and Alabama. ix + 734 pp., illus., Univ. of Georgia Press, Athens, Georgia, & London. ISBN 0-8203-1035-2. Price: $50 (cloth). I have known, respected, and liked the writer for many years now and must state that it came as no surprise that the work is so good. It is equal to if not better than Godfrey’s previous floristic works, Trees of Northern Florida (with H. Kurz) and Aquatic and Wetland Plants of Southeastern United States (with Jean W. Wooten). Several factors make this new work as good as is: Robert Godfrey is one of the most perceptive people in botany, an indefatigable field man with a stack of full field notebooks reaching back to the 1930s. i fewer still have matched this taxonomist in inter- pretation of such field experience, and no other writer possesses quite the same style. Each clearly Few have exceeded him as a collector; written description is a seemingly effortless ren- dition of observations accumulated the hard way over decades. Where there is a doubt or a differ- ence or a wonder about something, that comes across. The work is very well edited, primarily because there has been such consideration of the thorough and factually entertaining style unique to this man. The illustrations are superb. They are botanical art in the best sense. The artist for most of them (Melanie Darst) has pleased the eye without sac- rificing morphological detail. The teaching value of such work deserves highest praise. The keys work. Prediction of the popularity of such a book is a safe bet. This reference, handsomely and durably bound, will be mandatory equipment for and good company on any field trip for woody plants within the Southeastern U.S.A.—Robert Kral, Director and Curator of the Herbarium, Box 1705, Sta- tion B, Vanderbilt ш Nashville, Теп- nessee 37235, U. ANN. Missour! Вот. GARD. 77: 223. 1990. Rolla Milton Tryon, Jr., and Alice Faber Tryon at Harvard University, October 1989. The portrait is of Daniel Cady Eaton (1834-1895), student of Asa Gray and first American pteridologist. A Festschrift in Honor of Alice Faber Ттуоп and Rolla Milton Tryon, Jr. CONTENTS FOR THE FESTSCHRIFT Introduction: A Festschrift in Honor of Alice Faber Tryon and Rolla Milton Tryon, Jr. David S. Barrington, Organizer William Jackson Hooker and the Generic Classification of Ferns Cathy A. Paris & David S. Barrington m Studies of Neotropical /soetes L. I. Euphyllum, a New Subgenus R. James Hickey Two New Species of Cnemidaria (Cyatheaceae) from Panama Robbin C. Moran Pteridophytes of the Venezuelan Guayana: New Species Alan R. Smith ................. Observations on Ctenitis (Dryopteridaceae) and Allied Genera in America Robert G. Stolze Defense Strategies in Bracken, Pteridium aquilinum (L.) Kuhn Gillian A. Cooper- river Pityrogramma calomelanos (L.) Link (Adiantaceae) on Layers of Volcanic Ash in Los Tuxtlas, State of Veracruz, Mexico Ramon Riba & Irma Reyes J. ............... Observations on the Reproductive Biology of Alsophila Species and Hybrids (Cyathea- ceae) David S. Conant Hybridization and Allopolyploidy in Central American Polystichum: Cytological and Isozyme Documentation David S. Barrington Electrophoretic Evidence for Allotetraploidy with Segregating Heterozygosity in South African Pellaea rufa A. F. Tryon (Adiantaceae) Gerald J. Gastony .............. Biosystematic Analysis of the Cystopteris tennesseensis (Dryopteridaceae) Complex Christopher Н. Haufler, Michael Р. Windham & Thomas A. Ranker ................ The American Paradox in the Distribution of Fern Taxa Above the Rank of Species ramer Recurring Hybrid Formation in a Population of Polystichum X potteri: Evidence from Chloroplast DNA Comparisons Diana B. Stein & David S. Barrington ..... 225 228 MODERN SYSTEMATIC STUDIES IN AFRICAN BOTANY Peter Goldblatt and Porter Р. Lowry 11, Editors Modern Systematic Studies in African Botany, a Monograph in Systematic Botany from the Missouri Botanical Garden, contains about 60 papers ranging in content from systematics, to ethnobotany, to conservation. The papers were presented at the Eleventh Plenary Meeting of the Association for the Taxonomic Study of the Flora of Tropical Africa held at the Missouri Botanical Garden, St. Louis, Missouri, U.S.A., on 10-14 June 1985. Topics include: Modern Systematic Studies, Madagascar and the Mascarenes, Pollination and Breeding Systems, African Ethnobotany, Lichenology and Bryology, Conservation in Africa, Contrib- uted Papers, and Flora Reports. O Monographs in Systematic Botany from the Missouri Botanical Garden Volume 25, 710 pages, 1988 Price $90.00 To place an order, send check or money order in U.S. funds, payable to U.S. bank; U.S. Shipments: add $1.50 for one book and $0.50 for each additional book; non-U.S. shipments: _ add $2.50 for one book, and $0.50 for each additional book. Orders should be prepaid; a $1.00 fee will be added to orders requiring invoices. No shipments are made until payment a а Mail form with your check or money order, payable to Missouri Botanical raen, to: | eN es: St. Louis, MO 63166-0299, U.S.A. — Please send __copy(ies) of Modern Systematic Studies in African Botany to: —— ame ES N, ES C Payment enclosed. — Address = Fp Gall воке 00 вав Бе | added to total Mie аы Me et x. New Species ix New Combination in Can ipylone ES c Pia New durar of Palicourea ica) i же Costa Rica CONTENTS A Century of Scientific Publications at the Missouri Botanical Garden George K. | Rogers pur Conserving Biological Diversity: Prospects for the 21st Century, the Thirty-fifth Annual — ‘Systematics Symposium of the Missouri Botanical Garden ros The Real Work of Systematics Michael E. Soulé E The Genetic Consequences of Habitat Fragmentation Alan R. Templeton, Kerry Shaw, Eric Routman & Scott K. Davis ил Conserving Biodiversity in the Canary Islands David Bramwell i omnes Integrated Strategies for Conserving Plant Genetic Diversity Donald А. Falk .... Conserving Botanical Diversity on a Global Scale David Given .........— Introduction: А Symposium on the Biological Diversity and Evolution of the Тым E _ Adaptive Radiation of the Hawaiian Silversword Alliance (Compositae- Madiinae): Eco-. i logical, Morphological, and Physiological Diversity Robert H. Robichaux, | - Gerald D. Carr, Matt Liebman & Robert W. Pearcy — Flavonoid Diversity i Relation to Systemati 1 Evolution of the Tarweeds | William д " J. Crins & Bruce A. Мы ы су ос к е. — -Biodiversity ge тыына of the Tarweeds (Asteraceae: Helantheae Mad | inae) d W. Kyhos, Gerald D. Carr & Bruce G. Baldwin . 2 з Chloroplast DNA Evoluti d Adaptive Radiation in tl ás AR | (Asteraceae-Madiinac) | Bruce G. Baldwin, Donald W. Kyhos Ф ж Dvorak DM ud "Evolution і in the Madiinae: ‘Evidence from bius ЖО p Mer > : New Tiebypteris (Thelypteridacene) from Central America Alan f: Smith o The Grasses + ашы Santa Cruz, Bolivia ноу J. Killeen cc e Narda: oad Combination i їп Thelypteris for bee: iios R. Smith — т Steyermarkochloa angustifolia (Sprengel) Judziewiez, A New Combination (Poaceae-Arun- | dinoideae-Steyermarkochloeae) Emmet J. Judziewicz ......... AAA Cestrum neblinense (Solanaceae), A New Species from тыш | William 6. d з v Y 7. : Carmen Benitez de Rojas ........ A € A New Vitex (Verbenaceae) from Madagascar | ge E E. Schatz . — ende А = pues of Mactolobium m (Fabaceae: аа from Mesoamerica > 'arucchi ..... M bod ad. i aceae) Blanca León .... Charlotte M. i po Volume 77 Number 2 Volume 77, Number 2 Spring 1990 Annals of the Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, con- tributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be accepted. Authors should write the Editor for information concerning arrangements for publishing in the Annars. Instructions to Authors are printed in the back of the last issue of each volume. Editorial Committee George K. Rogers Editor (this issue), Missouri Botanical Garden Amy Scheuler _ Managing Editor, Missouri Botanical Garden Glenda Nau Magdalen Lampe Publications Staff Marshall R. Crosby Missouri Botanical Garden Gerrit Davidse Missouri Botanical Garden John D. Dwyer Missouri Botanical Garden & Saint Louis University Peter Goldblatt Missouri Botanical Garden . . Dale E. Johnson : _ Missouri Botanical Garden Fw Henk 0З дег Werff | Missouri Botanical Garden | For eS асе contact Бы . Eleven, P.O. Box 299, St. Louis, MO 63166. Sub- .. scription price is $75 per volume U. S., $80 Canada - . and Mexico, $100 all other countries. Four i issues = p bul volume, die "e © Missouri Botanical Garden 1990 A A The ANNALS or THE Missouri BOTANICAL GARDE! - (ISSN 0026-6493) is published quarterly by Ye i Missouri Botanical Garden, 2345 Tower Grove AY enue, St. Louis, MO 63110. Second class po -~ paid at St. Louis, MO and additional mailing 0 3 POSTMASTER: Send address changes to pied . THE Missouri BoranicaL GARDEN, Depar МА чес ы Р.О. Box 299, St. Louis, MO ie | stage so Volume 77 Number 2 1990 Annals of the Missouri Botanical Garden A FESTSCHRIFT IN HONOR OF ALICE FABER TRYON AND ROLLA MILTON TRYON, JR. David S. Barrington, Organizer' With the exception of Dryopteris and Ela plagas: sum and a few small genera, not a single g really indescribable, the synonymy exceedingly com plicated and intricate, the variation of many species amazing. ... Domin (1929). Karel Domin’s lament, now 60 years old, provides the context for an appreciation of the immense contribution of Alice and Rolla Tryon to twentieth- century pteridophyte systematics. Armed with an prhausdivn knowledge and innate appreciation of ‚ supreme patience, and a remarkable sient Tor omnc in nomenclatural tangles, these two botanists have provided monographic treat- ments for an impressive number of taxa in the New World (e.g., A. Tryon, 1957, 1962, 1970; R. Tryon, 1941, 1942, 1955, 1960, 1970a, 1971а, 1976). These and related works have been included in a monographic synopsis of the entire diversity of New World ferns, with notes on their relation- ships to the remainder of the world's fern flora, in the monumental Ferns and Allied Plants (Tryon & Tryon, 1982). Domin's indescribable chaos has been substantially reduced to order. The Tryons' contributions have not been limited to the field of systematics. Each has aggressively pursued basic problems in the evolutionary biology of pteridophytes that arose during the monographic work. Alice Tryon has specialized in fern spore architecture and evolution (e.g., A. Tryon, 1964, 1971, 1972, 1985, 1987), while Rolla Tryon has focused on biogeography and its role in the evo- lution of ferns (e.g., R. Tryon, 1944, 1956, 1970b, 1971b, 1972, 1986; R. Tryon & Gastony, 1975). Inaugurated by the Tryons in 1970, the New England Fern Conference has been for 20 years a productive intellectual setting for students of fern biology to discuss the issues. А remarkable array of pteridophyte research projects has been con- ceived and developed in the seminar room at the Harvard Forest during the New England Fern Con- ferences. In addition to many of the research pro- grams represented in this Festschrift, the work on genetic load and homoeologous pairing (Klekowski, 1973a, b), biochemistry of fern spore germination (DeMaggio & Raghavan, 1972; DeMaggio et al., 1980), autogamous allohomoploidy (Conant & Cooper-Driver, 1980), and secondary compounds in pteridophyte systematics, ecology, and evolution (e.g., Britton € Widén, 1974; Cooper-Driver, 1980; Cooper-Driver & Haufler, 1983; Richard- son, 1984) were all discussed early in their de- velopment at New England Fern Conferences. ! Pringle Herbarium, Department of Botany, Burlington, ANN. Vermont 05405-0086, U.S.A. Missouni Bor. GARD. 77: 225-227. 1990. 226 Annals of the Missouri Botanical Garden Throughout these conferences, Alice and Rolla Tryon's interest in all aspects of fern biology has provided a stimulus and a focal point for a diversity of research efforts The Tryons have influenced the professional de- velopment of students in another way. Their field trips have introduced numerous professional bot- anists to the ferns of the American tropics. These trips, which exposed students to the nature of fern research in the tropics, have enriched the litera- ture, in many cases as the students” first experi- ences of research and publication (e.g., R. Tryon et al., 1973; A. Tryon et al., 1975; R. Tryon & Conant, 1975; R. Tryon « Vitale, 1977). ough their work continues apace, as the fas- cicles of the Flora of Peru appear (R. Tryon « Stolze, 1989a, b) and the authoritative monograph on fern spores of the world is readied for press (A. Tryon, 1990), we, pteridophyte biologists pro- foundly influenced by our contact with the Tryons, take this opportunity to honor Alice Faber Tryon and Rolla Milton Tryon, Jr., with a Festschrift on the occasion of their retirement from Harvard Uni- versity. Our contributions are as diverse as the realm in which the Tryons have been influential in pteridology: history of systematics (Paris & Bar- rington), taxonomy and phylogeny (Hickey; oran; Smith; Stolze), ecology (Cooper-Driver; Riba & Reyes J.), reproductive biology (Conant), evolutionary systematics (Barrington; Gastony; Haufler, Windham & Ranker), biogeography (Kra- mer), and molecular biology of hybridization (Stein & Barrington). Dr. , who has contributed many high-quality photographs to Tryon publications, provided the portrait at the end of this Festschrift. We thank the Tryons for their contribution to our professional development and wish them all the best in their present endeavors. alter H. Hodge LITERATURE CITED BRITTON, D. M. & € WIDÉN. 1974. Chemotaxo- nomic studies on parey ris from Quebec and east- ern Nor th Pp s ug J. Mx 52: 627-638. CONANT, PER-DRIVE Autoga- mous alohomoploidy i in. Usophila and Nephelea (Cy- atheaceae): a new hypothesis for speciation in hom- ора homosporous ferns. Amer. J. Bot. 67 1288. CooPER-DnivER, G. 1980. The role of flavonoids and related d in fern systematics. Bull. Torrey Bot. Club 1 116-127. € C Hau FLER. The changing dx of hen in fern classification. Fern ( 3-294. DeMace IO, A. E. € V. RAGHAVAN. 1972. Germination of brac ‘ken fern spores. Exp. Cell Res. 73: 182-186. —— ——, €. GREENE & D. SrETLER. 1980. Biochemistry of fern spore germination: glyoxylate and glycolate cycle activity in Onoclea sensibilis L. Plant Physiol. (Lancaster) 66: 922-924. Domin, К. 1929. The Pteridophyta of the Island of Dominica. E. Grégr & Son, Pra KLEKOWSKI, E. J., JR. 1973a. Genetic load in Osmunda аг зш ке Amer. J. Bot. 60: 146-154 xual and subsexual кайа in the onera аз a new hypothesis. Amer. J. Bot. 60: 535-544 RICHARDSON, Р. М. 1984. The taxonomic significance of xanthones in ferns. Biochem. Syst. Ecol. 12: 1-6, Tryon, A. К. A revision of the genus Pe ага section Pillada Ann. Missouri Bot. Gard. 44: 193. 1962. A monograph of the fern genus Jame- sonia. un Gray Herb 1: 109-197. 964 Platyz гота а о a with ven heterospory. Amer. J. Bot. 51: 93 1970. A monograph of the iun genus Er- iosorus. Contr. Gray Her 54-174. 1971. Structure and variation in д of Thelypteris к Rhodora 73: 444-46 972. Spores, chromosomes, and ыш of ше fern Pe Пава баны, а. Rhodora 74: 220- = . Spores of myrmecophytic ferns. Proc. Roy. sor nur 80B: 10 19 Stasis, diversity, ad on an electron microsc iid study D p e Pter- T Linn. Soc. Symp. Ser. 12: 233-249. 90. Spores of the Preridophyta. pum n s Yor — — —., H. P. Bautista & 1. S. ARA AUJO. 1975. Chr mosome studies of Brazilian ferns. Acta Amazonica 5: 35-4: Ткүом, R. M., Jr. 1941. A P p of the genus Pterid- ium. Rhodora 43: 1-31, 61. (Contr. Gray Her 134 and function in spores A revision of the genus Doryopteris. Cons Giy Herb. 143: 3-80. 19: Dynamic phytogeography of Doryop- eens J. Bot Я Selaginella pupe stris and its allies. Ат Misour Bot. Gard. 42: 1-99. 956. A revision of de American e of я О Contr. Gray Herb. 179: 1-10 1960. review of the ere ш in America. Contr. Gray Herb. 18 3-52. . 19 The б ation of С z SN Contr. Gray Herb. 3-53. 1970b. лш апа evolution a fern feta on oceanic islands. Biotropica 2: 1971a. The American tree ferns “allied to Sphaeropte ris horrida. Rhodora . 1971b. The process of evolu сна migration in species of Selaginella. Brittonia 23: 89-100, 19 is ndemic areas and geographic с гй tion in tropical American со я 4: 121- 131. 976. А revision of the genus Cyathea. Contr. Gray Herb. 206: 19-98. 1086. The biogeography of spec ies ns ^d cial reference to ferns. Bot. Rev : 117 . CONANT. 1975. The ferns of Fu TN Грае Аш Amazonica о: 23-34 Volume 77, Number 2 1990 Barrington 227 Introduction & С. J. Gastony. 1975. The biogeography of endemism in the Cyatheaceae. Fern Gaz. 11: 73- 79, & К. С. STOLZE. I. 1. Ophioglossaceae — 12. Bot. N. S. 20: 1- Tu 1989a. Pteridophyta of Peru Cyatheaceae. Fieldiana, & 989b. Pteridophyta of Peru II. 13. Pteridaceae — (a Dennstaedtiaceae. Fieldiana, -128. Bot. N. S. 2 A. F. Tryon. 1982. Ferns and Allied Plants. Springer-Verlag, New York. — & L. Vitale. 1977. Evidence for antheridi- ogen production and its mediation of a mating system in natural populations of fern gametophytes. J. Linn oc., Bot. 74: 243-250. ‚ В. VOELLER, A. TRYON & R. Ripa. 1973. Fern biology i in Mexico (a class field program). BioScience 23: 28-33. WILLIAM JACKSON HOOKER AND THE GENERIC CLASSIFICATION OF FERNS' Cathy A. Paris? and David S. Barrington? ABSTRACT Twentieth- century pteridologists (e.g., Copeland, Pichi Sermolli) have criticized the work of William Jackson Hooker, K 841- Director of 65) and author was regressive, we compared it his subgenera and sections are similar to their genera. Кее, based on a comparison of the aut treatment of the group, but his work was alone had a substantial impact ог effectively challenged until the turn of the centur of Species Fi regressive and that he impeded 1 9th-century progress in pteridology. to his large genera based almost exclusively on characters of the sorus. To the contention that Hook with those of his contemporaries Presl, referred to use reproductive characters to circumscribe genera, he ma delimit natural groups within them. His large genera are i aaa to Ms iu of other authors of his y els in his time. Of the four did -century pteridology; only his system was widely a of not Ed claiming that his fern classification was rticular, Hooker's critics have objected Fée, eie Smith ood use of vegetative features to ay, an r's system as modern as those of Presl and е polypodivids Smith produced the most insightful authors we discuss here, Hooker adopted, and it was not e that many of the issues that surrounded generic ay. classification in Hooker's time, e.g., ot size of genus to accept, are still unresolved toda In the time of Smith and Fée and such contemporar- ies as Kunze, immediately preceding the Origin of Species, pteridology was perhaps the most yt anced division of botany. nity w common use, and, except that it gioi a арга! foun- dation, it seems to have meant very much чон t it means today, Under the inc Hm of Hoo lvanced position, and in forty years became one of the most backward divisions. With these words Copeland (1947) summarized the response of many 20th-century pteridologists to the life and work of William Jackson Hooker, 19th-century author of the monumental Species Filicum (1844-1864) and numerous other works on fern classification. Although Copeland's com- ments stand out in the severity of the criticism, they are by no means atypical. Pichi Sermolli (1973), for example, wrote: If Presl did not receive the attention he deserved, E was due to the influence exercise hat time '. J. Hooker, whose preference for large genera, based solely on the characters of the sori and the sporangia, is well known to all pteridologists. In general, Hooker's classification has been called regressive with respect to those of his contemporar- ies. To determine whether this is accurate, we compared Hooker's classification with those of Presl, Fée, and Smith, all of whom history has treated more favorably. The systems were compared with reference to hierarchical organization, characters used at various hierarchical levels, and the extent to which each author succeeded at producing a natural classification of a problematic group, the polypodioid ferns. We present evidence that Hook- er's modern critics have judged his work unfairly: Hooker's classification was as insightful as those of his more highly regarded contemporaries, but he has been criticized because he adopted a broad genus concept. ooker's system has also been described as an impediment to progress in pteridology. We ex- plored this idea by evaluating the impact of the various mid-19th-century fern classifications. To ! This paper was developed by the first author for a course in the History of Botanical Systematics, offered by Dr. oung, and an anonymous reviewer Harvard University Herbaria libraries was most helpful. We especially thank Dr. Rolla T Fern Classification" was the sou wledge of the history of pteridology and his critical insight on the development of fern phylogeny and 1952 paper “A Sket His extensive E ch of the History of thank him for helpful comments on early drafts of the m in the 19th century made discussions with him during the development of this work es specially valuable. Rolla Alice Tryon extended their generous hospitality and access to their private library durin g preparation of this ты ? Pringle Herbarium, Department of Botany, University of Vermont, Burlington, ao 05405, U.S ANN. Missouni Bot. Garb. 77: 228-238. 1990. Volume 77, Number 2 1990 Paris & Barrington 229 W. J. Hooker and the Generic Classification of Ferns do this we determined (1) how widely the various authors were followed in floristic treatments of the day, and (2) the effect of Hooker’s published work on the volume of subsequent pteridological research in the 19th century. Here we concur with the critics: Hooker’s scheme was followed more widely than any other mid-19th-century classification, and the volume of systematic work produced in pteri- dology was reduced following Species Filicum. To put the problem in context, we first present a brief account of Hooker’s life and the intellectual climate in which he worked. A SYNOPTICAL ACCOUNT OF THE LIFE AND Work OF WILLIAM JACKSON HOOKER William Jackson Hooker (1785-1865), born into a well-to-do Norwich family and raised in the tradition of the gentleman naturalist, made his first contributions to botany in the field of bryology. His early work on mosses brought him to the attention of Joseph Banks, Robert Brown, and other prom- inent naturalists of his day, and prompted Hooker’s election to the Linnaean Society in 1806. For additional information on Hooker’s early life, the reader is encouraged to consult Allan’s (1967) ex- cellent biography. ooker first turned his attention to the ferns during his tenure as Regius Professor of Botany at Glasgow University. His first major contribution to pteridology was Genera Filicum (1838-1842), a set of engravings by Bauer with accompanying text by Hooker. The circumscription and definition of the fern genera in that work follows the system proposed by the Czech pteridologist Presl in his Tentamen Pteridographiae (1836). In 1841, Hooker was appointed Director of the Royal Gardens at Kew. During his 25 years as director, Kew was built up from the Queen’s kitch- en garden to one of the world centers of botanical parable library and her- barium, acres of glasshouses and outdoor gardens, an arboretum, and a museum of economically im- portant plants. As Director of Kew, Hooker also did much to promote the study of botany in Great Britain. For instance, through his contacts around the world, often his former students at Glasgow or connections in business and government, Hooker amassed an enormous private herbarium, which he made freely available to the botanical community. Hooker secured his role as eminent pteridologist with the publication of Species Filicum, the mag- num opus of his pteridological career. Published in five volumes between 1844 and 1864, the work included detailed descriptions of all known ferns, research, boasting an i including distributional information, collection numbers, and illustrations. In Species Filicum Hooker abandoned the precedent set by Presl (1836) and produced his own classification of the ferns. Wishing to present a condensed version of the work that could be used in the field, Hooker began Synopsis Filicum in 1865. In August of that year, after just 22 pages of the manuscript had been submitted, Hooker died of diphtheria, then epidemic at Kew. Synopsis Filicum was com- pleted by Baker, an assistant curator at the Royal Gardens, who finished the manuscript using Hook- er’s notes, herbarium collections, and an annotated copy of Species Filicum. The Synopsis was fin- ished in 1868; a second edition containing an ap- pendix with many species newly described by Baker was published in 1874. HOOKER’S MILIEU: FERN CLASSIFICATION IN THE Mip-19TH CENTURY The years in which Hooker produced his major works on fern classification, 1844 to 1865, follow the death of Linnaeus (1778) and mostly precede the 1859 publication of Darwin’s Origin of Species. In this period, the Natural Method of classification developed by M. Adanson, B. de Jussieu, and A.-L. de Jussieu had become well established in flowering plant taxonomy and had largely displaced the Lin- naean Sexual System. Under the Natural System, characters from all parts of the plant body came to be used to define plant groups and to assess relationships among them. In contrast to Lin- naeus’s system, which emphasized characters of the reproductive body (the “fructification””), no characters were to receive a priori weighting; all were considered potentially informative. The value of any particular character was determined only following an evaluation of the constancy of that character and its correlation with other characters in the group of interest (Tryon, 1952). A natural taxon, then as now, was one that had a real existence in nature. However, in the years before Darwin, relationships among species and higher taxa were interpreted, not in the genealog- ical sense, but as variations on a pure form that existed in the mind of the Creator. The goal of Natural Classification was a system that revealed the Creator’s mind as evidenced in His creation. In the first half of the 19th century, changes in the approach to fern classification were taking place that paralleled those in the classification of flowering plants. At the start of the century pter- idology was still under the influence of the Linnaean system. The major classifications produced around 230 Annals of the Missouri Botanical Garden that time, Sir J. E. Smith's ‘Tentamen botanicum de Filicum generibus dorsiferarum” (1793) and Olaf Swartz's 1806 handbook Synopsis Filicum, were based almost exclusively on the shape and position of the sorus and the indusium. Using those characters, Smith and Swartz arranged the several hundred known fern species into 20 and 38 genera, respectively. By 1836, however, a new approach to the clas- sification of ferns was evident. That year marked the beginning of 40 years of great activity in pter- idology, a period in which the old Linnaean methods were set aside and the Natural System was adopted in fern classification. Four men figure prominently in this period: Karl Boriwog Presl, Antoine Laurent Apollinare Fée, John Smith, and William Jackson Hooker. In 1836, K. B. Presl of Prague published the Tentamen Pteridographiae. His system relied on attributes of the rhizome, petiole vasculature, in- dument, venation, and size and form of the paren- chyma cells, as well as those of the sorus. The use of many characters, both reproductive and vege- tative, permitted Presl to resolve 112 fern genera among the polypodioid (sensu lato), cyatheoid, and gleichenioid ferns. Subsequently, Presl (1843, 1845, 1847, 1852) expanded his classification to include the rest of the homosporous ferns, bringing the total number of his genera to 176 (Tryon, 1952) A second classification of ferns published during the middle years of the 19th century was the Gen- era Filicum (1850-1852) of A. L. A. Fée, a French pteridologist working at Strasbourg. In that work, Fée treated 181 genera of polypodioid and cy- atheoid ferns. The characters he used were pri- marily those used by Presl (1836), but Кее also noted the number of cells in the annulus and spore shape. These new characters were more useful in the delimitation of genera than of higher groups, so the resultant classification, while more elaborate than Presl’s, did not substantially improve upon it (Pichi Sermolli, 1973). John Smith, a gardener at Kew between 1822 and 1864, was another important contributor to pteridology in this period. His knowledge of ferns, based on daily observations of living plants in cul- tivation at Kew, led Smith to publish his ideas on their relationships in a treatise entitled “Ап ar- rangement and definition of the genera of ferns" (1841-1843). In that work, he used combinations of vegetative and reproductive characters to define 143 genera of filicalean ferns. Although Smith developed his system without having seen Presl's 1836 Tentamen, the overall structure of the two classifications and the number of genera recognized in each were remarkably similar. In 1875, Smith published his second major classification, Historia Filicum, reflecting the experience of more than 30 years since “An arrangement and definition of the genera of ferns" was d nop Using a combi- nation of vegetati cters, including vernation of the croziers, rhizome habit, petiole articulation, and leaf venation, together with the reproductive characters mentioned previously, Smith defined 220 genera of ferns, many constituted as they are to- da у. The fourth major figure in mid-19th-century pteridology was William Jackson Hooker. Hooker completed most of the the work on his Genera Filicum during his years at Glasgow. There is no evidence that he had great experience with the ferns at that time, so perhaps it is not surprising that he adopted Presl’s circumscription and ar- rangement of fern genera. Of the 135 genera in- cluded in Genera Filicum, the circumscription of 115 followed Presl (1836); the 20 not included in the Tentamen were adopted from Smith (1841- 1843). Hooker’s admiration of Presl's work is ev- ident in the preface of the Genera Filicum, in which he praises the Tentamen for the accuracy of the research and for the “clear and perspicuous” arrangement of the genera presented therein. That he nevertheless had some reservation about Presl’s genera and the characters used to circumscribe them is evident when he wrote that perhaps too much emphasis is placed on the number and po- sition of vascular bundles in the petiole, and that venation **holds too prominent a place in the ge- neric character" (Hooker, 1838-1842). Hooker reserved for himself the right to reconsider his endorsement of Presl's genera as he prepared Species Filicum (Hooker, 1838-1842). That Hooker did change his mind substantially is evident in the classification he presented in those volumes. In Species Filicum, he circumscribed genera al- most exclusively on the basis of soral characters. As a result, the 135 genera of the Genera Filicum were reduced to 63 The substantial change in Hooker's approach to generic classification between the completion of Genera Filicum and the initiation of Species Fil- icum is attributable to his change of mind about the kinds of characters useful in circumscribing genera. As a Glasgow professor, Hooker had en- dorsed Presl's use of vegetative characters; as chief botanist at Kew, he relied almost entirely on char- acters of the fructification. Although Hooker never published an explicit justification of his revised ap- proach to fern classification, it appears that con- Volume 77, Number 2 1990 Paris & Barrington 231 W. J. Hooker and the Generic Classification of Ferns venience was a major factor in his decision (Tryon, 1952). As curator of ferns at Kew Gardens, then the largest botanical research center in the world, Hooker had the task of processing the enormous volume of collections arriving at the herbarium from around the world. Given that task, it is no wonder he preferred to use characters that could readily be evaluated from herbarium specimens. Characters of the sorus, especially shape and po- sition, are easy to evaluate on herbarium speci- mens, whereas those of the petiole vasculature often are not. That it was specifically convenience Hooker was seeking in his choice of reproductive characters, and not some Linnaean concept of the essential nature of those characters, is evident in his comment that Presl had “laid too much stress on the number and other circumstances connected with the bundle of vessels in the stipe, which in the Herbarium are difficult of investigation” (Hook- er, 1838-1842 Hooker was also concerned that his works on fern taxonomy be useful to the amateur. Having argued against the usefulness of annulus position as a generic character, he went on to say that it would be inconvenient to maintain even a group or a section on the basis of that character in “а work whose main object is to assist the tyro in the verification of genera and species” (Hooker, 1838- Was HOOKER’S CLASSIFICATION OF THE FERNS REGRESSIVE? To determine whether Hooker’s 20th-century critics have been justified in calling his work re- gressive with respect to that of his contemporaries, n er & Baker, 1874), and Historia Filicum (Smith, 1875). Particular attention was given to the hi- erarchical structure of each author’s classification and to the characters used at various levels in the hierarchy. To determine how well each author’s chosen characters resolved natural relationships, we made a detailed comparison of the authors’ treatments of the polypodioids. This group was chosen first because its circumscription and defi- nition has a history of confusion, and second, be- cause there is now a reasonable consensus on how the polypodioids should be treated. HIERARCHICAL STRUCTURE To compare the hierarchical structure of the four classifications, we first reviewed each of the four principal works to determine which taxonomic categories were used and how they nested. To make the comparison, we ranked the supraspecific cat- egories by number, with I being the fundamental division of the Filicopsida and so forth (Fig. 1). We found that the authors differed substantially in their use of categories. This is perhaps not surprising, since the use of taxonomic categories had not yet been stabilized by the International Code of Bo- tanical Nomenclature (Lawrence; 1951). For ex- ample, the rank “section” was used above the genus level by Presl, below it by Hooker, а ү both above and below the genus level by Smith. We then compared the position of the genus in the hierarchy, and the use of subordinate categories to resolve н o (Fig. 1). Notice that, whereas both s and Fée's genera are at approximately the same level in the hierarchy, reflecting their similar concepts of the genus, Hook- er's genus occupies a higher level. He makes fre- quent use of infrageneric categories (often un- named) to designate natural groups within the genus, as many as four in the case of large genera such as Polypodium. Smith, by contrast, used the genus at a much lower level in his hierarchy and never used more than one infrageneric category above species. CHARACTERS USED AT EACH LEVEL IN THE HIERARCHY To determine what kinds of characters the au- thors used to circumscribe taxa at each rank, we ordered the characters according to the frequency with which each was employed at a given rank, omitting those characters that were seldom used (Fig. 1). We found that, with one exception, all four authors used characters of the reproductive body at the highest levels of their classifications. Smith was exceptional in his use of two vegetative characters, internode length and articulation of petiole to rhizome, to establish his fundamental division of the ferns. Although these characters proved to be powerful in elucidating relationships among genera and tribes, Smith's use of them at the highest level overemphasized their importance. His decision led to an improbable alliance of all species with leaf bases articulate to the rhizome, a distinct flaw in a system that was otherwise fairly insightful In their circumscription of genera, Presl, Fee, and Smith all used vegetative features such as venation, leaf architecture, and rhizome structure (Fig. 1). Hooker, however, made infrequent use of such characters. All of his genera were circum- 232 Annals of the Missouri Botanical Garden S FÉE HOOKER SMITH ANNULUS POSITION ANNULUS POSITION ANNULUS POSITION PETIOLE ARTICULATION RNATION INTERNODE LENGTH SPORANGIUM DEHISCENCE subordo -atae suborder division INDUSIUM PRESENCE SORUS SHAPE INDUSIUM PRESENCE ANNULUS POSITION /ABSENCE /ABSENCE cohors -taxiocarpeae -ae (unnamed) А INDUSIUM STRUCTUR SORUS SHAPE, POSITION SORUS SHAPE, POSITION INDUSIUM PRESENCE SORUS SHAPE, POSITION INDUSIUM PRESENC NDUSIUM STR / ABSENCE SPORANGIUM STAL 11 PRESENCE/ ANNULUS POSITION, FORM tribus -eae tribe (unnamed) * SORUS SHAPE, POSITION INDUSIUM STRUCTURE SORUS е int SORUS SHAPE, POSITION INDUSIUM STRUCTURE SORUS POSITION SPORANGI INDUSIUM STRUCTURE, INDUSIUM PRESENCE LEAF е PRESENCE/ABSENCE IV INDUSIUM POSITION LEAF ARCHITECTURE HABIT HABIT sectio -eae genus tribe VENATION VENATION PETIOLE ARTICULATION VENATION INDUSIUM STRUCTURE LEAF ARCHITECTURE SORUS POSITION SORUS POSITION V INDUSIUM STRUCTURE genus] enus series series LEAF ARCHITECTURE SORUS SHAPE, POSITION VENATIO VENATION VENATION INDUSIUM STRUCTURE SORUS sesion POSITION SORUS POSITION LEAF ARCHITECTURE LEAF ARCHITECTURE VI INTERNODE LENGTH INDUSIUM STRUCTURE section (unnamed) I section section VENATION - LEAF ARCHITECTURE BIT INTERNODE L LEAF ARCHITECTURE SORUS SHAPE, POSITION VII LEAF ARCHITECTU VENATION RECEPTACLE FORM (unnamed) (not used) (unnamed) * genus VENATION - LEAF ARCHITECTURE INTERNODE LENGTH VIII LEAF ARCHITECTURE (unnamed) (not used) (unnamed) + section FIGURE a Comparison of the hierarchical organizations of the aria produced by our l9th-century authors. Cha The genus is highlighted in bold. 18 (1850 cters are 52); Hooker = Hooker ranked according to the frequency with which they w The classifications being c & Baker (1874); Smith = scribed mostly with reproductive features, presum- ably because of the ease with which these could be evaluated. In his use of generic characters, Smith is again exceptional. Basing his genera on a re us ompared are as follows: Smith (1875). ed at ea Presl = ch level in the тайин Presl (1836); Fée = Кее combination of reproductive and vegetative fea- tures and weighting them equally, Smith more nearly approximated the ideals of the Natural System than did his contemporaries. Volume 77, Number 2 90 Paris 8 Barrington 233 W. J. Hooker and the Generic Classification of Ferns RESOLUTION OF NATURAL GROUPS: THE POLYPODIEAE To evaluate the success of our authors in re- solving natural groups, we compared their treat- ments of the polypodioid ferns. In each treatment that alliance was defined as the group of ferns with round, exindusiate sori; by choosing such a broadly defined assemblage, we had the opportunity to de- termine how well each author used vegetative char- acters to order variation within the alliance. In particular, we attempted to determine how well the 19th-century pteridologists drew the distinction be- tween the polypodioid ferns and the phegopteroids, a group aligned with the thelypteroids in modern classifications (Copeland, 1947; Tryon & Tryon, 1982). The two share the character of round, exindusiate sori, but the phegopteroids differ in a suite of other important attributes, such as petiole articulation, leaf indument, and spore morphology. ineteenth-century treatments of the polypo- dioids were compared with reference to two criteria: (1) Did the author in question avoid the inclusion of extraneous elements within his Polypodieae? In particular, did he draw the distinction between the polypodioids and the phegopteroids? and (2) Did he assemble the polypodioids—as we currently un- derstand them—together in one place, or were they scattered among other tribes? Our analysis was made with reference to Copeland (1947) and Tryon & Tryon (1982), two well-known and widely fol- lowed 20th-century treatments. Within Presl’s sect. Polypodieae, elements of diverse affinities are intermixed, sometimes without apparent pattern (Table 1). Phegopteris, for ex- ample, is included in Presl’s genus Polypodium. This is surprising, since he was alert to the problem of distinguishing the exindusiate aspidioids (i.e., the phegopteroids) from the polypodioids (Presl, 1836, р. 167). Using round, exindusiate sori with punc- tiform receptacles located on veins, Presl assem- bled in the tribe Polypodieae approximately half of the elements placed there by modern authors. Oth- er polypodioids were included in his Acrosticheae (those with sporangia distributed over the abaxial surface of the leaf—a character that has arisen in several different lineages), Grammitideae (those with more or less elongate receptacles), and Taenitidae (those with sporangia borne on long, inframarginal commissures). Fee followed Presl quite closely, as is evident in their treatments of the polypodioids (Table 1). Like Presl, Fée included dipterids, dryopteroids, gym- nogrammoids, and thelypteroids within his Poly- podieae. Fée recognized Phegopteris as a separate genus, following Polypodium. Also like Presl, Fée included about 50% of the modern polypodioids in his Polypodieae. Because he accorded great im- portance to the position of the sporangia relative to the fertile veins, the remaining polypodioids are scattered throughout tribes of diverse affinity in Fée's system. The monotypic Polypodieae of Hooker com- prised Polypodium, which included more than 400 species. Within Polypodium, however, Hooker made extensive use of infrageneric categories, en- abling him to circumscribe some reasonably natural groups. For instance, by drawing an early distinc- tion between the Eremobryoid series, with ar ticulate petiole bases, and the Desmobryoid series, without them, he appropriately separated Phegop- teris (as a section) from Polypodium. He placed true polypodioids in the Eremobryoid series, in sect. Eupolypodium. Although several grammitidoid species are also included here, they are a cohesive group of species placed at the beginning of the section, indicating that Hooker recognized their distinctiveness. His emphasis on sorus shape led Hooker to place a number of polypodioid genera in the Grammitideae (the next tribe in his sequence, with sori at least twice as long as broad) or the Acrosticheae. Nevertheless, he correctly classified almost two-thirds of modern polypodioid genera. Hooker did not always circumscribe groups so suc- cessfully, however. For example, his Grammitideae comprised a diverse assemblage of thelypteroids, cheilanthoids, vittarioids, and misplaced polypo- dioids. By far the most modern treatment of the Po- lypodieae was that of John Smith. Although he mistakenly included two grammitidoid genera, his tribe was a remarkably natural assemblage, com- prising more than 80% of the modern polypodioids. Smith's use of petiole-base articulation at the high- est level in his classification was in this case an advantage: it separated Phegopteris from the po- lypodioids right from the start. We conclude that Hooker's perception of nat- ural relationships was at least as advanced as Presl's and Fée's. This is evident in a comparison of the three treatments of the phegopteroids relative to the polypodioids. All three erroneously included the phegopteroids in their Polypodieae. Hooker, how- ever, indicated the distinctiveness of the phegop- teroids by placing them, together with several other exindusiate thelypteroids and dryopteroids (taxa that have until recently been placed together in the Aspidieae), at the beginning of the Polypodieae. Furthermore, by placing them immediately follow- ing the Aspidieae, Hooker apparently implied an Missouri Botanical Garden Annals of the 234 sue dogeqq ‘EZI (1) aea1ojdoSouq —) "129g oeapuojdoSoeuq сє] equ] ([еәцзәл snpnuuy) ‘| VAYEOWSA ‘П NOISIAIG PUIWIBIZOIN “ZTE sujedeuy Те snsXorday 1б (d) sidajoydo7 'oz (d) ararjados[g — 9 1998 (рәпициоэ) с̧281 'Чишс ѕшәшојәціту “GZ (d) eydioworsy "$c (d) umnÁqoejsoKiq ‘ez (d) eueu&iq “zz (4) әеиеиќі] — С "1998 suejdoue»eT “[Z (4) sisdoydiy :0с (d) sisdoyeuIdyg “61 (4) оа “at d cr яа eanes “21 (д) кы аса —@ 1999 eudesoig Ss (9) avapoddjog — [ “1998 avaipoddjog “€ equr ([£9u1e4 sn[puuy) “y УАНЧОЙЯНЧЧ I NOISIAIG S/gI ‘KUS eueuAXq ++ (suejdiq = (peureuun), x (umipuno[q = 'peureuun) ++ (peureuun) 4- (peureuun), (suredoa]4 Surpnout) (snjoqoydin = uomauo¡Ádure”) 1998 umipoqo[qq "1935 umiqay domos) 1998 umipodA¡odny :1998 sous pio&1qouise( y umipodA[oq “gf averpod4jod "TX 9quip әрұвләп[олшхя “Y avaoeipoddjog үү 19p10qng 281 exeg Y 19xooH TET ELEELEA E X — ~ — — UOÁYIBISOÁAL] “9Z1 suaidiq `65@1 ширипәд “Pcl ецеиќі ‘EZI UINIIOSOLMIA “ZZ sədo pq "IcI eydioworpszsy ‘OZI ‘oll euepedser) 'gII snjoqoydin “LTT ѕшәјіоиғэәт “OTT — — — — — — 2 E o [ыл a © Y Р ey = ы ~ ~ — uoinauo|Adure?) ‘стт umtqopqdomo?) ТІ sueidomo2) ‘етт sueidoSeugd “ZI E 3 3 © 2. = © A — — =ч A a pa 2 D E E Е 5 “© — a C5 (5 B E Oy Ba А Ge А А А АА, А СА, М2 М5 МУ МУ МУ МУ МУ МУ ae чш sueidoose[q “SOT ~ avatpoddjog `Z BLIOSQUUIÁS) “y aead.1esorxe |, 3812143019418) sasyerpodÁ]o gy 7581-0581 `әәя uopourul[e?) “gg (9) suajdoueo»eT “18 (4) aeapia1doue9a7] "II 00995 suajdoiKog UONI (d) sniogdo[o&^?) иоцэәс (4) snjoqoydiy иоцэәс (4) snjoqoydin “98 ummudésq “Gg — (4) eueuÁXT иоцэәѕ (d) штрипәү uonooeg (d) sapoyeurdydny uonoog (d) ѕәројешАц “PQ suadoAowq eg (№) sujedoeqd 38 (4) umaneuo|Ádure?) *18 (4) шешер uong (4) WINIUOZOINe[q uon2ec (4) eueuiZieW 08 umiqaydomos “62 (4) етшәиәоәјд `// (V) suajdomo-) :9;7 (1) suaido33y g uonoeg (1) stuajdouay) uonoag umrmodÁ[oq с; aearpodá[oq `I] 0829S әвәәвтроЧАтод "А SOUL ABIJOSQUUIÁS) J] s1040) әвуелАЗотәцтег) ']I Opsoqns 9681 “89:4 'suoneor[qnd [euiguo əy} ш sioyine ayi Aq pasn әѕоці әле (yy ‘+ 8:9) sjoquiAs Jayjo pue siaquinyy “aes9epuardig = (q) їәвәзрриәздоА щт] = (y) “aea9epila1dÁA[9y = (1) “Sea9epunuuels = (9) (2861 'uoKip x чоА1ү nsues) aeatpoddjog = (d) (2861) U0A1] x» U0A1] 10/pue () $61) puefedoy ut 1ueureove|d s ПОХР] juasaida sosoqjuaied ш sia1197 `цогшЧо oneurejsÁs 1цәзә1 0} aoualaja Чим 'sprorpodA[od ay) jo siuaunean oneurojsÁs AIN}U9I-Y1G 1-PUA `1 318v[ Volume 77, Number 2 1990 Paris 8 Barrington 235 W. J. Hooker and the Generic Classification of Ferns affinity of the phegopteroids to the aspidioids. (AI- though he asserts that the phegopteroids shoul certainly be maintained within Polypodium, Hook- er notes that the aspidioid genus Lastrea, when its indusium has fallen off, is “impossible to distin- guish" from the phegopteroids [Hooker, 1844- 1864, vol. 4, p. 231]). Presl’s and Fée's phegop- teroids, by contrast, were interspersed among the true polypodioids. Although it is true that Hooker’s choice and weighting of characters led him to form some assemblages that we recognize today as poly- phyletic, e.g., his Grammitideae, his contemporar- ies made similar errors for similar reasons. It seems, then, that the criticism of Hooker’s work has been elicited primarily by the breadth of the genera he recognized, not by his failure to perceive natural groups within them. At the generic level, Smith’s system was by far the most modern. His faithful use of equally weighted characters from all parts of the plant body yielded genera that are quite natural by today’s standards. Note that, although the 19th-century pteridol- ogists explained their taxa differently (i.e., as the products of creation, not evolution), they had the same goal as 20th-century systematic biologists: circumscribing groups with a real existence in na- ture. We therefore maintain that it is appropriate to evaluate the systems of the 19th-century pter- idologists by comparing them to modern schemes. It is a tribute to all of the mid-19th-century pter- idologists that major segments of their systems survive in modern classifications, though they had neither the data sets (e.g., chromosomal and bio- chemical) nor the analytical methods available now. Dip HooKEr’s SYSTEM IMPEDE THE PROGRESS OF. FERN CLASSIFICATION IN THE 19TH CENTURY? To determine whether Hooker in fact impeded the development of fern classification in the 19th century, we first had to determine whether his system was adopted and, second, whether there was a reduction in the amount of work done on pteridophyte classification in the years after his publication of Synopsis Filicum. To assess the impact of the various mid-19th- century classifications on the systematic commu- nity, we surveyed 22 floras published between 1876 and 1900 to determine which of the major clas- sifications their authors followed in their treatments of the ferns (Table 2). We were able to do so because each system has a unique sequence of higher taxa; e.g., the polypodioids are placed near the end of the sequence in the systems of Hooker, FIGURE 2. wich Museum Portraits series b Sir William Jackson Hooker, 1851. (/ps- Maguire, artist and engraver. Courtesy of Hunt Institute for Botanical Documentation, Carnegie Mellon University, Pittsburgh, Pennsylvania.) Fée, and Presl, whereas Smith placed them near the beginning. Our review showed that there were two systems in common use during the period: Hooker's system and that of Endlicher, set forth in his Genera Plantarum (1836-1840). (Endlich- er's system is almost identical to that used by Mettenius in his Filices Horti Botanici Lipsiensis [1856], but Endlicher's work predates Mettenius by almost 20 years. This suggests that the system presented in Mettenius and attributed to him by various authors [e.g., Bower, 1926, pp. 57-58; Pichi Sermolli, 1973] was not original. For this reason we have not included Mettenius in our com- parison of the 19th-century pteridologists.) We found that, of the 22 floras we surveyed, 11 followed Endlicher (North America and conti- nental Europe), seven followed Hooker (British Em- pire and Latin America), and three used, to a greater or lesser extent, the older system of Swartz (1806). Smith's system was adopted in only one of the floras, and none followed Presl or Fee. Hook- er's influence, then, if not global, was at least more widespread than that of his contemporaries, whose efforts seem to have been largely overlooked by 236 Annals of the Missouri Botanical Garden TABLE 2. Fern classifications followed in floristic works, 1876-1900. Endlicher = Endlicher, 1836-1840; Hooker = Hooker & Bak 1874; Swartz = Swartz, 1806; Smith = Smith, 1875. Author Date Geographic region Taxonomic system Eaton, D. C. 1877-1880 North America Endlicher Orcutt, С. R. 1885 california Endlicher Tracy, 5. М. 1886 Missouri Endlicher : nderwood, L. M. 1888 North America Endlicher y A. 1889 northeast North America Endlicher | Н. С. 1891 New Hampshire Endlicher Rand, E. L. & J. H. Redfield 1894 Maine Endlicher Deane, W. 1896 Massachusetts Endlicher Dodge, R. 1896 New cam Endlicher Garcke, F. A. 1885 Germar Endlicher Koch, W. D. 1892-1907 боол. « Switzerland Endlicher Hooker, J. D. 1878 Great Britain Hooker Beddome, R. H. 1883 India Hooker Druce, С. С. 1886 ngland Hooker Jenman, G. S. 1890-1898 Jamaica Hooker iley, E 1882 Hawaii Hooker Alfaro, A. 1887 Costa Rica Hooker Sodiro, A. 1890-1895 Ecuador Hooker Hillebrand, W. F. 1888 Hawaii Swartz Knuth, P. E. 1887 Schleswig-Holstein similar to Swartz Noldeke, J. I 1890 northern Germany similar to Swartz Potonié, Н. 1889 Germany similar to Smith floristicians and others who might have put them to practical use. Another indication of the extent of Hooker’s influence on pteridology is the extent to which publication in the field was affected by his output. s we have seen, 1836-1865 was a period of great activity in fern systematics: major treatments were published every few years. Following the com- pletion of Hooker’s Synopsis Filicum (1868), how- ever, Smith’s Historia Filicum (1875) was the only major contribution published until almost the end of the century. Christ’s 1897 publication of Die Farnkrauter der Erde struck the first blow at the hegemony of the Hookerian system, but it was in Diels’s treatment of the ferns (1899-1900) in Eng- ler and Prantl’s Naturlichen Pflanzenfamilien that a new classification “far more in accordance with the modern view of phylogeny as the basic principle of systematics” was set forth (Christensen, 1938; see also Tryon, 1952). Thus it appears that Hook- er’s prolific output in the years 1846-1865 did indeed have a stifling effect on new work in pter- idology. Why Was Hooker MORE INFLUENTIAL THAN HIS CONTEMPORARIES? Given that Presl, Fée, Hooker, and Smith all contributed substantially to the mid-19th-century renaissance in pteridology, it is not at first clear why Hooker’s work was the most influential. The triumph of Hooker’s system may be attributable to several factors. First, as Hooker was Director of the Royal Gardens at Kew, his opinions carried the weight of authority. Also, he wrote in English at a time when British influence spanned the globe. Finally, because the classification presented in Species Filicum was summarized in the portable Synopsis, Hooker’s views were readily accessible to collectors; the utility of the Synopsis was thus a major factor in the widespread acceptance of his system (Tryon, 1952). By contrast, consider the positions of Hooker’s contemporaries and the impact that they might be expected to produce on pteridology in their time. There was Presl, whose work was highly regarded among a small group of experts, but who worked in Prague and wrote in Latin at a time when sci- entific prose was being written more and more commonly in the vernacular. There was also Fée, whose writings were in French and Latin. Finally there was John Smith, gardener at Kew, who spent 20 years working in the service of its celebrated director. Although Smith's generic classification was easily the most modern among the systems we compared, Smith was overshadowed by the persona and reputation of Hooker, and his work was largely ignored. Only in this century has Smith been rec- Volume 77, Number 2 1990 Paris & Barrington 237 W. J. Hooker and the Generic Classification of Ferns ognized as a pioneer of modern pteridology (Chris- tensen, 1938; Tryon, 1952) CONCLUSIONS Having evaluated the contentions of Hooker’s critics, we are in a position to ask whether they judged his works fairly. On the whole, we conclude that they did not: critical analysis of Hooker’s work demonstrates that he had a reasonably advanced perception of natural relationships relative to most of his contemporaries. His system cannot fairly be called regressive just because of his broad genus concept. Today, more than 120 years since Hook- er’s death, generic concepts still vary widely among systematists, as at least two recent symposia on the subject attest ("Generic Concepts in the Compositae,”” American Institute of Biological Sci- ences [A.I.B.S.], 1983; “The Concept of the Ge- nus," A.I.B.S., 1987; see also Stevens, 1985). Arguments about the criteria by which genera should be recognized and the kinds of characters useful in generic delimitation are as yet unresolved (Davis, 1987). Also unsettled is the question of whether (or under what circumstances) it is more useful to recognize large, often polymorphic genera that are well established in the taxonomic literature, or smaller, distinctive, segregate genera (Wagner, 1987; Young, 1987). Despite more than a century of systematic research since Hooker, there is no more of a consensus on generic classification than there was in his time. It is thus superficial to com- pare Hooker’s system unfavorably with those of his contemporaries because he recognized fewer genera. LITERATURE CITED Lista de las plantas ntradas hasta ү” territorios limitrofes, " An- ALFARO, A. 1887. ahora en Costa Rica y en extractada de la “Biologia Centrali Americana. ales Mus. Nac. Costa Rica 1: 1-101. 1967. The жен of Kew. M. Joseph, " 1889. Hawaiian Ferns; A Synopsis. T. G. Thrum, Honolulu, Hawaii. BEDDOME, R.H. 1883. Handbook to i» EM of British 20., 1926. The Ferns icc Volume 2. Cambridge Univ. Press, Cambri T zt 1897. Die Farnkrauter pu Erde. Fischer, Coni C. 1938. Filicinae. Pp. 522-550 in F. Verdoorn сл. Manual of Pteridology. Nijhoff, The Hague. COPELAND, E. B. 1947. Genera Filicum, the Genera of Ferns. Chronica Botanica, Waltham, Massachusetts. Davis, J. I. The generic concept in the grass family (Poaceae). Amer. J. Bot. 74: 716. DEANE, W. did Flora of the Blue Hills. C. M. Barrows & ston. Diets, L. "1899- 1900. Polypodiaceae. Pp. 139-339 in А. Engler & К. Prantl (editors), Die Natürlichen Pflanzenfamilien, 1(4). DobcE, R. . The Ferns and Fern Allies of New England. W. Clute & Co., Binghamton, New York. DRUCE, a C. 1886. The Flora of Oxfordshire. Parker & Co., Oxfor ae р. С. 1877-1880. The Ferns of North Amer- a. S. E. Cassino, Salem, Massachusetts. eonun S. 1836-1840. Genera Plantarum. Beck, Jienna. КЁЕ, A. L. A. 1850-1852. Genera Filicum (Cinquième Mémoire sur la Famille des Fougères). Berger-Lev- lt, Strasbourg. GARCKE, F. A. 1885. Flora von Deutschland. P. Parey, Berlin. Gray, A. 1889. Manual of the Botany of the Northern United States, 6th edition. American Book Co., New rk. Yo HILLEBRAND, W. 1888. Flora of the Hawaiian Islands. Williams € Norgate, London Hooker, J. D. 1878. The Student s Flora of the British Islands. Macmillan, Londor Hooker, W. J. -1842. pom Filicum. Bohn, London. . 1844-1864. Species Filicum. 5 volumes. Pamplin, London. & J. G. BAKER. 1865-1868. Synopsis Filicum. m London. Second edition, 1874. Har icke, London. Tapa G. S. 1890-1898. Synoptical list, with de scriptions of the ferns and fern-allies of Jamaica. Bull. Bot. Dept. [Numerous short parts.] Jesup, Н. С. . Catalogue of the Flowering Plants and Higher Cryptogams Found Within About 30 Miles of Hanover, New ова Н. С. Jesup, Hanover, New Hampshir Кмотн, P. E. 1887. Flora m Province Schleswig- Holstein. Otto Lenz, Leipzig. Косн, W. D. 1892-1907. Synopsis der Deutschen und Schweizer Flora, 3rd o E. Hallier (editor). F. ilmans, Frankfurt am Mai Чү m G. H. M. 1951. ds of Vascular s. Macmillan, New York. Мека, С. 5 Filices Horti Botanici Lipsiensis. Vos zig. e J. 8. 1890. Flora des Fúrstentums Liineberg. E. Spangenberg, Celle. Orcutt, С. R. 1885. Flora of Southern and Lower California. Torrey Botanical Club, New York. > SERMOLLI, В. E. С. 1973. Historical review of e higher classification of к жечи Рр. 11- pe in А. C. Jermy, J. A. Crabbe & B. A. Thomas (editors), The Phylogeny me Cusco of the Ferns. Suppl. 1, J. Linn. Soc. Illustrierte Flora von Nord- und Berlin. 36. Tentamen Pteridographiae. Haase, 1843. Hymenophyllaceae. Haase, Prague . 1845. cn Tentarninis ia phiae. br» Pra 847. Die Cefissbündel i im Stipes der Farrn. Hasse, Prague 852. Epimeliae Botanicae. Haase, Prague. 238 Annals of the Missouri Botanical Garden Rann, E. L. & J. Н. ReprieLD. 1894. Flora of Mt. Desert Island, Maine. J. Wilson & Son, Cambridge, Massachusetts. 1841-1843. An d ges and definition 4 (1842): 38-70, 668. 1843; 1843. [Reprint. рр. l- 131. A 378-394. ———. 75. Historia Filicum. iles p a SMITH, J. 3 1 Tentamen botanicum de Filic generibus айгай: Mem. Acad. Roy. Sci. Turin 5: 401-422 SODIRO, А. 1890. 1895. Cryptogamae Vasculares Qui- tenses. Typis Universitatis, Quito, Ecuador STEVENS, P. F. 1985. The genus concept in practice— but for what practice? Kew Bull. 40: 457-405. Swartz, O. 1806. Synopsis Filicum. Bibliopolium Aca- demicum, Kiel. RACY, S. M. 1886. Catalogue of the Phaenogamous and Vascular Cryptogamous Plants of doo Tri- bune Printing Co., Jefferson City, Miss Tryon, R. M., Jr. 1952. A sketch of fhe a of fern classification. Ann. Missouri Bot. Gard. 39: 255- 262 & A. F. Tryon. 1982. Springer-Verlag, New York. Unperwoop, L. M. 1888. Our Native Ferns and Their Allies. Henry Holt, New York. WAGNER, W. H., JR. 1987. The ene concept in pteridophytes. Amer. er uni Younc, D. A. 1987. Concept of the genus: introduction i писти al perspective. Amer. J. Bot. 74: 718. Ferns and Allied Plants. STUDIES OF NEOTROPICAL /SOÉETES L. I. EUPHYLLUM, A NEW SUBGENUS! R. James Hickey’ ABSTRACT А 1 new cad in Isoetes, Euphyllum, is proposed for three extant species and for members of the fossil genus is distinguished fro Isoetites. This subgenus om subg. [soete ssession of a number of plesiomorphic s by the po Шоо. чаб ай completely alate leaves and the absence of peripheral fibrous bundles. Descriptions and a ke to the three modern species of Euphyllum are presented. Our perception and understanding of the mor- phology and anatomy of /soétes is primarily a result of studies of north temperate species. Such a bias is a reflection of the availability of material and of the historical distribution of plant anatomists, mor- phologists, and taxonomists. Much of the diversity in this genus, however, occurs among the largely unstudied tropical representatives. Recently, Hick- ey (1986a) described a new /soetes with a com- pletely alate (= laminate) leaf. In that paper, the genus was hypothesized as comprising two basal evolutionary groups: an alate grade association, consisting of three recent taxa plus the fossil genus Isoetites, and a subulate (= lineage comprising the rest of the genus recognized here as subg. /soetes. The former group merits recognition as a separate subgenus based on its unique morphology and distinctively primitive leaf architecture and leaf base characters. This sub- genus, Euphyllum, is described here and its three extant species are reviewe nonalate) monophyletic KEY TO THE SUBGENERA OF /SOÉTES la. m completely laminate, without a distal ubula subg. i ax loa partially laminate, with eri alae sub a distal subula = ч pe Isoétes subg. Euphyllum Hickey, subg. nov. TYPE: Isoetes gigantea U. Weber, Nova Hed- wigia 63: 245. 1922. Subgenere /soéte a quo foliis omino lamellatis differt. Corms essentially globose, 2- or 3-lobed; di- chotomous roots arising synchronously within the circumbasal fossa(e). Leaves numerous, spirally ar- ranged, flattened to elliptic in cross section, com- alate; fibrous bundles absent; scales and face view, superficial or slightly embedded in leaf tissue, basal or elevated, the walls concolorous. Velum poorly developed. Ligule delicate, the mar- gins ephemeral. Labium large, covering the ligule. Megaspores globose, trilete, variously patterned. Microspores ellipsoid, monolete, variously pat- terned. Subgenus Euphyllum is a paraphyletic assem- lage characterized and y the presence of completely alate leaves, a character considered to be plesiomorphic on the basis of its presence in the sister group Selaginellaceae and the outgroup Lycopodiaceae. Members of this subgenus fre- quently show a number of other plesiomorphic con- ditions, such as superficial sporangia (in /. gigantea and /. bradei), little or no velum coverage (all members), a lack of peripheral fibrous bundles (all members), and a distinctly narrowed leaf attach- ment (I. gigantea and І. bradei). In addition, megaspores of the extant members of the subgenus have surface morphologies roughly referable to the tuberculate morphology class, a condition also con- sidered to be ancestral (Hickey, 1986b). The alate character condition of the leaves in subg. Euphyllum contrasts sharply with the con- dition found in the rest of the genus. In subg. Isoetes, the lamina is confined to the proximal ! This research was supported in part by NSF grant BSR 86-0672. 2 Botany Department, Miami University, Oxford, Ohio 45056, U.S.A. ANN. MISSOURI Вот. GARD. 77: 239-245. 1990. 240 Annals of the Missouri Botanical Garden Leaf base O in /soetes subg. Euphyllum. — А. Microsporophyll of /. gigantea showing large labium (Luetzelburg 298 M). . Microsporophyll of /. bradei, ligule and labium absent (Brade 8119 AAU).— C. Volume 77, Number 2 1990 Hickey 241 Studies of Neotropical /soétes L. portions of the leaf and a terete extension of the leaf, the subula, represents the major photosyn- thetic organ. The nonlaminate subula and the pres- ence of peripheral fibrous bundles are synapomor- phies that unite subg. /soetes and support its recognition as a monophyletic assemblage (Hickey, 1986a). The majority of extant /soetes species, including those previously placed in subg. (or the genus) quc (Gómez, 1980) are properly placed in subg. /soe Subgenus Euphyllum, as currently construed, contains three modern species, /. baculata, 1. bradei, and І. gigantea. It also appears that cer- tain fossil taxa segregated as the genus /soetites fall within this group (Brown, 1939, 1958; Bock, 1962; Bose & Roy, 1964). While the fossil mem- bers are worldwide in distribution, the extant taxa are restricted to Brazil and the Colombia—Brazil border. This range, coupled with the plesiomorphic, grade-association nature of subg. Euphyllum, sug- gests that the extant taxa are survivors of a for- merly cosmopolitan group and have survived only in isolated refugia of South America (Simpson & Haffer, 1978) KEY TO THE SPECIES OF SUBG. EUPHYLLUM la. Sporangium elevated above leaf base; mega- spores essentially saccate, proximal ridges nar- sii 1. аны m basal; megaspores baculate or tu орти proxima ridges as wide as or maily as wide a 2a. Mi tuberculate; leaves broad, 1.8- : wide at mid-length, the alae 3.0- mm wide at the sporangium; p of Soda ены Brazil 2 . L bradei Meg purum baculate; leaves narrow, 0.6- ; wide at mid-length, the alae 0.6- 5 mm ideis at the sporangium; пш of western Brazil . L baculata — = un 5 © ^ @ d ко = == Isoétes gigantea U. Weber, Nova Hedwigia 63: 245. 1922. TYPE: Brazil. Rio de Janeiro: Serro dos Orgáos, Morro Assu, Hochmoor bei Isabelloca, 2,000 m, 1915, v. Luetzelburg 6438 (lectotype, here pure M; isolec- totype, M). Figures 1A, 2A, 2B, Corm globose, 25-45 mm wide, 22-35 mm high, 2-lobed; dichotomous roots arising synchro- nously within the continuous circumbasal fossa. Leaves 80+, dark green, laxly ascending, flattened to narrowly elliptic in cross section, 200-300 mm long, 8.0-10.0 mm wide at the base, 14.0-20.0 mm wide above the sporangium, 3.0-5.0 mm wide at mid-length, apex acute; alae light green to nearly translucent and chartaceous proximally, dark green and membranaceous distally, 5.0-6.0 mm wide at the sporangium, 1.0-2.0 mm wide at mid-length, extending to the apex; fibrous bundles absent; scales and phyllopodia absent. Sporangia obovate in face view, superficial, microsporangium walls buff, me- gasporangium walls light brown, 14.0-20.0 mm long, 3.0- 7.0 mm wide, positioned 3.0-8.0 mm distal to the leaf base. Velum minute, 0.2-0.5 mm wide along the lateral edges of bu sporangium, absent along the upper edge of the sporangium. Ligule odor to widely Mut delicate, the margins ephemeral, the persistent cushion dark, 4.0-5.0 mm long, 2.0-3.0 mm wide, hidden be- hind the more massive labium. Labium persistent, depressed ovate, cordate, entire to laciniate, buff- colored, 3.0-5.0 mm long, 5.5-7.0 mm wide. Megaspores gray, not lustrous, 590-690 (X — 616) um diam., (verrucate to) saccate, the elements occasionally merging laterally to form meandriform rugae near the proximal ridges; equatorial ridge straight or undulate, distinct, broad, proximal ridges straight, very narrow, appearing more distinct be- cause of the slight separation between them and the proximal ornamentation. Microspores light gray, 38.0-48.0 (X = 43) um long, 30.0-37.0 (X = 33.0) um wide, surface morphology of confluent papillae. Additional specimens examined. BRAZIL. BAHIA: Quellgebiet des Rio de Contas, Minas de Contas, in einem Sumpf, Aug. 1913, Luetzelburg 298 (M); Rio Ramalho, Feb. 19136 Са zelburg 15057 (M). Isoetes gigantea is known only from the type locality north of the city of Rio de Janeiro and from two localities in the mountains of Bahia, Bra- zil. While the collections of /. gigantea are inad- equate to describe its habitat, the overall mor- phology suggests that it is a completely submerged aquatic, perhaps of mid-elevation (ca. 2,000 m) swamps and bogs. Fertile collections are known from February and August; there is no evidence of vegetative reproduction. eber’s description of /. gigantea cites the type s “Ph. у. Lutzelburg 1915.” This collection is represented by two sheets at Munich, both of which <— Megasporophyll of /. baculata showing pronounced labium (Luetzelburg 23769 M). Labium i bar applies to Figure C, both = 2 mm. regions; upper bar applies to Figures A and B, lower = heavily stippled Annals of th Missouri Botanical Garden FIGURE. 2. Spore morphology in /soetes subg. Euphyllum.— A. Meg oe aa of 1. en near proximal view (Luetz elburg 15057 M). — B. Microspore of /. gigantea, equatorial view (Lue zelburg 150: ).—C. Me жнр of I. bradei, distal surface (Brade 8119 S). —D. Microspore of I. bradei, um al view (Brade 8119 AAU). е Megaspore of I. baculata, near proximal view arena 23785 M). —F, Microspore of /. кт Po view (Luetzelburg 23785 M). Bars in Figures A, C, and E = 500 шп; bars in Figures B, D, and F = 10 u Weber annotated. One sheet has a single specimen sequently annotated the sheet with two plants as while the other has two. The number 1915 as the holotype. However, the sheet with only a single marked on these sheets represents the year of plantis more diagnostic and presents the distinctive collection. Lutzelburg’s actual collection number is characters of the species in greater detail; for this indicated on both sheets as 6438. Launert sub- reason I designate it as the lectotype. Volume 77, Number 2 1990 Hickey 243 Studies of Neotropical /soétes L. $t ALAS 3 FIGURE 3. Collection sites of /. gigantea (triangles), Isoetes gigantea is one of the most easily rec- ognized species of Isoetes. The broadly laminate leaves, massive sporangia, which are virtually su- perficial on the leaf surface, and saccate mega- spores are immediately distinctive. The large, per- sistent labium and the narrow, knifelike proximal ridges are also characteristic. The saccate megaspores and the microspores with confluent papillae are uniquely derived char- acter states (= autapomorphies) and do not provide evidence of relationships. Ancestral states, such as a completely laminate leaf, a lack of scale leaves and fibrous bundles, a poorly developed velum, and a large labium, suggest a grade level relationship with 7. baculata and I. bradei. 2. Isoétes bradei Herter, Rev. Sudamer. Bot. 8: 19. 1949. TYPE: Brazil. Sào Paulo: Villa Emma, Rio Mooca, 1921, Brade 8119 (ho- lotype, С; isotypes, AAU, S, UC, US). Figures 1B, 2C, 2D, 3. Corm ellipsoid, ca. Z0 mm wide, ca. 10 mm high, 3(?)-lobed; dichotomous roots arising syn- VR. 9 є I. bradei (diamond), and I. baculata (dots). chronously within the continuous circumbasal fos- sale). Leaves 60-100+, medium to dark green, flexuous, flattened to elliptic in cross section, 630— 700 mm long, 5.0-8.0 mm wide at the base, 8.0- 15.0 mm wide above the sporangium, 1.8-4.5 mm wide at mid-length, apex attenuate; alae medium to dark green, membranaceous, 3.0-5.0 mm wide at the sporangium, extending to the apex; fibrous bundles absent; scales and phyllopodia absent. Spo- rangium narrowly obovate in face view, superficial to slightly embedded, the walls tan to light brown, concolorous, 10.0-15.0 mm long, 4.0-5.0 mm wide, basal. Velum minute, 0.0—0.5 mm wide along the lateral edges of the sporangium, absent along the upper edge of the sporangium. Ligule not seen. Labium ephemeral, frequently represented by a widely ovate brown fragment, 2+ mm long, 2.2 mm wide. Megaspores white to gray when dry, dark when wet, slightly lustrous, 500-660 (Х = 609) um diam., tuberculate; equatorial ridge un- dulate, much higher than wide, proximal ridges straight, distinct, much higher than wide; occa- sional dumbbell spores and very small (350-360 244 Annals of the Missouri Botanical Garden um) spores produced. Microspores light gray, 40.0— 51.3 (X = 45) um long, 32.5-38.8 (X = 35) um wide, echinate or the echinae broken off leaving a papillate surface. Additional specimen examined. BRAZIL. SAO PAULO: Rio Mooca, Villa Emma, Brade 7946 (SP). No ecological or phenological data accompany any of the collections but in the protologue Herter stated that the plant is submersed. This statement is supported by the general morphology and habit of the plants; that is, they are extremely large, lack scales and phyllopodia, and have long flexuous leaves lacking supporting fibrous bundles. There is no evidence of vegetative reproduction. Isoetes bradei is one of several poorly known, low- to medium-elevation species from Brazil. It is known only from the two Brade collections made at the Villa Emma population in Sáo Paulo which, according to the label data on the isotype (AAU), has been completely destroyed. Despite the limited amount of material available, there is no doubt that I. bradei is distinct. The broadly alate leaves cou- pled with the tuberculate megaspores are charac- teristic. The presence of small, and occasionally dumbbell shaped spores is interesting in that no accompanying microspore abnormalities have been observed. Such observations suggest that an eco- logical stress may have resulted in abnormal mega- sporogenesis in the type specimen. y Isoëtes baculata Hickey € Fuchs in Hickey, Syst. Bot. 11: 310. 1986. TYPE: Brazil. Ama- zonas (Uaupés): im fluss auf d. Colombian- isches Seite, 28 Nov. 1916, Luetzelburg 23785 (holotype, M). Figures 1C, 2E, 2F, 3. Corm globose, 13-18 mm diam., 2-lobed; di- chotomous roots arising synchronously within the continuous circumbasal fossa. Leaves 26-209, light green, flexuous, flattened to elliptic in cross section, 600-650 mm long, 5.0-8.0 mm wide at the base, 0.6-1.0 mm wide at mid-length, apex attenuate; alae light green to nearly colorless and chartaceous proximally, light green and membranaceous dis- tally, 0.6-1.5 mm wide at the sporangium, ex- tending to the apex as two minute lateral ridges; fibrous bundles absent; scales and phyllopodia ab- sent. Sporangium elliptic in face view, embedded, the walls pale tan, concolorous, 5.5-7.0 mm long, 2.5-4.5 mm wide, basal. Velum minute, 0-1.5 mm wide along the lateral edges of the sporangium, absent along the upper edge of the sporangium. Ligule margins delicate, ephemeral, the persistent cushion tan, membranaceous, deltate to widely del- tate, cordate, 1.4 mm long, 1.8 mm wide. Labium partially persistent, widely ovate to very widely ovate, entire to erose, tan, 0.7-1.6 mm long, 1.3- 1.6 mm wide. Megaspores light gray, not lustrous, (470-)580-730 (X = 667) um diam., baculate to subverrucate, the bacula isolated or fused into groups of two or three, 60-80 um in height, 30- 40 um wide; equatorial ridge straight or undulate, 2 to 3 times as high as wide, proximal ridges straight, distinct, as wide as or slightly wider than high. Microspores light gray, 35.0-43.0 (X = 39.0) um long, 27.0-34.0 (X = 31.0) um wide, echinate or smooth, their surface covered with a fine, fibrous perispore. Additional specimen examined. COLOMBIA. VAUPÉS: Uapes [sic], submersus auf Colombianisch Ufer, 28 Nov. 1918, Luetzelburg 23769 (M). Isoetes baculata is, as far as is known, confined to the Rio Vaupes amd Rio Negro drainages along the Colombia - Brazil border. According to label data it is a submerged aquatic; this is substantiated by the lax, flexuous habit of its leaves. Both collections of this species were made in November and both have abundant megaspores and microspores. There is no evidence of vegetative reproduction. Isoetes baculata is readily identified by its long flexuous leaves, which are narrowly laminate for their entire length, and by the large, baculate mega- spores. There is a distinct phenetic relationship between /. baculata and 1. panamensis Maxon, particularly in the morphological similarity of megaspores, micropsores, velum, and labium. Many of these similarities, however, appear to be the result of convergence or are the result of the re- tention of plesiomorphic character states in the otherwise derived /. panamensis (Hickey, 1985). Convergence in the echinate morphology of the microspores is obvious when they are viewed at high magnifications: the perispore of /. panamensis microspores is smooth and lacks the loosely fibrous morphology of /. baculata. Preliminary phyloge- netic analyses on the genus (Hickey, 1985, 1986b) also indicate that the tuberculate surface mor- phology is plesiomorphic within subg. /soetes and that the baculate megaspores of /. baculata are convergent with respect to /. panamensis. The lack of velum coverage and the large labia are plesiomorphic states in each taxon and cannot be used as indicators of phylogenetic relationship. In the absence of obvious synapomorphies, 1 have included /. baculata in subg. Euphyllum. It deviates most significantly from /. gigantea and /. bradei in having poorly developed alae and par- tially embedded sporangia. Both /. gigantea and Volume 77, Number 2 1990 Hickey 245 Studies of Neotropical /soétes L. І. bradei have sporophyll bases widest above the sporangium with a subtruncate narrowing of the leaf above the labium; /. baculata has leaves widest at or below the sporangium with a gradual reduction in leaf base width distally. In this regard, /. bac- ulata is more similar to members of the subg. Isoetes Тынай The various teris for organs and their character states are d in Hickey (1985, and 1986a); megaspore morphology follows the terminology presented in Hickey (1986b). Two- dimensional shape characterizations for the ligule, labium, and sporangium follow the terminology of Radford et al. (1972). LITERATURE CITED Bock, W. 1962. A study of fossil /soetes. J. Paleontol. 36: 53-59. Bose, M. М. & S. К. Roy. 1964. Studies on the u Gondwana of Kutch — 2. Isoetaceae. Palsobot&tist 12: 226-228 BRowN, R. W. 1939. Some American fossil plants ТР. to the Isoetales. J. Wash. Acad. Sci. 29 е 958. New s of the fossil quillwort called Fon. 2 Wash. Acad. Sci. 48: 358-361. GÓMEZ P., Vegetative reproduction in a Саша! ена Ж» (Isoétaceae). Its morpho- logical, Эи and taxonomical significance. Bre- -14. 1985. idera studies of neotropical Isoctes. Ph.D. Dissertation. The Univ. of Connecti- cut, om Connectic 1986 he early evolutionary and morpho- logical diversity of Isoétes, with descriptions of two new neotropical species. Syst. Bot. 11: 309-321 19 soetes megaspore ‘surface morphol- ogy: nomenclature, variation, and systematic impor- bd 1 -16. RADFORD, A. E., W. C. Dickison & С. R. BELL. 1972. Vascular Plant Systematics. "Univ. of North "Carolina; Chapel Hill, North Caro SIMPSON, B. К. & J. HAFFER. 1978. Speciation patterns in the Amazonian forest biota. Annual Rev. Ecol Syst. 9: 497-518. TWO NEW SPECIES OF CNEMIDARIA (CYATHEACEAE) FROM PANAMA! Robbin C. Moran? ABSTRACT Cnemidaria suprastrigosa and Cnemidaria varians are named as new species in the fern family Cyatheaceae. In 1974, Robert G. Stolze of the Field Museum of Natural History, Chicago, published a mono- graph of Cnemidaria based on a study of nearly 2,000 specimens from 17 herbaria. He was helped and encouraged throughout his study by Rolla and Alice Tryon, who provided the original impetus for his research as well as many other studies in the Cyatheaceae. Since the publication of Stolze's ex- cellent monograph, new collections have turned up two new species of Cnemidaria in addition to the 23 species recognized by Stolze (1974). Both are from Panama. Cnemidaria suprastrigosa H. C. Moran, sp. nov. TYPE: Santa Fé, ridge up from former Escuela Agri- cola, 1,000-1,300 m, Hamilton & Dressler 3049 (holotype, MO). Figure le. anama. Veraguas: Cerro Tute, Lamina adaxialiter ‘strigosa, apice pinnatifida, pinnis distantiae ad costam ca. 2% incisis; venis basalibus junctis; rhachidi 2 soris таи indusiis semicircularibus. Stem not seen; leaves ca. 1.5 X 0.3 m, the veins pubescent adaxially, with hairs 0.2-0.4 mm, subulate, strigose, scattered, the apex evenly ta- pered, pinnatifid; petiole lacking spines, with only a few scattered hairs and scales; rachis nonalate or slightly winged below the bases of the distal pinnae, the abaxial surface pubescent, the hairs hyaline, retrorsely strigose; pinnae sessile, cut ca. 24 to the costa; costae scaly and pubescent abax- ially, the scales scattered, brown, amorphous, the adaxial surface pubescent; veins 1-2-forked, the basal ones anastomosing to form costal areoles; sori medial; indusia semicircular, positioned on the cos- tular side of the receptacle This species differs from all others in the genus by its adaxial pubescence along the veins (Fig. le). It was formerly thought that the glabrous adaxial surface of the lamina was one of the distinctions between Cnemidaria and Cyathea. Cnemidaria suprastrigosa is known only from the type collection. Cnemidaria varians К. С. Moran, sp. nov. TYPE: Panama. Panama: Cerro Jefe, 1,000 m, Val- despino & Aranda 139 (holotype, МО; iso- types, UC, PMA not seen). Figure la-d. Lamina adaxialiter glabra, abaxialiter glandibus рипс- incisis, paribus soris medianis vel supramedianis; indusiis circularibus vel semicircularibus. Stem to 90 cm; leaves to 1.4 х 0.32 m, lan- ceolate, the abaxial surface punctate with reddish glands, the apex attenuate or abruptly reduced, pinnatifid; petiole base smooth, the scales bicolo ous; rachis alate throughout, the wing to glabrous to sparsely pubescent abaxially, brown; pinnae 12-15 x 2-2.3 cm, sessile or the basal ones stalked, entire or cut 16-14 to the costa, the mm, 2 lowermost pairs reduced, ca. half the length of the longest medial pinnae; costae glabrous to sparsely pubescent abaxially, sparsely scaly, the scales dark; veins not branched, mostly anasto- mosing, occasionally free; sori medial to inframe- dial; indusia circular, completely surrounding the base of the receptacle, or semicircular and posi- tioned on the costular side of the receptacle. Paratypes. PANAMA. DARIEN: Panama/Colombia border, Parque Nacional del Darien, near gold mine at heädwaters of N branch of Rio Pucuro, slopes of Cerro Tacarcuna, ca. 6 km N of Cerro Mali, 1,300-1,500 m, ! I thank Robert G. Stolze for helpful comments on the manuscript. 2 Missouri Botanical Garden, ANN. P.O. Box 299, St. Louis, Missouri 63166, U.S.A. Missouri Bor. GARD. 77: 246-248. 1990. Volume 77, Number 2 Moran 247 1990 New Species of Cnemidaria from Panama FIGURE 1. a-d. Cnemidaria varians. —a-c. Variation in shape of the leaf apex. — d. Abaxial surface of leaf. — е. Cnemidaria suprastrigosa, adaxial leaf surface showing strigose hairs. a. Mori & Kallunki 3798 (MO). b, с, d. Valdespino & Aranda 139 (UC). e. Hamilton & Dressler 3049 (MO). 248 Annals of the Missouri Botanical Garden de Nevers et al. 116531 (COL, MO, PMA, UC). PANAMA: Cerro Jefe, along trail on ridge running NE from summit, cloud forest dominated by Clusia spp. Que Colpothrinax cookit, ca. 1,000 m, Mori & Kallunki 3798 (MO). This species is extremely variable in characters that are usually constant in other species of Cnemi- daria, thus the specific epithet varians. For ex- ample, the shape of the apex, even on the same plant, varies from evenly tapered (Fig. lb) to abruptly contracted (Fig. 1с); intermediates are known (Fig. 1a). The basal veins are anastomosing or are free (Fig. 1d). The type collection had a stem 90 cm long, but the Darién collection was described as a This variation in several characters suggests hybrid origin, but I cannot find suitable parent species. The spores are normal, i.e., not aborted. “terrestrial rosette with no stem." This new species might be confused with Cnemi- daria decurrens (Liebm.) R. Tryon because both have alate rachises and entire or shallowly lobed pinnae. The new species differs, however, from C. decurrens by its spineless petiole, reduced basal pinnae, and minute, round, reddish glands on the abaxial surface of the lamina. Cnemidaria decur- rens is known from southern Mexico to Honduras, whereas C. varians is endemic to eastern Panama. This new species also resembles C. glandulosa Stolze because both have relatively small leaves, pinnae that are not deeply lobed, and minute red- dish glands on the abaxial surface of the lamina. Cnemidaria varians differs, however, from C. glandulosa by having alate rachis, shallowly lobed pinnae, dark costal scales, unbranched veins, basal veins that frequently anastomose, and sometimes fully circular indusia that completely surround the base of the receptacle. Both species are known only from Panama. LirERATURE CITED SroLZE, R. G. 1974. A taxonomic revision of the 2 Cnemidaria (Cyatheaceae). Fieldiana, Bot. : l- 98. PTERIDOPHYTES OF THE VENEZUELAN GUAYANA: NEW SPECIES Alan R. Smith! ABSTRACT Studies on = dp ot n of the Venezuelan Guayana have resulted in the d = following 31 new blin species: Asplenium chiman Sticherus tepuiensis (Gleicheniaceae) Ceradenia arthro Doryopteris cyclophy. S. cyclophylla, S e, А. cowanii (Aspleniaceae); Cyathea аме. C. neblin thrix, C. microcystis, licata, С. sinuosa (Grammitidaceae); Pleopeltis repanda (Po ао, Adiantum nudum, А lla, D. davidset ea Selagine , S. puben . imbrican ns, э. maranuacae, э. ne inae eps (Cyatheaceae); paar бега. С. peritimundi nella albolineata . arrecta, S. beitelii, S. versatilis ding aug Ru e eris peradenia (Thelypteridaceae). In addition, new combinations are made for the following six taxa: Salpic chlaena lomarioidea (Blechnaceae); Cyathea macrosora var. reginae, C. macrosora var. vaupensis, C. "s danalepis Grammitis blanchetii, and Sticherus tomentosu ASPLENIACEAE Asplenium chimantae A. R. Smith, sp. nov. TYPE: Venezuela. Bolivar: Dtto. Piar, Macizo del Chimanta, sector centro-noreste del Chi- manta-tepui, cabeceras orientales del Cano Chimanta, alrededor del comienzo E del Canon recto del Rio Chimantá superior, 5?18'N, 62?09'W, ca. 2,000 m, 26-29 Jan. 1983, Huber & Steyermark 6927 (holotype, UC; 1 isotypes, AAU, VEN not seen). Figure A-C Ex affinitate 4. dissecti Sw., pinnis latioribus, pau- cioribus, minus incisis йшй ab А. serra Langsd. & nibus palearum majoribus hyalinioribus differt. Rhizome short-creeping, densely scaly; rhizome scales spreading, blackish, lanceolate with a long uniseriate tip, clathrate, up to 8 X 0.5 mm; fronds clustered, to ca. 35 cm long; stipe to ca. 12 cm x 1 mm, blackish or atropurpureous, shining, gla- brescent or with a few filiform scales; rachis sim- ilarly colored, bearing scattered filiform scales up to 2 mm long, 1-3 cells wide at the base; lamina ovate, 1-pinnate with a pinnatifid apex; pinnae to ca. 10 pairs, + bilaterally symmetrical or with the basiscopic side slightly excised, to 5 X 1.3 cm, broadly cuneate or rounded at the base, attenuate at the tip, margins serrately or biserrately incised up to half the distance to the costa; veins simple or 1-forked, strongly ascending at ca. 20° from the costa; lamina glabrous on both sides or abaxially with inconspicuous, + appressed, septate hairs to 0.3 mm; sori borne near the base of the veins or at vein forks, nearer the costa than the margin, mostly 2—5 mm long, with a linear or oblong white or gray indusium ca. 0.7 mm wide. Paratype. Same locality as type, Steyermark, Hu- ber & Carreño 128211 (VEN). This species is perhaps most closely related to A. dissectum Sw., known from the Greater Antilles and southern Mexico to Ecuador, Venezuela, and southern Brazil, but the new species differs in the wider, less incised pinnae in fewer pairs. It is also related to A. serra Langsd. & Fischer, from which it differs in the more delicate, thin-textured lamina and more attenuate rhizome scales with larger and clearer lumina and long-filiform tip. The type was described as growing in deep grottoes. Asplenium cowanii A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Ser- гата Parú, Rio Ventuari, dry crevices of cliff, talus forest, 1,800 m, 12 Feb. 1951, Cowan & Wurdack 31393 (holotype, US). Figure D-E A. repenti Hook. primo adspectu maxime simile, sed rhizomate breviter repenti vel suberecto, frondibus fas- ciculatis, stipite et rhachidi rubrobrunneis vel atropur- pureis. Rhizome short-creeping to suberect; rhizome scales dark reddish brown, lanceolate with a uni- seriate tip, clathrate, ca. 2-3 x mm; fronds densely clustered, to ca. 6 cm long; stipe to 1.5 mm, reddish brown to atropurpureous, ' University Herbarium, University of California, Berkeley, California 94720, U.S.A. ANN. Missouri Вот. GARD. 77: 249-273. 1990. 250 Annals of the Missouri Botanical Garden shining, glabrous or with a few filiform scales 1- 3 cells wide; rachis similarly colored, glabrous; lamina pinnate-pinnatifid to barely 2-pinnate with a pinnatifid or cleft apical segment; pinnae 5-10 pairs, to 6 X 4 mm, with at most | pair of lateral pinnules and a terminal segment, the lateral ones bifid or trifid, the terminal 3-6-cleft; one vein per ultimate lobe; lamina glabrous on both sides; sori 1-3 per pinna, the indusium oblong to semicircular, 1 x 0.4 mm, brownish. This very distinct new species, known only from the type, is most closely related to 4. repens Hook., from Ecuador and Peru, from which it differs by the short-creeping or suberect rhizome (vs. long- creeping in A. repens) with fasciculate fronds and by the reddish brown or atropurpureous (vs. green- ish to tan) stipe and rachis. In laminar dissection, the two species are very similar. BLECHNACEAE Salpichlaena lomarioidea (Baker) A. R. Smith, comb. nov. Blechnum volubile Kaulf. var. m Baker in C. Martius, Fl. Bras. 1(2): 428. 1870. Distribution. Guianas to Peru. Specimens examined. VENEZUELA. TERRITORIO FED- ERAL AMAZONAS: 15 km SE de San Fernando de Atabapo, Stergios et al. 11605 (UC); Dpto. Rio Negro, Neblina Base Camp, Beitel 85222 (UC). COLOMBIA. VICHADA: ca. 25 km E of Cumaribo on road between Las Gaviotas and Santa Rita, Davidse 5315 (MO). PERU. LORETO: Mec- Daniel A a 21515 (MO), Klug 1126 (МО), Moran 3672 ( CYATHEACEAE Cyathea macrosora (Baker) Domin var. regi- nae (Wind.) A. R. Smith, comb. nov. Sphae- ropteris macrosora (Baker) Wind. var. re- ginae Wind., Bradea 1: 374. 1973. Cyathea macrosora (Baker) Domin var. vau- pensis (Wind.) A. R. Smith, comb. nov. Sphaeropteris macrosora (Baker) Wind. var. vaupensis Wind., Bradea 1: 374. 1973 Cyathea thysanolepis (Barrington) A. R. Smith, comb. et stat. nov. Trichipteris demissa (С. Morton) R. Tryon var. thysanolepis Barring- ton, Rhodora 78: 1. 1976 Cyathea liesneri A. К. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dpto. Rio Negro, Cerro Aracamuni, summit, Popa camp, 01°26'N, 65°47'W, 1,550 m, 16 Oct. 1987, Liesner & Delascio in dwarf forest, 21964 (holotype, UC 2 sheets; isotype, MO not seen). Figure 2K-N. Probabiliter C. williamsü (Maxon) Domin et C. cyclo- bus, paraphysibus densioribus paleis costarum ovatiori- us, brunneis, nitentibus Rhizome not known, plants probably acaulescent or with a short trunk; fronds ca. m; stipe lacking spines, dark brown to atropurpureous, ca. 90 x 1.2 cm, the basal third with spreading, lan- ceolate scales to 1.5 cm х 2 mm, these with a dark brown to blackish central portion and narrow tan to whitish margins, cells toward the margin flaring outward and ending in short cilia; stipe scurf (microscales) seemingly lacking o or sparse; rachis tan, becoming alate distally; lamina 2-pinnate at the base, pinnate-pinnatifid distally, the pinnae pro- gressively shortened and less divided, the lamina ultimately ending in a pinnatifid apex; pinnae non- articulate, largest ones to ca. 35 X 9 cm, alate and pinnatifid distally; pinnules elliptic, to 5 x 1.2 cm, entire or crenulate (especially toward the tip), narrowed at the base, sessile or becoming adnate distally; veins free, main ones from costules with pairs of strongly ascending veinlets, these running + parallel to margin, slightly raised and readily visible on both sides; costules lacking hairs abaxially, with scattered, shiny, brown, ovate-lan- ceolate scales, these uniseriate at the tip, the mar- gins with ascending, ciliate teeth; sori in a single row on each side of the costules, nonindusiate, borne on the veins ca. 14-14 the way toward the margin, with persistent, rather dense, brownish, filiform paraphyses equal to or slightly exceeding the sporangia; spores whitish, strongly 3-lobe This species is probably most closely related to С. williamsii (Maxon) Domin and С. cyclodium (R. Tryon) Lellinger but differs from both in the bipinnate lamina and longer, more numerous, and less sharply bicolorous stipe scales, denser paraph- be indicative of a hybrid origin, but I c other species from the area that might serve as the other parent. The distinctive blade dissection is sufficient to allow recognition of С. liesneri. Cyathea neblinae A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dpto. Rio Negro, 0-1 km E of Cerro de la Neblina Base Camp on Rio Mawarinuma, 0?50'N, 66°10'W, 140 m, 9 Feb. 1984, on steep Volume 77, Number 2 1990 mith 251 S Pteridophytes of the Venezuelan Guayana: New Species T le WEA oS 22 Ñ ES i We 72, Г a A FIGURE 1. Habit. a Pinna, abaxial surface.—C. Scales from rhizome apex. 31393, US. — D. Habit. — E. Pinna, abaxial surface moss-covered bank, Liesner & Funk 15781 (holotype, UC; isotype, MO not seen). Figure 2F-J Differt a C. pungenti (Willd.) Domin trunco minus quam 30 cm longo, frondibus plerumque 0.5-1.0 m lon- gis, pinnato-pinnatifidis vel bipinnatis, pinnulis integris vel crenulatis, segmentis obliquioribus, falcatis, axibus abax- ialiter epilosis, paleis costarum sparsis Rhizome erect, caudex 7-30 x ca. 1-2(-2.5) cm; fronds 50-100(-150) cm; stipe with a few spines, 8-30(-60) cm x 2-10 mm, brown to atropurpureous, with lanceolate scales 5-10 mm at the base, these strongly and sharply bicolorous with a dark brown central portion and whitish mar- gins, the marginal cells flaring outward and ending in short cilia; stipe scurf of sparse brownish mi- croscales; lamina chartaceous, drying dark, pin- nate-pinnatifid to bipinnate or less often 2-pinnate- pinnatifid (if the latter then with only 1—4 pairs of lower pinnae pinnate-pinnatifid), apex gradually re- duced, not pinnalike; pinnae nonarticulate, largest New species of Asplenium. A-C. Asplenium chimantae, Huber & Steyermark 6927, UC.—A. D-E. Asplenium cowanii, Cowan & Wurdack ones 5-15(-30) x 1.5-4(-13) cm, segments or innules entire to pinnatifid; ultimate segments ide, falcate, acutish at the tip; costae and cos- tules lacking hairs abaxially, with scattered whitish to tan, bullate scales ca. 0.5-1.5 mm, costae winged; veins mostly simple in pinnatifid pinnae, often forked in pinnate pinnae; sori nonindusiate, medial on the veins, with paraphyses much shorter than the spo- rangia. Paratypes. | VENEZUELA. TERRITORIO FEDERA AMAZONAS: same general locality as type, 17 July 1984, river bank with nearly pure bamboo stand, Davidse & Miller 27456 (UC); same general locality, 25-26 Nov. 1984, Liesner 17274 (UC); same locality, 9 Feb. 1984, е: 15798 (UC); same general locality, 160 m, 26 ‚ 1984, Kral 13. (ОС); same general locality, 1.5 sd E of Base Camp, 2-3 Dec , Liesner 17419 (UC); same e locality, 25 Nov. 1984, Croat 59303 (UC); same general locality, 26 Nov. 1984, Bell 310 (UC); same general locality, 3 Dec. 1984, Croat 59596 (UC). This species is most closely related to C. pun- 252 Annals of the Missouri Botanical Garden B 2. лу, > PR ANAL ү лд Sa, VD і pi NS A 0 S (NA In, <) 7 Lun S V Жул ЛЛ Y м US ly. у; x Жум, A MU EN (Ud tage y Lu If PIES RSS 77^ KAP Wirz FIGURE 2. New species of Cyathea. A-E. Cyathea praeceps, Stein et al. 1655, UC.—A. Stipe base. — B. Stipe scale, with detail of margin. — C. Pinna. —D. Pinnule.— E. Abaxial surface of ultimate segments, showing sori and indusia. F-J. Cyathea neblinae, Liesner & Funk 15781, UC. —F. Rhizome and stipe bases. — C. Stipe scales, with detail of margin. — H. Proximal pinna.— I. Distal pinna. —J. Ultimate segments, showing sori and venation. K-N Cyathea liesneri, Liesner & Delascio 21964, UC.—K. Proximal pinna. detail of margin. — N. Base pinna, showing sori and venation. —L. Distal pinna.—M. Costal scales, with Volume 77, Number 2 1990 Smith 253 Pteridophytes of the Venezuelan Guayana: New Species gens (Willd.) Domin, with which it shares very similar bicolorous stipe-base scales, nonindusiate sori with short paraphyses, and mostly unbranched veins. Barrington has determined two of the col- lections cited, including the holotype, as young fertile plants of that species as Trichipteris procera (Willd.) R. Tryon. Cyathea neblinae differs from C. pungens in the smaller size of the plants with fronds mostly 0.5-1.0 m; the short, thin trunk less than 30 cm; the lamina commonly merely pinnate- pinnatifid or bipinnate with entire to crenulate pin- nules; the more oblique and falcate ultimate seg- ments (segments only slightly oblique, nonfalcate, and rounded at the tip in С. pungens); the essen- tially hairless, often darkened axes abaxially; and in the rather sparse costal scales. The last three paratypes cited have 2-4 lowermost pairs of pinnae pinnate-pinnatifid before the pinnae become merely deeply pinnatifid. All of the collections seen are from the base of Cerro de la Neblina. Cyathea praeceps A. R. Smith, sp. nov. TYPE: enezuela. Territorio Federal Amazonas: Dpto. Rio Negro, Cerro de la Neblina, 6.5 km SSW of base camp, southern extension of range, 0%47'N, 66°11’W, 1,600 m, 18 Арг. 1984, dense rain forest with Euterpe palms, Clusia, and Cyclanthaceae, Stein, Thompson & Gen- try 1655 (holotype, UC; isotype, MO not seen). Figure 2A- A C. platylepis (Hook.) Domin indusiis hemitelioidiis, axibus abaxialiter glabris vel sparsim pilosis, frondibus monomorphis differt Weakly arborescent, with trunk ca. 50 cm; fronds ca. 1-1.4 m long; stipe lacking spines or muricate, brownish to atropurpureous, at the base with dense, persistent, lustrous ovate scales to 25 6 mm, with a brown central portion and with slightly and gradually lighter brownish mar- gins, cells toward the margin flaring outward and ending in tan to whitish cilia; stipe scurf lacking; lamina subcoriaceous, 2-pinnate-pinnatifid to bare- ly 3-pinnate; pinnae nonarticulate, the largest ones up to ca. 30 cm; pinnules pinnatifid to barely pinnate, largest ones 3.5-6 x 0.6-1.7 cm, sessile or the largest stalked to 2 mm; segments entire to crenulate; costae abaxially lacking hairs or sparsely long-hairy, bearing scattered, brownish, ovate scales; veins forked at the sori; sori inframedial to medial, at the base with a small, tan, hemitelioid indusium not surrounding the sorus, also with tan, filiform paraphyses equal to or exceeding the spo- rangia; spores beige. Paratypes. VENEZUELA. TERRITORIO FEDERAL AMAZONAS: Dpto. Atabapo, Cerro de Marahuaca, between Rio Yameduaka arriba and base of cliff, 3238'N, 65°28'W, 1,225 ш, 20 Feb. 1985, Liesner 17742 (UC); Dpto. Rio Negro, Cerro de la Neblina, Camp 3, NW Plateau 13.5 km ENE of Base Camp, 0%54'N, 66*04' W, Liesner 16082 (UC); Dpto. Atabapo, Cerro Marahuaca, SE slopes, below Salto Los Monos, 3°35'N, 65?23'W, 1,500-1,600 m, 20 Oct. 1988, Liesner 25120 (UC). The second paratype cited was determined by R. Tryon in 1986 as **Cyathea— very close to or a variant of C. platylepis." This may indeed be its closest relative, but it differs from that in the hemitelioid indusium, general lack of hairs on the axes abaxially, and monomorphic fronds. The first paratype cited contains the following note, which suggests the specific epithet: “Please forgive condition [of specimen]. When the bag and I started falling down a 10 т, 70 degree slope, I was more concerned about me.” GLEICHENIACEAE Sticherus tepuiensis A. R. Smith, sp. nov. TYPE: Venezuela. Bolivar: Meseta del Jaua, Cerro Sarisariñama, porción nor-este, bosque enano por encima y al borde de la Sima Mayor, 4?41'40"N, 64?13'20"W, 1,320 m, 13 Feb. 1974, Steyermark et al. 109036 (holotype, US). Figure 3E-J. Ex affinitate S. bifidi (Willd. ) Ching et specierum af- i l finium, lamina a a vel rubro- P paleis costarum rubro-brunneis, denticulatis, dentibus < mm distinguendus. Plants to 4 m or more long, sprawling; stipe brown or purple-brown toward the base, exceeding 3 mm diam., hard and brittle, glabrous; lamina with 2 or more pairs of opposite branches, the internode between them ca. 25 cm, primary branches forking again to give rise to secondary branches, each of these forking yet again to form deeply pinnatisect tertiary branches; each fork with a usually dormant bud at the base; dormant-bud scales dark brown to purplish brown, concolorous, lanceolate, ca. 2 x 0.2-0.3 mm, weakly ciliolate along the margin, cilia less than 0.1 mm; axes proximal to the penultimate segments tan to brown- ish, lacking indument or with very sparse small scales, commonly lacking segments, or sometimes with a few segments below the most distal forks; penultimate segments to ca. 40 cm X 2 cm; ultimate segments linear, perpendicular to the costa, ca. 2 mm wide at their base, separated by 254 Annals of the Missouri Botanical Garden FIGURE 3. New species of Sticherus в. ан е A-D. dua B. Portion of lamina, showing venat E-J. Sticherus gs sis, Steyermark et al. 1090. A US. —E. lamina at fork. . Segment, adaxial surface. scale. half to nearly their entire width, acute at the tip; costae reddish brown, bearing persistent, reddish brown, lanceolate scales to ca. 10 х 2 mm, these denticulate at their margin, teeth < 0.1 mm; lam- inar tissue glabrous adaxially, abaxially with tan, matted, crispate hairs nearly obscuring the lamina; veins 1 -Ѓогкеа at or near their base, with up to ca. 25 pairs per segment; sporangia 3 or 4 per sorus. BOLÍVAR: Paratypes. VENEZUELA. H. Segment, abaxial binos. Meseta del Jaua, Cerro Sarisarinama, porción sur-oeste, entre la sabana y UC.— Pleopeltis repanda, Steyermark 88177, :ale from rhizome ape D icre oscales on abaxial Шаш: H Portion of abaxial a . I. Bud scale at fork. —J. Costal el salto arriba del vun al este del А del Rio al 4?48'50"N, 34'10"W, 1,800 m, 26 Feb. 1974, Steyermark et 2 109527 (US). TERRITORIO FEDERAL AMAZONAS: summit of Cerro Duida, on moist ridge top, 1,820-2,075m, 4 Sep. 1944, Steyermark 58338 IS). ~ The lamina of this new species dries a rather shiny dark brown or reddish brown adaxially, as shown by all three collections seen. Sticherus te- puiensis has as its nearest allies members of the S. bifidus (Willd.) Ching alliance, which bear mat- Volume 77, Number 2 1990 Smith 255 Pteridophytes of the Venezuelan Guayana: New Species ted hairs on the lamina abaxially. A close relative may be S. brevipubis (Christ) A. R. Smith from southern Mexico and Central America. That species differs in having pectinate or sinuately winged axes proximal to the ultimate forks; bud scales with longer, whitish setae; and costal scales with longer setae often exceeding the width of the scale body. Sticherus tomentosus (Cav. ex Sw.) A. R. Smith, comb. nov. Mertensia tomentosa Cav. ex Sw., Kongl. Vetensk. Acad. Nya Handl. 25: 177, t. 5, fig. 4. 1804. Gleichenia tomentosa (Cav. ex Sw.) Sprengel, Syst. Veg. 4: 27. 1827. Sticherus velatus (Kunze) Copel., Gen. Filic. 28. 1947. GRAMMITIDACEAE? Ceradenia arthrothrix L. E. Bishop & A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Atabapo, Cerro Marahua- ca, above Salto Los Monos, on tributary of headwaters of Rio Iguapo, 3°37'N, 65°23'W 2,555 m, Liesner 18008 (holotype, UC). Fig- ure 4D-G. Filicula parvula paleis rhizomatis minutis ciliis setuli- formibus carentibus etiam lamina perpinnata setis con- spicue septato-nodulosis instructa. Rhizome very small, dorsiventral, with brown scales up to 0.6 x 0.3 mm, cordate at base, these (at least when young) with marginal glands but no setulose cilia; stipe black, 5-10 x 0.1-0.2 mm, set with scattered glandular hairs; rachis similar to stipe, terete, lacking setae; lamina probably pen- ent, perpinnate, up to at least 5 cm long, some- what narrowed toward base, with prolonged apical growth (the complete apex not seen), soriferous from base or with a few basal pinnae sterile; pinnae suboblong or triangular-oblong, 3-5 x 1.5-2.0 mm, rather irregular due to the 2-4 pairs of sinuate lobes, set at 70—85? to the rachis, at base basi- scopically constricted, straight, or shortly decurrent, acroscopically constricted or abruptly notched, at apex broadly rounded, at margin with scattered, conspicuously septate-nodulose, castaneous setae up to 0.7 mm long, otherwise glabrous, neither the costa nor the (always?) simple veins PE on either surface; stomata 38-4 —42 um; sori up to 3 pairs per segment, си medial or slightly supramedial; sporangia 120-140 x 85- 100 um, with 13-15 annulus cells, the distal ones 18-21 um high; spores globose or subglobose, 20— 24 um diam. Paratype. Same locality as type, Liesner 18009 MO). This remarkable little fern was noticed by Alan Smith associated with a specimen of Selaginella. It is apparently a reduced member of the C. semi- adnata group. Like those species, it is perpinnate, and the lamina apex shows extended growth. The tendency of the setae on those species to be septate- nodulose is strongly developed in C. arthrothrix, and these setae constitute its most characteristic feature. Ceradenia microcystis L. E. Bishop & A. R. Smith, sp. nov. TYPE: Venezuela. Bolivar: Dtto. Piar, Macizo del Chimanta, altiplanicie en la base meridional de los farallones superiores del Apacará-tepui, sector Norte del Macizo, 520'N, 62°12’W, 2,200 m, Steyermark, Huber & Carreño 128257 (holotype, UC; isotypes, MO, VEN not seen). Figure 5. Est filix monticola quae in terra ad saxos perve m in ramis v 2-4plo ioribus, pinnis setas marginales et aliquot pilos interspersos gerentibus Rhizome scales reddish brown, 2.5-4.0 x 0.3- 0.7 mm, cordate or cordate-auriculate at base, fairly abrupt at tip, with marginal, hyaline or stra- mineous ciliae, the cells unusually small, medially 25-35-um wide; stipe brownish black, nitid, near base with a few setae and many hyaline or brown- ish, branched hairs, 1.5-3 times as long as the lamina; rachis flexed basally, with sclerenchyma + exposed on both sides, abaxially with scattered, short setae, adaxially with rather crowded, longer setae to 2 mm long; lamina deeply pinnatifid, nar- rowly oblong-triangular, usually broadest at base, with pinnae oblong, 2.5-3.5 mm wide, up to 25 mm long, fertile to the rounded apex, slightly di- lated at base, at the entire or repand margin bearing antrorse setae intermixed with somewhat fewer and shorter branched hairs, on surface mostly glabrous and with the costa prominulous on either side; stomata 48-56 x 44-56 um; each pinna with up to 10 pairs of sori, with capsules 195-220 x 140- 160 um, with annulus of 11—13 bow cells, these 28-32 шт high distally; spores hemispherical or subtetrahedral, 40-48 um in longer diam. Earl Bishop, University Herbarium, University of oia Berkeley, California 94720, U.S.A., is principal rL, author of the descriptions of three species in Grammitidace 256 Annals of the Missouri Botanical Garden es AL, Са NN ` Sd BN СА Dan => NE Р y Nal FIGURE 4. New species of Grammitidaceae. A-C. Grammitis sinuosa, Maguire 24556, NY.—A. Habit. — B. Scale from rhizome apex.— C. Pinnae, abaxial surface. D-G. Ceradenia arthrothrix, Liesner 18008, UC.— D. Нари. — E. Scale from rhizome apex. —F. Pinna apex and margin.—G. Pinnae, abaxial surface. H-J. Grammitis plicata. —H. Habit. —I. Scale from rhizome apex.— J. Pinnae, abaxial surface. K-L. Grammitis peritimundi, Steyermark et al. 128719, UC. —K. Habit, with details of indument. — L. Scale from rhizome apex. — M. Lamina, abaxial surface. N-Q. Grammitis liesneri, Liesner 18241, UC.— N. Habit. —O. Scale from rhizome apex.— P. Pinnae, abaxial surface. — Q. Pinnae, adaxial surface. Volume 77, Number 2 1990 Smith Pteridophytes of the e G 257 uayana: New Specie FIGURE 5. Ceradenia microcystis, Steyermark et w 128257, UC.— A. Habit. — B. Rhizome scale, with detail. — C. B abaxial surface, with details of marginal hai The epithet refers to the unusually small cells of the rhizome scales. Paratypes. VENEZUELA. TERRITORIO FEDERAL AMAZONAS: Dpto. Atures, 8 km NW of Yutaje, 5?41'N, 66?10'W, 1,500-1,760 m, Liesner & Holst 216394 (MO, UC). BOLÍVAR: Chimantá Massif, upper falls of Rio Tirica, 1,950 m, n & Wurdack 557 (F, NY, UC); Dtto. Pia cizo del Chimantá, Apacará-tepui, ca. 2,200 m, Stey о et al. 126524 (UC, VEN not seen). This species is most distinctive in the small cell size of its castaneous rhizome scales and the long- stiped fronds whose laminae bear setae and branched hairs on the margin. In the Guayana area, C. spixiana, mostly of lower elevations, is most similar and differs in its rhizome scales with larger cells and concolorous cilia, its lack of exposed scleren- chyma on either face of the central rachis, and the lack of branched hairs among the marginal laminar setae. Grammitis blanchetii (C. Chr.) A. R. Smith, comb. nov. Polypodium blanchetii C. Chr., Bot. Tidsskr. 25: 78. 1902 Polypodium nanum Fee, Gen. Filic. 238. 1852, not Grammitis nana Fée, 1853, nec Brackenr., 1854. Grammitis liesneri A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dpto. Atabapo, slope of Huachamacari, 3?39'N, 65?43'W, 1,000-1,300 m, 5 Mar. 1985, Liesner 18241 (holotype, UC). Figure 4N-Q 258 Annals of the Missouri Botanical Garden Ex affinitate С. (ui den (Jenman) Proctor et spe- cierum afhnium pi vatis, adiu us, 4- Splo lon- ioribus quam latioribus, lamina a sparsim setosis, paleis rhizomatis aureis, setis marginalibus nm coloratis dis- tinguenda. Plants on cliffs; rhizome ascending, up to 3.5 cm long, ca. 0.5-1 mm diam., at apex with ovate- lanceolate, stramineous to golden, nonclathrate, setose-margined scales 1 - 1.5 х mm, scales decidedly cordate at base; fronds 3-5 cm long; stipe up to ca. ¢ mm, atropurpureous, glabrous or with scattered dark reddish brown setae to 2 mm; lamina deeply pinnatisect to fully pinnate, proximal pinnae somewhat reduced; rachis reddish brown, glabrous or sparsely setose with dark red- dish brown hairs to 2 mm; pinnae entire, linear, parallel-sided for most of their length, to ca. 5 x mm, adnate and strongly decurrent at the base, ascending ca. 70—80° from rachis, with sinuses up to 2 mm wide; costae unbranched, glabrous or with an occasional single seta toward the base, ending adaxially in а clavate hydathode; laminar tissue glabrous on both sides, chartaceous; sori one per pinna, round, superficial, on the basal !4 of costa near rachis, without paraphyses. This species is a most distinct member belonging to the group of С. taenifolia (Jenman) Proctor and G. blanchetii but differs from those and other similar species in the one-nerved pinnae, linear segments 4-5 times longer than wide, and golden rhizome scales with similarly colored marginal setae (setae and often scale margins darkened in the two allied species). Grammitis liesneri is less hairy than most of its closest relatives. It is named for its collector, who has made numerous important col. lections of pteridophytes from the Guayana. Grammitis peritimundi L. E. Bishop & A. R. Smith, sp. nov. TYPE: Venezuela. Bolivar: Dtto. Piar, Macizo del Chimanta, pequeñas altiplani- cies en la base septentrional de los farallones superiores del Amuri-tepui, 5?10'N, 62°07'W, m, Steyermark, Huber & Carreño 128719 (holotype, UC; isotypes, MO, VEN not seen). Figure 4K-L. Grammitis major epiphytica vel lithophytica quae in Scuto Guayano endemica est, paleis rhizomatis brunneis, lamina plerumque lineari-elliptica usque ad 26 cm longa, hydathodis laminae adaxilaris distincte et regulatim evi- dentibus. Rhizome scales medium brown, up to 4 x 0.6 mm, cordate-auriculate, entire, medial cells 40- 80 x 15-30 um; stipe brown or blackish brown, glabrous, 0.3-0.6 mm wide at very base, soon alate distally; costa prominulous on either side, at times with the sclerenchyma exposed abaxially; veins well sclerenchymatized, somewhat prominu- lous on either side, ascending 35-50" to the costa, the sterile ones simple, adaxially with distinct hy- dathodes regularly developed at their distal ends; lamina 3-7 mm wide, fertile from the middle, oc- casionally to the tip but more often with a pro- longed, acuminate, sterile apex; marginal scleren- chyma substriolate, 1.0-1.5 mm wide; hairs on margin when young usually branched, at maturity usually lost, those on midrib mostly simple, cate- nate or brown, 3-5-celled with terminal cell often enlarged; stomata 52-62 x 50-55 um; sori ex- tending from near costa to less than halfway to margin; capsules broadly obpyriform, 80-90 x 72-80 um, with 9-11 annulus cells, the distal ones 30-38 um high; spores minutely papillate, subglo- bose, 22-26 um diam. Paratypes. VENEZUELA. BOLÍVAR: T Piar, Muri- sipan-tepuí, summit, lithophyte in deep s " 300 m, Holst 3531 (UC); Mt. Roraima s summit, 8,600 McConnell & Quelch 568 (US); ; Roraima Ex- pe s 307 (US) Chimantá oe ннн in wet soil in waterfall spray, 1,880- 1,970 m, Steyermark 4 6 (US); ds e epiphyte on slender tree along stream, 1,922 n, Steyermark 98005 (US); о сгеуісеѕ of e 2,360-2,420 m, Stey- Е 8 et al. 115689 (ОС); Macizo del Chimantá, Apa- -tepui, crevices of bluff, 2,200 m, Steye ond et al. | 285. 34 (UC); Macizo del Chimantá, Amuri tepui, epi- phyte in low forest along stream, 1,850 ш et al. 128631 (MO, UC); Auyan-tepui, 1, 850 m, ед chi & Foldats 4817 (US); Auyan-tepuí, 2,300 m, Va- reschi 4928 (US). TERRITORIO FEDERAL UC); Cerro de la Neblina, epiphyte, 1 1,850 m, Liesner 16080 (UC); Atabap TO epiphyte, 1,720 m, Liesner 18133 ^t 1); Cerro de la Neblina, epiphyte ie tree tru man & Thomas l. 36: 35 (UC r crevices, 2.520-2.650 m, Steyermark & Holst E 3081 2 (UC); Atabapo, Cerro Marahuaca, epiphyte, 1,560 , Steyermark 129618 (UC). This large species is most similar to G. flumi- nensis, with which it shares large lamina size, large stomata, and linear-elliptic fronds. It differs in the timundi differs from the sympatric G. bryophila in its medium brown (vs. golden brown) scales. The epithet is derived from peritus (lost) and mundus (world), in reference to a fantasy by H. G. Wells set on Mt. Roraima. Grammitis plicata A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Ser- гаша Parú, Caño Asisa, Rio Ventuari, cumbre Volume 77, Number 2 1990 Smith 259 Pteridophytes of the Venezuelan Guayana: New Species just S of valley head of Camp Stream, 2,000 т, 2 Feb. 1951, Cowan & Wurdack 31164 (holotype, UC; isotype, NY). Figure 4H-J. G. blanchetii (C. Chr.) A. R. Smith affinis, a qua imprimis differt pinnis кийер, lamina densius setosa subcoriacea plicata conduplicata Plants epiphytic; rhizome suberect, ca. 1 mm diam., at apex with lanceolate, golden brown, non- clathrate scales ca. x 0.2-0.3 mm, scales setose along margin, especially toward apex, setae dark reddish brown; fronds mostly 3-5 cm, tightly clustered; stipe ca. 5 X 0.2 mm, with numerous ark red-brown setae to 3 mm; lamina often con- duplicately folded, pinnatisect nearly or quite to rachis, with gradually reduced and more distant proximal segments; rachis brownish to blackish, with numerous setae to 3 mm on both sides; pinnae oblong, 1.5-2 х 1 mm, strongly adnate and de- current, not gibbous, cut to rachis but with bases contiguous, ascending ca. 60° from rachis, sepa- rated by less than 0.5 mm distally to more than 1 mm proximally; lamina tissue subcoriaceous, with stiff setae on both sides; veins simple and un branched, ending in obscure hydathodes adaxially; sorus one per pinna, occupying the basal half of a vein, confluent with sori of adjacent pinnae and across the rachis, intermixed with stiff setae. es. VENEZU JELA. TERRITORIO FEDERAL pa Cam Delascio 21999 (UC); Dpto. Rio Negro, Cerro Void muni, summit, Proa Camp, 1?32'N, 65%49"W, 1,400 30 Oct. 1987, Liesner & Carnevali 22662 (UC); poo Atures, valley of Rio Coro- Coro, W of Serrania de Yutaje, W of river, 5?41'N, olst & Liesner 3456 (MO, UC); Dpto. cu slope of Cerro de Mara- 3°38'N, 65°28'W, 1,225 (UC). BOLÍVAR: Cerro Marutani, Dtto. Her 'N, 62°15’ W, 1,200 m, 11- 14 Jan. 1981, Die uia et al. 123952 (MO). This species seems most closely related to G. blanchetii but differs by having one-nerved pinnae, more densely and longer-setose lamina, and sub- coriaceous, usually conduplicately folded lamina. Grammitis plicata grows at higher elevations than С. blanchetii, which is typically a lowland species. Grammitis sinuosa A. R. Smith, sp. nov. TYPE: Surinam: Tafelberg, 29 Aug. 1944, Maguire 24556 (holotype, NY). Figure 4A-C. Ex affinitate C. suspensae (L.) Proctor et specierum p margine pinnarum anai soris ега еср аа nec restri a in naru E latiore et breviore, textura Da dae Plants epiphytic; rhizome suberect, ca. 1 mm diam., at apex with lanceolate, brown, clathrate, glabrous, entire scales ca. 2-3 x 0.2-0.3 mm; fronds tightly clustered, 8-12 cm long; stipe 0.2- 0.3 mm diam., faintly winged, glabrous; lamina deeply pinnatisect to barely pinnate, with gradually reduced proximal pinnae, ultimately reduced to a narrow, sinuate wing extending to the rhizome, distally reduced gradually to a pinnatifid apex; rachis atropurpureous to dark reddish brown, glabrous or with a few tan hairs 0.1—0.3 mm; pinnae sinuately lobed or pinnatifid, mostly 8-15 strongly adnate and decurrent, rounded at the tip, cut to the rachis but with bases contiguous, as- cending ca. 60° from rachis, separated their width or sometimes slightly more; costae 3-5 mm apart, brownish, sinuous, glabrous; laminar tissue gla- brous on both sides; veins 3-5 pairs per segment, ending well before the margin in clavate hydathodes adaxially; sori 2-4 pairs per pinna, round or slightly oblong, slightly sunken or superficial, without pa- raphyses mm, Paratypes. GUYANA: Essequibo River, 1899, Jen- man s.n. (NY). VENEZUELA. BOLÍVAR: Dtto. Piar, summit of Amaruay-tepui, S side of E half, 5955'N, 62°13'W, 950-1,100 т, 11 May 1986, Liesner & Holst 20812 (MO, UC). TERRITORIO FEDERAL AMAZONAS: Cerro 3 ap Maguire & Politi 27710 (UC); Dpto. М зас near base of Duida, 3°34’N, 65°32’W, 1,000 m, 25 Oct. 1988, Liesner 25394 (UC); pis Atabapo, Cerro Hua- chamacari, E slope, 3?49'N *42'W, 600-700 m, З Nov. 1988, Liernar 25778 TA This species may be related to G. suspensa (L.) Proctor, but the latter differs in having entire pin- nae, sori more deeply sunken in pits and tending to be localized at the pinna tips, thicker texture, and longer, narrower laminae. At a higher level, G. sinuosa is a member of the group with entire, clathrate rhizome scales comprising such species as G. moniliformis (Lag. ex Sw.) Proctor, G. firma (J. Smith) C. Morton, and many others. POLYPODIACEAE Pleopeltis repanda A. R. Smith, sp.nov. TYPE: Venezuela. Bolivar: Altiplanicie de Nuria, E of Miamo, upper part of W-facing wooded slopes, 8 Jan. 1961, Steyermark 88177 (ho- lotype, UC). Figure 3A-D. a P. macrocarpa (Bory ex Willd.) Kaulf. lamina ыша glabra vel sparsim squamata, soris paraphy- sibus et squamis carentibus, margine laminae undulato vel repando. Plants epiphytic; rhizome creeping, ca. 3 mm diam., bearing dense, lanceolate scales ca. 2-3 x 260 Annals of the Missouri Botanical Garden 0.3 mm with a broad, blackish central stripe and narrow, tan, erose-denticulate margins; fronds to 30 cm long, undivided; stipe up to ca. 5 cm X mm, brownish, + glabrous; lamina coriaceous, nar- rowly elliptic but broadest in the basal 14, cuneate at the base, 1.7-2.8 cm wide, acute at the tip, with a sinuate or repand margin, especially in fertile portion of lamina, crenations slightly broader than the sorus length, 1-2 mm deep; rachis stramineous to tan, glabrous or with a few minute scales to 5 mm long; laminar tissue glabrous on both sides, or abaxially with a very few minute, tan to brownish, ovate-attenuate scales to ca. 0.5 x 0.1-0.2 mm; sori confined to the distal 14-24 of the lamina, in a single supramedial to submarginal row on each side of the rachis, nonindusiate, oblong, to 5 x 3 mm, lacking scales and paraphyses, even when young. Paratype. | VENEZUELA. BOLÍvAR: Río Samay, Holst & бо 2434 (МО). This species differs from all other simple-bladed Pleopeltis species in the Neotropics by the abax- ially nearly scaleless lamina and by the lack of scales or paraphyses in the sori. It also has a more undulate or sinuate margin than its closest relatives, 1 e.g., Р. macrocarpa (Bory ex Willd.) Kaulf. PTERIDACEAE Adiantum amazonicum A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazo- nas: Dpto. Atabapo, SE bank of middle part of Cano Yagua at Cucurital de Yagua, 03°36'N, 66°34’ W, ca. 120 m, 8 May 1979, нең forest with Leopoldinia palms very ominant, Davidse, Huber & Tillett 17429 VAR UC; isotypes, MO, VEN not seen). Figure 6C- A. cajennensi Klotzsch et 4. Tetraphyllo Humb. $ Bonpl. ex Willd. maxime simile, sed pinnulis integris, ad apicem rotundatis, mae dindi squamis rhachidis valde dentatis aut seto Rhizome short-creeping, ca. 3-4 mm diam., with internodes 2-3 mm apart; rhizome scales dark brown or blackish with lighter margins, denticulate on the margin, 1-1.5 x 0.2-0.3 mm; fronds to . 65 cm long; stipe to ca. 40 cm x 1-2 mm, dark reddish brown to atropurpureous, panei glabrescent; rachis similarly colored and w merous reddish brown hairs and linear- pius nu- setose-margined scales to ca. 2 x 0.1 mm; lamina 2-pinnate, with 2—5 pairs of lateral pinnae and a similar terminal one; pinnae l -pinnate, the pinnules sessile or stalked less than 1 mm at the pinna base, oblong, dimidiate, sterile margins entire or cren- ulate, rounded at the tip, straight; laminar tissue abaxially with scattered reddish brown scales with an expanded, stellate base; veins forking, free; sori in oblong patches along acroscopic margin and tip of pinnule, mostly 8-12 per pinnule; indusium adaxially with reddish brown hairs. Paratypes. iris ири track from Mutum- paraná to Rio Madei O Nov. 1963, Prance et al. 9012A (NY not seen, EN (A 90- 93, Madeira-Mamore ere near Jaciparaná, 30 June 1968, Prance et al. 99 (NY not seen, UC). This is most similar to 4. cajennense Klotzsch, A. fuliginosum Fée, and A. tetraphyllum Humb. & Bonpl. ex Willd. From all three it differs in having oblong, more or less entire pinnules with rounded tips and indusia with reddish brown hairs adaxially. From 4. fuliginosum, А. amazonicum also differs in the less densely scaly rachis but it agrees in having the rachis scales toothed or setose. This contrasts with the condition in А. cajennense and A. tetraphyllum, which have nearly entire, linear or hairlike rachis scales. Adiantum nudum А. К. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dpto. Rio Negro, E of Cerro de La Neblina Base Camp, along Rio Mawarinuma, 0?50'N, 66°09'W, 140 m, 2 May 1984, along stream banks, Thomas 3326 (holotype, UC; isotypes, MO, NY not seen). Figure 6A-B Differt a 4. petiolato statura minore stipitibus graci- la stipite et sen glabra, rhizomate breviter re- | penti internodiis brevissimis, pinnis oe plerumque 4-jugatis, etiolulis innarum glabris “Jug р à 8 — Rhizome long-creeping, 1.5-2 mm diam., with internodes ca. 1-2 mm apart; rhizome scales cas- taneous, shining, entire, ca. 1.5 X 0.1-0.15 mm; fronds 8-25 cm long; stipe 3-12 cm x 0.4-0.8 mm, atropurpureous to blackish, shining, glabrous; rachis similarly colored, also glabrous; lamina l -pinnate, with 1-4 lateral pinna pairs and a tri- angular terminal segment; pinnae oblong, short- stalked 1-3 mm, 1.5-3.5 x (0.5-)1-1.6 cm, ob- tuse at tip, sterile margins evenly serrulate, at the base inequilateral with the basiscopic side excavate, acroscopic side rounded, pinnae becoming equilat- eral toward the tip; laminar tissue glabrous on both sides; veins forking, free; sori in oblong to linear patches along acroscopic margin, tip, and most of basiscopic margin except for excised basal portion. VENEZUELA. TERRITORIO FEDERAL 1.5 km E of Cerro de la Paratypes. AMAZONAS: Dpto. Río Negro, 261 Pteridophytes of the Venezuelan Volume 77, Number 2 Smith 1990 i Guayana: New Species (1 АЙ Ie, FE, ——G ve TN Y x APIS Po d J ( і C FIGURE 6. New species of Pteridaceae. A-B. Adiantum nudum, Thomas 3326, UC.— A. Habit. — B. Pinnae, abaxial surface. C-E. 4diantum amazonicum, Davidse et al. 17429, UC.—C. Habit. — D. Pinnule, with detail of i is. F-H. Doryopteris cyclophylla, Davidse & Miller 27341, UC.—F. abit. —G. Frond, with detail of sorus, abaxial surface. — H. Frond, with detail of margin to show hydathodes, adaxial surface. 262 Annals of the Missouri Botanical Garden Neblina Base Camp on the Rio Mawarinuma, 0-3 km 5 of main river, 140 m, 2-3 Dec. 1984, stream bank, Liesner 17407 (MO, UC), Link 17416 (MO). BARINAS: along Rio Caparo, 2-4 km up river from dam side, 07941'N, 71?28'W, c Liesner & González 9452 (MO not seen, UC). BRAZIL. MATO GROSSO: 20 km S of Xavatina, 400 m, 15 June 1966, Irwin et al. 17092 (MO). This appears to be most closely related to 4. petiolatum Desv. but differs by the completely glabrous stipe, rachis, and pinna stalks; rhizome with very short internodes (vs. internodes often more than 1 cm in 4. petiolatum); smaller stature of the plants with thinner stipes; fewer pinna pairs (mostly 5-10 in 4. petiolatum); and longer pinna stalks. Doryopteris cyclophylla A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazo- nas: Dpto. Rio Negro, Neblina Massif, Canyon Grande, along Rio Mawarinuma between the mouth of canyon and first major fork of river, ca. 7 airline km ENE of Puerto Chimo, 0°50- 51'N, 66%02-06'W, 400-700 m, on cliff, in cracks, in trickle of water, 9-14 July 1984, Davidse & Miller 27341 (holotype, UC; iso- type, MO not seen). Figure 6F-H Species frondibus minimis rotundatis vel subcordatis, laminis membranaceis a congeneribus diversa. Rhizome compact, suberect, caudex ca. 5-8 rhizome scales tan, concolorous, entire, ca. 1 X mm; fronds monomorphic, 1-3 cm long, densely tufted; stipe to ca. 25 x 0.2-0.3 mm, terete, atropurpureous to blackish, shining, glabrous, widened at the point of attachment to the lamina; lamina subrotund to subcordate, up to 10 x 13 mm, usually a little broader than long, entire or faintly crenulate on the margin, lacking buds at the base; laminar tissue glabrous, thin; veins forking, free, ending short of the margin in distinct, clavate hydathodes adaxially; sori in elongate patches or subcontinuous at the margin, the in- dusium scarious and ca. 1 mm wide, strongly re- mm diam.; flexed This is readily distinguishable from all other Do- ryopteris species by the very small frond with its round or subcordate rather thin lamina. Doryopteris davidsei A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dpto. Atures, 23 km NE of Puerto Ayacucho and 10 km E of hwy., hills and base of hills near Cachama, 5°51'N, 67?24'W, 90 m, 17-19 Apr. 1978, on large, mossy boulders, Davidse & Huber 15315 (holotype, MO Frondes 5-10 cm longae; paleae rhizomatis bicolores; stipes adaxialiter complanatus, pilis minutis 0.1 mm longis praeditus; lamina profunde cordata, 1.8 apicem acuta; venae areolatae, terminantes adaxialiter in hydathodos. Rhizome compact, suberect, caudex 5-8 mm diam.; rhizome scales bicolorous, with a castaneous central stripe and narrower, tan margins, entire, 2.0-2.5 x 0.2 mm; fronds monomorphic, 5- 10 cm long, clustered; stipe ca. 4-7 x 0.3-0.4 m, dark brown to atropurpureous, shining, adax- ially flattened and bearing numerous glandular cas- taneous hairs 0.1 mm long on the flattened portion; laminae narrowly and deeply cordate, up to 3 X 1.8 cm, nearly twice as long as broad, basal lobes rounded, acute or acuminate at the tip, crenulate along the sterile margin, lacking buds at the base; laminar tissue glabrous, thin; veins forming elon- gate polygonal (often hexagonal) areoles in 2-4 series, the ultimate veins ending short of the margin in distinct, clavate hydathodes adaxially; sori con- tinuous along the basal and outer margins but with distal % of lamina sterile, the indusium scarious and ca. 0.4 mm wide. The affinities of this species are uncertain. Be- cause of the simple, deeply cordate blades, it will key to near D. sagittifolia (Raddi) J. Smith, but D. davidsei has much smaller fronds, round rather than acute basal lobes, and thin texture. The small, thin fronds suggest relationship with D. cyclo- phylla, but that species differs in having free veins; lamina as broad as or broader than long; tan, con- colorous rhizome scales; and a wider indusium. SELAGINELLACEAE ШЕ albolineata А. К. Smith, sp. nov. YPE: Venezuela. Territorio Federal Amazo- nas: Dpto. Atures, Rio Coro-Coro, W of Ser- гаша de Yutaje, 6-8 km N of settlement of Yutaje, 05°41'N, 66%07'30"W, 320 m, 23 Feb. 1987, Liesner & Holst 21341 (holotype, UC; isotype, MO). Figure 7A-E. Ca repentes, 0.2-0.3 mm diam., non articulati, apicibus flagelliformibus et sobolibus carentes, rhizophoris filiformibus 0.1 mm diam. per totam longitu nai caulis; folia ubique dimorpha, membranacea, utrinque glabra, striis albidis vel dnd entibus longitudinalibus folia i in- termedia өз, neque а ad 1. 0,7 larg [ ; folia le x ].2 mm, ad apicem е integra; ѕро- "ооа pauca lanceolata plerumque patentia laxe ag- gre Plants on moist rocks, ca. 5 cm long, branch system anisotomous and ca. 3 times divided, branches arising along length of main stem; main Volume 77, Number 2 1990 Smith Pteridophytes of the prom Guayana: New Specie 263 eral leaf, with deti Habit. ac ае Selaginella arrecta, Holst & bur 3332, portion. N-R. Selaginella stems 0.2-0.3 mm diam., creeping, Ea bearing filiform rhizophores 0.1 mm diam. alon their length; leaves dimorphous throughout, light green, dull, thin-textured; medial leaves ovate, to ca. 1.0 x 0.7 mm, acuminate at apex, truncate at base, denticulate from a very narrow and not . Hab ubens, Maguire & Politi 27634, UC.—N. surface. — Q. Adaxial surface, detail. — R. Fertile portion. abit. — O. peer surface.— P. Abaxial very prominent whitish margin, with numerous, narrow, whitish or translucent, longitudinal streaks; lateral leaves to 1.8 1.2 mm, rounded at tip and base, inequilateral, planar, entire or nearly so, also with streaks like those of medial leaves; strobili consisting of only a few sporophylls, not tetrago- 264 Annals of the Missouri Botanical Garden nous; sporophylls lanceolate, + spreading, loosely aggregated; megaspores ca. 150 um diam., light yellowish, + smooth; microspores not seen. This is perhaps related to 5. potaroensis Jenman or some similar species, but 5. potaroensis differs in having acute and larger lateral leaves that are often minutely pubescent on the basiscopic side adaxially and in having median leaves less than half the length of the lateral leaves. Another close relative may be 5. porelloides (Lam.) Spring, which has thicker stems and larger, firmer leaves. Both of these possible relatives lack the streaking in the leaves that is characteristic of 5. albolineata. Such streaking has been seen to a lesser extent in other species of Selaginella. The systematic significance of this character is unknown Selaginella arrecta A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dpto. Atures, valley of Rio Coro-Coro, W of Serranía de Yutaje, 05?42'N, 66?09'30" W, 1,250 m, 6 Mar. 1987, Holst & Liesner 3332 (holo- type, UC; isotype, MO not seen). Figure 7J-M . vernicosae affinis, a qua imprimis differt marginibus filiorum intermediorum et lateralium late albidis ca. 6 cellulis latis, foliis intermediis denticulatis ad apicem ari- statis, foliis lateralibus acutis vel breviaristatis. Plants on moist banks and on mossy rocks in or adjacent to streams, to ca. 20 cm; main stems ca. 0.5 mm diam., erect, width with leaves 2-3 mm, sparingly branched, branches simple, strongly ascending, nearly parallel to main stem, nonarticu- late, bearing only a few, short, filiform rhizophores 2 mm diam. at the very base of main stem; leaves dimorphous throughout, dark olive-green above, yellow-green below, firm and subcoriaceous; medial leaves ovate, to 1.8 x 1 mm, short-aristate at apex, truncate at the hidden base, strongly over- lapping, denticulate from a broad whitish margin ca. 6 cells wide; lateral leaves broadly ovate, to 1.8 x 5 mm, strongly overlapping, acute to short-aristate at the apex, denticulate from a broad, whitish margin; strobili tetragonous, to ca. 10 x 1.5-2 mm at tips of lateral branches; sporophylls ovate-lanceolate, to 2 mm long; megaspores not basal, ca. 200-300 um diam., light yellowish, with faintly raised ridges; microspores shed singly, or- ange. 'aratypes. VENEZUELA. BOLÍVAR: cumbre del Cerro Guaiquinima, sector suroeste-central, a lo largo del af- luente suroc aie del Rio Carapo, 05*45'N, 63?35'W, 950 m, 26 May 1978, Steyermark et al. 117478 (MO, UC). TERRITORIO FEDERAL AMAZONAS: Serranía Yutaje, Río Manapiare, Cerro eru along left fork of Cano Yutaje, 1,250 m, 12 Feb. 1953, Maguire & Maguire 35204 (UC); Cerro EM (Paráque), lower Cano Negro, 1,400 m, 25 Dec. , Maguire & Politi 27920 (UC). This is most closely related to S. vernicosa Ba- ker, which differs in lacking broad whitish margins on the nonaristate median and lateral leaves and in having shortly and finely ciliate median leaves. Selaginella beitelii A. R. Smith, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Cer- Phelps, along Cano Gardner, 0?50'40"N, 65°58'10"W, 1,735 m, 30 Jan. 1985, Beitel 85079 (holotype, UC; isotype, NY not seen). Me 8l- aules ca. 0.7 mm bn non articulati, apicibus fla- gellformibus et soboli rhizophoris filiformibus 0.1 mm diam. per totam a caulis Hane folia ubique dimorpha, chartacea, veneta adaxialiter in stat nque glabra; folia We to late ovata, D l. 1, ad apicem acuta, non arista- ta, ad marginem obscure albida vel margine virello; folia laterales late ovata, 4.0 x 1.6-2.5 mm, basi i inae- qualiter rotundata, marginem acroscopicum denticu- ad apicem rotundata vel subacuta; strobili trago sporeghalla ovato-lanceolata, margine denticulata vivo, utri а. 1.5-2.5 x 1.2-1.8 mr ec Plants terrestrial or epiphytic, branch system anisotomous and 2 or 3 times divided, branches arising perpendicularly all along main stems; stems to ca. 0.7 mm diam., prostrate, lacking flagelliform tips or sobols, nonarticulate, bearing filiform rhi- zophores 0.1 mm diam. along their length; leaves dimorphous throughout, bluish green adaxially when living, silvery green abaxially, chartaceous, gla- brous on both surfaces; medial leaves broadly ovate, with a very narrow and inconspicuous whitish bor- der, or whitish border lacking, ca. 1.5-2.5 x 1.2- 1.8 mm, truncate at base, acute at apex, lacking an aristate tip; lateral leaves broadly ovate, 3.0- 4.0 -2.5 mm, unequally rounded at base with acroscopic portion strongly overlapping stem, denticulate along acroscopic margin, rounded or subacute at apex; strobili tetragonous; sporophylls ovate-lanceolate, to ca. 1.5 mm long, denticulate along margin. TERRITORIO FEDERAL AMA , 1,950 m, Boom Paratype. VENEZUELA. Nas: Cerro de la Neblina, Camp 12 al. 6012 (NY not seen, UC). This seems most similar to S. muscosa Spring in С. Martius but differs in the acute, nonaristate median leaves and bluish green color. It is also similar to S. rhodostachya, from which it differs in its broader branches, ca. 5 mm wide, and more overlapping lateral leaves, especially toward the Volume 77, Number 2 Smith 265 1990 Pteridophytes of the Venezuelan Guayana: New Species P $2 ae "E 4 2: E) СА Ws Ir «x QE | ац т m E г 5 2 м2 e. = == VS са; = 2 * NU; TX 2 Aou SE 4 ENG z $ € VE Y Wy >) y = 3 2 a NS = NS = M “2 ES AA E Ye к=, CYS W “2 V JAS 2 S Куз N? Уа ¡E SA YE 2 сў y WZ М Wy, NZ = AN ANS (^ We N : NE < PEU A, Миии, SE n NS PTI CN (i EF WA WW ne SE * un Eo VV i VE pu. = AY = eh = WAYS NE W E : es E ^ A SÉ ub к> ST dem S4 i С * oy ч, yo | 2 : ay, - MANI Re NS SEK, Ne voee SS NES SS I ЖОМ. NE а > > д? ОЙ Ж р 7 " wm A Y F Sat! ү Un (9 А g. FIGURE 1. Ctenitis scales and trichomes. a, b. Ctenitis microchlaena —a. Segment margin, with trichomes (Peru, Belshaw 3503, F).—b. Rachis scale (Peru, Belshaw 3503, F).—c. С. lo: pee scale (Guatemala, Standley , F).—d. C. lanceolata, costa scale (Guatemala, Standley 91628). —e. С. grisebachii, petiole scale (Guatemala, Steyermark 35102, ever, where the tips end in relation to the margin the segment midrib (simple vs. one-forked), as well does not appear to be very important in distinction as some difference in vein tip termination (nearly of species within Ctenitis. Between a few species at the margin vs. well short of the margin). How- there is some variation in the forking of veins from ever, in neither my Guatemalan nor Peruvian stud- 278 Annals of the Missouri Botanical Garden ies have I found these features to be constant or greatly effective. Ctenitis microchlaena has basal veins of segments terminating at the margin well above the sinus, which helps separate it from C. refulgens and С. submarginalis, in which basal veins end short of the margin or reach it at or near the sinus. Compared with some other diag- nostic features in the genus this is admittedly a subtle one. Nevertheless, I suggest it is worth more careful scrutiny in future research in Ctenitis. LAMINA Most regional treatments of the genus have used the degree of dissection of lamina as an important key character. Least dissected laminae in Ctenitis are one-pinnate-pinnatifid, such as C. refulgens, С. salvinii, С. submarginalis, and С. vellea (Willd.) Proctor (Greater Antilles); whereas some of the larger-leaved species can be as much as four-pin- nate-pinnatifid: С. equestris and С. excelsa. This separates species into artificial groups, since lamina dissection is intermediate in other species, varying from two- to three-pinnate-pinnatifid, and the di- viding line can be shifted nearly anywhere an au- thor wishes. Many species of Ctenitis have enlarged, as well as more highly dissected, basal pinnae, such as С. hemsleyana (Hemsley) Copel., in which all except the basal pinnae are essentially one-pinnate-pin- natifid. Basal pinnae may be nearly three-pinnate, with basiscopic pinnules about the size and shape of adjacent pinnae. Consequently, caution should be used in writing and/or interpreting keys to such species. Furthermore, an herbarium specimen of a larger-leaved species can be misleading if it con- tains only the basal or apical portion of the lamina, without label data explaining the remainder of the lamina architecture. SUMMARY All of the taxonomic characters discussed above may be variously effective in identifying the species of Ctenitis if the indicated cautions are heeded. However, groups may be artificial ones if separated by degree of lamina dissection, forking of veins, type of trichome, presence or absence of indusia, and soral position. My studies of Ctenitis in Central and South America, appear to corroborate Christensen’s the- ory that scale type is one of the best indicators of relationship. Regardless of lamina dissection or presence or lack of indusia, species with any of the following scale types probably have strongest af- finity: (1) dark, filiform, with abruptly expanded base (Fig. 1с); (2) broad, flat, delicate, with fine cell walls and conspicuously clathrate (Fig. 1b); (3) broad, bullate, with attenuate tips (Fig. 1d); (4) linear, rigid, with thick cell walls, not or scarcely clathrate (Fig. le). Holttum (1985) believed that two natural groups of Old World species can be distinguished by their scales. Several other characters might well be strong indicators of species affinity but need further com- parison across the full range of distribution and should be assessed for degree of constancy: (1) basal veins of segments extending to near, or well above the sinus, and/or reaching or terminating short of the margin; (2) glands (color, shape, orientation) present or lacking on indusia or laminar surface; 3) segment margins with trichomes present or _ lacking; (4) spore morphology. Since Megalastrum and Triplophyllum have been separated from Christensen's five original groups of Ctenitis, it is now necessary to determine if the remaining three constitute natural divisions. By far the largest of these is the submarginalis group, which was distinguished chiefly by simpler lamina architecture, i.e., leaves no more than one- pinnate-pinnatifid. However, this may be an arti- ficial grouping, for too many of the species widely differ in the more significant characters mentioned above. The next largest group, Airta, contains species with three different scale types (1, 3, and 4. above), both ciliate and eciliate segment margins, and different types of glands. The smallest group, ampla, is distinctive at least in that its species all have type 2 scales. It thus appears that the latter may be the most natural group, and species com- plexes of the others should be examined to rede- termine their proper relationships. Among all the different scales, types 2 and 3 seem to be most closely related. Although the for- mer is flat and the latter bullate, both are con- spicuously clathrate with broad cells. The other two types have very small, or very narrow and crowded cells. Future studies should ascertain whether species of type 2 (ampla) and 3 (some species of hirta) share other significant features. Tryon & Tryon (1982) concluded that Christen- sen's hirta group had too many diverse characters and divided its species between the other two, prin- cipally supported by features of the spore surface. Therefore, it will be important to learn whether species with scale types 2 and 3 have the same kind of spore ornamentation. If so, this would ef- fectively delimit two discrete species groups in the place of the three polymorphic groups earlier rec- ognized by Christensen. Volume 77, Number 2 1990 Stolze Ctenitis and Allied Genera in America 279 CONCLUSION The last six decades have produced dramatic results in the understanding of species groups once gathered under the umbrella of “Dryopteris.” With the impetus furnished by Christensen, such genera as Ctenitis, Cyclodium, Lastreopsis, Megalas- trum, Stigmatopteris, Thelypteris, and Triplo- phyllum have been correctly set apart, and many of their species have been clearly delineated. Def- inition of Ctenitis, in turn, recently has been sim- plified by the separation of Megalastrum and Trip- lophyllum. Yet this important pantropical e of ca. 80 species still needs major revision. Maj diagnostic characters have been highlighted жаз based on knowledge I have gained from the studies of the genus in two floristic treatments, which in- cluded a canvass of most of its neotropical species. These characters have been subjected to close anal- ysis and evaluated for efficacy, and suggestions have been made for their future correlation. It is hoped that this information will help in floristic studies and facilitate monographic work on the species remaining in Ctenitis. LITERATURE CITED CHRISTENSEN, C. species of Dryopteris. 8: Warmin Biol. Arbej On a natural classification of j. til. Eug 13. A monograph of the genus Dryopteris, Part 1. The tropical American pinnatifid-bipinnatifid species. Kongel. Danske Vidensk. Selsk. Skr. Na- dp Afd. Ser. 7, 10: 55-282. A monograph of the genus Dryopteris, EN 2 The tropical — Mt decompoun p Kongel. Dan nsk. Selsk. Skr. Na- turvidensk. Ad. Ser. 8. 6: 3-13 2. 1938. Filicinae. Pp. 522-550 in Verdoorn, Manual of Pteridology. Nijhoff, The Hague. HoLTTUM, В. E. 1985. Pa on in the fern- "genera allied to Tectaria Cav., 4. The genus pu in Asia, Malesia and the western Pacific. Blumea 31: 1-38. . Studies in the fern- pii. allied to Tectaria Cav., 5. Triplophyllum, a new genus of Africa and America. Kew Bull. 41: 237-260. Studies in the fern- -genera allied to Ticiaria Cav World regarded a as related to Tectaria, tions of two genera. Gard. Bull. Straits Settlem. 39: 6T. "er T. BEITEL. 1988. Ctenitis. Pp. -135 in ^ Prop Flora Oaxaca, Mexico. Mes. New n t. Gard. 46: 8. MORAN, В. C. 1986. "The vesical fern genus Ol- fersia. елы Fern J. 76: 161- 1 987. Monograph of the а fern ge- nus Poly botrya (Dryopteridaceae). Bull. Ш. Nat. Hist. Surv. 34: 1-138. SMITH, A. R. Ctenitis. Pp. 77-85 in D. Breed- love (editor), йек qe Part 2, Flora of Chiapas. Calif. Acad. S -370. . C. Moran. 1987. New combinations in Megalastrum (Dryopteridaceae). Amer. Fern J. 77: 24-130. STOLZE, R. G. 6. Ctenitis. Pp. 143-161 in Ferns and Fern Allies of M pU Part 2. Polypodiaceae. Fieldiana, Bot., n.s. 6 Е Tryon, К. М. € К. С. STOLZE. Pteridophyta of Peru, Part IV: a Fieldiana, Bot., n.s. (in prep.). A. К. Tryon. 1982. Ctenitis. Pp. 5 567 in Ferns and Allied Plants, with Special Ref. erence to Tropical America. Springer-Verlag, New York. EPILOGUE Although I had been acquainted with Alice and Rolla Tryon earlier, it was not until 1970 that our relationship developed into friendship. I had come to Harvard to ask Rolla’s advice in selecting a genus for monographic study, my briefcase laden ый potentials. We discussed them, one after another, and dismissed them, one after another. Finally he suggested one I hadn't considered: “Bob, here's a small, oi ies genus that really needs work, of just the right size, scope, and diffi perfectly!” Once P'd been convinced, had then as I do now, I would have properly interprdiel the twinkle in his eye to mean, **Aha! Another candidate to cooperate on tree-fern research!" e genus was Cnemidaria, and my ultimate publi- cation was to be but one of many prompted by Rolla on the tree-ferns, a group that had long fascinated him. He has been the catalyst for dozens of papers on the subject, from standpoints of morphology, anatomy, and cytology, to virtually жее other botanical discipline. With these studies also has come controversy — students and experts wie theories on the phylogeny or taxonomy of tree- A sult? Probably no group of ferns has received so muc e attention from so many workers in so many е this with ids that had pce been almost e ignored and scarcely understood My per ы visits to the Gray Herbarium also dem- onstrated de unusual warmth 2 selflessness in Rolla, promptly cleared a space on the desk, pushed back his chair, lit his pipe, and became immediately receptive to my problems. I've observed this iege dp times, with his visitors, students, volunteers — everyone. How he was able to give so much of himself to others pee ES publish scores of significant fern works is still a myster This warmth, cooperation, and Mn excellence was two-fold for the Gray haria visitor. In the room adjacent to Rolla’s quiet office was Alice Tryon’s beehive of pteridological activity, where this dynamic, hustle-bus- tle, perpetual-motion lady held sway. In many visits to the Tryon offices I have few recollections of Alice sitting, but I know she must have, as she spent hours at the microscope and scanning electron microscope. Neverthe- less, she, like Rolla, always had time to answer questions. She was a walking encyclopedia of information about fern spores and freely shared her read with anyone. А number of times she pointed out to me spore characters that helped confirm or deny haad pui eue dn More- over, s ever concerned with ferreting out and suggesting lunchtime or evening activities and excellent 280 Annals of the Missouri Botanical Garden restaurants in the Harvard area, all with the purpose of E guests feel welcome ortunate I pteridology that Rolla and Alice took time off from their individual projects to collaborate on their magnum a pns and Allied Plants. This mas- sive and impressive work sums up the knowledge of all genera (and many species) of neotropical pteridophytes in a way accessible to all. Nothing quite like it has been produced in nearly half a century, and : Mig as a synthesis of the Tryons’ historic contributions to pteri- dology. They have unselfishly dedicated ое to the study of a major group of organisms. Our science is much richer for their efforts; and for those of us fortunate enough to be ee among their friends, our lives are much richer for their being here. DEFENSE STRATEGIES IN BRACKEN, PTERIDIUM AQUILINUM (L.) KUHN Gillian A. Cooper-Driver' ABSTRACT In 1877 Francis Darwin expressed his belief that the primary role of nectaries in bracken is as waste glands, and continues to intrigue us nemies, over 100 years later the question of whether the ant-bracken association is mutualistic My first acquaintance with Alice and Rolla Tryon was a wonderful day we spent together at Down House in Down, Kent, approximately 20 miles south of London. It was here in Down House that Charles Darwin lived with his family from 1842 until his death in 1882. It was here that he wrote The Origin of Species by Means of Natural Se- lection and his papers on plants: “The Fertilization of Orchids””; ““The Variation of Plants and Animals under Domestication”; “Insectivorous Plants”; “Climbing Plants”; “The Effects of Cross and Self Fertilization in the Vegetable Kingdom”; “Differ- ent forms of Flowers in Plants of the Same Species”; and “The Power of Movement of Plants” (Darwin, 1892). The last of these was jointly published with his third son, the botanist of the family, Francis Darwin (Rendle, 1925). There are many ferns growing in the vicinity of Down House, including the ubiquitous bracken fern, Pteridium aquilinum (L.) Kuhn. Bracken was of particular interest to Charles Darwin and his son because it has extrasoral nectaries, which exude secretions attractive to ants (Fig. 1). In an appendix to a paper on glandular bodies in Acacia sphaero- cephala and Cecropia peltata, Francis Darwin (1877) recorded his father’s fascination with the nectaries: ““Examined several ferns with Myrmica on them, and found the glands quite dry. Brushed off the ants, and in from 5 to Ó minutes distinct drops of secretion were formed.” Especially inter- esting in this paper are Francis Darwin’s own ob- servations on the nectar glands of bracken and his views on their function. Contrary to the then widely held opinion (Del- pino, 1886-1889) that extrafloral nectaries serve as protection to a plant by attracting ants, which then ward off marauders, Francis Darwin believed that the primary role of bracken nectaries was as waste glands or excretory organs. Аз a result of his observations, he felt that bracken did not re- quire „Жаран by ants. He wrote, “This plant is singularly free from enemies not being eaten by the larger шг by rodents or by grasshoppers” (F. Darwin, 1877). his view expressed by Francis Darwin raises several interesting questions. Is it true that bracken has few natural enemies and therefore does not require the biotic protection of ants? Is the brack- en—ant interaction mutualistic, i.e., do both part- ners benefit, the plants by increased protection against natural enemies and therefore by increased fitness, and the ants from feeding on the nectar secretions, which provide a rich source of sugars and amino acids? Or is this a commensalistic as- sociation, as Francis Darwin suggested, in which y the ants benefit, and the association results in no fitness benefit to the plant? If possession of nectaries does not confer a fitness advantage to the plant, why is this trait maintained? Although research in the last 100 years refutes Francis Darwin's statements that bracken lacks natural enemies, the same controversy over the ' Department of Biological Sciences, Boston University, Boston, Massachusetts 02215, U.S.A. ANN. Missouni Вот. GARD. 77: 281-286. 1990. 282 Annals of the Missouri Botanical Garden FIGURE 1. Fiddlehead of bracken showing the position of the nectaries (g!, g?). From J. Linn. Soc., ne 15. 1877 Bot. Volur exact function of bracken nectaries continues as it did in Delpino’s and Francis Darwin’s time. This paper provides evidence that bracken does have natural enemies and considers why the role of nectaries in bracken appears to be so difficult to determine. Does BRACKEN HAVE NATURAL ENEMIES? Bracken has always been a long-lived (Oinonen, 1967), aggressive, opportunistic weed (Page, 1976, 1982a), but probably because of increased defor- estation bracken has spread extensively during this century, resulting in a severe loss of grazing land in many parts of the world (Page, 1982b). Even 60 years ago, bracken was described as “‘the ghost stalking silently at our side which nobody dares to discuss”” (Tryon, 1941). Because bracken is so invasive, there is much interest in finding ways to eradicate it, particulary through developing meth- ods for biological control (Perring & Gardiner, 1976; Dyer & Page, 1985; Kirkwood & Hin- shalwood, 1985). This has resulted in an active search for the natural enemies of bracken, i.e., bracken pathogens, parasites, and herbivores. Many species of fungi are known to be parasitic upon bracken (Gregor, 1932, 1938; Hutchinson, 1976; Webb & Lindow, 1981; San Francisco & Cooper-Driver, 1984; Bennell & Henderson, 1985), causing a range of different diseases. The most promising fungi in terms of biological control are three fungi that cause ““curl-tip” disease in brack- en. Combinations of Phoma aquilina Sacc. & Penz. (Coelomycetes), Ascophyta pteridis Bres. & Syd., and Septoria Sacc. have shown enough potential as biological control agents to be further investi- gated (Burge & Irvine, 1985; Burge et al., 1986). Francis Darwin claimed that larger animals and rodents do not eat bracken. In fact, cattle, sheep, horses, deer (Braid, 1959), and humans (Hodge, 1973; L.A. Times, 1988) are now known to feed on bracken. Grazing by animals is one of the major concerns caused by the increased spread of brack- en because feeding on bracken can often have drastic effects. Bracken can cause digestive tract and bladder tumors in humans, tumors and fatal hemorrhagic effects in sheep and cattle, and avi- taminosis in pigs and horses (Evans, 1976; Taylor, 1989) Francis Darwin’s view that ferns, including bracken, are resistant to grasshoppers and other insect pests is not unique (Schneider, 1892; Brues, 1920; Erlich & Raven, 1964), but until recently there was little experimental evidence to support or refute this belief. During the 1970-19805, many reports on phytophagous insects and arthropods colonizing ferns appeared in the literature showing that, in fact, ferns support a rich insect fauna (Balick et al., 1978; Hendrix, 1980; Ottosson & Anderson, 1983a). Balick et al. (1978) listed well over 100 species of arthropods associated with bracken fern. Many of the insect species found feeding on bracken are bracken specialists (Lawton, 1976; Cooper-Driver, 1978). Common phytophagous in- sects discovered on bracken in Britain and North America include hymenopteran sawflies (Stron- gylogaster, Aneugmenus, and Tenthredo), lepi- dopteran larvae, e.g., species of Callopstria and Papaiperma (Noctuidae), dipterans that mine the stems and tips of the pinnules (Chirosa sp. (Agro- myzidae)), and Dasyneura species (Cecidomyi- idae), which induce galls. Sucking insects, such as а (Aphidae), are also common in the (Lawton & MacGarvin, 1985). dum has studied insects associated with bracken in North Yorkshire, England (Lawton, 1976, 1982; Lawton & MacGarvin, 1985), Ha wail, New Mexico, and Arizona, U.S.A., and Papua New Guinea (Lawton, 1982, 1984), and he has conducted preliminary surveys in South Africa, Australia, and New Zealand. As expected, these surveys have shown that the types of phytophagous insects found on bracken vary in each geographic area. Lawton has shortlisted two South African moths, Conservula cinisigna and Panotima, as possible ee control organisms. on bracken pathogens and her- norther bivores, ken regional and somewhat limited, cer- Volume 77, Number 2 1990 Cooper-Driver 283 Defense Strategies in Bracken tainly suggests the Francis Darwin was wrong in saying that bracken has no natural enemies. Bracken has been described by Harper (1977) as one of the five most common plants in the world, suggesting a fair degree of ecological success. How then does bracken deal with its natural enemies, by chemical defenses and/or from protection by ants? THE ROLE OF CHEMISTRY For a single plant species (Tryon & Tryon, 1982), bracken contains an unusually large and diverse number of secondary compounds (Cooper-Driver, 1976; Lawton, 1976; Jones, 1983; Ottosson & Anderson, 1983b). Compounds include sesquiter- pene indanones (pterosides and pterosins); carcin- ogenic and mutagenic norsesquiterpene, pta- quiloside; the enzyme thiaminase, a vitamin B,-decomposing enzyme; that release cyanide (Cooper-Driver & Swain, 1976; Schreiner et al., 1984); phytoecdysteroids or insect molting hormones; phenolic compounds, phenolic acids, flavonoids, and tannins (Tempel, 1981; Cooper-Driver, 1985); lignins; and silicates. Effects of phytochemical variation on colonization by insect communities has been reviewed by Jones (1983). Many of these compounds are toxic or deterrent and defend bracken against potential pathogens and predators, suggesting that chemistry must have played an important role in the ecolog- ical success of bracken. Do other factors play a role in the protection of bracken from its natural enemies? Do ants pro- vide an alternative defense strategy as they do in many other plants (Bentley, 1977; Beattie, 1985; McKey, 1988)? cyanogenic glycosides THE ROLE OF THE NECTARIES Relationships with ants are not uncommon in ferns as evidenced by the conspicuous stem mod- ifications of ant-inhabiting ferns (Wagner, 1972; Gomez, 1974), and the morphological adaptations of fern spores for ant dispersal (Tryon, 1985). Nectaries occur in a number of fern genera in addition to bracken, namely Polybotrya, Poly- podium, and Drynaria (Koptur et al., 1982), but the nectaries of bracken have been the most studied (Darwin, 1877; Figdor, 1891; Potonie, 1891; Lloyd, 1901; Luttge, 1961; Schremmer, 1970; Page, 1982c; Power & Skog, 1987). Many different types of ants have been recorded as feeding on bracken nectaries (Lawton & Heads, 1984). Tryon (1941) wrote, “Large red and black ants were attracted by the exudation and observed in considerable numbers feeding upon them (the bracken nectaries).”” Ants are not the only arthro- pod species that have been observed visiting the nectaries; parasitic wasps, coccinellid beetles, flies, and spiders also visit (Douglas, 1983; Tempel, 1983). The assumption has always been that the as- sociation between ants and bracken nectaries is facultative; the ants are opportunistic feeders, drawn to a source of sugars and proteins. However, we know relatively little about the behavioral ecology of the ants that visit bracken nectaries, and perhaps this aspect should be further scrutinized. Does the presence of ants provide protection for bracken? There are conflicting reports in the lit- erature. Tempel (1983), studying ants on bracken in New Jersey, could find no evidence for ants laying a protective role. She did not observe any aggressive behavior by the ants, nor did she find that there was any difference in herbivore damage on control leaves, plants in which ants were present at the nectaries, as compared with on leaves from which ants had been excluded. Lawton & Heads 1984), in an extensive study of North Yorkshire and New Mexican bracken, reached the same con- clusion. Field experiments designed to test the role of ants by excluding them from bracken leaves failed to produce convincing evidence that ants depress insect herbivore populations (Heads & Lawton, 1984). Further studies in South Africa (Rashbrook et al., 1989) also could not prove this. However, there is evidence that in some cases ants do protect bracken. Douglas (1983), studying bracken in Michigan, noticed that different ant species behave differently toward herbivorous in- sects. Douglas observed that the larger ants, For- mica and Camponotus species, defend the nec- taries; any marauders are bitten, stung, or killed. Two smaller species of ants, Tapinoma and Lep- tothorax, apparently only feed and do not play a protective role. A survey of matched sites with and without wood ants (Formica lugubris Zett) re- vealed consistently lower populations of external feeders when wood ants were present (Heads, 1986). However, the only statistically significant differ- ences between the insects at the experimental sites were for guilds of sucking insects. When foreign lepidopteran larvae of Plodia were experimentally placed on the leaves of bracken, the ants imme- diately removed them (Lawton & Heads, 1984). Although in some cases ants obviously attack generalist phytophagous insects, what about the insects that normally feed on bracken, bracken — 284 Annals of the Missouri Botanical Garden specialists? Lawton & Heads (1984) have shown that some of the most common specialist insect herbivores of bracken possess characteristics that make them immune to predation by ants. ample, bracken sawflies (Strongylogaster, Aneug- menus, and Tenthredo) have a distasteful, viscous hemolymph that repels ants (Pasteels et al., 1983; Heads & Lawton, 1985). Sawfly eggs are also apparently immune to ant attack. Other bracken specialists, such as gall formers and mining insects, feed inside the tissues and are inaccessible; others hide or jump away from patrolling ants. It appears that over periods of evolutionary time bracken insects have evolved and adapted to avoid ants (Heads & Lawton, 1984). Because the majority of insects found feeding on bracken are specialists, the generalist natural enemies of these herbivorous insects, i.e., the ants, may have provided the major selective pressure for restricted host-plant range (Bernays & Graham, 1988). Jeffries & Lawton (1984) also predicted that predators (in this case ants) are often potent selective agents and are likely to select for prey (insects) characteristics that re- duce the impact of predators. The insect species thus evolves to occupy enemy-free space. The high or ех- proportion of specialist sawfly species, gall formers, and miners that make up the bracken herbivore community could therefore have been markedly influenced by ant predation. CONCLUSION Bracken defense and control still pose a number of evolutionary and ecological questions—for ex- ample, was the bracken-ant association once mu- tualistic? Unfortunately, we cannot always know what took place in the past. Are nectaries an ex- ample of a structure that was once useful but whose function is no longer obvious, just as phytoecdyste- roids may have once acted as a defense against insects (Cooper-Driver, 1976), or are they exam- ples of structures whose role is yet to come? The exact nature of the association is difficult to resolve until we know more about the ants and associated insects in different parts of the world. Almost all the detailed information we have comes from tem- perate areas, and we do not know what role ants associated with the nectaries play in defense in tropical regions. There is also still no conclusive evidence that ants improve the fitness of bracken. hus, it appears that Francis Darwin was prob- ably right in believing that ants do not protect racken, not because bracken does not have nat- ural enemies, but because specialist bracken her- bivores have evolved defensive mechanisms to deal with the ants. Whenever I see ants at the nectaries of bracken I am reminded of the day at Down House and realize that in another 100 years bracken will still be around to puzzle and intrigue us. My heartfelt thanks go to the Tryons for introducing me to bracken and for stimulating a lifelong interest in the community ecology of ferns. LrrERATURE CITED Влиск, M. J., D. G. FURTH & С. COOPER-DRIVER. 1978. Biochemical and evolutionary aspects of arthropod predation on ferns. Oecologia (Berlin) 35: 55-89. BEATTIE, А. The evolutionary ecology of ant- plant mutualisms. Cambridge Univ. Press, Cam- ridge. BENNELL, A. P. & D. M. sn 1985. Rusts and other fungal i eds as aids usi ci tax- c. Edinburgh 86B: 115-124. BENTLEY, B. L. ei Ext rafloral nectaries t pro- tection by ara bodyguards. Annual Rev. Ecol. Syst. 8: 407-427 BERNAYs, E. & M. GRAHAM. 1988. On the evolution of host specificity in phytophagous arthropods. Ecol- ogy 69: 886-892. Braip, K. W. 1959 1 A Review of the Liter- ature. 3: 1-69. Cone, Agricultural Bureau, Hurley, Berks. Brues, C. T. 1920. The selection of food-plants by insects, with special reference to lepidopterous larvae. Amer. Naturalist 54: 313-332 BURGE, J. А. IRVINE. | 1985. Recent studies onthe potential for biol l of bracken беш fungi. Proc. xs Soc. күн! 86B: 187- ———— . MCELWEE. 1986. The ema for biological UE of bracken with causal a for curl-tip disease. Pp. 455-460 in R. T. Smit J. A. Taylor (editors), Bracken: Ecolo and Control Technology. Parthenon Press, Lancs, United Kingdom COOPER-DRIVER Chemotaxonomy id Ro tochemical ecology of bracken, Bot. J. Linn. Soc. (3: 35- — ” ‘ — pus Insect- fern associations. Entomol. Exp. & Appl. 24: 310-316 985. Anti- -predation strategies in oe Sc phytes—a VER C approach. Proc. Roy. Edinburgh 86B: 402. ——— ‚ SWAIN Cyanogenic polymorphism in bracken in relation to hahae predation. Nature 200: 604. DARWIN, F. 1877. On the p bodies on Ac acia sphaerocephala and Cecropia pel food for ants. Appendix on common Brake Fern (Pteris aquilina). J. Linn. Soc, B -409. Charles Darwin, His Life Told in an Autobiographical Chapter and in a Selected Series of His Published Prod D. Appleton, New York. DELPINO, К. 1886-1889. Funzione mirmecofila nel reg- no 10 vegetale. Мы Reale Accad. Sci. Inst. ANN r 323; 8: 601-650; 10: 115 Volume 77, Number 2 1990 Cooper-Driver 285 Defense Strategies in Bracken DoucLas, M. М. 1983. Defense of bracken fern by arthropods attracted to axillary nectaries. Psyche 90: 20 Dyer, A. F. & С. №. PAGE (editors). 1985. Biology of Pteridophytes. Proc. Roy. Soc. Edinburgh 86: 179- ERLICH, Р. К. & Р. Н. Raven. 1964. Butterflies and 8: 586- plants: a study in coevolution. Evolution 18: 5 Evans, W. C. 1976. Bracken thiaminase-mediated neu- rotoxic pr Bot. J. Linn. Soc. 73: 113-131. Fenwick, С. В. 1988. The toxicity ud chemical com- position of bracken. J. Sci. Food Agric. 46: 147- 17 FIGDOR, W. 1891. Ueber die extranuptialen Nectarien von Pteridium aquilinum. Oesterr. Bot. Z. 9. Gomez, P. L. 1974. Biology of the potato-fern, Solanopteris brunei. Brenesia 4: 37-61. GREGOR, M. J. F. 1932. The possible utilization of disease e a factor in bracken control. Scott. Forest. J. 46: 5 TE Associations with fungi and other lower plants. Pp. 141-158 in F. Verdoorn ux SE of Pteridology. Martinus Nijhoff, The Ha HanPER, J. L. 1977 denen Biology of Plants. Ac- 7 Bracken, ants and extrafloral nectar- ‚ Do wood ants (Formica lugubris) protect de plant against insect herbivores? J. Animal Ecol. 55: 795-809. J. H. Lawron. 1984. Br ин: ants and extrafloral nectaries. II. The effect of ants on the insect herbivores of bracken. J. Animal Ecol. 53: 1015-1032 & ——. 19 Bracken, ants and extra- floral nectaries. III. How insect herbivores avoid ant predation. Ecol. Entomol. 10: 29-42. HENDRIX, S. D. 1 An evolutionary and очоко perspective . bd insect fauna of ferns. Amer. Nat uralist 115: à HopcE, W. Н. Fern foods of Japan and the problems of bracken toxicity. Amer. Fern J. 63: 77- 80. poe S. A. 1976. The s > fungi on brack- t. J. mea Soc. 73: Жош, М. 1. Н. i M Enemy-free space and ha structure of ecological communities. Biological J. Linn. Soc. 23: 269-286. Jones, C. С. 1983. Phytochemical variation, coloni- zation and insect communities: in the case of bracken fern (Pteridium aquilinum). Pp. 513-558 in R. F. Denno McClure (editors), Variable Plants re Herbivores in Natural and Managed Systems. cademic Press, New York. кы ‚ С. & А. HINSHALWOOD. 1985. Recent studies on the biology of bracken and the mode of n of herbicides used Y its control. Proc. Roy. oc. Edinburgh e 179-185. «Eus S., A. К. SMITH & I. DAKAN 1982. Nectaries in some neotr кш species of Pol podium (Poly- podiaceae): preliminary pat and Mrs i =113, 1976. The structure of the arthropod community on bracken. Bot. J. Linn. Soc. 73: 187- 1982. Vacant niches and unsaturated com- munities: a comparison of bracken herbivores at sites on two continents. J. Animal Ecol. 51: 1984. Non-competitive populations, noncon- vergent communities, and vacant niches; the herbi- vores on bracken. Pp. 67-101 in D. R. Strong, Р. Simberloff, L. G. Abele & A. B. Thistle (editors), Ecological Communities: Conceptual Issues and the Evidence. Princeton Univ. Press, Princeton, New Jersey. Heaps. 1984. Bracken, ants and extra- floral nectaries. I. The components of the system. J. Animal Ecol. 53: 995-1014. 8 M. MACGARVIN. 1985. Interaction between bracken and its insect herbivores. Proc. Roy. So Edinburgh 86B: 125-131. LLovp, F. Е. The extra-nuptial nectaries in the ommon brake, Pteridium aquilinum. Science 13: 885-890. Los ANGELES TIMES. 1988. Fern: Far East Tastes. June 3th 8. 1 р: 1, LurrGE, U. 1961. Ши йе сое des Nek- tars und den ner Sekretion. I. Planta 56: 189-212. McKey, р 1988. Promising new directions in the study of ant-plant mutualisms. /n W. Greuter immer (editors), Proc. XIV International Botanical Con- gress, Berlin, 24 July-1 August 1987. Koeltz Sci- OINONEN, E. 67. The correlation between the size of Finnish bracken clones and certain periods of site history. Acta Forest Fenn. 83: 1-51. OTTOSSON, J. . M. ANDERSON. 1983a. Number, seasonality and feeding habits of insects attacking erns in p ү ecological consideration. J. Animal Ecol. 52: с 1983b. Seasonal and interspecific variation in the biochemical composition of some British fern species and their effects on Spodoptera littoralis larvae. J. Linn. Soc. 19: 305-320. Pack, С. N. 76. The taxonomy and phytogeography of bracken—a review. J. Linn. Soc., Bot 1982a. The history and the spread of ken in Britain. Proc. Roy. Soc. Edinburgh 81B: 3-10. —. 1982b. The Ferns of Britain and Ireland. Cambrige Univ. Press, Cambridge. 1982c. Field observations on the nectaries of bracken, Pteridium aquilinum, in Britain. Brit. Fern Gaz. 12: 233-240. PASTEELS, J. M., 1983. The chemical ecology of defence in arthro- pods. Annual Rev. Entomol. 28: 263-289. PERRING, Е. H. € В. С. Pe ара ur The biology of bracken. Bot. J. L 1-302. PoTONIE, Н. Die е ена beim Adlerfarn. Naturwiss. Wochenschr. 6: 401-440 Power, М. S. & J. E. Skoc. 1987. Ultrastructure of the extrafloral a of Pteridium aquilinum. Amer. Fern J. 77: 1-15. RASHBROOK, V. K., d G. CoMPTON & J. H. 1989. Bracken and ants: is there a теба Meeting of the Zool. Soc. July, Oxford. [Abstract.] RENDLE, A. В. (editor). 1925. Francis Darwin (1848- 1925). The Journal of Botany, British and Foreign. 63: 333-334. [Obituary San FRANCISCO, М. & С. A. COOPER-DRIVER. . GREGOIRE & М. ROWELL-RAHIER. 1984. 286 Annals of the Missouri Botanical Garden Anti-microbial activity of phenolic acids in Pteridium aquilinum. Amer. Fern J. 74: 87-96. SCHNEIDER, G. Ferns, еее, British апа exotic. Pp. 169-178 in The "аш of Choice Ferns, Volume 1. Upcott Gill, ions SCHREINER, I., D. Narus & D. PIMENTEL. 1984. Effects of лано n bracken fern (Pteridium aquili- num) on associated insects. Ecol. Entomol. 9: 69- SCHREMMER, F. 1970. Extranuptiale Nektarien: Beo- bachtungen an Salix eleagnos Scop. und Pteridium aquilinum (L.) Kuhn. Oesterr. Bot. Z. 117: 205- 2999. TAYLOR, J. 1989. Bracken toxicity and carcinogenicity as related to animal and human health. International Bracken Group Institute of Earth Sciences, Field studies on the relationship be ban herbiy ore e damage and tannin concentration in bracken (Pteridium aquilinum L. Kuhn) Oecologia ймы a 97-10 а: 1983. Bracken fer (Pteridium aquilinum) and nectar-feeding ants: a nonmutualistic interaction. Ecology de 1411-1422, Tryon, A. К. 1985. Spores of е ferns. Proc. Roy. ro Edinburgh 86B: 105-110. 9 TrYON, В. М. A revision of с genus Pteridium. Rhodora 43: 37-67. & A. F. TRYON. s Ferns and Allied Plants. E. Verlag, New ‘ W AGN W. H., Jr. 1972 Solanopteris brunet, litte. known fern epiphyte dil dimorphic stems. Amer. Fern 2 7 43. LiNbow. 1981. Evaluation of Ascochy ta ser as a potential biologic al control agent of bracken fern. Phy topathology 71: 911. [Ab stract PITYROGRAMMA CALOMELANOS (L.) LINK (ADIANTACEAE) ON LAYERS OF VOLCANIC ASH IN LOS TUXTLAS, STATE OF VERACRUZ, MEXICO Ramon Riba' and Irma Reyes J.’ ABSTRACT The selective establishment of Pityrogramma calomelanos (L.) Link on alternate layers of volcanic ash is described in this paper. The p ants grow in layers of fine sand where the texture of the pyroclastic material and its moisture- lants. holding capacity favors spore deposition and development of the plants Ferns tolerate diverse and extreme environ- mental conditions, particularly climatic and edaph- ic factors. There are very narrowly distributed species (e.g., Adenoderris sororia Maxon) while others have broad distributions (Anogramma lep- tophylla (L.) Link). The ranges are related pri- marily to dispersal and germination of the spores, and to the fitness of the species within various habitats. Fern species with wide ranges grow in ecologically open habitats where they frequently row as pioneer species, experiencing low com petition (Page, 1979b). After release, the spores are exposed to physical factors such as gravity, wind, temperature, and humidity. With wind as the main dispersal agent, widespread dispersal to an array of differing hab- itats is possible. Fern spores can germinate at tem- peratures as low as 1—2°С or as high as 30-35%C (Miller, 1968), although the most favorable tem- peratures for germination of some species are 18- 25°C (Pérez-Garcia & Riba, 1982). High humidity helps reactivate the metabolism of the spore and promotes germination. Spores germinate and prothallia usually grow on horizontal surfaces, but they can grow even on vertical slopes and on the roofs of caves (Page, 1979b). Since dispersal capability seems to be equivalent among homosporous ferns, establish- ment of the sporelings and plantlets must depend on ecological factors and fitness of the species in a particular niche (Tryon, 1972). Ferns include species that invade disturbed hab- itats. Pityrogramma calomelanos is native and widely distributed in tropical America and has nat- uralized in the Old World. It is an effective dis- turbance colonizer throughout the tropics. This fern may have been introduced deliberately into the Old World as an ornamental plant or accidentally (Schelpe, 1975; Pérez-Garcia et al., 1982). Pityrogramma calomelanos has numerous fea- tures of a disturbance colonizer. It tolerates short dry periods (Tryon & Tryon, 1982), and it grows in open sites; it may be designated a heliophyte. Its spores have long viability, as is usual in non- chlorophyllous spores of homosporous ferns. The rhizomes resist such extreme conditions as high temperatures and high sulphur concentration at the substrate, as was shown by the plant regen- eration after the 1982 eruption of the Volcan Chi- chonal in Chiapas, Mexico. Rapid regeneration un- der adverse conditions was evidenced by specimens with fertile leaves 1 m high, and abundant young sporophytes 5-7 cm high were found within 2 km of the crater of the volcano two years after the eruption (Spicer et al., 1985). Here we describe the selective establishment of Pityrogramma calomelanos on a roadcut through successive strata of volcanic ash. The roadcut is located at km 106 of the Federal Highway 180, 14 km NW of Santiago Tuxtla, state of Veracruz, Mexico, at an altitude of 340 m. At this site, hills of Miocene volcanic ash have nearly horizontal alternating layers of different colors, textures, and compositions. The slope is strongly eroded, and almost vertical walls of ash are visible; the layers with ferns are clearly delimited (Figs. 1, 2). The characteristics of the two kinds of strata are as follows: (a) layers of pale brown, fine ash Met lit Tot I ! Universidad Autónoma ANN. Iztapalapa, Apartado Postal 55-535, México, D. F. 09340. Missouni Bor. GARD. 77: 287-289. 1990. Annals of the Missouri Botanical Garden Volume 77, Number 2 1990 Riba & Reyes J. 289 Pityrogramma calomelanos (L.) Link 1-10 cm thick with fine ash and basaltic tuffs predominating, moderate cohesiveness, and great absorption and retention of water; and (b) layers of gray ash, 1-50+ cm thick, made of coarser, non- cohesive particles with basaltic gravel predominat- g. Both layers react strongly to alophane (NaF and phenolphthalein), and they have weakly alkaline (7.5) in aqueous solution and slightly acidic pH (6.2-6.5) in saline solution (KCl 1 N pH 7). The qualitative analysis for soluble nutrients showed that the brown fine ash is richer in phosphorus and calcium than the gray coarse ash. Pityrogramma calomelanos grows in the layers of fine ash, even where they are 1 cm thick. The species is restricted to specific layers, even where there is no soil in the strict sense, but regolite. Why do these plants prefer the thin substrate of the fine ash to that of the coarse one, even when this is more abundant? While Pityrogramma cal- omelanos is considered a colonizer of well-drained substrates, in this case its restricted establishment in the thin layers of fine ash is affected by factors beyond drainage. The pH of the substrate is critical for establishment of some species, but Pityro- gramma calomelanos can grow on rocks, slopes, river banks, and lava flows, and it has even been considered as a pumicicolous fern (Page, 1979a), colonizing well-drained volcanic habitats. There are some studies about the edaphic pref- erences of ferns in which pH and calcium content of the substrate (Wherry, 1920; Hevly, 1963), kind of rock (Kruckeberg, 1964; Page, 1979a), pH, concentration of specific nutrients, percentage of humidity, the soil texture, and organic matter (Graves & Monk, 1982) are considered. Petersen (1985) assessed selected literature on edaphic ad- aptation of pteridophytes. The results of the physical and chemical analysis of the two kinds of ash show that the preference for the fine ash is not determined by the pH nor by the chemical composition. The texture of the fine ash does favor aggregation and cohesiveness of the particles, which permits greater retention of water and nutrients than in the coarser substrate. The fine layer erodes more slowly than the gravel layers. These features facilitate establishment, and the rhizoids of the gametophytes attach easily to the aggregated particles of fine sand. In the same roadcut there are ash walls that have not been disturbed recently. There other es- tablished liverworts, mosses, and lichens occur without selectivity; they cover both kinds of ash (Fig. 3). In those walls we find some individuals of Nephrolepis cordifolia (L.) Presl (Fig. 4) with long wiry roots. In the undisturbed roadcuts, decay products of bryophytes and other organic debris concentrate organic material, accumulate in the walls, and thus permit establishment of other vas- cular plants. LITERATURE CITED Graves, J. Н. 8 С. D. Monk. 1982. Herb-soil rela- tionships on a lower north slope over marble. Bull. Torrey Bot. Club 109: 500-507. HevLy, R. H. . Adaptations of cheilanthoig ferns to desert environments. J. Arizona Acad. Sci KRUCKEBERG, A. R. 1964. Ferns associated with ultra- maphic rocks in the Pacific Northwest. Amer. Fern Fern gametophytes as experi- 61-440. da The diversity of ferns. An ecological . 10-56 in A. F. Dyer (editor), The Q e Biology of Ferns. Academic Press, New York. 1979b. Experimental aspects of fern ecology. Pp. 552-589 in A. F. Dyer (editor), The Experi- mental Biology of Ferns. Academic Press, New York. Pérez-Garcia, B. & R. Ripa. 1982. Germinación de esporas de Cyatheaceae bajo diversas temperaturas. Biotropica 14: 281-287. ROZCO-SEGOVIA & R. . 1982. El banco de esporas de helechos en К suelo de Los Tuxtlas, Veracruz. Bol. Soc. Bot. México 43: 89- 92 PETERSEN, R. L. 1985. Towards an appreciation on fern edaphic niche A Proc. Roy. Soc. Edinburgh 86B: 93- SCHELPE, E. A. С. L. E. on Observations on the spread of the American fern Pityrogramma calo- melanos. Fern Gaz. 11: 101-104. Spicer, R. A., R. J. BURNHAM, P. GRANT & Н. GLICKEN. 1985. Pityrogramma calomelanos, the primary, post-eruption colonizer of Volcán Chichonal, Chiapas, México. Amer. Fern J. 75: 1-5. Tryon, В. 1972. Endemic areas and geographic spe- ciation in American tropical ferns. Biotropica 4: 1 131. . TRYON. 1982. Ferns and Allied Plants, with Ж Reference to Tropical America. Spring er-Verlag, New York. баш. E. Т. 1920. The soil reactions of certain rock ferns—I, II. Amer. Fern J. 10: 15-22, 45-52 — FIGURES 1-4.— 1. Ash walls showing ferns Vita] distributed. — 2. Pit . Ash fine ash. —3. Ash walls with liverworts and moss roots. tyrogramma calomelanos in layers o walls showing Nephrolepis cordifolia with long wiry OBSERVATIONS ON THE REPRODUCTIVE BIOLOGY OF ALSOPHILA SPECIES AND HYBRIDS (CYATHEACEAE) David S. Conant? ABSTRACT The percent spore inca рк gross morphology, and cis structure of sex organs on gametophytes x ec Alsophila hybrids were with thos in the hybrids md in pd 5s 10Wever, germination rates as hig than in specie gross morphology was more variable in hybrids t of Alsophila species. The ean percent spore germination was low = 5 43% we were observed among the hybrids. Prothallial any gametophytes of the hybrids, however, grew . Ma into normal, M E prothallia that later тек hermaphroditic. Detailed studies revealed no difference between the structure of sex organs of hybrids and those of species. Sperm an nd egg appeared to mature at about the same time indicating that species and hybrids have the potential for self-fertilization. The Cyatheaceae are unique among the ferns because of the apparent lack of polyploidy (Tryon & Tryon, 1982) and the existence of fertile diploid hybrids (Conant, 1975, 1983; Conant & Cooper- Driver, 1980). The family is of further interest because it has been hypothesized that tree ferns form new species by a specialized form of diploid hybrid speciation called autogamous allohomoploi- dy (Conant & Cooper-Driver, 1980). This mode of speciation begins with the formation of a fertile diploid F, hybrid. At meiosis, this plant presumably produces genetically recombinant spores. If these spores grow into bisexual gametophytes, and if intragametophytic selfing occurs, a completely omozygous F, recombinant sporophyte would re- sult. Its spores will be genetically identical, barring mutation and homoeologous chromosome pairing, and it could give rise to a colony of individuals that are recombinant for the morphological char- acters of the progenitor species. These plants would be morphologically distinct from the progenitor species. If selfing is the norm and outcrossing is rare, the colony would also be reproductively iso- lated from the сеат pes and would potentially constitute a new spec Conant & Cooper- Driver (1980) presented two kinds of evidence that are consistent with this hy- pothesis. The first was the existence of five species of Nephelea (later treated as Alsophila by Conant, 1983) that recombined or blended unique mor- phological characters of sympatric or nearby species. The second was the occurrence of fertile diploid hybrids and morphologically recombinant hybrids in the mountains of Puerto Rico. Although spores of the hybrids were shown to germinate, the degree of hybrid fertility and the development of these spores into bisexual gametophytes capable of selfing was not documented. The determination of whether the sex organs of hybrid gametophytes are functional can be made only in comparison to the gametangia on the ga- metophytes of species. A brief review of what is nown about the prothallia of the Cyatheaceae will help to put our results into perspective. Stokey (1930) described the prothallia and sex organs of 14 species of Cyatheaceae, including three mem- bers of Alsophila. In general, antheridia do not appear on the prothallia of the tree ferns until they have formed a notch meristem and well-developed wings. There was no evidence of antheridia pro- -celled), as is frequently found in polypodiaceous (sensu lato) ga- metophytes. Archegonia appear on the gameto- phyte after development of the cushion, which in the Cyatheaceae is always four or more cells thick. The mature archegonium has a neck, which is duction at very early stages (5-2 ! [ thank two former Lyndon State College students for assistance with this project. Ms. Lynn Farrell helped in - e cule о Жет, and Ms. Janice > the Dr. Finley А, Bryan and one anonymous reviewer for = Rucker prepared the E fes arl L. Man for help in embedding and sectioning the gametophytes, to Dr. Da constructive criticism of the manuscript, and to Dr. for assistance with “ke M analysis of the spore a data. This research was supported in part by Advanced Study Grants from the Vermont State College ? Department of Natural Sciences, Lyndon State Сойере, Lyndonville, Vermont 05851, U.S.A. ANN. MISSOURI Bor. GARD. 77: 290-296. 1990. Volume 77, Number 2 1990 Conant Reproductive Biology of A/sophila Species and Hybrids A o N o Percent Germination 39026 39036 o > ROA о N о о m m 3859.) > о m о № 3937.) O о о e m Alsophila bryophila FIGURE 1. signify percent germination. of Conant; J = Monte Jayuya, G = Monte Guilarte. slightly curved toward the base of the gametophyte, the apical side of the neck usually being 6-7 cells, the basal side 5-6 cells. The ventral canal cell and the egg are situated directly below the neck canal nuclei. Occasionally, divisions of the venter cells (the cells surrounding the egg) were observed as the archegonium matured. Stokey (1918) de- scribed such divisions adjacent to eggs on game- tophytes of Trichipteris aspera (L.) Tryon (as Cy- athea muricata Willd.) and Alsophila tussacii (Desv.) Conant (as Cyathea tussacii Desv.) as apo- gamous developments; however, in no case were these structures reported to give rise to sporo- phytes. This paper compares the percent spore germi- nation, prothallial gross morphology, and structure of sex organs of Alsophila hybrids with those of species in order to demonstrate the degree of hybrid fertility and to show that the hybrids have normally developed bisexual prothallia possessing the poten- tial for intragametophytic se MATERIALS AND METHODS Leaves of two Alsophila hybrids (А. amintae х A. portoricensis and A. bryophila х A. por- toricensis) and two Alsophila species (A. bry- ophila and A. portoricensis) were collected on Monte Guilarte and on Monte Jayuya (ca. 10 km west and ca. 20 km east of Adjuntas, respectively) in the central mountains of Puerto Rico. Vouchers are cited in Figure 1 and are deposited at LSC. A. portoricensis a 15 : 1 Em LA. Oo & o мо - мо 8 д а y о => © = QA 4 сч к о с с с с o0 a 0 H omm mmn м m m A. amintae X A. bryophila X . portoricensis A. portoricensis Comparison of percent spore germination of Alsophila species and hybrids. Numbers at tops of bars Numbers in boxes indicate mean germination. Numbers below bars are collection numbers Spores were collected, sterilized, and sown in Puer- to Rico according to the following protocol: fertile pinnules were washed under running water, blotted dry, and allowed to dehisce overnight in clean petri dishes. The following day, spores were surface ster- ilized with 1% sodium hypochlorite (commercial laundry bleach: Clorox), rinsed three times with sterile distilled water, and sown on sterile nutrient agar. These spore cultures were taped shut and transported back to the laboratory where they were kept in a growth chamber at 25°C for 12-hour days for three months. Percent germination was determined after six weeks by making three counts of 200 spores for different quadrants of each plate. The data was then subjected to the non-parametric Mann-Whitney test. Line drawings were prepared to compare the overall form of the gametophytes of hybrids with those of species. Gametophytes were grown to sexual maturity, and ten individuals of each hybrid and species were fixed in Randolph’s af. The gametophytes were then dehydrated, embedded in paraffin, and sectioned serially at 10 microns. Sections were stained in safranin and fast green and mounted in Permount. RESULTS AND DISCUSSION The results of the spore germination experiments are shown in Figure 1. The Mann-Whitney test indicated a significant difference between the per- cent germination of A. bryophila and that of each of the hybrids. In A. bryophila the mean percent 292 Annals of the Missouri Botanical Garden germination was 50% compared with 18% and 29% for A. amintae х A. portoricensis and A. bryophila X A. portoricensis, respectively. There was also a considerable range in the percent ger- mination of species and hybrids, especially in A. portoricensis where germination varied from 6% 2%. This large range rendered the percent germination data for A. portoricensis statistically indistinguishable from those of each of the hybrids even though the mean percent germination (36%) is higher for A. portoricensis. The lower mean percent germination in hybrids suggests that there is some spore inviability; however, spores appeared uniform in shape and size in all samples studied. The variation in percent germination among indi- viduals of species and hybrids is probably a reflec- tion of the maturity of the different spore samples when collected. None of the spore samples had 100% germination. This may be because the ster- ilization process killed a fraction of the spores in each sample. It should be noted that in a study of the germination of spores of four species of Cy- atheaceae (Pérez-Garcia & Riba, 1982), percent spore germination was temperature dependent. Cultures maintained at 25°С, however, had 100% germination. In contrast, among species of Dryop- FIGURES 2-5. Gametophytes of Alsophila species and teris, germination ranged from 78% to 95%; how- hybrids. — 2. 4. bryophila. —3. A. portoricensis. —4. — ever, one plant of D. celsa had only 37.3% ger- . amintae X A. portoricensis. —5. A. bryophila x А. portoricensis. Dotted lines indicate regions where ga- metophytes were more than one cell thick. SP — spore from which gametophyte developed. All scale bars = 0.25 of a species (Whittier & Wagner, 1971). mm. Gametophytes of the species studied (Figs. 2, 3) mination indicating that the percentage of spores that germinate is not always high among individuals —э FIGURES 6-10. Gametangia of Alsophila species. 6-8. Micrographs from same section of A. bryophila game- tophyte grown from spores taken from Conant 4118. —6. Bisexual thallus; scale bar = 0.1 mm.— 7. Enlargement of antheridium shown in Figure 6; dark cells within are spermatids; scale bar = 0.025 mm.—8. Enlargement of archegonium shown in Figure 6: left, younger; right, mature; scale bar = 0.025 mm. 9, 10. Sections of different 4. ortoricensis gametophytes grown from spores from Conant 4058; scale bars = 0.025 mm.—9. Antheridium containing sperms. — 10. Archegonia: left, nonmedial section; right, medial section, mature. Key: AN, antheridium; AR, archegonium; E, egg; VC, ventral canal cell; NC, neck canal cell. FIGURES 11-16. Gametangia of Alsophila hybrids. 11-13. Micrographs from same section of 4. amintae х А. portoricensis gametophyte grown from spores taken from Conant 4104. —11. Bisexual thallus; scale bar = 0.5 mm.— 12. Enlargement of antheridium shown in Figure 11; coiled cells within are sperms; scale bar = 0.025 mm.— 13. Enlargement of archegonium shown in Figure 11; near medial section; scale bar = 0.025 mm. 14-16. Same section of A. bryophila х A. portoricensis gametophyte grown from spores taken from Conant 4106. — 14. Bisexual thallus, scale bar = 0.5 mm. — 15. Enlargement of antheridium shown in Figure 14; dark cells within are spermatids; scale bar = 0.025 mm.—16. Enlargement of archegonium shown in Figure 14; nonmedial section, ventral canal cell indistinct; scale bar = 0.025 mm. See caption for Figures 6-10 for key to lettering on photographs. FicunES 17-20. Gametangia of Alsophila hybrids. 17, 18. Sections of different А. amintae х A. portoricensis gametophytes grown from spores taken from Conant 4104. —17. Antheridia containing sperms. — 18. Archegonium, medial section. 19, 20. Sections of different 4. bryophila x A. portoricensis gametophytes grown from spores taken from Conant 4106. —19. Antheridia containing sperms. — 20. Archegonium, medial section. See caption for Figures 6-10 for key to lettering on photographs. All scale bars — 0.025 mm. Volume 77, Number 2 1990 Conant Reproductive Biology of Alsophila Species and Hybrids 293 294 Annals of the Missouri Botanical Garden Volume 77, Number 2 1990 Conant Reproductive Biology of Alsophila Species and Hybrids 295 296 Annals of the Missouri Botanical Garden rew into normal, cordate prothalli and were de- velopmentally similar to those Stokey (1930) re- ported. and appeared retarded in their development (not shown), although many grew into normal, cordate prothalli (Figs. 4, 5) and were closely similar to those of the species. The growth of these normal hybrid gametophytes appears to be free from the kinds of developmental abnormalities (Pray, 1971 that may be observed in gametophytes derived from the small percentage of viable spores often pro- duced by "sterile" fern hybrids (Wagner et al., 1986; Whittier & Wagner, 1971). Gametophytes of species and hybrids were sec- tioned serially to investigate their sexuality, to com- pare the structures of their gametangia and ga- metes, and to look for evidence of apogamous embryo formation. The gametophytes of all species and hybrids studied first produced antheridia. Аз a rule, antheridia were not formed until after ga- metophytes were beyond the stage of development shown in Figures 2-5. The gametophytes of species (Figs. 6-10) and hybrids (Figs. 11-16) became hermaphroditic after 2-3 months in culture. The gametangia and gametes of hybrids (Figs. 17-20) were similar in structure to those of the species observed in this study (Figs. 7-10) and to those described by Stokey (1930). Divisions of venter cells were observed in species and hybrids, although in no case did such divisions result in a mound of tissue resembling an apogamous outgrowth. fter three months of culture in the absence of water, no sporophytes had appeared. After this, the gametophyte cultures were maintained with occasional flooding for an additional six months. During this period, sporophytes appeared in cul- tures of both species and hybrids. It is presumed that these sporophytes arose as a result of normal sexual reproduction; however, whether these mat- ings were the results of self- or of cross-fertilizations remains to be shown Gastony & Houser (1976) documented well- developed archegonia on the apogamous triploid Bommeria pedata (Sw.) Fourn. After archegonia were mature, the egg degenerated and became ybrid gametophytes were smaller x necrotic. Sporophytes arose apogamously from vegetative cells peripheral to the archegonium. This apogamous life cycle included the Döpp-Manton type of sporogenesis, in which the chromosome number of spore mother cells is doubled prior to meiosis, resulting in the formation of diploid spores. Cytological examination of hybrid Cyatheaceae (Conant & Cooper-Driver, 1980; Conant, unpub- lished data) has demonstrated that the hybrids have an unaltered sporogenesis. The number of spore mother cells per sporangium (4) and the number of bivalents at diakinesis (N = 69) is the same in hybrids and species. Furthermore, necrotic eggs were not observed in the archegonia of the hybrids examined in this study. It therefore seems unlikely that the sporophytes appearing in cultures of hybrid gametophytes arose apogamously. Although the mating system of tree ferns re- mains unknown, the bisexual condition and the synchronous maturation of male and female ga- metes indicate that the gametophytes of tree fern hybrids have the potential to self-fertilize. If this potential is realized in hybrid gametophytes, such as those shown here, and if the embryos and youn sporophytes develop and grow normally, complete- ly homozygous fertile F, hybrid sporophytes would result. If intragametophytic selfing is more common than outcrossing among gametophytes derived from these plants, it is possible that such homozygous hybrids would give rise to new, recombinant species. The mechanism that would maintain the repro- ductive isolation between these new allohomoploid species and their progenitors would be a mating system based on selfing. LITERATURE CITED CONANT, D. S. 1975. Hybrids in American Cyathea- ceae. n 77: 441-455. 83. A revision of the genus Alsophila (Cy- nold Arbor. 64: o. in the Americas. J. Ar -382. OPER-DRIVER. 1980. Autogamous al- a in Alsophila e Nephelea (Cyathea- ў а new hypothesis for speciation іп homoploid homosporous ferns. Amer. J. Bot. 67: 1269-1288 ;. Н. HAUFLER. 1976. Chromosome йы añ apomixis in the fern genus Bommeria Gymnogrammaceae). up a l- PÉREZ-GARCÍA, B R. Ripa. 1982. Germinación de esporas de Cyatheaceae bajo diversas temperaturas Biotropica 14: 281-287. Pray, T. R. 1971. The gametophytes of natural hie in the fern genus Pellaea. Amer. Fern J. 61: 128 136. STOKEY, A. С. Apogamy in Hi саш Bot. Gaz. бау ог) 65: 97- 1930. Prothallia of the ur RN Bot. бау: (Crawfordsville) 90: 1-45. Tryon, R. М. & A. F. Tryon. 1982. Ferns and Allied Plants. Springer- Verlag, Berlin. Wacner, W. Н S. WAGNER & W. C. TAYLOR. 1986. Detecting abortive aes in herbarium specimens of sterile hybrids. Amer. Fern J. 76: 129-140. WHITTIER, D. Р. & W. Н. Wacner. 1971. The vari- ation in spore size and rud in Dryopteris «127, . Amer. Fern J. 61: HYBRIDIZATION AND ALLOPOLYPLOIDY IN CENTRAL AMERICAN POLYSTICHUM: CYTOLOGICAL AND ISOZYME DOCUMENTATION! David S. Barrington? ABSTRACT Cytological di two diploid progenitors of the allopolyploid P. talamancanu ta for four Polystichum species and four hybrids between them fr rom the Sierra Ee: pra olystichum talamancanum and P. orbiculatum, also an allopolyploid, share an unidentified diploid progenito Polystichum hybrids was documented for in north te emperate these mona vana hybrids. The demonstrated pattern of reticulation is similar to such patterns in ferns of north temperate region The origin of fern species from sterile hybrids via chromosomal nondisjunction and polyploidiza- tion is supported almost entirely by work in nort temperate regions (Manton, 1950; Lovis, 1977; Barrington et al., 1989). Studies in north temperate Polystichum (Dryopteridaceae) have played a crit- ical role in demonstrating the contribution of hy- bridization and polyploidy to fern evolution (e.g., Daigobo, 1972; W. Wagner, 1973; Vida & Reich- stein, 1975; D. Wagner, 1979; Barrington, 1986; Soltis et al., 1987). Recently, a similar pattern of reticulate relationships among fern species has been proposed for Polystichum in the Sierra Talamanca of Costa Rica and Panama (Barrington, 1985a, b, c; Barrington et al., 1986; Barrington, 1989; Bar- rington et al., 1989). In this paper, I use chro- mosome and isozyme data to document the pro- posed relationships and provide insights into hybridization and hybrid speciation of Polystichum in montane tropical regions. Four Polystichum species are involved in hy- bridization in the Sierra Talamanca (Fig. 1). They comprise two broad-leafed diploid species, P. con- cinnum Lell. ex Barr. and P. speciosissimum (Kunze) R. Tryon & A. Tryon, and two narrow- leafed tetraploid species, P. talamancanum Barr. and P. orbiculatum (Desv.) Gay (Barrington, 1985a, b; Barrington, 1989). The forest-dwelling P. concinnum is endemic to the Sierra Talamanca, as is the low-alpine P. talamancanum. In contrast, the high-alpine P. speciosissimum is known from Mexico and Guatemala as well as Costa Rica, and P. orbiculatum is common in the high-alpine from Mexico to Bolivia. Four hybrids between these species have been proposed (Fig. 1). Hypotheses for hybridity and hybrid parentage as duo as s dofinifions of species were based on characters and degree of spore abortion em 1989). Morpholog- ically, the hybrids are noteworthy in that they include combinations of both morphologically dis- parate species such as Polystichum concinnum and P. speciosissimum and morphologically similar species such as P. orbiculatum and P. talaman- ! Rolla Tryon originally suggested that Polystichum in the American tropics was in need of systematic study. I thank Luis Diego Gome z P. for his support of my fieldwork in the Sierra Talamanca. Steven R. Hill, Pet er T. Hope, Bruce Howlett, and жен Zika all provided assistance in the field. Cathy Paris provided nue insights in the laboratory and careful readings of the manuscript. David S. Conant also provided a critical rev anonymous reviewer provided extensive, ud suggestions that substantially improved this work. Electrophoretic and cytological work reported here was begun at the University o leave in Kansas oe Ыр spring of 1985, п I spen Haufler's lab. [y Dr. Haufler for training in nt a sabbatical e electrophoresis. The релеи of Ver Pp support for laboratory and fieldwork through its кш ода! Grants BRSG-79 and В5С185-1; this contribution s the specific goal of the latter grant ? Pringle Herbarium, Department of Botany, University of Vermont, Burlington, Vermont 05405-0086, U.S.A. ANN. MISSOURI Вот. GARD. 77: 297-305. 1990. Annals of the Missouri Botanical Garden tala Emancanum 1 CCX X™ OX X Y-34—XX Y Y CCX ¿+ CS-——8SS ES COS 5 46 сопсіп Мес пит speciosis E simum , > D à (t PD atio ы NN | Proposed relationships a a and hybrids of Polystichum at Cerro B la Muerte, Sierra im e of undetermined : P. concinnum; S: P. speciosiss ас апит; Ү: genome of аа аи Баа unique to s of species (all Barrington collections deposited at . talamancanum, 1252. Scale bar, 1 leave speciosissimum, 1700; omes are as follows > orbiculatum, Talamanca, С osta Rica. derivation shared only by P. S ке and | Р. orbiculatum. Collection numbers for diit жон 316; I 1273; concinnum, 5 CI lowe r right, is Volume 77, Number 2 Barrington 299 Hybridization and Allopolyploidy in Central American Polystichum FIGURES 2-5. VT).—3. Р. speciosissimum, Meiosis 1 of Polystichum species from Costa Rica.—2. Р. concinnum, 41 pairs (Barrington 822 41 pai ҮТ) pairs (Barrington 976 .—4. P. talamancanum, 82 pairs (Barrington 1261 i ) VT).—5. P. orbiculatum, 81 pairs and 2 univalents indicated with arrows (Barrington 1273 VT). canum. The hybrids between closely allied species of Polystichum are especially difficult to distinguish morphologically because of overlap in ontogenetic and environmentally induced variation (Barrington, 1985a). A combination of morphological, cytolog- ical, and isozyme data is necessary to establish evolutionary relationships among these Polysti- chum taxa. METHODS FIELDWORK Species and hybrids of Polystichum were col- lected during several trips to the Cerro de la Muerte, San José and Cartago provinces, Costa Rica, be- tween 1980 and 1989. The study set comes from the montane rainforest and subalpine rain páramo at altitudes between 2,800 and 3,400 m; details of habitat and altitude for species and hybrids have been reported elsewhere (Barrington, 1985a, b, 989). CYTOLOGY Field-collected species and hybrids for cytolog- ical work were cultivated at the University of Ver- mont greenhouses. Young sporangia for study of meiotic chromosome number and pairing were fixed in freshly mixed Farmer’s solution (3 : 1 absolute ethanol : glacial acetic acid) for 18-24 hours, rinsed in 70% ethanol, and stored in 70% ethanol at 0°C. 300 Annals of the Missouri Botanical Garden TABLE 1. Fixed and common allozyme bands char- to nearest the origin. Allozymes are reported as acteristic of Costa Rican Polystichum Vnd Both bands of fixed heterozygotes reported. Bands are reporte percent of fastest migrating band. Bands зө of proposed diploid db as labeled: C = concinnum, = speciosissimum, X = genome of unknown origin shared by talamancanum 23 orbiculatum, Y — genome of unknown origin unique to orbiculatum. IDH РСІ-2 SkDH TPI-1 concinnum 82 100 C 100 C 68 speciosissimum 100 S 83 S 84 68 talamancanum 82 100 € 100 C 68 63 X 70 X orbiculatum 82 7 Y 84 86 Y 63 X 70 X 68 Material thus treated remained useful for analysis for at least three years. Staining of crushed spo- rangia was in 2% ferric acetocarmine at room temperature for about five minutes; Hoyer’s so- lution was thoroughly mixed into the staining mix- ture before squashing. ISOZYME ELECTROPHORESIS To document the heritage of hybrids, fresh leaf samples from greenhouse-grown species and hy- brids were ground in the phosphate extraction buff- er of Haufler (1985) in preparation for electro- Electrophoretic in Soltis et al. (1983). Triosephosphate isomerase (TPI), phosphoglucomutase (PGM), and phospho- glucose isomerase (PGI) were resolved on buffer system 6 of Soltis et al. (1983). Leucine amino- peptidase (LAP), aspartate aminotransferase (AAT), and hexokinase (HK) were assayed using buffer system 8 of Haufler (1985). System 11 of Haufler (1985) was used to resolve fructose 1-6 diphos- phatase (F 1 6DP), shikimate dehydrogenase (SkDH), isocitrate dehydrogenase (IDH), and malate de- hydrogenase (MDH). All isozymes migrated ano- dally; isozymes were numbered from most anodal percent of fastest-migrating band. RESULTS CHROMOSOME NUMBERS OF SPECIES Here I provide chromosome numbers for each of the four species as a basis for inferences about parentage of the hybrids encountered. The counts are all first records for the species. On the basis of counts for nine individuals, P. concinnum is diploid with 41 pairs of chromosomes at meiosis I ig. 2). Polystichum speciosissimum is also dip- loid, with 41 pairs in meiosis I, based on several counts of a single individual (Fig. 3). Polystichum talamancanum is tetraploid, since it showed 82 pairs of chromosomes in meiosis I, based on a sample of five individuals (Fig. 4). Material from two sporophytes of Р. orbiculatum collected on Cerro de la Muerte yielded counts of 82 pairs or 81 pairs and two univalents (Fig. 5), documenting it as tetraploid. ISOZYME PATTERNS OF THE SPECIES Unique combinations of fixed (or in one case very common) bands characterize the four Poly- stichum species from the Cerro de la Muerte (Table 1, Figs. 12-15). Two bands are diagnostic of P. concinnum: SkDH'” (present in all 14 sporophytes sampled) and PGI-2'” (present in 13 of 14 spo- rophytes sampled). Polystichum speciosissimum also has two diagnostic bands: IDH-2'" and PGI- 2%, both fixed in a sample of nine sporophytes. PGI-2'” and SkDH'"”, shared with P. concinnum, were present in 25 sampled plants of P. talaman- canum; the other alleles at each of these two loci (PGI-2% and SkDH”) have so far not been seen in diploid Polystichum species from Costa Rica. РС1-2° and SkDH", fixed in P. talamancanum but absent in P. concinnum, were also present in all seven sampled sporophytes of P. orbiculatum. Two bands at fixed heterozygous loci, PGI-2** and PI-1*, were unique to P. orbiculatum among Costa Rican species. x s 6-11. d 18 univé alents (Barrington 1274V Т). concinnum X p. speci Barrington 1274 VT). ~ 8 univalents (Barrington 1278 VT).— 10. 126: Meiosis I of Polystichum pone аш Costa Rica 9. Pune concinnum X P. talam Polystichum speciosissimum X P. ta nS 973 VT). — 11. Polystichum orbiculatum X P. talamancanum, 51 pairs and 62 univalents (Barrington "D. — .— 6. Р. PORC innum X P. браса каш mancanum, 41 pairs and 41 univalents lamancanum, 31 pairs and 60 univalents Volume 77, Number 2 1990 Barrington Hybridization and Allopolyploidy in Central American Polystichum 301 302 Annals of the Missouri Botanical Garden 12 o9» 2405-3.1185. * 729910465 7 ooooxttttxxcccxx ssssxxttt FIGURES 12-15. a { = Р. talamancanum, x = are average percents of fastest bar x, 1244, 1260; с, e 6, 822; x, 1274, 1510; s, 3. PGI-2: P (ит X ү talamancanum note r нше from (marker inherited from P. speciosissimum). — heterozygote combines bands of ME оғ mes 68 and 86 id 100 with all possible heterodimers. ТРІ-1% co-migrate Isozymes of Costa Rican е = P. rid between S ace id at each end Barrington order left to right (vouchers deposited at VT): o, 1273, 1504, 1509, 978; 81 1513, 1708, 1704, a х, " 1500. 973; t, concinnum shows three e heterodimer combining PGI-2 P. concinnum); in P. speciosissimum х P РС1-2'° (P. talamancanum vnnd inherite orbiculatum, s — P. nt es. Mobilities ee ciel to right of gels collection numbers 2 source plants for isozymes in , 1263, 1155, 1270, 1261; 1250, 1280, enotypes, two with the P. concinnum marker РСІ-2'%; ^ (P. orbiculatum marker) with PGI-2!° con cinnum, o d from P. concinnum) with : P. orbiculatum band interpretation- — fixed their respective modified versions at 86 and 's with the modified variant of TPI-1°. Other taxa have only the slower triplet. Alternatively 86 and 100 are the unmodified products Each of the two tetraploid species consistently showed a single heterozygous banding pattern at loci with more than one allele. Polystichum tala- mancanum showed fixed heterozygous patterns at HK, LAP, MDH, PGI-1, PGI-2, and SkDH in all sampled plants. All sampled plants of P. orbicu- latum showed fixed heterozygous expression at AAT, HK, LAP, PGI-1, PGI-2, PGM, SkDH, and TPI-1. Complexity of the fixed heterozygous pat- tern at ТРЇ-1 is hypothesized to be increased by Volume 77, Number 2 1990 Barrington 303 Hybridization and Allopolyploidy in Central American Polystichum post-translational modification, as discussed in Gas- tony (1988); see caption to Figure 15 for inter- pretation. CHROMOSOME NUMBERS AND PAIRING BEHAVIOR OF HYBRIDS The four putative hybrids from the Cerro de la Muerte were documented with information on chro- mosome number and pairing behavior. The hybrid between the two diploid species, P. concinnum and P. speciosissimum, was documented to be diploid based on counts of two hybrid individuals (Bar- rington 705 and 1274). Pairing was notably vari- able within individuals, ranging from 32 pairs and 18 univalents to 40 pairs and 2 univalents (Figs. 6, 7). Trivalents were also observed (Fig. 8). The hybrid between P. concinnum and P. talaman- canum, based on chromosome counts of six indi- viduals, is triploid with 41 pairs and 41 univalents at meiosis I (Fig. 9). The hybrid between Р. spe- ciosissimum and P. talamancanum is represented in this study by two individuals (Barrington 973 and 1500), the first of which yielded a triploid count of 31 pairs and 61 univalents in meiosis I (Fig. 10). The fourth hybrid, Р. orbiculatum x P. talamancanum, is represented by a single in- dividual, Barrington 1283. This hybrid yielded a tetraploid count of 51 pairs and 62 univalents in meiosis I (Fig. 11). ISOZYME PATTERNS OF HYBRIDS Hybrids have isozyme profiles consistent with their parentage as proposed from morphological criteria (Figs. 12-15). Based on two hybrid plants, Barrington 1274 and 1516, Polystichum con- cinnum X P. speciosissimum includes the marker bands of P. concinnum (PGI-2' and SkDH'”) and P. speciosissimum (IDH-2'% and PGI-2*). The hybrid between P. concinnum and P. talaman- canum has the same diagnostic bands and fixed heterozygous loci as P. talamancanum, based on electrophoretic analysis of three individuals. Both individuals of P. speciosissimum X P. talaman- canum include the marker bands reported for the proposed parents (IDH-2' and PCI-2* for P. spe- ciosissimum; РС1-2'%° and SkDH'” for P. tala- mancanum). Polystichum orbiculatum X P. tala- mancanum includes the diagnostic bands for P. talamancanum and P. orbiculatum (that is, PGI- 21% and SkDH'” shared with P. talamancanum, PGI-2* and SkDH” shared with both P. talaman- canum and P. orbiculatum, апа PGI-2** and TPI- 1% shared only with P. orbiculatum). DISCUSSION INFERENCES ABOUT EVOLUTIONARY RELATIONSHIPS Consideration of the cytological and electropho- retic data presented here leads to hypotheses for evolutionary relationships between Polystichum species on the Cerro de la Muerte (Fig. 1, Table 1). Polystichum talamancanum is an allotetra- ploid, since it has 82 pairs of chromosomes in meiosis I and its heterozygous banding patterns are all fixed. Corroborating evidence is available: the triploid P. speciosissimum X mancanum consistently shows fewer than 41 pairs of chro- mosomes, suggesting that the hybrid combines three nonhomologous sets of chromosomes. Hence the two sets of chromosomes contributed by P. tala- mancanum are not homologous, so P. talaman- canum must be an allotetraploid. Furthermore, because all marker allozymes for P. concinnum are included in the fixed heterozygous banding pat- terns of P. talamancanum, one of the progenitors of P. talamancanum is almost certainly P. con- cinnum. This interpretation is supported by the pairing pattern (equal number of univalents and pairs) in P. concinnum x P. ae typ- ll ls and their diploid progenitors. Polystichum filaman т 11 is almost certainly recent in origin, since its geo- graphic range is highly restricted and coterminous with one of its progenitors (Stebbins, 1971). Polystichum orbiculatum, a common alpine species from Bolivia to Mexico, is allotetraploid at Cerro de la Muerte; it shows the fixed pala y 33 pattern typical of allopolyploids that combine ferent alleles from each of their diploid rose Polystichum orbiculatum, as broadly circum- scribed by recent taxonomists (Smith, 1981; Stolze, 1981), may also include one or both of its diploid progenitor species, since small-spored plants have been encountered in Venezuela (Barrington et al., 1986). A similar species perhaps involved in the ancestry of P. orbiculatum is P. sodiroi Christ of Ecuador, for which a diploid count of 2n = 82 has been reported (Sorsa in Fabbri, 1965). Polystichum orbiculatum evidently shares one parent with P. talamancanum, since a subset of marker allozymes is consistently shared by the two allotetraploid species. Corroborating evidence for this conclusion is the pairing pattern in P. orbi- culatum X talamancanum. lts 51 pairs and 62 univalents are interpreted as representing two homologous sets of chromosomes, one contributed by each of the two tetraploids reported here, and two nonhomologous sets (some members of which ical of backcrosses between 304 Annals of the Missouri Botanical Garden pair), representing a concinnum genome contrib- uted by P. talamancanum and a genome of un- known origin contributed by P. orbiculatum. The two homologous sets, which undergo allosyndetic pairing in this hybrid, are presumably derived from a single diploid progenitor, so far not documented from the Cerro. This implication of a single diploid progenitor in the origin of more than one allote- traploid is a pattern already documented in the north temperate zone (e.g., Dryopteris cristata and D. carthusiana, Werth, 1989) GENERAL SIGNIFICANCE There is no evidence that the species on Cerro de la Muerte constitute a monophyletic group; on the contrary, there is ample morphological evi- dence that the diploid species present or implicated in the origin of hybrid species on the Cerro are members of different species groups with geograph- ically disparate affinities. For instance, Polysti- chum concinnum is allied to the diverse complex of species from southern Mexico (Barrington, 1989), but P. orbiculatum is presumably Andean in origin (Barrington et al., 1986). Although P. speciosis- simum is sian ушын distinct enough to have been recognized as a genus Plecosorus, its conie behavior in mei- notypic species of the osis I suggests that it is not strongly differentiated from the remainder of the taxa here. Even in a large and morphologically diverse genus like Po- lystichum, hybridization and polyploidy depend on geographic and ecological proximity. The scope of Polystichum species involved in secondary inter- actions is not limited by phylogenetic proximity (as inferred from morphology). high frequency of homoeologous pairing is documented here for hybrids between tropical- montane Polystichum species, just as it was for north temperate hybrids (Wagner, 1973). There is also notable variation in pairing in these hybrids. Furthermore, degree of homoeologous pairing in hybrids is independent of morphological similarity in these species. Especially telling is the hybrid between Polystichum concinnum and P. speciosis- simum, which shows virtually complete pair for- mation in some cells although the progenitor species are strongly differentiated morphologically (Bar- rington, 1985b). This emerging picture of unusual residual homology (homoeology), typical in Poly- Polypodiales, may reflect one of two cytological phenomena. It is possible that divergence of whole genomes is not so great in Polystichum species as in other genera of ferns, in which case the genus has a noteworthy stichum but rare elsewhere in the place in studies of fern evolution. Alternatively, pairing may be under less stringent genetic control in Polystichum than in other fern groups, thus permitting the relatively high frequency of homoeo- logous pairing observed in Polystichum hybrids (Wagner, 1973; Barrington, 1986) The pattern of interactions among Polystichum species in high-montane Costa Rica emerging from this work and from earlier papers is notable for its similarity to patterns of hybridization and specia- tion in north temperate ferns in the Dryopterida- ceae, Aspleniaceae, and other groups (Lovis, 1977; Barrington et al., 1989). The ecology of interaction among species (Barrington, 1985b), the morphol- ogy of the hybrids and hybrid species (Barrington, 1985a, b, 1989), the basic features of chromosome behavior at meiosis I, and isozyme patterns are all similar to those seen in north temperate polyploid complexes. Since Pleistocene climatological change was significant in tropical America (van der Ham- men, 1974), vegetational change and ecological disturbance in the Sierra Talamanca may have been qualitatively like that in the north temperate zone. Hence, similar Pleistocene climatological his- tory may be the underlying determinant of specia- tion phenomena in the genus Polystichum in both regions. LITERATURE CITED BARRINGTON, D. S. 1985a. The morphology and origin of a new Polystic Xo hybrid from Costa Rica. Syst. Bot. 10: 199-204. 1985b.- Hybridisation in Costa Rican Polys- ис hum. Proc. Royal Soc. Edinburgh 86B: 335- iid ——, 5c. The present evolutionary and ta nomic statu the fern genus Polystichum: le 1984 баш] Society of America оо 5ес- tion ives Amer. Fern J. 75: 2 The morphology and к of Po- ly "T hum potter hybr. nov crostichoides x P. braunii). Rhodora 88: 297-313 New species and nahian i in trop- ical American Polystichum (Dryopteridaceae). Ann. Missouri Bot. ү 65-373. ;C. LER & C. R. WERTH. 1989. Hy- břidization, aa and species concepts in ferns. Amer. Fern E 55-64 ! 5& T. A. RANKER. 1986. System- atic Я Bom spore and stomate size in the erns. Amer. Fern J. 76: 149-159. DAIGOBO, "s. 1972. Taxonomical studies on the fern genus Polystichum in Japan, Ryukyu, and s Sci. Rep . Tokyo Bunrika Daigaku. Sect. B. 57 0 + FABBRI, Е. 1965. Secondo supplemento alle Tavole cro- mosomiche delle Pteridophyta di Alberto Chiarugi. Caryologia 16: 675-731. Da G. da The Pellaea glabelle complex: electrophoretic evidence for the derivations of aga- Volume 77, Number 2 1990 Barringto 305 Hybridization and Allopolyploidy in Central American Polystichum mosporous taxa and a revised taxonomy. Amer. Fern ‚ 78: 44-6 HAUFLER, C. Н. 85. Enzyme variability and modes of evolution in Bommeria (Pteridaceae). Syst. Bot 10: 92-104 1950. ns of quus and Evolution in the Pteridophyta. Cambridge Univ. Press, Cam Part 2. Pteridophytes. /n: D. E. ), Flora of А California Acad- my of Sciences, “San u ra была, D. E., С. Н. HaurL n. c Darrow & С. J. ASTONY. 1983. Starch ‘cel elecirophoresi of ferns: a compilation of grinding buffers, ge electrode buffers, and staining schedules. pea Ten J. 73 SOLTIS, . P. 8. D. E. SoLris & Е. К. ALVERSON. 1987. Electrophoretic and morphological confirmation of interspecific hybridization between Polystichum kruckebergii and P. munitum. Amer. Fern J. 77: 42-49. STEBBINS, С. L. Chromosomal Evolution and > Plants. Addison Wesley, Reading, Massachu- . Ferns and fern alli a R. G. 1981 f Guatemala. Part 2. Polypodiaceae. Fieldiana, Bot. n.s. 6: 1-522 VAN DER HAMMEN, Т. 1974 e Pleistocene changes in a m climate in tropical South America. J. Biogeogr. 1: 3- Vipa, С. & Т. y on HSTEIN. 1975. Taxonomic problems in the fern genus Polystichum caused by hybridiza- tion. Pp. 126-135 in S. M. Walters (editor), Eu- ropean Floristic and Taxonomic Studies. E. W. Clas sey, itas England. WAGNER, D. Н. 1979. Systematics of Polystichum in western North America and north of Mexico. Pterido- logia 1: 1-64. WAGNER, W. H., JR. 1973. Reticulation in holly ferns (Polystichum) i in the western United States and ad- jacent Canada. Amer. Fern J. 63: 99-115. WERTH, C. R. . Isozyme evidence on the origin of Dryopteris cristata and D. carthusiana. Amer. J. Bot. 76(6, supplement): 208. ELECTROPHORETIC EVIDENCE FOR ALLOTETRAPLOIDY WITH SEGREGATING HETEROZYGOSITY IN SOUTH AFRICAN PELLAEA RUFA A. F. TRYON (ADIANTACEAE)' Gerald J. Gastony? ABSTRACT Pellaea rufa is a sexually reproducing tetraploid species with a chromosome number of n = I at meiosis. Unlike previously reported tetraploid ferns, sporophytes of P. rufa exhibit substantial intrapopulational electrophoretic variation id respec ctive enzyme phenotyp es, and many of their heterozygous phenotypes are segregating rather than fixed. Two hundred and bun pope gametophytes from a tetraploid sporophyte ои for four alleles at tk PCL: 2 locus were subjected to elec steph id expectations under the hypothesis that Р. г intragenomic heterozygosity. tetraploid Р. rufa via gametes from unreduce spores from intermediate allodiplo ed spore trophoresis to determine whether inheritance is tetrasomic as expected i or disomic as oe ected in an allotetraploid. Results reject the hypothesis of anal, but fit rufa is an allotetraploid with fixed intergenomic heterozygosity and segregating bserved intragenomic d is consistent with a direct hybrid origin of allo- oids. Disomic ае ance of intragenomic he їп an variation in P. rufa that is less than that in autopolyploids with polysomic inheritance e greater than that previously reported in allopolyploids. Pellaea rufa A. F. Tryon is endemic to xeric areas of the southern Cape Province of South Af- rica (Anthony, 1984; Schelpe & Anthony, 1986) and is the only species of Pellaea sect. Pellaea with a non-American distribution (Tryon, 1957). Until Tryon (1955) recognized their morphologi- cally distinctive features, these African plants were considered conspecific with Р. andromedifolia (Kaulf.) Fée from California and Baja California Norte. Tryon (1955, 1957) had no living material of P. rufa from which chromosome counts could be made. She noted, however, that sporophytes have 64-spored sporangia, the number typical of sex- ually reproducing diploid species of Pellaea, as opposed to the 32-spored condition characteristic of agamosporous taxa that are typically triploid or tetraploid in this genus (Tryon & Britton, 1958; Tryon, 1968). Three recent floristic works (Jacob- sen, 1983; Anthony, 1984; Schelpe & Anthony, 1986) included treatments of P. rufa, but offered no observations regarding its chromosome number or reproductive biology. Tryon’s (1955, 1957) and Anthony’s (1984) observations of 64 spores per sporangium therefore stand alone in suggesting that this species is a normal, sexually reproducing dip- loid like most populations of P. andromedifolia, its putative close relative in California. This paper presents both cytological and isozymic evidence that Р. rufa is tetraploid and analyzes its segre- gating electrophoretic phenotypes to determine whether it had an autopolyploid (intraspecific) or allopolyploid (interspecific) origin. MATERIALS AND METHODS Live sporophytes of Pellaea rufa A. F. Tryon grown in the Indiana University greenhouses were ' This paper is dedicated to Dr. Rolla M. Tryon, Jr., and Dr. Alice F. Tryon to whom I am grateful for the graduate training that makes my research possible. I thank Nicola C. Anthony and P. Anne of Pellaea rufa, Valerie R. Savage for meticulous help as laboratory technician, Marcus С. Rhoades, Ellen De стресу апа Drew ee for discussions of cytogenetics, and the National Science ee for support under grant d С. R. Werth. acknowledge helpful reviews of this BSR 8516666 auller an b > Department of Biology, Indiana University, Bloomington, Indiana 47405, U.S.A. ANN. Missouni Вот. Garb. 77: 306-313. 1990. Volume 77, Number 2 1990 Gastony 307 Electrophoretic Evidence for Allotetraploidy in Pellaea rufa obtained from four native populations in the Cape Province of South Africa (““sporophytes”” below). Additional accessions of sporophytes from these populations did not survive transport to Indiana and were instead analyzed via families of game- tophytes grown from their spores (“‘gametophyte families”” below). These four populational sources are (1) ee (grid relereico 3222AD), five (grid reference 3320BA), 14 sporophytes and 10 gametophyte families; (3) Die Hel (grid reference 3321BC), seven sporo- phytes and 18 gametophyte families; (4) Boonstevlei (grid reference 3220DD), 22 sporophytes and 13 gametophyte families. Specimens of P. androme- difolia used in Figures 5 and 6 are from popula- tions 2, 5, б, 7, апа 11 of Gastony & Gottlieb (1985). Voucher specimens are deposited at IND. For meiotic chromosome squashes, portions of spo- rophylls were fixed at 21°C in 3: 1 absolute alco- hol : glacial acetic acid for at least 24 hours. Spo- rangia were then removed individually and squashed following Manton’s (1950) acetocarmine tech- nique. For mitotic chromosomes, root tip squashes were prepared as in Roy & Manton (1965) using glusulase obtained from DuPont pharmaceuticals. Electrophoresis and staining methods were as in Soltis et al. (1983), using 12.8% starch and the tris-HCl grinding buffer with 12% PVP. Single gametophytes were subjected to electrophoresis as in Gastony & Gottlieb (1982) except that each was ground with one or two drops of grinding buffer from the tip of a Pipetman 100 in its own well in a porcelain spot plate (mortar) using a small glass test tube as a pestle. Shikimate dehydrogenase (SkDH) was resolved on gel and electrode system 2, and phosphoglucoisomerase (PGI) was resolved on system 7. Allozymes are labeled with the most anodally migrating one designated A, the next most anodal labeled B, etc., and with the corresponding coding alleles labeled a, b, etc. RESULTS AND DISCUSSION CHROMOSOME COUNTS Recently obtained living sporophytes of P. rufa have provided the first chromosome counts for this species, showing that it is tetraploid, not diploid, with п = 58 II at meiosis (Figs. 1—3). Counts of approximately 116 mitotic chromosomes (Fig. 4) in several cells from root tips confirm this tetra- ploidy and demonstrate the normal alternation of 116 mitotic chromosomes and 58 meiotic bivalents expected with 64-spored sexually reproducing tet- raploids. ELECTROPHORESIS Studies of electrophoretic isozyme patterns were initiated as part of an ongoing study of the light- stiped species of Pellaea sect. Pellaea. In P. rufa, these revealed the complex enzyme phenotypes expected from tetraploids with up to four alleles per enzyme locus. Results for the monomeric en- zyme SkDH and the dimeric enzyme PGI illustrate these findings. For SkDH, diploid sporophytes of Р. andro- medifolia (Fig. 5, lanes AND) exhibit a single al- lozyme band when they are homozygous and two bands of equal staining intensities when they are heterozygous at this locus (Gastony & Gottlieb, 1982). All other lanes in Figure 5 are a random selection of sporophytes of tetraploid P. rufa from four different populations. Only four of these 23 sporophytes of P. rufa (17%) are one-banded as expected of plants homozygous at this locus, where- as 83% of them are at least two-banded and there- fore heterozygous at this locus. Allozymes A, B, C, and D are coded at this locus (Fig. 5), and two of these sporophytes encode three allozymes at the Skdh locus, manifesting the extra heterozygosity per locus expected in tetraploids. PGI is expressed as chloroplastic and cytosolic isozymes (Gastony & Darrow, 1983). PGI-1, the chloroplastic more anodal isozyme (Fig. 6), is not well resolved in species of Pellaea and is not dis- cussed further. A diversity of patterns is expressed for the cytosolic cathodal isozyme (PGI-2) in P. rufa. Lanes 7—11 (AND) of Figure 6 are from diploid sporophytes of Californian P. andromedi- folia, which expresses the normal diploid comple- ment of two copies of the locus coding PGI-2, one in each of its chromosome sets. Diploids homozy- gous at the locus coding this dimeric enzyme (lanes 7, 8, 10, 11) express a single homodimeric allo- zyme band, whereas diploids heterozygous at this locus (lane 9) exhibit a three-banded pattern of symmetrical staining intensities with outer homo- dimeric bands and a central heterodimeric band combining the monomeric subunits coded by both alleles (Gastony & Gottlieb, 1982). Tetraploid spo- rophytes of P. rufa have four copies of the Pgi-2 locus, one copy in each chromosome set. These tetraploids express up to four homodimeric PGI-2 allozymes (Figs. 6 and 7, bands AA, BB, CC, DD) plus heterodimeric bands in all possible combina- tions (six bands: AB, AC, AD, BC, BD, CD), so that a sporophyte heterozygous for all four alleles expresses ten bands very closely spaced, as in Figure 6 lane 4 and Figure 7 lane 2 (in both figures, heterodimeric bands AD and BC are contiguous Annals of the Missouri Botanical Garden FIGURES 1-4. Meiotic chromosomes showing n = 58 II, from Boonstevlei Chromosomes of Pellaea rufa. — 1, 2 sporophytes B-39 and B-1 respectively. — 3. Camera lu- cida drawing interpreting areas of overlap in Figure 2. — and partially overlapping, and heterodimeric band CD is contiguous with faint homodimeric band CC). Allozymes specified by alleles b, c, and d are often more weakly expressed than those specified by allele a even when these alleles are present in equal doses, as in the sporophyte in Figure 6 lane 4 and Figure 7 lane 2 whose Pgi-2 genotype is abcd as shown below. The occurrence of tetraploid sporophytes with varying electrophoretic phenotypes for SkDH and PGI in P. rufa was unexpected. In previous studies of polyploid ferns, all sporophytes in a given pop- ulation have been monomorphic for each enzyme pattern, except for occasional individuals with vari- ant phenotypes attributed to independent allopoly- ploid origins from parental diploids of variant ge- notype, or to mutations, or to rare recombinational events involving pairing among homoeologues in the polyploids (e.g., Werth et al., 1985b). Such phenotypic monomorphism was explained in the classic study of Tragopogon by Roose & Gottlieb (1976), where allotetraploids exhibit additive expression of the bands coded by the divergent genomes of their diploid progenitors. The allodip- loid hybrid inherits the variant alleles at the enzyme loci of both progenitor species and expresses both allozymes. By doubling its chromosomes, the al- lodiploid becomes an allotetraploid in which the chromosomes of each parental genome have iden- tical mates with which they preferentiallly pair at meiosis. Intragenomic pairing of the doubled chro- mosomes precludes genetic segregation, all meiotic products maintain the intergenomic heterozygosity of the allotetraploid sporophyte, and the fixed het- erozygous enzyme phenotypes in the population are thereby perpetuated generation after genera- n. Similar fixation (nonsegregation) of enzyme phe- notypes in previously studied polyploid ferns is reported in the study of the Appalachian Aspleni- um complex by Werth et al. (1985a, b) and in Haufler's (1985) study of tetraploid Cystopteris fragilis. In both of these studies, as in that of Roose & Gottlieb (1976), lack of segregation of heterozygous enzyme phenotypes in allotetraploid taxa is attributable to the fact that heterozygosity is intergenomic whereas chromosome pairing is in- tragenomic, between the doubled chromosomes within each of the two divergent genomes inherited from the diploid progenitor species. -— 4. Mitotic chromosomes showing 2n = ca. 116 in Skeiding sporophyte 5-8. Scale bars = 0.01 mm. Volume 77, Number 2 1990 Gastony 309 Electrophoretic Evidence for Allotetraploidy in Pellaea rufa FIGURES 5-10. figures, the anode is toward the to Monomeric allozymes A, B, C, Dc p and origin constituting variant s at the and s t SDH "oe of randomly а tetraploid sporophytes from four populations of P. rufa, except for diploid sporophytes of P. andromedifolia in lanes Electrophoretic SkDH and А һи cuam of баер, rufa and Е ен мм НА In all cm —5. es Figures 5-10. omodimeric and heterodimeric allozymes constituting variant phenotypes of PGI-2 (PGI-1 inadequately resolved) in randomly E hands P. andromedifolia AD and BD are not labeled in the right lane but are identifiable by comparison with es counterparts in the left lane. Bands AD and BC are contiguous and partially overlapping, and bands contiguous. — 8. Variant At phenotypes of randomly selected sporophytes from the Skeiding (S) and Boonstevlei (В) populations of P. rufa. — 9. Three-banded SkDH phenotype i in a sporophyte (S) of P. rufa and in an extract of mixed gametophytes (MG) from that sporophyte, with segregation of two-banded patterns in single gametophytes (G).—10. Segregation of homodimeric d PGI-2 phenotypes AA/CC, AA/DD, BB/CC, BB/DD (with respective ree y р bands AC, AD, BC, BD unlabeled) in gametophytes from a sporophyte with the phenotype of Figure 6 lane 4 a Figure 7 lane 2. Heterozygous electrophoretic phenotypes are also fixed in polyploid agamosporous fern taxa with the Dópp-Manton type of sporogenesis (Gastony & Gottlieb, 1985; Gastony, 1988; Gastony & Wind- ham, 1989). Sporophytes of these taxa produce viable spores via modified sporogenesis (Manton, 1950) in which the final mitotic division before meiosis is endomitotic. This yields restitution nuclei of doubled ploidy in which each chromosome has a mitotically derived, duplicate sister chromosome with which to pair. Because pairing is restricted to these genetically identical sister chromosomes (Lovis, 1977), there is no genetic segregation at meiosis. As reported in allopolyploid Asplenium species and allotetraploid Cystopteris fragilis, the banding patterns of agamosporous sporophytes are invariantly expressed in their individual gameto- phytes and in the sporophytes they in turn beget. Contrary to the findings of these previous stud- ies, diverse enzyme phenotypes are seen among : he Ta single populations of tetraploid P. rufa (F and many of the heterozygous E A ai leerte do segregate, indicating an intralocus component to the heterozygosity. For example, three-banded (BCD) SkDH patterns of P. rufa sporophytes (Fig. 9, S), maintained when sev- eral randomly selected gametophytes are combined for electrophoresis (Fig. 9, MG), segregate phe- notypes BD and BC in single gametophytes (Fig. 9, G). Similarly, single gametophytes from sporophytes with the PGI-2 phenotype seen in Figure 6 lane 4 and Figure 7 lane 2 segregate diverse banding patterns for PGI-2 (Fig. 10). Thus, unlike previous reports for polyploid ferns, heterozygous banding patterns in tetraploid P. rufa are coded by up to four alleles per enzyme locus, and these alleles and their coded allozymes do segregate. Meiotic seg- regation of these genotypes into spores and deriv- ative gametophytes and the subsequent random 310 Annals of the Missouri Botanical Garden recombination of gametophytic genotypes into spo- rophytes at fertilization accounts for the diversity of PGI-2 phenotypes seen among sporophytes with- in single populations. Tetrasomic inheritance of segregating allozyme markers is increasingly invoked as evidence that tetraploid populations of various angiosperm species are autotetraploid (Soltis 8 Rieseberg, 1986; Soltis & Soltis, 1988, 1989; Rieseberg € Doyle, 1989; Wolf et al., 1989). Tetrasomic inheritance is to date unreported in pteridophytes (Werth, 1989). Because P. rufa provides the first reported instance of regularly occurring segregation of electropho- retic phenotypes in natural popu lations of tetraploid ferns, it is of interest to determine whether this species exhibits tetrasomic inheritance, a corollary of autopolyploidy (intraspecific polyploidy) as dis- cussed in Soltis & Rieseberg (1986), Soltis & Soltis (1989), and Wolf et al. (1989), ог r disomic inher itance as expected from all polyploidy). In ferns, autotetraploidy may result when a nor- mal diploid sporophyte produces unreduced spores whose diploid gametophytes self-fertilize or cross- fertilize with another diploid gametophyte from the same sporophyte or from a different sporophyte of the same species. Alternatively, an unreduced dip- loid gametophyte may cross with a normal haploid gametophyte of the same species to yield an au- totriploid sporophyte. If the autotriploid produces unreduced spores, the triploid gametophytes may cross with a normal haploid gametophyte of the same species to yield an autotetraploid by this less direct route (see discussion of these processes in Haufler et al., 1985; Gastony, 1986). In such cases, the progenitor diploid sporophytes have two homologues for each kind of chromosome, and those homologues pair regularly as bivalents at meiosis. Gametophytes from the unreduced diploid spores of those sporophytes carry the same pairable sets of homologous chromosomes, and when these unreduced gametophytes cross as noted above, the resulting triploid or tetraploid sporophyte has three or four homologues respectively for each chro- mosome. Because each of these homologues is ca- pable of pairing with the others, some degree of multivalent chromosome pairing is generally ex- pected in autotetraploids (Stebbins, 1947; Jackson & Casey, 1982). Although only bivalent pairing has been observed in P. rufa, lack of multivalents cannot be regarded as evidence against autopoly- ploidy in this species, because bivalent formation appears to be under genetic control in autopolyploid ferns (see review in Lovis, 1977; Gastony & Win ham, 1989). Genetic control is shown, for example, I in an agamosporous triploid species in which some sporangia have one endomitotic division and others have two successive endomitotic divisions preced- ing meiosis (Windham, pers. comm.). Chromosome pairing is strictly bivalent in the latter sporangia, although the same genome is represented at least four and possibly twelve times in them. Although only bivalents form in autotetraploid ferns, pairing partners are expected to be randomly chosen from among the homologues, with resultant tetrasomic inheritance. Such random pairing is expected be- cause the multiple homologues resulting from un- reduced spores should be indistinguishable except inasmuch as they may carry different alleles at a given locus—a factor that does not affect pairing potential in Mendelian genetics. If P. rufa is autotetraploid, sporophytes should exhibit tetrasomic inheritance when single game- tophytes derived from their spores are analyzed electrophoretically. If it is allotetraploid, sporo- phytes should exhibit disomic inheritance with seg- regation restricted to whatever intralocus hetero- zygosity may be present within each ancestral genome. Because autopolyploidy and allopolyploidy predict different genotypic results in gametophytic progeny, critical analysis of genetic segregation can be used to test the nature of polyploidy in P. rufa. Tetrasomic inheritance in an autotetraploid with random bivalent pairing and no crossing over be- tween the Pgi-2 locus and the centromere would yield six classes of gametophytic genotypes from a tetraploid sporophyte like that in Figure 6 lane 4 and Figure 7 lane 2 with genotype abcd: ab, ac, ad, bc, bd, cd. If the Pgi-2 locus is sufficiently distant from the centromere that crossing over can occur betwen them, chromatid recombination could generate four additional gametophytic genotypes аа, bb, cc, dd) so that tetrasomic inheritance should then yield ten classes of gametophyte ge- notypes: ab, ac, ad, bc, bd, cd, aa, bb, cc, dd. Because the position of the Pgi-2 locus relative to the centromere is unknown in P. rufa, six to ten classes of gametophytes should be expected if a sporophyte of P. rufa with Pgi-2 genotype abcd is autotetraploid. In the absence of recombination, the six classes noted above should occur with equal frequencies. With disomic inheritance in an allotetraploid, the genotypes of the gametophytes derived from an abcd sporophyte will depend on how the four alleles are distributed within the sporophyte's two ancestral genomes. Three different distributions are possible, each with its respective set of gameto- phytic genotypes: (1) аб in one progenitor genome and cd in the other would yield gametophytic classes — Volume 77, Number 2 1990 Gastony Electrophoretic Evidence for Allotetraploidy in Pellaea rufa ac, ad, bc, bd; (2) ac in one genome and bd in the other would yield gametophytic classes ab, ad, cb, cd; (3) ad in one genome and bc in the other would yield gametophytic classes ab, ac, db, dc. No mat- ter how these alleles are distributed in the ancestral sporophytic genomes, chromosome pairing will be intragenomic, and this will yield only four classes of gametophytes. Chromatid recombination will not create more than four classes of gametophytes in the case of an allotetraploid with bivalent pairing. An allotetraploid not entirely restricted to prefer- ential intragenomic pairing might engage in some random bivalent pairing or quadrivalent pairing (if bivalent pairing were not under genetic control), with segregational results approaching those of an autotetraploid. In that case, an allotetraploid could generate six or more classes of gametophytes and would be indistinguishable from an autotetraploid. Because only bivalents have been observed at mei- osis in the sporophytes examined (e.g., Figs. 1-3), quadrivalent pairing in P. rufa is not likely. It is possible, however, that brief quadrivalent pairing at an early stage of prophase has gone unnoticed. Therefore if six or more classes of gametophytes are observed, further effort would be required to distinguish between the hypotheses of autopolyploi- dy and allopolyploidy. On the other hand, if no more than four classes of gametophytes are found in a large sample, inheritance is disomic and P. rufa is allotetraploid with intragenomic bivalent pairing. Two hundred and twenty single gametophytes derived from a sporophyte with PGI-2 genotype abcd and the phenotype seen in Figure 6 lane 4 and Figure 7 lane 2 were subjected to electropho- resis and assayed for PGI-2. Only four classes of gametophytic phenotypes were observed: AA/CC, AA/DD, BB/CC, and BB/DD (each with its re- spective heterodimeric band; Fig. 10). The classes that were not observed are noteworthy. No ga- metophytic phenotypes corresponding to genotypes aa, bb, cc, or dd were observed. Those four phe- notypes should have been observed if P. rufa were an autotetraploid with random bivalent pairing and with crossing over between the PGI-2 locus and the centromere. Furthermore, no gametophytic phenotypes corresponding to genotypes ab and cd were observed. Those two classes should have been observed if Р. rufa were autotetraploid with random bivalent pairing and with no crossing over between the PGI-2 locus and the centromere. The absence of these six progeny classes rejects the hypothesis that P. rufa is autotetraploid. Moreover, because paired homologues separate at the first division of meiosis, unobserved classes AA/BB and CC/DD are precisely those that should be lacking if alleles ab are in one ancestral genome and alleles cd are in the other ancestral genome of an allotetraploid with bivalent pairing restricted to intragenomic homologues. The observed absence of gametophytic classes AA/BB and CC/DD is therefore predicted by the hypothesis that P. rufa is allotetraploid with disomic inheritance resulting from exclusively intragenomic bivalent pairing. Fi- nally, if the allotetraploid hypothesis is correct, the four observed classes of gametophytes are the only ones expected, and they should occur in equal frequencies in a large enough sample. The fre- quencies of the four observed classes of gameto- phytes were AA/CC = 49, AA/DD = 55, BB/ CC = 55, and BB/DD = 61. A Chi-square test indicates that these results are not statistically dif- ferent from the ratio of 1:1:1:1 expected for these 220 gametophytes (x? = 1.309, P > 0.70). Pellaea rufa is therefore accepted as an allotetra- ploid species. Furthermore, it is the first electro- phoretically demonstrated example of an allopoly- ploid fern species in which fixed intergenomic heterozygosity is associated with segregating intra- genomic heterozygosity. Several species of Pellaea that might have been involved in an allopolyploid origin of P. rufa occur in South Africa (Anthony, 1984), but they are currently placed in sect. Holo- chlaena (Tryon & Tryon, 1982). Although the segregational data from P. rufa appear to fit all expectations for an allotetraploid, it has been suggested (Haufler, pers. comm.) that the abcd Pgi-2 genotype of P. rufa may never- theless have had an autopolyploid (intraspecific) origin. If a single diploid species ancestral to P. rufa contained alleles a, b, c, and d at Pgi-2, a given diploid sporophyte may have been hetero- zygous for alleles ab and another may have been heterozygous for alleles cd. Endomitotic divisions prior to sporogenesis in these sporophytes would yield unreduced diploid spores, gametophytes, and gametes such that an egg carrying Pgi-2 alleles ab could be fertilized by a sperm carrying alleles cd, yielding an autotetraploid sporophyte with Pgi-2 genotype abcd. The four observed classes of ga- metophyte progeny discussed above (Fig. 10: AA/ CC, AA/DD, BB/CC, BB/DD) could be imagined to result from such an autotetraploid sporophyte if meiotic chromosome pairing were genetically re- stricted to bivalents in which pairing partners are not random, as expected in autotetraploids (Wolf et al., 1989), but invariably limited to chromosome pairs a/b and c/d. Such restricted pairing would be identical to what is expected and observed in allotetraploids, where preferential intragenomic 312 Annals of the Missouri Botanical Garden pairing is attributable to differences between the diverged genomes of the two progenitor species. This hypothesis that P. rufa may be an autotetra- ploid with preferential bivalent pairing seems im- plausible for two reasons. (1) In an autotetraploid, the four homologous chromosomes would be de- rived from the pairable parental diploid chromo- somes that differ only in their alleles at the Pgi- locus. Thus there is no reason to assume that pairing would be restricted to only one of the three combinations possible with four homologous chro- mosomes. On the other hand, the observed seg- regations are readily explained by а airing behavior well known in ferns. (2) This hy- pothesized autotetraploid origin of the Ren abed sporophyte involves the improbable simultaneous occurrence of three rare events: two independent cases of unreduced spore formation in sporophytes with complementary variant genotypes and inter- gametophytic crossing between the resulting un- reduced gametophytes. If the interspecific diploid sporophytic progeni- tors of allotetraploid P. rufa were heterozygous at the enzyme loci examined in this study, their un- reduced spores and derivative gametes would also be heterozygous (heterozygosity at enzyme loci is common in sporophytes of sexually reproducing diploid Pellaea species that have been examined N by Gastony & Gottlieb, 1985; Gastony, 1988). Fertilization involving such het- erozygous кын ч from different species would instantly yield an allotetraploid taxon with fixed intergenomic heterozygosity and segregating intra- genomic heterozygosity. This derivation of P. rufa is improbable, however, because it entails the si- multaneous occurrence of the same three rare events as discussed above, except that here the sporophytes producing unreduced spores would be- ong to different species. A simpler explanation for the origin of the seg- regating intragenomic heterozygosity in P. rufa is based on the commonly encountered method of allotetraploid formation in ferns, chromosome dou- bling of an allodiploid hybrid sporophyte. An al- lotetraploid that arises in this way, however, should show intragenomic homozygosity and only inter- genomic heterozygosity, as in the previously re- ported allotetraploid taxa discussed above. Only through gradual accumulation of mutations will intragenomic heterozygosity arise in such an al- lotetraploid taxon, and in this case different variant alleles might be expected to arise in different pop- ulations of the allotetraploid, unlike the situation in P. rufa where all populations examined express the same four alleles at Pgi-2. However, if tetra- ploid P. rufa had two or more independent origins from allodiploids with differing alleles at the various enzyme loci, as reported by Werth et al. (1985b) in Asplenium, random matings among the game- tophytic progenies of the eae th variant allo- tetraploids would generate the various banding pat- terns of Р. rufa sporophytes in Figures 6-8. Rieseberg & Doyle (1989) reviewed the idea that allotetraploids may have a fitness advantage over their diploid progenitors because of the in- creased biochemical diversity they achieve through fixed heterozygosity resulting from the addition of two differentiated genomes. They contrasted this with an autotetraploid’s ability to maintain as many as four alleles at a single locus and noted that resultant high levels of heterozygosity maintained by polysomic inheritance in autopolyploid popu- lations are of prime importan Pellaea rufa shows that allotetraploids also can maintain as many as four alleles per locus and can ce in their evolution. generate as many isozyme variants per enzyme locus per individual (ten bands) as can autotetra- ploids. Meiotic segregation does generate more ge- netic variation from autotetraploids with four alleles per locus than from equally heterozygous allote- traploids. Nevertheless, Р. rufa demonstrates that even disomic inheritance allows natural populations of allotetraploids to maintain higher levels of het- erozygosity than has previously been reported. LITERATURE CITED ANTHONY, М. С. A revision of the southern Af- rican species of С айий Swartz and Pellaea Link (Pteridaceae). Contr. Bolus Herb. 11: 1- 293. GAsTONY, G.J. 1 Electrophoretic sidence for the origin of fern species by unreduced spores. Amer. J. Bot. 73: 1563-1569. The Pellaea Faa E elec- trophoretic exento for the deriva e aga- mosporous taxa and a revised a je Fern ‚ (OF 4-67. . Darrow. 1983. t and с ytosolic i un qe e the homosporous fern Athyrium des ia a L. Amer. J. Bot. 70: 1409-1415. E E 1982. Evidence for genetic i О in a homosporous fern. Amer. J. Bot. -637. enetic variation in the homosporous fern Pe ae а andromedifolia. Amer. J. Bot. 72: 257-267. ——— D. WINDHAM. 1989. Species concepts in pteridophytes: the treatment and definition of aga- mosporous species. Amer. Fern ]. 79: 65-77. HaurLER, C. Н. 1985. Pteridophyte i aa а ology: the oe approach, Proc. Roy. S Edinburgh 86B: 1 SE . BRITTON & S. J. M. D. ROBINSON. 1985. Triploidy and its par saka significance in Cystopteris protrusa. Canad. J. Bot. 63: 1855-1863. Volume 77, Number 2 1990 Gastony 313 Electrophoretic Evidence for Allotetraploidy in Pellaea rufa JACKSON, R. C. & J. CASEY. 1982. Cytogenetic analyses of si ое шгек models and aa у triploids to octoploids. Amer. J. Bot. 69: 4 JACOBSEN, W. B. G. The Ferns pue Fern Allies of Southern -e Butterworths Publishers, Wo- burn, Massachusett Lovis, J. р. 1977. Evolutionary саре к processes in ferns. Pp. 229-415 in R. Р. Preston & Н. W. Woolhouse (editors) Advances in эы Ке- search, Volume 4. Academic Press, New York. Manton, I. 1950. Problems of Cytology and Evolution in the Pteridophyta. Cambridge Univ. Press, London. RIESEBERG, L. H. & М. Е. роті. 1989. Tetrasomic segregation in the naturally occurring autotetraploid Allium nevil (Alliaceae). Hereditas 111 sd 36. Roose, M. L. & L. D. GOTTLIEB. 1976. Genetic and biochemical consequences of n in Trago- pogon. a Ba 818-8 Roy, 5. K. € L N ew base number . 1986. Flora of Southern Africa. к of Agriculture and Water Supply, Preto SoLTIS, D. E. & L. H. ише. 1986. порочен іп Tolmeia menziesii (Saxifragaceae): genetic in- sights from enzyme electrophoresis. Amer. J. Bot. 73: 310-318. & P. S. SoLris. 1988. Electrophoretic evi- ence for tetrasomic d n in Tolmeia men- ziesii (Saxifragaceae). Heredity 60: 375-382. & 19 etr и. gag е їп енн женш (Saxifragaceae). J. Her 80: 123- С. Н. Hau IFLER, D. C. DARROW & С. J. GASTONY. 1983. Starch gel electrophoresis of ferns: a com- pilation of grinding buffers, gel and жас pes and staining schedules. Amer. Fern J. 7 ; STEBBINS, С. L. Types of al ik clas- sification and significance. Advances Genet. 1: 403- 429. Tryon, А. F. 1955. A new Pellaea pa ад Аїгїса. Ann. Missouri Bot. Gard. 42: 101- 1957. A revision of the m p Pellaea iati Pellaea. Ann. Missouri Bot. Gard. 44 3 1968. Comparisons of sexual and apogamous races in the fern genus Pellaea. Rhodora 70: 1-24. & D. M. BRITTON. 1958. Cytotaxonomic stud- ies on the fern genus Pellaea. Evolution 12: 137- 145 Tryon, R. M. & A. F. TrYon. 1982. Ferns and Allied Plants with Special Reference to Tropical America. Springer- are New Yor WERTH, C. 1989. The use of isozyme data for inferring ancestry of ЖҮ? pteridophytes. Bio- chem. Syst. Ecol. a 117 — ——, S. I. Guttman & BAUGH. 1985a. Electrophoretic dh of dio мА wr b the Appalachian Asplenium complex. Syst. Bot 84-192. , & 85b. Recurring origins of allopolyploid species in Von ha Science 228: 731-733. Worr, P. G., S. Sorris & D. E. Sorris. 1989. Tetrasomic inheritance and chromosomal pairing be- havior in the naturally occurring autotetraploid, Heu- chera grossulariifolia (Saxifragaceae). Genome 32: 655-659 BIOSYSTEMATIC ANALYSIS OF THE CYSTOPTERIS TENNESSEENSIS (DRYOPTERIDACEAE) COMPLEX! Christopher H. Haufler,? Michael D. Windham?? and Thomas A. Ranker* ABSTRACT The allotetraploid Cystopteris tennesseensis and its putative diploid progenitors, C. bulbifera and C. constitute the C. tennesseensis complex. Although previous studies provided evidence of morpholo nd chromosomal differ e шешн of this complex, puzzling morphological variability pre protrusa, ological, dl ecluded d oa isozymic, and gametophytic nod and supported past treatments of the complex as three separate i i ined, and meiotic alley of triploid hybrids . ennesseensis ig provided о evidence that the diploid genomes are nonhomologous. Because C. tennes- and contains isozymic profiles that are consistently аво of diploid patterns, readily when sympatric with the tetraploid. In part because of these characteristics, precise identification of species and hybrids in this complex is difficult and depends on evaluation of cryptic features. The cosmopolitan genus Cystopteris Bernh. has been called “perhaps the most formidable biosys- tematic problem in the ferns" (Lovis, 1977, p ). Although Cystopteris species are primarily north temperate and therefore readily accessible to pteridologists, complex patterns of morpholog- ical variation have thwarted satisfactory taxonomic treatments. The most recent monograph (Blasdell, 63) recognized ten species in two subgenera. Subgenus Acystopteris includes the Asian species C. japonica and C. tenuisecta, which are so dis- tinctive that some have placed them in a separate genus (Nakai, 1933; Pichi-Sermolli, 1977). Blas- dell divided the remaining species (all in subg. Cys- topteris) into two sections based on the pattern of vein termination in the leaves. Species having veins directed into teeth were placed in sect. Cystopteris, while those with veins directed into sinuses were assigned to sect. Emarginatae. Blasdell e tered variability for this vein termination character in some specimens and suggested that such anom- alies could result from introgression. Lovis (1977), on the other hand, considered it more likely that venation features are not as stable in some species as in others and should not be used to define sec- ncoun- tions. Our observations suggested that Lovis was cor- rect and, "ther than organizing the North Amer- ican y based sections, we I hien ahoan to көп кшй оп шїег- active species complexes involving allopolyploid species and their putative diploid progenitors. One ' CHH is profoundly indebted to Alice and Rolla Tryon for PEE the opportunity through a Gray Herbarium Postdoctoral Fellows Tic ES - E 2 e £e іа) e ship to delve deeply into the world of ferns David Barrington for drawing Figure 1, and the officers of the following herbaria for loans of specimens: COLO, DAO, n ILLS, IND, KANU, KE, KY, MICH, MIL, MO, NY, PAG, OAC, OS, SIU, TENN, UARK, US, VPI, VT, and Ге are especially thankful to Ralph Brooks and Ron McGregor of KANU for housing the many hundreds of ets Living унш were generously supplied by D. M. Britton, R. E. Brooks, S. C. Churchill, J. J. Doyle, Sac e . Moran, T. Reeves, D. E. Soltis eg aut of жез U niversity of Kan з Present Address: Utah Museum of Natural ale, and W. H. ай ence, Kansas 66045, U.S.A. ] ae. University of Utah, Salt Lake City, Utah 84112, U.S.A. S.A. t University of Hawaii, Hawaiian Evolutionary Biology Program, Honolulu, Hawaii 96822, U. ANN. Missouni Вот. GARD. 77: 314-329. 1990. Volume 77, Number 2 1990 Haufler et al. 315 Analysis of the Cystopteris tennesseensis Compl mplex such complex centers on С. tennesseensis, a species first recognized by Shaver (1950) who described it as a hybrid between C. bulbifera and C. protrusa. In 1954 Shaver developed an extended discussion of the morphology and ecology of the complex which still stands as an excellent summary of the natural history of this difficult group. From the moment it was named, there was dis- agreement concerning the proper status and dis- position of С. tennesseensis. McGregor (1950), in collaboration with C. A. Weatherby, reduced this species to a variety of C. fragilis. In addition, citing specimens that did not conform to the pro- tologue of C. fragilis var. tennesseensis and that seemed to be close to Weatherby’s С. fragilis f. simulans, McGregor recognized C. fragilis var. simulans. This approach seemed justified because some specimens tend to bridge the morphological gap between these two varieties as well as between C. fragilis var. fragilis and C. fragilis var. ten- nesseensis. However, Blasdell (1963) revealed C. tennesseensis as a tetraploid with n = 84 chro- mosomes. His analysis showed that hybridization between С. protrusa and С. bulbifera was the most reasonable explanation for the origin of this tet- raploid. Therefore, Blasdell resurrected С. tennes- seensis as a species, including the former С. fra- gilis var. simulans The subtlety of abaa features separat- ing the species as well as the continued discovery of intermediates has perpetuated the systematic confusion of the C. tennesseensis complex. Even though the accumulated data demonstrate conclu- sively that C. tennesseensis is isolated from its congeners, this tetraploid has not been widely ac- cepted as a distinct species. As recently as 1982, Moran (1982, p. 94) found in a study of Cystop- teris specimens from Illinois that “all 90 herbarium specimens of C. tennesseensis . . . were originally misidentified. n this paper, we report a series of analyses designed to investigate the patterns and the pro- cesses behind the systematic confusion in C. ten- nesseensis. We describe studies of biogeography, morphology, chromosomes, and isozymes that clar- ify the origin and current status of C. tennesseensis and its progenitor diploids. MATERIALS AND METHODS Observations of morphological variability, geo- graphic distribution, and habitat diversity were ob- tained from surveys of herbarium specimens. If specific locality data were supplied, collection sites were plotted by county and used to obtain distri- E l. Mean spore sizes and standard deviations calculated by measuring the long axis of 25 spores. Spec- imens whose ploidy level has been verified by meiotic chromosome squashes are indicated by asterisks. States listed in parentheses are those from which the specimens were collected. C. bulbifera (Illinois) 37.48 + 1.949 C. bulbifera* (Oklahoma) 35.21 + 3.408 C. bulbifera* (Arizona) 36.27 + 2.560 C. bulbifera* (Indiana) 35.96 + 2.325 C. bulbifera* (Ohio) 39.81 + 2.944 C. bulbifera (Ohio) 35.71 + 3.038 C. bulbifera (Kentucky) 34.39 + 1.764 C. bulbifera (Indiana) 36.78 + 2.336 mean = 36.45 + 1.650 C. protrusa (Illinois) 33.65 + 1.788 C. protrusa* (Kansas) 35.22 + 4.714 C. protrusa (Missouri) 33.33 + 2.421 C. protrusa* (Kansas) 30.77 + 1.868 C. protrusa* (Kansas) 30.98 + 2.260 C. protrusa* (Michigan 33.21 + 2.332 C. protrusa* (North Carolina) 34.22 + 3.042 mean = 33.05 + 1.634 C. tennesseensis* (Kansas) 40.02 + 3.250 C. tennesseensis* (Missouri) 41.40 + 1.907 С. tennesseensis* (Illinois) 41.29 + 2.416 C. tennesseensis* (Arkansas) 40.65 + 2.084 mean = 40.84 + 0.639 butional information. When the gross morphology of specimens was not sufficient to assign them readi- ly to species, spores were removed and mounted in Permount on glass slides. The longest diameter of the monolete scores (Table 1) and evidence of abortion (e.g., shrunken or malformed spores) was noted. Spore slides were placed in en- velopes and attached to herbarium sheets. To supplement herbarium collections and supply living material for chromosomal, isozymic, and common garden morphological comparisons, spec- imens were collected over much of the range of the genus in North America and were donated as noted in footnote 1. Living collections were main- tained in the University of Kansas greenhouses. Collection sites for materials used in this study are listed in 1 ] 1 1 e 1 Ecologically i maucea obscure genetically based species distinctions. Therefore, we conducted discriminant analyses of morphological variation based on plants cultivated in the greenhouse. All individual plants were iden- tified and assigned to a group on the basis of chro- mosome number and isozymic composition. Fea- 316 Annals of the Missouri Botanical Garden ABLE 2. Locality data for collections providing material for electrophoretic and cytogenetic analyses. Asterisks indicate localities used for population genetic analyses (see Table 4). Chromosome mber Locality Collector verified? Cystopteris bulbifera Arizona: Coconino Co., Rio de Fla Windham 194 Yes, 421 Arizona: Coconino Co., Oak Creek Canyon Windham 314 Yes, 4211 Arizona: Coconino Co., Lower West Fork W. H. Wagner 82113 Yes, 42II Indiana: Fountain Co., Portland Arch* C. H. Haufler & R. C. Moran s.n. No Indiana: bi Co., Clifty Falls C. H. Haufler & R. C. Mor No Indiana: Monroe Co., Cedar Bluffs* C. H. Haufler & R. C. ia s.n. No Eu. Powell Co., Natural Bridge* C. Н. Haufler & R. C. Moran s Мо Ohio: Adams Co., near Stout* C. H. Haufler & R. C. Moran s.n. Yes, 4211 Oklahoma: Ottawa Co., Dripping Springs C. H. Haufler & C. K. Teale s.n. Yes, 4211 Wisconsin: Door Co., Peninsula State Park W. C. Taylor s.n. Yes, 4211 C. protrusa Iowa: Fremont Co., SW of Sidney R. E. Brooks 14929 Yes, 42II Illinois: Cook Co., McGinness Slough R. C. Moran s.n. No Conservation Area Illinois: Union Co., Shawnee ТЕВЕ Forest C. H. Haufler s.n. Yes, 4211 Indiana: Monroe Co., Cedar C. H. Haufler & R. C. Moran s.n. No Indiana: Monroe Co. Cascades Park* C. H. Haufler s.n. No Indiana: Monroe Co., Farr Road* C. H. Haufler & R. C. Moran s.n. No Indiana: Perry Co., N of Tell City* C. H. Haufler s.n. No Kansas: Douglas Co., Breidenthal Woods* Windham 579 Yes, 4211 Kansas: Chautauqua Co., W of Elgin R. E. Brooks 16162 Yes, 42II Kansas: Miami Co., S of Homewood C. H. Haufler s.n. Yes, 4211 Michigan: Wachienaw Co., Homer Woods C. H. Haufler & W. H. Wagner s.n. Yes, 4211 Missouri: Boone Co., SW of Columbia R. E. Brooks 15298 s, 421] Missouri: Cooper Co., Arrow Rock К. E. Brooks & C. Н. Haufler 15343 Yes, 4211 Missouri: Franklin Co., Meramec State Park К. E. Brooks & C. Н. Haufler 15341 Yes, 4211 Missouri: St. Louis Co., near Allenton* R. E. Brooks & C. H. Haufler 15332 Yes, 4211 North Carolina: Swain Co., Nantahala Gorge W. H. Wagner s.n. 421 С. tennesseensis Illinois: Jackson Co., Fountain Bluff C. H. Haufler & К. C. Moran s.n. Yes, 841 Kansas: Doniphan Co., N of Wathena R. E. Brooks 14898 Yes, 841 Missouri: Franklin Co., E of Sullivan R. E. Brooks & C. H. Haufler 15334 Yes, 8411 Missouri: LaClede Co., Big Niangua River В. C. Phillips s.n. 8411 Nebraska: Richardson Co., NE of Shubert R. E. Brooks 14925 Yes, 8411 Oklahoma: Ottawa Co., Dripping Springs C. H. Haufler & С. K. Teale s.n. Yes, 841I С. tennesseensis х С. bulbifera Missouri: Boone Co., SW of Columbia Nebraska: Richardson Co., МЕ of Shubert С. tennesseensis X С. protrusa Kansas: Chautauqua Co., W of Elgin Missouri: St. Louis Co., near Allenton К. E. Brooks & С. Н. Haufler 15302 R. E. Brooks 14925a R. E. Brooks 1616 К. E. Brooks & С. P Haufler 15331 Yes, 421 + 42] Yes, 4211 + 421 Yes, 42П + 421 Yes, 4211 + 421 tures surveyed are listed in Table 3 and depicted — species-specific leaf attributes or combinations in Figure 1. Because the species are often difficult thereof. Involved in this analysis were 9 plants of to discern based solely on leaf features (often the С. bulbifera, 21 of C. protrusa, and 8 of C. ten- only part preserved on herbarium specimens), we — nesseensis. Because our first analysis could not first used only quantitative leaf features to seek completely discriminate the species (see Results), Volume 77, Number 2 1990 Haufler et al. 317 Analysis of the Cystopteris tennesseensis Complex a second analysis was conducted that included qual- itative characters (Table 3) as well as mean spore sizes per plant (Table 1). This second analysis in- cluded 8 plants of С. bulbifera, 14 of С. protrusa, and 4 of С. tennesseensis. The data from both sets were analyzed with the BMDP7M stepwise dis- criminant analysis program (BMDP Statistical Soft- ware, 1981, UC Press). For each step in the anal- yses we allowed an experiment-wise error rate of = 0.05 and judged the significance of each approximate F-ratio as the criterion for entry of a variable into the discriminant models. The critical values of F were estimated using Bonferroni's in- equality (Ranker & Schnabel, 1986 Unlike angiosperms, the gametophyte genera- tion of fern species is independent of the sporo- phyte. Whereas the sporophyte is important in po long-term survival of individuals, the ophyte carries out sexual reproduction and haa porcum of populations. Thus, in de- veloping a complete picture of the biology of fern species as well as characterizing evolutionary ten- dencies, analyses of gametophyte reproductive bi- ology are necessary. Їп a previous study Haufler & Ranker (1985) determined which Cystopteris species produced and/or responded to antheridi- ogen. In the present study 100 gametophytes of each species were reared individually, employing the culture conditions described in Haufler & Ranker (1985), and assayed for genetic load fol- lowing the procedures outlined by Lloyd (1974). Analyses of chromosomal behavior at meiosis were especially important in characterizing sus- pected hybrid individuals. The procedures followed in obtaining and photographically documenting stages in sporogenesis were described in Haufler et al. (1985). Voucher specimens for each chromo- some count will be deposited at KANU. Electrophoretic analyses of isozyme variability were performed as described by Haufler (1985). Leaf samples were ground in the phosphate grind- ing buffer of Soltis et al. (1983) using a mortar and pestle or by placing them in spot plate wells and using a round-bottomed centrifuge tube as a pestle. Each of the 22 enzymes listed in Soltis et al. (1983) was surveyed and was subjected to a variety of gel and electrode buffer conditions (large- ly those of Soltis et al., 1983). Enzymes that pro- vided consistent, interpretable results were hexo- kinase (HK), isocitrate dehydrogenase (IDH), leucine amino peptidase (LAP), the more cathodal (presumably cytosolic—see Weeden, 1983) bands of phosphoglucoisomerase (PGI-2), two sets of bands (presumably representing cytosolic and chloroplas- tic enzymes—see Weeden, 1983) for phosphoglu- TABLE 3. analysis. Asterisks indicate those illustrated in Figure 1. Plus marks indicate features that were statistically sig- nificant in discriminating species. List of features included in morphometric 1—5. FEATURES OF ENTIRE LEAF Length of petiole (cm) . Length of blade (cm . Angle of departure of median pin . Length of rachis between lowest pica of first pinna pair and lowest pinna of second pinna pair (mm *5. Length of rachis between lowest pinna of second pinna pair and lowest pinna of third pinna pair хк ox + wn re (mm) 6-17. FEATURES OF LONGEST PINNA 6. Number of pinna pairs from base (first one = number 1 *7. Length of ТА stalk (petiolule) (mm) *8. Length of pin a (mm) *9. Width of us (mm) +*10. Number of segments along acroscopic edge of pinna (including pinnatifid ones) *11. Length of first acroscopic nnule (mm) "12. тана of major sinuses E distal edge of first acroscopic pinnule *13. Length of stalk of basiscopic pinnule (mm) *14. анна н а rachis between lowest two pinnule pu *15. Angle formed by apex of pinna +*16. An gle formed by basiscopic pinnule axis with pin- na ax *]7. Angle “of тока = of basiscopic pinnule base with pinnule 18-20. QUALITATIVE FEATURES 18. Presence or absence of bulblets . Presence or absence of glandular trichomes 20. Color of petiole . Spore size (see Table 1) comutase (PGM-1, the more anodal set, and PGM- 2, the more cathodal set), shikimate dehydrogenase (SkDH), and two sets of bands (presumably rep- resenting cytosolic and chloroplastic enzymes—see Weeden, 1983) for triosephosphate isomerase (TPI- 1, the more anodal set, and , the more cath- odal set). HK, PGI-2, LAP, TPI-1, and TPI-2 showed the best resolution on either system 6 of Soltis et al. (1983) or the modified system 8 dis- cussed in Haufler (1985). IDH, PGM-1, PGM-2, and SkDH were resolved best on the modified sys- tem 11 of Soltis et al. (1983) discussed in Haufler (1985). Identification of bands shared among pop- ulations and species was accomplished through co- electrophoresis of the samples on the same gel. Individual gametophytic progeny were used to assess the genetics of sporophytic banding patterns. We followed the electrophoretic procedures de- Annals of the Missouri Botanical Garden У 2:94 с | т чү. “ЖУТТУ г Volume 77, Number 2 1990 Haufler et al. 319 Analysis of the Cystopteris tennesseensis Complex scribed by Gastony & Gottlieb (1982) as modified by Haufler & Soltis (1984). For the enzymes sur- veyed, gametophytes express the same isozymes as do sporophytes. Because gametophytes are mul- ticellular, haploid individuals derived from single meiotic products (spores), gametophytic progeny arrays from single sporophytes can be used to determine directly the genetic constitution of com- plex sporophytic banding patterns. In this way, segregational analyses of putatively heterozygous banding patterns can be performed without pur- suing more time-consuming crossing programs. After allelic determinations were made for each of the nine putative loci, allozymic data from a set of representative populations (Table 4) were ana- lyzed. The proportion of polymorphic loci (P) and the mean number of alleles per locus (A) were calculated from these populational data. A statis- tical program (LYNSPROG) written by Marilyn Loveless, The College of Wooster, Wooster, Ohio, was used to calculate levels of heterozygosity, the fixation index (F), and Nei’s coefficients of genetic identity and distance. Because C. protrusa tends to be clonal, separate calculations were made for populations consisting of (1) each leaf sampled (ra- mets) and (2) only the number of different geno- types (genets). The first should overestimate the number of individual organisms while the second should underestimate it. RESULTS SPORES Blasdell (1963) and Moran (1982) noted the value of spore features for identifying hybrids and in determining the ploidy of specimens. Despite Lovis’s (1977) caveat that it may not be appro- priate in Cystopteris to infer ploidy levels from spore measurements, we were able to verify in- dependently through analyses of meiotic chromo- some behavior that spore size did correlate with several critical genetic features of species (Table 1). We agree with Lovis that spore size comparisons alone do not substitute for direct chromosome anal- yses, but once a strong correlation between spore features and genetic condition is established, spores can be extremely valuable in surveying specimens < 0.002). Although C. bulbifera spores were larg- FIGURES 2-4. Geographic distributions of members of the Cystopteris tennesseensis complex based on data 15 wn were rrt — nesseens sa. as rcles on Figures 2 and 4 i" о ‘oft div bid "on hybrids. er than those of C. protrusa, spores of tetraploid C. tennesseensis showed the further size increase commonly associated with higher ploidy. MORPHOLOGICAL ANALYSES OF SPOROPHYTES The taxa in the Cystopteris tennesseensis com- plex are among the most distinctive in the genus. There are more unique features that characterize URE l. Leaf features included in morphometric analysis. Numbers correspond to descriptions of features in 0.5 cm Table 3. Scale bars: for whole leaf = 2.0 сш; for pinna base = 320 Annals of the Missouri Botanical Garden each species than there are in any other group of Cystopteris species. Cystopteris protrusa, for ex- ample, has long internodes and a peculiar, pro- truding rhizome apex whose growing point extends past the current season's leaves. Given the evi- dently rapid growth of the rhizome, we were not surprised to find extensive clones of this species in mature woodlands. Cystopteris bulbifera, com- monly found on moist cliffs, has short internodes and bears asexual tive bulblets on its leaves, and most individuals have tack-shaped, glandular trichomes that are particularly prominent on the indusia and along the rachis between pinnae. The allotetraploid C. tennesseensis combines the ge- nomes of its progenitor diploids and, perhaps as a result of this genetic amalgamation, is morpholog- ically variable. In many features, the tetraploid is intermediate between the diploids, e.g., it has poor- ly formed bulblets and a reduced frequency of glandular trichomes. Its rhizome features, however, are not intermediate. Perhaps because it inhabits somewhat disturbed, often dry cliffs rather than forest floors, it has the short internodes typical of C. bulbifera. Even though past workers (Shaver, 1954; Blas- dell, 1963) provided well constructed species de- scriptions and indicated the distinctive features of each species, taxonomists not familiar with the genus continue to encounter problems when iden- tifying members of the C. tennesseensis complex. To generate additional discriminating features and in trying to separate environmentally induced vari- ability from that based on genetic differences, we performed a discriminant analysis of leaf mor- phology on plants grown in a common garden. In the first analysis, employing only quantitative leaf characters, only two features were entered into the model as being statistically significant, the number of acroscopic segments on the longest pinna (#10, Table 3, Fig. 1) and the angle formed by the basiscopic pinnule axis with the pinna axis (#16). The species differed significantly from one another in the space defined by the two canonical axes (P < 0.01). The first axis accounted for 85% of the total variance and separated C. bulbifera from the other two species. The two variables in the model exhibited similar character loadings (standardized coefficients) on the first axis, 0.608 for #10 and 0.720 for #16, indicating a nearly equal contri- bution to the canonical variable. Inspection of the original data revealed that C. bulbifera is distin- guished from the other species by a larger number of acroscopic segments on the longest pinna and a larger angle formed by the basiscopic pinnule. The second axis accounted for the remaining 15% of the variance. The jackknifed classification ma- trix correctly identifed 100% of the C. bulbifera individuals, 71.4% of C. protrusa, and 87.5% of C. tennesseensis. Six individuals of C. protrusa were incorrectly identified as C. tennesseensis, and one individual of C. tennesseensis was incorrectly identified as C. protrusa. A second discriminant analysis was conducted in which qualitative leaf features (listed in Table 3) and spore size were added to the data set in an attempt to identify combinations of characters that would allow a better discrimination of species than was obtained in the first analysis. In this second analysis, three characters were entered into the model as statistically significant: angle of the basi- scopic pinnule (#16), presence or absence of glan- dular trichomes, and mean spore size (Table 1). The first canonical axis accounted for 82% of the total variance, and the character loadings were 0.809 (#16), 0.875 (glandular trichomes), and 0.485 (spore size). The first axis primarily sepa- rated the two diploids from one another. As indi- cated by the relative value of the loadings, the diploids were primarily distinguished by the angle of the basiscopic pinnule and the presence or ab- sence of glandular trichomes. On the second axis, the character loadings were — 0.081 (#16), 0.429 (glandular trichomes), and —0.887 (spore size). The large spore size of the tetraploid, therefore, was the most important factor in discriminating it from the diploids. The jackknifed classification ma- trix produced 100% correct classifications for all individuals in the analysis. BIOGEOGRAPHY The distribution of species is shown in Figures 2-4. Cystopteris bulbifera is the most northern species, extending well into Canada. Cystopteris protrusa is most common in the east-central United States and is rare in southern Canada (Britton et al., 1984; Haufler et al., 1985). The allotetraploid derivative C. tennesseensis is found primarily in the region of overlap of the two diploids. Using spore abortion as an indicator of sterility, we iden- tified interspecific hybrids between the tetraploid and its diploid parents. The locations of these hy- brids are indicated as open circles on Figures 2 nd 4. Because the morphological features are quite plastic, it is not always possible to determine which of the diploids hybridized with C. tennes- seensis in forming the sterile plants. Using chro- mosomal and isozymic data, we were able to doc- ument that backcrosses to both parents do occur in nature (see below). feb) Volume 77, Number 2 1990 Haufler et al. 321 Analysis of the Cystopteris tennesseensis Complex FIGURES 5-9. n = 8411.—7. C. protrusa, п = 4 Identification of hybrids was based on morphological and isozymic data. —8. Hybrid C. bulbifera, п = 4211 + 421.—9. Hybrid between C. tennesseensis and C. protrusa, п = 4211 + 421. All x 1,000. GAMETOPHYTES Summarizing previous studies of Cystopteris ga- metophytes (Blasdell, 1963; Profumo, 1969; Au- quiere & Moens, 1972), Blasdell (1963, р. 6) stated, ““All of the taxa examined are similar in the bulk of their gametophytic features.” Our study of gametophyte ontogeny confirmed Blasdell's as- sessment of the level of morphological variability. However, Haufler & Ranker (1985) demonstrated significant interspecific variation in response to the pheromone antheridiogen. Genetic load studies have been used to look into aspects of reproductive biology (Lloyd, 1974). Be- cause inbreeding brings about the expression of recessive alleles, species with low genetic load should be more inbred, whereas those having high genetic Representative meiotic chromosome squashes.—5. C. bulbifera, п = 4211.—6. C. tennesseensis, 211. 8, 9. Backcross hybrids between C. tennesseensis and its diploid progenitors. between C. tennesseensis and load are presumed to be outcrossers. In homo- sporous ferns, levels of genetic load can be deter- mined by assessing the ability of isolated gameto- phytes to produce sporophytes. In the present study, none of the 100 isolated gametophytes of C. bul- bifera and C. protrusa produced sporophytes, even after repeated waterings over a four-month period. However, 25 of the 100 isolated gametophytes of С. tennesseensis did form sporophytes. Although we did not examine each isolated gametophyte, we have verified that gametophytes of the diploid species do become hermaphroditic in culture. Thus, genetic load in the tetraploid is significantly lower than that in the diploids (3 x 2 contingency table of species vs. sporophyte production: G = 59.63, P « 0.001). 322 Annals of the Missouri Botanical Garden Ficures 10-13. Anode = top of photographs, ca = bottom of pho- tographs. 10-12. Interspecific comparisons of banding patterns. Lanes a-d = otrusa; е-) = . pro . tennes = С. bulbife ra. Lanes f and h (starred) are үс isozyme profiles. seensis; k-m from triploid backcross hybrids between С. tennesseensis and C. protrusa. — H (monomeric, single com e C. bulbifer ud mobilities, i = : m pattern cC ennesseensis appea broad fuz scien fo aia of the parental bands n an een he terodimeric band. — 12. TPI dimeric; in both diploids, anodal, multiple-banded жей proba x rep- resents a post-translational ү of the yme ~ 2 "ауа parents. ane 1 m к lane j received the more cathodal band. — 13. TPI e MEIOTIC CHROMOSOME BEHAVIOR Analyses of meiosis from plants collected across the species ranges generated consistent counts of n = 42 in C. bulbifera (Fig. 5) and C. protrusa (Fig. 7), and л = 84 in С. tennesseensis (Fig. б). Table 2 lists the collection localities of the plants yielding new chromosome counts. In addition to analysis of the sexual species, plants verified via isozyme analysis to be backcross hybrids were also studied cytogenetically. As shown in Figures 8 and 9, all hybrids showed a complement of 42 bivalents plus 42 expected from backcross hybrids of an allotetraploid to its diploid progenitors. These results indicate little or no chromosomal homology between the diploid species C. protrusa and C. bulbifera and suggest that C. tennesseensis con- tains one genome of each. ISOZYMES Isozyme patterns of the tetraploid C. tennes- seensis are perfectly additive of those from the two diploid species. Gametophytic progeny arrays dem- onstrated that the complex banding patterns did not segregate during meiosis (Fig. 13), suggesting that the isozymes comprising these patterns were coded by genes situated on nonhomologous chro- mosomes contributed by the two diploid parents. Given that all tetraploids showed fixed heterozy- gosity that always combined patterns from each diploid progenitor, no evidence of “orphan alleles” or gene silencing could be detected. All isozymic variability observed in the allotetraploid apparently шой irom the ш орао. of various diploid derivative via recurring lalala тни (Figs. 10-12, and see Hauf- ler & Soltis, 1986). O chromosomal and isozymic data made it possible to identify hybrids and determine their parentage. Sporogenesis was studied in plants hav- ing aborted spores. Such plants were shown to be triploids having 42 pairs and 42 univalents during meiosis and thus were determined as backcrosses etween С. tennesseensis and either С. protrusa or С. bulbifera. Even though the variability of morphological features could preclude precise iden- tification of parents, the triploids always contained unequal enzyme contributions from the two par- nts. Thus, dosage effects could be used to show 2 sion in individual gametophytic progeny of a ate i having the banding pattern seen in lanes e and j of Figu 12. "There is no segregation for the ES m indicating that the sporophyte was a fixed heterozygote. Volume 77, Number 2 Haufler et al. 323 1990 Analysis of the Cystopteris tennesseensis Complex TABLE 4. Locations and sizes (л) of populations of diploid Cystopteris species surveyed in generating genetics statistics (see Tables 5-7). Cystopteris bulbifera Population #1—Fountain Co., Indiana, Portland Arch, N of Covington—n = 18 Population #2 — Adams Co., Ohio, limestone cliffs near Stout—n = 41 Population #3— Monroe Co., Indiana, Cedar Bluffs, S of Bloomington—n = 11 Population #4— Powell Co., Kentucky, along trail in Natural Bridge State Park—n = 12 C. protrusa Population #1—Douglas Co., Kansas, Breidenthal Woods, near Lawrence Population #2—Monroe Co., Indiana, Cedar Bluffs, S of Bloomington Population #3— Monroe Co., Indiana, Cascades Park, NNW of Bloomington Population #4— Monroe Co., Indiana, Farr Rd., N of Bloomington Population #5 —St. Louis Co., Missouri, near Allenton Population #6—Perry Co., Indiana, N of Tell City Population sizes for C. protrusa calculations First set of calculations Population #3—n = 18 Population #4—n = 12 Population #5—n = 10 Population #6—n = 24 Second set of calculations based on genets Population #1—n = 15 Population #2—n = 3 Population #3—n = 2 Population #4—n = 6 Population #5—n = 7 Population #6—n = 10 which backcross had formed the hybrids. Hybrids formed between C. tennesseensis and C. protrusa had the C. protrusa bands stained more intensely than the C. bulbifera contribution (Figs. 10-12, lanes f and h). Conversely, C. tennesseensis х C. bulbifera hybrids contained a double dose of С. bulbifera. Although all nine enzymes helped corroborate that C. tennesseensis was an amalgam of its pro- genitor diploids, the most complex and informative enzyme was TPI (Fig. 12). For both diploids, band- ing patterns indicative of two genetic loci were resolved. The cathodal bands showed banding pat- terns typical of dimeric enzymes (single-banded in homozygotes and three-banded in heterozygotes). The anodal zone of bands were expressed as fixed, multiple-banded patterns which probably represent post-translational modifications (Gastony, 1988; Hickey et al., 1989). In C. protrusa (lanes a-d), the strongly staining, most cathodal band was an invariant marker for the species. The more weakly staining, multiple-band pattern was variable. As shown in Haufler & Soltis (1986), three patterns were detected among the sporophytes surveyed: a slow-migrating triplet (as in lanes a and b of the present paper), a fast-migrating triplet (not illus- trated here, but see Haufler & Soltis (1986, fig. la)), and the complex, five-banded heterozygote between them (as in lanes с and д). In С. bulbifera (lanes К-т), the rapidly migrating triplet pattern was invariant while there were two alleles expressed for the cathodal locus. Lane | is the heterozygote formed from outcrossing between homozygotes such as k and m. When these diploid variants are com- bined in C. tennesseensis (lanes е—)), some re- markably complex patterns result. Lanes е and j appear to be relatively simple additive patterns between plants such as those in lanes a (C. pro- trusa) and k (C. bulbifera). Lanes g and i combine the bands found in lanes a and m. Lane f (identified via morphology and meiotic chromosomal behavior as a triploid backcross between C. protrusa and С. tennesseensis) combines one dose of a profile like lane m with two doses of lane b. Lane h (also a С. protrusa х С. tennesseensis triploid) appears to combine one dose of a plant like lane k with two doses of one like lane b or c. Note also the banding patterns for the triploids in SkDH (Fig. 10). For lanes f and h, the two cathodal bands are shared with C. protrusa while only the most anodal band is found in C. bulbifera. These SkDH patterns help confirm that both triploids contain two genomes from С. protrusa and only one from С. bulbifera. The slowest migrating SkDH band in С. tennes- seensis (lanes f, h, and i) is from C. protrusa even though it is not seen in the representative C. pro- trusa plants on this gel. The rapidly migrating SkDH band in lane с is a rare С. protrusa variant 324 Annals of the Missouri Botanical Garden TABLE 5. polymorphic loci, А = average number of alleles per locus. Population genetics statistics. Р = proportion A were identical for the ramet and genet samples of C. protrusa. Cystopteris bulbifera Mean i heterozygosity num - he be Р А served pected 1 0.6% 1.89 0.435 0.470 2 0.56 1.78 0.138 0.219 3 0.56 1.7 0.394 0.334 4 0.67 1.78 0.347 1273 Mean 0.62 1.81 0.329 0.324 Cystopteris protrusa Mean Mea Popu heterozygosity heterozy ae pal r ramets lor genets num- - x- » x- ber E A served pected served pected 1 0.56 1.67 0.128 0.153 0.152 0.223 2 0.22 1.22 0.042 0.104 0.048 0.134 3 0.56 1.56 0.524 0.302 0.429 0.286 4 0.78 1.88 0.357 0.342 0.286 0.345 5 0.56 1.67 0.157 0.235 0.204 0.273 6 0.44 1.56 0.185 0.176 0.200 0.198 Mean 0.52 1.59 0.232 0.219 0.220 0.243 not incorporated in any of the C. tennesseensis plants shown here A series of беш па was surveyed (Table 4) and used to generate calculations of Р, А, an levels of heterozygosity (Table 5), Nei’s coefficients of genetic identity and distance (Table 6), and the fixation index (+) (Table 7). Results derived from ramet and genet views of individuality in popula- tions of C. protrusa did not differ markedly. Ob- served heterozygosity differed by 0.012 and mean genetic identities were within 0.013 of each other. Although the calculations of F derived from the ramet samples appear to have a large number of significant departures from random mating, these values are both positive and negative. A mean S, — 0.0687 for these significant values suggests ilio the ramet figures may be influenced by oversam- pling of certain clonal genotypes. Cystopteris bul- bifera and C. protrusa shared no alleles at the loci examined; thus, based on nine putative loci (most of which code for "conservative" enzymes (Gilles- pie & Kojima, 1968)), the genetic identity between these congeneric species was zero. DISCUSSION BIOGEOGRAPHY AND ECOLOGY Fieldwork and mapping of species distributions provided new perspectives on the biology and evo- lution of the Cystopteris tennesseensis species complex. Often two and occasionally all three species were sympatric. In such situations, C. protrusa was confined to the forest floor, C. bulbifera was on moist cliffs and among the talus at the base of cliffs, and C. tennesseensis could be found on drier cliffs and/or disturbed sites such as old rock walls. When plants of C. protrusa and C. bulbifera were intermingled, a special effort was made to locate the primary diploid hybrid between them. However, primary hybrids could not be found. Others have noted that primary hybrids of well established al- lopolyploids can rarely be discovered (e.g., the Ap- palachian Asplenium complex (Wagner, 1954; Werth et al., 1985a) and the Dryopteris carthu- siana complex (Wagner, 1971)). Perhaps in these situations, the intermediate niche necessary for establishment of a diploid hybrid is occupied by the allopolyploid species. In contrast to the primary diploid hybrid situ- ation, when C. tennesseensis was sympatric with either parent, triploid backcross hybrids could be identified. Although an equivalent number of sta- tions having putative hybrids has been identified (Figs. 2, 4), fieldwork demonstrated that at any one locality, hybrids with С. protrusa were more frequent than those involving С. bulbifera. This result may reflect the greater specificity of ga- metophytic safe sites for the latter species. Alter- natively, differences in hybrid frequencies could be related to variation in response to antheridiogen. Haufler & Ranker (1985) showed that C. protrusa responded to antheridiogen while C. bulbifera did not, and that C. sitivity to this antheridia-inducing pheromone. Un- tennesseensis had a reduced sen- der these circumstances, it is likely that crosses involving С. protrusa would be more frequent. In a given safe site containing spores of C. protrusa and C. tennesseensis, it is likely that C. tennes- seensis gametophytes would be mostly female while neighboring C. protrusa gametophytes would be primarily male. If this scenario is accurate, о would predict that C. protrusa should be the ids parent in most crosses. Extending this argument to the origin of € tennesseensis, it may be predicted that C. bulbifera would be the egg parent in most cases. This hy- pothesis could be tested using information from chloroplast DNA (cpDNA) sequence variability. Volume 77, Number 2 1990 Haufler et al. 325 Analysis Я the Cystopteris tennesseensis Complex TABLE 6. Population genetics statistics. Calculation of Nei's genetic identity and distance between infraspecific populations. Identity measures above the diagonal, distance measures below the diagonal. Cystopteris bulbifera Population number 1 P^ 3 4 1 == 0.7356 0.7716 0.8103 2 0.3071 == 0.9203 0.8734 3 0.2592 0.0830 — 0.8474 4 0.2104 0.1353 0.1656 Mean genetic identity — 0.8264 Cystopteris protrusa—RAMET POPULATIONS Population number 1 2 3 4 5 6 1 — 0.9857 0.9161 0.8518 0.8940 0.7917 2 0.0144 == 0.9219 0.8313 0.9191 0.7937 З 0.0876 0.0813 ак 0.8578 0.847 0.7599 4 0.1604 0.1848 0.1534 ns 0.7791 0.8271 5 0.1120 0.0844 0.1655 0.2496 — 0.7280 6 0.2336 0.2310 0.2745 0.1898 0.3174 -— Mean genetic identity — 0.8470 Cystopteris protrusa — GENET POPULATIONS Population number 1 2 3 + 5 6 1 — 9280 0.9449 0.8253 0.8769 0.7670 2 0.0747 == 0.9275 0.7534 0.9364 0.7708 3 0.0567 0.0753 == 0.8840 0.8674 0.7765 4 0.1920 0.2832 0.1233 ses 0.7285 0.7886 5 0.1313 0.0651 0.1423 0.3167 == 0.7371 © 0.2652 0.2603 0.2530 0.2375 0.3050 — Mean genetic identity = 0.8342 Recent analyses (Stein & Barrington, 1990) in- dicated that cpDNA is inherited uniparentally in ferns (but see Andersson-Kotto, 1930). If cpDNA is carried in the egg, most C. tennesseensis plants should contain the chloroplast genome of C. bul- The available data indicate that C. tennesseensis is a relatively young species. The range of the derived allotetraploid C. tennesseensis extends only slightly beyond that of its diploid progenitors (Figs. 2-4). Further, C. tennesseensis shows no evidence of gene silencing. For the enzymes surveyed, all plants displayed fixed heterozygotic banding pat- terns that were additive for bands found in the diploid progenitors. All variability in the tetraploid could be attributed to recurring hybridization be- tween genetically different diploid ancestors. All of these features are those anticipated for allopoly- ploids having recent origins (Haufler & Werth, GENETICS The mean levels of polymorphism and allelic variability for C. protrusa and C. bulbifera (Table 5) are similar to those expected of long-lived pe- rennials (P — 0.66, A — 2.07) having a primarily outcrossed breeding system (P — 0.51, A — 1.85) (Hamrick et al., 1979). Based on calculations of mean observed and expected heterozygosity values (Table 5), neither diploid species appears to have an excess or deficiency of aen: relative to Hardy-Weinberg expectations. netic identity among populations (Table ^i o. 83 59) is at the lower end of that found among other fern species (I = 0.912, range = 0.78-0.99 in Soltis & Soltis, 1989). Table 7 lists the values of F and indicates which of these values represents a signif- icant departure from zero. Those values that do not depart from zero are consistent with random mating for the species. Only the ramet populations 326 Annals of the Missouri Botanical Garden TABLE 7. Population genetics statistics. Calculation of the fixation index by population. Values not calculated for invariant loci. Asterisks after values = chi- MI test. No asterisk indicates that the locus did not differ significantly from О and thus suggests Е mating. * indicates significance at the Р < 0.05 level; ** indicates Р < 0.01 level; *** indicates P < 0.001 lev Cystopteris bulbifera opu- lation num ber PGI-2 IDH SkDH PGM-2 TPI-2 LAP 1 0.1806 0.0941 — 0.0606 —0.2500 0.1923 0.4582 2 0.3836** 0.5220*** —0.0167 0.2458 0.3324* 0.0000 3 0,8129 — 0.3208 0.0000 — 0.1053 0.0000 0.3000 4 – 0.0952 0.3450 0.0952 0.2778 0.0000 0.0455 Cystopteris protrusa—RAMET POPULATIONS Popu lation num- ber PGI-2 IDH SkDH PGM-1 PGM-2 TPI-1 LAP 1 0.0437 —0.0781 0.0000 —0.1228 0.0000 0.5071*** 0.6697*** 2 0.0000 0.0000 0.0000 0.0000 0.0000 0.3478 1.0000*** 3 – 0.0606 0.0000 —0.6667** —0.9444*** 0.0000 —0.6667** —0.6667** 4 —0.5972* 0.0336 0.0000 0.5175 =(),3529 0.0000 0.4202 5 0.5476 0.0556 0.4328 0.0000 0.0000 0.7467* —0.1343 © 0.0627 0.0000 0.0784 0.0000 0.0000 —0.2703 0.0000 Cystopteris protrusa—GENET POPULATIONS Popu lation num ber PGI-2 IDH SkDH PGM-1 PGM-2 TPI-1 LAP 1 0.0114 0.0000 0.0000 0.1944 0.0000 0.6420* 0.7100** 2 0.0000 0.0000 0.0000 0.0000 0.0000 0.4444 1.0000 3 0.0000 0.0000 0.0000 — 0.5000 0.0000 0.0000 0.0000 4 0.22020 0.5926 0.0000 0.5926 =0:2222 0.0000 0.6857 5 0.3500 —0.0833 0.3953 0.0000 0.0000 0.7111 —0.1143 © 0.2400 0.0000 0.1204 0.0000 0.0000 —0.2667 0.0000 of С. protrusa show a large number of significant figures. Two facts suggest that these values should not be used as evidence that С. protrusa deviates from random mating. First, the significant values are both positive and negative with a mean of .0687, a value that is very close to zero. Sec- Ne deviations from zero can be caused by over- sampling of cloned genotypes. Examining the F values for the genet populations shows that the number of significant figures drops to only two. Thus, the overall F values (Table 7) indicate that breeding systems for both diploid species conform to random mating models enetic analyses of sporophytic populations can be correlated with antheridiogen data and measures of genetic load to draw conclusions about the breed- ing systems of diploid members of the C. tennes- seensis complex. Antheridiogen response data in- dicate that C. protrusa should be outcrossing while C. bulbifera should be inbreeding. Genetic load data suggest that both diploids should be outcross- ing. As discussed above, chi-square tests of fixation index values (Table 7), however, indicate that nei- ther diploid species deviates significantly from ran- dom mating. Thus it may be that genetic load has a stronger control over fern mating systems than does possession of an antheridiogen system. By combining laboratory study of gametophytes and isozymic analyses of natural sporophytic popula- tions we can obtain a clearer understanding of the factors controlling breeding systems among these Cystopteris species (Schneller et al., 1990) Volume 77, Number 2 90 Haufler et al. 327 Analysis of the Cystopteris tennesseensis Complex Although the lack of allelic variability in C. tennesseensis precluded the application of isozyme data to analysis of breeding systems, genetic load data do provide clues to possible reproductive modes of this tetraploid. Other studies of diploid /polyploid complexes (Masuyama, 1979; Masuyama et al., 1987), have shown that although diploid progen- itors may have high genetic loads, polyploids de- rived from them do not. Our studies of the C. tennesseensis complex provided another example of this phenomenon. Isolated gametophytes of both diploids were incapable of forming sporophytes, presumably because of post-zygotic lethal genes. However, 25% of isolated С. tennesseensis ga- metophytes ae yield sparephytes, It has besa hy- pothesized t provide buffers against expression of lethals aM thereby allow polyploids to inbreed even if their progenitors cannot (Haufler, 1989). ur data indicate that genetic identity between the diploids is zero. Given that the enzymes we surveyed are considered to be evolutionarily con- servative (Cillespie & Kojima, 1968), the lack of similarity between Cystopteris congeners is quite remarkable. Angiosperm congeners have much higher mean genetic identity values (0.67 in Craw- ford, 1983) than those of most ferns that have been examined (0.33 in Soltis & Soltis, 1989). For Bommeria, Haufler (1985) suggested that the low genetic identity among congeners could indicate that iting the species resulted through convergent evolution ati, there- fore, that this genus was not monophyletic. Few would dispute, however, the clear relatedness of Cystopteris species. Apparently, the lack of ho- mology between diploid genomes observed in the cytogenetic study extends to the isozyme data as ell. Thus, these data support alternative hypoth- eses that either (1) speciation in ferns may be ac- companied by greater isozymic divergence than is typical for angiosperms (Haufler, 1987), or (2) that most congeneric fern species diverged prior to most angiosperm congeners (Soltis & Soltis, 1989). he extraordinary genetic divergence of the dip- loids ensures a fixed heterozygote pattern for all enzymes in the derived allotetraploid. Yet, as in- troduced above, there is variability among individ- uals of C. tennesseensis (Figs. 10- 12), These vari- tic progeny (Fig. 13), and are a at reflection ‘of the vari- ability observed among diploid populations. In As- p Werth et al. (1985b) used such infor- ation to propose recurring origins of the allopolyploids. It seems clear that C. tennesseensis originated more than once; but, in contrast to the amano ants do not seg e situation in Asplenium, there does not seem to be any geographic pattern to these variants. The lack of evidence for gene silencing and the fact that no orphan alleles were detected also suggests recent origins. SYSTEMATICS The current investigation did not result in re- vision of the current systematic treatment of mem- bers of the C. tennesseensis complex. It was pos- sible, however, to clarify the genetics of species boundaries and identify biological factors that have contributed to confusion in species circumscrip- tions. Perhaps the most significant factor blurring the boundaries between species is the formation of interspecific backcross hybrids. We collected and positively identified hybrids between C. tennes- seensis and each of its diploid progenitors. These sterile triploids are morphologically intermediate between the sexual species and, if they are not removed from consideration, can bridge the mor- phological gaps between what are otherwise rea- sonably distinct taxonomic entities. When not confounded by hybrids, a suite of qualitative features clearly discriminates C. pro- trusa from C. bulbifera (presence or absence of glandular trichomes and bulblets, clear differences in rhizome characteristics). In addition, we have demonstrated genetically regulated differences in quantitative leaf features (Table 3, Fig. 1) and significant differences in spore size between diploids and tetraploids (Table 1). Thus, although the char- acters are subtle, it is possible to identify each species through awareness of cryptic morphological characteristics and recognition of backcross hy- brids. If pteridologists hope to resolve significant evolutionary units it will be necessary to consider suites of qualitative differences, which are some- times cryptic. Cryptic features are even more important in the identification of tetraploid C. tennesseensis than they are in the diploids. Especially significant are spore characteristics, both in terms of detecting spore abortion in backcross hybrids and calculating size measurements for positive identification of ploi- dy level (Table 1). Somewhat problematic is our observation that spores of autotriploid C. protrusa can appear normal and can be quite large (Haufler et al., 1985). It is therefore important to consider the mean and standard deviation in developing an accurate assessment of the spore size (and thus the ploidy) of individual specimens. Given the large genetic distance between the diploid congeners in the C. tennesseensis complex, 328 Annals of the Missouri Botanical Garden it is probable that they have been phylogenetically isolated from each other for a long time and may represent systematic poles of the genus. The mag- nitude of the genetic distance between the two diploid species might also lead to the conclusion that hybrids between them should be rare and allopolyploid derivatives unsuccessful. (1980) suggested that most successful polyploids are not strict autopolyploids or allopolyploids, but occupy a position somewhere between these ex- tremes. Yet, our evidence indicates that even the highly differentiated C. protrusa and C. bulbifera have a history of recurring hybridizations and al- lopolyploid initiation. The derivative allotetraploid C. tennesseensis appears to be a successfully fledged and vigorous young species. It is quite widespread, is beginning to extend beyond the region of origin, nd appears to be exploiting a niche that is not inhabited by either parent: neither diploid is found in the drier, more disturbed localities characteristic of C. tennesseensis. CONCLUSIONS Morphometric, chromosomal, and isozymic analyses of the C. tennesseensis complex support he taxonomic treatment originally proposed by Shaver (1954). The present studies document the genetic distinctness of the diploids C. protrusa and C. bulbifera and confirm the hybrid origin of the tetraploid C. tennesseensis. Considering the limited geographic range of the tetraploid, its sympatry with its parental diploids, and the of gene silencing, it appears that this allotetraploid is of relatively recent origin(s). Our studies have also pinpointed why these three species continue to be misidentified and misinterpreted taxonomically. First, the diploid species have a rather plastic mor- phology that, though genetically determined, is subtle and may be modified by environmental con- ditions. Second, there are autotriploid individuals of C. protrusa that expand the range of variability for that diploid (Haufler et al., 1985). Third, out- crossing breeding systems (driven by antheridiogen systems and/or high genetic loads) may contribute to frequent formation of backcross hybrids. These sterile triploid individuals span the morphological gaps between species, obscuring the boundaries that separate these genetically discrete taxa. Fourth, through recurring allopolyploid events, C. tennes- seensis has incorporated much of the genetic vari- ability of the diploids into its populations. As a result, positive identification of taxa in this complex depends on observations of cryptic spore and leaf features. LITERATURE CITED ANDERSSON-KorrO, I. 1930. Variegation in three species e Z.LA.V. (Molec. Gen. Genetics) 56: 115- 9 AUQUIERE, J. P. & P. Mogens. 1973. Développment due gamétophyte et vascularisation du jeune T phyte dans le genre Cystopteris. Cellule 70: 13 59. BLASDELL, R. 19 A monographic study of the fern genus C ystopteris. Mem. Torrey Bot. Club 21: 1-10 BRITTON, p. M., W. G. Stewart & W. J. Copy. 1984. Cystopteris protrusa, creeping fragile fern, an ad- dition to ps flora of Canada. Canad. Field-Naturalist 99: 380-382 CRAWFORD, p. 1 1983. Phylogenetic and systematic inferences from electrophoretic studies. Pp. 257- 287 in S. Tanksley & T. Orton ea Isozymes in Plant Genetics and Bisdding: Par Elsevier Science Publishers, Amsterdam. Gastony, C. J. The Pellaea glabella complex: electrophoretic sida for the derivations of B osporous taxa and a revised taxonomy. Ame . D. Gor TLIEB. 1982. Evidence for genetic heterozygosity in a homosporous fern. Amer. J. Bot. 69: 634-637 GILLESPIE, J. г К. Kojima. 1968. The degree of polymorphisms in enzymes involved in energy pro- duction compared to that in nonspecific enzymes in two Drosophila ananassae popi ulations. Proc. Nat. Acad. Sci., U.S.A. te 582-585. Hamrick, J. L., Y. ART & J. B. Mitton. 1979. Relationships ee life history characteristics and electrophoretically detectable genetic variation in plants. Ann. Rev. Ecol. Syst. 10: 173-200. HavrLER, C. H. 1985. Enzyme variability and modes of evolution in the fern genus Bommeria. Syst. Bot 10: 92-104. 1987. Electrophoresis i is modifying our con- cepts of evolution in homosporous pteridophytes. Amer. J. Bot -9660. 1 owards a synthesis of evolutionary заны and Ме жа in Pan eau pterido- 1) 21 Syst. Ecol. 109-115 ÍT NKER. 1985. p of differential жуса ridiogen aan on evolutionary и п Cystopteris. Amer. Ј. Bot. 72: 659-66 & D. E. Sorris. 1984. Obligate ee ш а кенин fern: field confirmation of a labo- ratory prediction. Amer. J. Bot. 71: 878-881 & ———. 1986. Genetic evidence suggests that homosporous ferns with high Ие num- bers are di Proc. Nat. Acad. ., U.S.A. 83 4389 4393 ~i ~ d. ERTH. 1986. Deciphering differ- ences in the ev aog ‘histories of polyploid fern Amer. J. : D. SA. D. M. BRITTON & $ ROBINSON. 1985. Triploidy and its area и in Cystopteris protrusa. Canad. J. Bot. 63: 1855-1863. spec ie 5. Hickey, R. J., S. I. GUTTMAN & W. Н. EsuBAUGH.. 1989. Evidence for post-translational modification of triose I phosphate isomerase (TPI) in /soetes (Isoétaceae). Amer. J. Bot. 76: 215-221. Volume 77, Number 2 Haufler et al. 329 1990 Analysis of the Cystopteris tennesseensis Complex мин R. M. 1974. Reproductive biology and evolution with Special Reference to Tennessee. General Pub- Te А Ann. Missouri Bot. Gard. 61: lishing Co., Toronto Sorts, D. E., С. Н. Haurter, D. C. Darrow & G. J. Lovis, Т y GASTONY. 1983. Starch gel electrophoresis of ferns: on Evolutionary patterns and processes in ferns. Adv. Bot. Res. 4: 220-415. MasuYAMA, S. 1979. Reproductive biology of the fern opteris decursive-pinnata l. The dissimilar mating systems of dua and tetraploids. Bot. Mag (Tokyo) 92: 275-2 , K. Mitur & N. Nak 1987. Studies on intraspecific polyploids of н ne fern Lepisorus thun- bergianu (2 ). Mating system ud the ploidy. J. Jap. ot. . MCGREGOR, R. L 1950. p 5l of Cystopteris ragilis. Amer. Fern J. 40: 201-207. MORAN, В. C. 1982. C) tote tennesseensis in 11- linois. Amer. Fern J. 72: 93- Nakal, T. 1933. Notes on | Japanes se ferns. l. А new e of Japanese И with special view to the spores and connective cells. escrip- tions ad amendments for East- yee ferns. Bot. Mag. (Tokyo) 47: 151-186 PicHI-SERMOLLI, R. E. С. 1977. Tentamen Pterido- phytorum genera in taxonomicum ordinem redigendi. Webbia 31: 313-512. PnoruMo, Р. 1969. Не Aametoplytes of C phot н запа Cystopte . Webbia 23: 317- bd Y ‚ К. SCHNABEL. 1986. Allozymic and id evidence a progenitor-deriv- ative species pair in Camassia (Liliaceac) Syst. Bot. SCHNELLER, J. J., C. Н. HAUFLER & T. A. RANKER. 1990. Norge Pa natural gametophyte populations. . Fern v (in press). - ER, nT M. A new fern, Cystopteris tennes- seensis, n nov. from Tennessee. J. Tennessee Acad. Sci. 25: 13. Ao. E of the Eastern Central States a compilation of grinding buffers, gel and electrode buffers, and staining schedules. Amer. Fern J. 73: -21, P. S. Sorris. 1989. Polyploidy, breeding systems, and genetic differentiation in omosporous pteridophytes. Pp. 241-258 in D. E. Soltis & P. S. Soltis (editors), scu in Plant Biology. Dioscorides Press, Portland, Oregon. STEBBINS, C. k 1980. Polyploidy in plants: unsolved roblems and prospects. Pp. 495-520 in W. Lewis (editor), Polyploidy: Biological Relevance. Ple- num Press, New York. STEIN, D. & D. S. BARRINGTON. 1990. Recurring hybrid formation in a population of Polystichum X potteri evidence from chloroplast DNA comparisons. Ann. . Gard. (this volume). Жыны. W. H., Jr. 1954. е evolution in the Appalachian p le Evolutio 11 1971. Evolution of Dry opteris in relation the ‘Appalachia ns. Pp 147-192 in P. Holt Фаг), The Distributional History of the Biota of the South- art П iv. Mono. ern Appalachians. Part II. Flora. Res. Di no 2, Virginia Polytechnic Inst. and State Univ WEEDEN, №. Е. 19 Plastid isozymes. Pp. 139-156 Jr in S. D. Tanksley & T. J. n (editors), Isozymes in Plant Genetics and Breeding, Part A. Elsevier Science ом Amsterdam WERTH, C. R., 1. GUTTMAN & W. ESHBAUGH. 1985a. E ea evidence of геси еуо- lution in the Appalachian Asplenium complex. Syst Bot. 10: 184-192. 5b. Recurring origins & 98 of allopolyploid species in n Asplenium. Science 228: 731-733. THE AMERICAN PARADOX IN THE DISTRIBUTION OF FERN TAXA ABOVE THE RANK OF SPECIES: K. U. Kramer? ABSTRACT Whereas the numbers of fern species in the eastern Paleotropics and the Neotropics are comparable, the diversity at the subgeneric and especially the generic level is much higher in in the present paper. phenomenon is termed the "American Paradox” the eastern Paleotropics. This unexplained In editing or preparing accounts of the various fern families for the treatment of pteridophytes in the series “The Families and Genera of Vascular Plants" (Kramer et al., in press), I gave special attention to the distribution of the genera and in- frageneric taxa throughout the world, especially on the major continents. The areas where genera had their major concentrations were specially not- ed. А somewhat unexpected picture emerged. The poverty of species as well as superspecific taxa in the African fern flora compared with that of Madagascar, tropical Asia-Australasia, and the Neotropics is well known; it has also been docu- mented in many groups of flowering plants. For comments and citations of further literature, see Parris (1985) for ferns and Richards (1973) for flowering plants. Another striking phenomenon, largely unnoticed, is the relative poverty of fern taxa above the rank of species in the warmer parts of the New World. In this paper 1 po docu- mentation for this American Paradox SPECIES RICHNESS Comparing the species richness of the New World fern flora with that of tropical Asia-Australasia (leaving out Africa as much as possible for reasons stipulated above) is not easy, as few genera or families of ferns have been critically revised re- cently on a worldwide scale. The figures in Table 1, unless they are derived from recent monographs, should be regarded as rough approximations, most- ly based on the data assembled for the "Families and Genera" referred to above. But they are prob- ably correct regarding order of magnitude. (For the classification underlying this and other tables and lists of taxa in the present paper, see the Appendix.) These figures show that the New World does not emerge with a poorer fern flora as a whole. In species richness the two areas are fairly bal- anced, the difference in the sum being undoubtedly fortuitous. When large genera such as Asplenium and Dryopteris are revised, the balance will prob- ably be in favor of the eastern hemisphere. GENERIC DIVERSITY The picture is very different when instead of the number of species, the number of genera present in the western vs. the eastern hemisphere (exclud- ing purely African taxa) is compared. (See Table Note that in Lomariopsidaceae, Cyatheaceae, and Polypodiaceae, for example, where the eastern Old World and the New World have equal or at least comparable numbers of species, the number of (sub)genera is much higher in the eastern Old World. When families with fewer than four genera, where the figures are less meaningful, are left out of consideration, the difference is particularly strik- ing: out of 13 families, 10 are considerably more diversified at superspecific levels in the Old World than in the New World; two families, Pteridaceae and Vittariaceae, are about equal; only one, Hy- menophyllaceae, is more diversified in the New World. In Hymenophyllaceae the supraspecific classification is not yet completely worked out be- cause Iwatsuki’s (1984) system is based chiefly on the Asiatic representatives, so the picture may change to become even more favorable for the New World. A comparison of Tables 1 and 2 leads to the 'T thank an anonymous reviewer for valuable suggestions on the manuscr ? Institut für Systematische Botanik der Universitát Zürich, Zollikerstr. T CH. 8008 Zirich, Switzerland. ANN. MISSOURI BOT. GARD. 77: 330-333. 1990. Volume 77, Number 2 1990 Kramer American Paradox in Fern Distribution 331 conclusion that the fewer genera present in the New World contain relatively more species than ld. This suggests an imbalance between specific vs. generic representation. This picture is again found when pantropical genera with very uneven distributions are com- pared. In Table 3 such genera are listed; they have five or fewer species in one hemisphere, the re- mainder in the other hemisphere. In addition, two genera should be mentioned that, although being well represented in the Old orld, have many more species in the New World, viz. Elaphoglossum and Polypodium. the more numerous genera of the O INFRAGENERIC DIVERSITY Infrageneric diversity is more difficult to assess because many subgenera and sections of ferns are treated as full genera by authors more inclined toward narrow generic circumscriptions. Here 1 refer to subdivisons of genera that are rarely, if ever, treated as distinct genera by contemporary authors. It proved difficult to develop reliable fig- ures for a basis; few cosmopolitan or pantropical genera have recently been revised complete with natural infrageneric classifications. Conclusions from the following figures must be a as prelim- inary, as the number of cases is Anemia a see ieee 1981) and Tryon & Tryon (1982). Three subgenera, one pantropical, one American with a single African species, one purely American. The genus was, how- ever, more widespread in the Old World in the Tertiary (see, e 1981, p. 24). Bolbitis (Lomariopsidaceae): see Hennipman ., Huang, and its islands, three exclusively in the Old World (excluding strictly African taxa), one only in the New World. Elaphoglossum (Lomariopsidaceae): see Mickel & Atehortua (1980). Nine sections were recog- nized; four are strictly neotropical, not surprising in view of the enormous species diversity in the New World, but the five others are also represented in the Old World. Dryopteris (Dryopteridaceae): see Fraser-Jen- kins (1986). This comprises 16 infrageneric units (sections, or subgenera not further divided into sections); one is north-temperate, six are on both sides of the Atlantic, eight only in the eastern and one (almost) confined to the western hemisphere. Lindsaea (Dennstaedtiaceae): see Kramer (1957, 1967, 1972), Kramer & Tindale (1976). roximate numbers of species in some widely distributed fern families and genera. Tropical sia- Family / genus Australasia Neotropics Cyatheaceae ca. 200 ca. 205 Polypodiaceae ca. 350 ca. 250 Thelypteridaceae ca. 440 ca. 240 Adiantum (Pteridaceae) ca. 40 ca. 100 Bolbitis 12 14 (Lomariopsidaceae) Cyclopeltis ca. 5 1 (Dryopteridaceae) Elaphoglossum ca. 50 ca. 350 (Lomariopsidaceae) Lindsaea 62 48 (Dennstaedtiaceae) Lomariopsis 5 ca. 15 (Lomariopsidaceae) Lygodium (Schizaeaceae) 14 ca. 8 Nephrolepis ca. 12 ca. 10 (Nephrolepidaceae) Odontosoria (including 7 8 Sphenomeris) (Dennstaedtiaceae) Oleandra (Oleandraceae) ca. 25 ca. 8 Schizaea (Schizaeaceae) 7 8 Totals ca. 1,229 са. 1,265 Twenty sections were recognized; two are only in the southwestern Pacific, two in Madagascar; of the others, four are present in the Old and the New World, ten in Asia-Australasia only (or, in very few cases, extending to Africa); only two are confined to the Neotropics. Hymenophyllum (Hymenophyllaceae): see Iwatsuki (1984). Fifteen sections or subgenera with three to the New World; here the distribution is approximately balanced. smunda (Osmundaceae): various sources for classification. There are three subgenera, one worldwide, one north-temperate, and one in eastern Unf p no sufficiently reliable data are available for ollowing large, pantropical to Bons genera where information on the distribution of their subdivisions would be use- ful: Adiantum, Asplenium, Blechnum, Chei- lanthes, Ctenitis, Diplazium, Grammitis, Poly- odium, Pteris, Paleogean preponderance for supraspecific diversity seems and Tectaria. Annals of the Missouri Botanical Garden TABLE 2. Distribution of the genera of the fern families (excluding probable relict families). Genera not present in the tropics of Asia-Australasia and the Neotropics appear in the overall sums of genera but not in the three columns at the right. Number of In both Old World New World Family genera hemispheres only only Blechnaceae 9 2 4 1 Cyatheaceae (genera or sections; see Appendix) 4 3 1 0 Dennstaedtiaceae 16 11 4 1 Dry е sens. lat. 45 18 18 9 Gleicheniaceae 5 3 2 0 ups eaque 1 3 0 Hymenophyllaceae 8 5 1 2 Hymenophyllopsidaceae 1 0 0 1 Lomariopsidaceae 6 4 2 0 larattiaceae 6 1 4 1 Nephrolepidaceae 1 1 0 0 Oleandraceae 3 1 2 0 Ophioglossaceae 3 2 1 0 Joi еее (see Appendix) 32 5 21 6 idaceae sens. lat. 33 10 11 12 Thelypteridaceae (see Appendix) 29 9 17 3 Vittariaceae _ 6 _2 _2 2 Totals 211 78 93 38 likely for Pteris and Tectaria, neogean for Adian- tum and Polypodium. At infrageneric levels, the preponderance of Asia- Australasia also seems to be present, but more weakly. This was to be expected in view of what was found for the distribution of species over the two hemispheres. DISCUSSION Is the “American Paradox” perhaps an artifact of different approaches of systematists working wit Ol as opposed to New World ferns? It is true that certain taxonomists concentrating on Asiatic ferns have a much narrower concept of taxa on all taxonomic levels than most others; a prominent example is Ching (e.g., 1978). But the classification worked out for the "Families and enera” from which the above figures have largely been taken is the result of a critical reevaluation of all genera. Even if this classification is not ac- ceptable to others, it may be said with some con- fidence that it does not ошаш а bias toward either * for the classification of Е "splitting" or “ the Old World or New Wor Is there an explanation for ils "American Par- adox"? At the moment I prefer not to make an attempt. Assuredly, the New World tropics must have been habitable for ferns as long as the Pa- leotropics, and they harbor relict taxa like Loxo- mataceae, Gleicheniaceae, and Marattiaceae that go far back in geological history. Yet the assort- ment of ferns belonging to the so-called higher leptosporangiates gives the impression of relatively recent speciation in comparatively few taxa of higher rank; many of them are also present in the Old World and perhaps have originated from there. An explanation will eventually have to come from ge- ology, paleogeography, and paleoclimatology data. In light of what we know about the breaking up BLE 3. Genera of wide distribution but very un- evenly distributed over the world, with most species in one hemisphere, fewer than six in the other hemisphere. Predominantly eastern hemisphere (9 taxa): Platycerium Thelypteris sect. Cyclosoriopsis (Christella as genus) Thelypteris sect. Stegnogramma (or Stegnogramma as genus) Diplopterygium Lomagramma Microlepia (native in New World?) Predominantly western hemisphere (4 taxa): Hemionitis (the sole Old World species perhaps misplaced in the genus) Microgramma Pleopeltis Trichomanes (sensu Iwatsuki, 1984) Volume 77, Number 2 1990 Kramer 333 American Paradox in Fern Distribution of Gondwanaland, the paradox is even more per- plexing. For the moment it may suffice to bring the phenomenon, duly documented, to the attention of the botanical community, without indulging in insufficiently founded speculation. LITERATURE CITED s . A classification of the genus Dryopteris ми с атти Bull. Brit. Mus. (Nat. Hist.) Bot. 14(3): 183-218. HENNIPMAN, E. A Monograph of the Fern Genus Bolbitis (Lomariopsidaceae). Leiden Bot. Series 2, Leiden es Press, Leiden. HoLTTUM, R. E. & P. J. Epwarps. 1983. The tree- rile of a Roraima with comments on the family Cyatheaceae. por Bull. 38: 155-188. How т С. 1981. Spore Flora of Taiwan. Nat. Taiwan , Tai Irsa; K. ams Studies in the systematics of filmy I. А scheme of classification based chiefly on E Asiatic species. Acta Phytotax. Geobot. 35: -179. KRAMER, K. U. 1957. A revision of the genus Lindsaea in the New World with notes on allied genera. Acta Bot. Neerl. 6: 97-290. 67. The lindsaeoid ferns of the Old World III. Notes on Lindsaea and Sphenomeris in the Flora Malesiana area. Blumea 15: 557-574. . 1972. The lindsaeoid ferns of the Old World IX. Africa and its islands. Bull. Jard. Bot. Etat 42: 305-345 M. D. TINDALE. of the Old World VII. Australia and New Zea Telopea 1(2): 91-128 MickEL, J. T. 1962. A monographic study of the fern genus Anemia, subgenus Coptophyllum. lowa State Coll. J. Sci. 36: 349-482 1981. Revision of Anemia subgenus Ane- Mu cda (Schizaeaceae). Brittonia 33: 413-429. —— — — & L. ATEHORTÚA G. 1980. urge of the genus Elaphoglossum. Amer. Fern J. 70: 47-68. 1976. The lindsaeoid ferns land. Parris, B. S. 1985. Ecological aspects of distribution and speciation in Old World tropical ferns. Pp. 341- in А. er & C. N. Page (editors), Biology of du ur ds ee Edinburgh. RICHARDS, P. W. . Africa, the “Odd man out.’ Pp. 21-26 in B. b Meggers D. Duckworth (editors), Tropical Forest Ecosystems in Africa and South America. Smithsonian Inst. Press Generic and family boundaries in the ү arg че and Athyriaceae. Bot. J. Linn. Soc. 67 Suppl. 1 0. Tryon, В. M., Jr. . К. Tryon. 1982. Ferns and Allied Plants a ‘Specia ial Reference to Tropical America. Springer-Verlag, Heidelberg APPENDIX. Notes on the classification. As the treatment of pteridophytes for the "Families and Genera of Vascular Plants” referred to above has not appeared as this paper goes to press, certain points in the system that underlie the figures given above are briefly indicated. Cyatheaceae: the к de caso by Holttum in Holttum & Edwards a 983) of R. M. Tryon’s school of tree-fern students Dryopteridaceae: the family is broadly circumscribed, fol- lowing Sledge (1973); it includes bs minore and the onocleoid fern: as well as Woodsi Oleandraceae: excluded from Davalliaceae (like Nephro- lepis). E Crore certain species groups often treated as enera of Polypodium i in the “Families and Genera," counted as equivalent to genera are here Pteridaceae: these are broadly circumscribed and include Adiantum and all so-called gymnogrammoid ferns Thelypteridaceae: only five genera are retained in the “Families and Genera," but many subgenera are rec- nized; these are here counted as full genera in ac- cordance with other authors” views. RECURRING HYBRID FORMATION IN A POPULATION OF POLYSTICHUM XPOTTER Г. COMPARISONS' Diana В. Stein? and David S. Barrington? ABSTRACT The origins of a population of the hybrid fern Polystichum X potteri (P. acrostichoides х P. braunii) were explored through an examination of restriction fragment 5 polymorphisms in chloroplast DNA. We demonstrate that the hybrid individuals in this population each had o of the two possible parental chloroplast genomes. This inheritance of two chloroplast g genomes in three or more different descendants of a paleopolyploid species. Plant organelle inheritance has been the focus of many investigations. The vast majority of pub- lished reports has examined reciprocal hybrids pro- duced by crossing (Gillham, 1978) or somatic hy- brids produced by protoplast fusion (Pelletier, 1986), but some studies have used natural hybrids (Palmer et al., 1983). Patterns of chloroplast in- heritance were first traced by using white and green plastids as parental markers; later they were traced using specific chloroplast gene products (Gillham, 1978). More recently, chloroplast DNA fragment profiles produced by digestion with restriction en- donucleases (e.g., Conde et al., 1979; Hachtel, 1980) have proven valuable as genetic markers of chloroplasts. Surprisingly little has been reported about the inheritance of organelles in ferns. Kirk & Tilney- Bassett (1978) cited only one study: the chloro- plasts of Phyllitis scolopendrium (L.) Newm. (as Scolopendrium vulgare Sm.) were found to be inherited biparentally. However, in mature plants of the rare fern hybrid Osmunda X ruggii R. Tryon (Tryon, 1940; Wagner et al., 1978), the chloro- plast genome of only one parent was found (Stein, 1985). Hence, the pattern of chloroplast DNA inheritance in ferns warrants further investigation. Interspecific hybrid plants may be difficult to detect by their morphology (Doyle & Doyle, 1988), or the parentage of apparent hybrids may be un- certain (Doyle et al., 1985; Hilu, 1988; Yatskie- vych et al., 1988). Evidence from molecular bi- ology has been helpful in resolving tl lti Electrofocusing of the small subunit of the enzyme ribulose-1,5-bisphosphate carboxylase, a nuclear DNA encoded polypeptide, has identified parents of tobacco hybrids, and examination of the large subunit (which is chloroplast DNA encoded) has determined maternal parentage (Wildman, 1983). Hybrids have also been documented by demon- strating summation of two allozymes, each fixed in one of the parent species (Werth et al., 1985a). The combination of diagnostic lengths of ribosomal DNA fragments from each parent in an interge- neric hybrid has been used as evidence for hybridity (Doyle et al., 1985). This approach was particu- larly helpful in the detection of Claytonia hybrids, where morphology is variable (Doyle $ Doyle, 1988). In addition, restriction site mutations in We greatly appreciate the advice and P uin of Jeffrey Palmer. Part of this study was accomplished in his laboratory. We also thank Elsbeth Walker exper Foundation Grant of Researc MER to D. B.S S. and David S. —— is gratefully acknowledge d ? Department of Biological Sciences, Mount t technical assistance. Funding by a William and Flora and a National Science Foundation Grant (BSR-8818459) to Hewlett yoke College, South Hadley, Massachusetts 01075, U.S.A. * Pringle Herbarium, со оѓ Botany, ets of Vermont, Burlington, Vermont 05405-0086, U.S.A. ANN. Missouni Bot. GARD. 77: 334-339. 1990. Volume 77, Number 2 1990 Stein 8 Barrington 335 Recurring Hybrid Formation in Polystichum x potteri chloroplast DNA have identified one parent of a hybrid in a number of studies (e.g., Palmer et al., 1983; Hilu, 1988; Yatskievych et al., 1988). Allopolyploidy is a prevalent mode of hybrid speciation in plants. Klekowski 6 Baker (1966) documented the frequent occurrence of high chro- mosome numbers in the homosporous ferns. Recent estimates suggest that 95% of pteridophytes are polyploid and that the majority of these are allo- polyploids (Grant, 1977). Although some of these reported polyploids act as genetic diploids as mea- sured by isozyme analysis (Haufler & Soltis, 1986; Wolf et al., 1987), numerous allopolyploid taxa have been documented by a variety of approaches (Wagner & Wagner, 1980; Barrington et al., 1989). One of the best-studied fern genera in which allopolyploidy has played an important role in spe- ciation of the group is the Appalachian Asplenium complex. Morphological analysis, chromosome number and pairing behavior (Wagner, 1954), fla- vonoid studies (Smith & Levin, 1963), and isozyme analysis (Werth et al., 1985a) have all been used to demonstrate the allopolyploid nature of members of this genus. In this complex, recurrent origins of allopolyploidy have been demonstrated by Werth et al. (1985b). Their isozyme analysis revealed several different fixed heterozygotes, strongly sug- gesting that the allopolyploids have formed several times. We have examined chloroplast DNA (cpDNA) and ribosomal DNA (rDNA) of members of a pop- ulation of Polystichum X potteri Barrington, which is the triploid hybrid between P. acrostichoides (L.) Roth (a diploid, n = 41) and P. braunii (Spen- ner) Fée (an allotetraploid, п = 82) (Barrington, 1986). This fern population was studied to address several questions: 1) Are chloroplasts inherited uni- parentally in Polystichum species? 2) Does a pop- ulation of hybrids have cryptic variation with re- spect to its chloroplasts? If both chloroplast APT can be demonstrated in the population but in ferent hybrid individuals, then recurrent hybrid formation would be demonstrated. 3) Are there restriction fragment polymorphisms in the ribo- somal DNA of the parents, and are these different fragments both present in the hybrid offspring as would be expected from the combining of two dif- ferent nuclei? 4) Do differing cytoplasms alter the morphology of hybrids with the same nuclear ge- no In this study we show that chloroplasts appear to be inherited uniparentally in Polystichum and that recurrent hybrid formation can be documented by the presence of two chloroplast genomes in different individuals of a hybrid population. The hybrids also show the expected additivity of re- striction fragment length polymorphisms in their ribosomal DNA. No effect of cytoplasmic inheri- tance on morphology of the hybrids was detected. MATERIALS AND METHODS Leaves of individuals of Polystichum acrosti- choides, P. braunii, and four of their hybrids, P. X potteri, were gathered from the talus below the limestone-rich outcrops of Barnard Gulf, Barnard, Windsor Co., Vermont, in July 1984. There, ca. 150 hybrid individuals are distributed over at least 0.4 km of east-facing slope from the foot to near the rim of the small steep-sided valley. The hybrid individuals included in this study, Barrington 1091, 1094, 1097, and 1098, were widely separated (at least 2 m apart). They were all morphologically documented as hybrids in earlier work (Barrington, 1986). At the same time, Barrington 1094 was documented cytologically as triploid. The irregular meiotic pairing he observed is typical of a hybrid between an allotetraploid and a nonancestral dip- loid. Collections of leaves from the two parents from the same site provided the material used to document the progenitor species in this study. Chloroplasts and nuclei were isolated from the leaves as described by Palmer & Stein (1982) and Palmer (1986). Briefly, the fronds were disrupted in a large volume-to-tissue ratio (450 ml to 25- 40 g of tissue) of an isolation buffer containing polyethylene glycol (PEG; M.W. 3,500) using first a blender and then a Polytron. Unbroken cells and other cellular debris were removed by filtration through a sandwich of two layers of cheesecloth and one layer of Miracloth. Nuclei and chloroplasts were collected by centrifugation in a GSA rotor at 2,500 RPM and washed with a buffer containing PEG. The re-collected organelles were suspended in a small volume of wash buffer and applied to sucrose gradients. The gradients were prepared by layering 30% on 50% sucrose. Both steps con- tained PEG 6000, and the gradients were allowed to diffuse overnight at 4?C before use. Centrifu- gation at 25,000 RPM for one hour using an SW 277 rotor separated the chloroplasts, which banded at the gradient-step interface, from the nuclei, which pelleted. Both organelles were recovered, lysed, and the DNAs were prepared in CsCl.ethidium bromide gradients. After removal of the CsCl and the ethidium bromide, the nuclear DNAs were ex- tracted with tris-buffered phenol and ether and exhaustively dialyzed. Chloroplast and nuclear DNAs were digested with restriction endonucleases according to the 336 Annals of the Missouri Botanical Garden manufacturer’s instructions. For the rDNA analysis the following eight enzymes were used: Apa I, BamH I, Bgl I, BstE П, EcoR I, Hinc П, Sac I, and Xho I. Hind III and Pst I were used to digest the cpDNAs. Our methods for electrophoresis, blot- ting, and Southern hybridizations are described in Yatskievych et al. (1988), Palmer (1986), and Palmer & Stein (1982). The cloned ribosomal DNA used was рНА1 (Jorgensen et al., 1987), which contains an entire repeat unit from pea; it was kindly provided by R. Cuellar. RESULTS AND DISCUSSION Chloroplast DNAs from Polystichum acrosti- choides, P. braunii, and the four hybrids were digested with Hind III and the resultant fragments separated by gel electrophoresis (Fig. 1). It is clear that the chloroplast DNA fragments of the two parental species are very similar, yet the chloro- plast DNA of P. braunii differs from that of P. acrostichoides in having a band of 11.5 kb, and Р. acrostichoides has a band of about 9.4 kb, which is lacking in P. braunii. The 9.4 kb fragment was most likely produced by an additional Hind III site in the 11.5 kb fragment, but the presence of the other fragment (which is likely to be 2.1 kb) could not be verified in this gel. Nonetheless, the two large fragments provide a clear criterion to distinguish the parental genomes. The fragment profiles reveal that two of the hybrids inherited the P. braunii chloroplast genome, whereas the other two contained the P. acrostichoides chloroplast DNA. Pst I digests demonstrated the same pattern of inheritance of parental cpDNA genomes in the hybrids. Since only one parental chloroplast genome is found in each of the hybrids examined, it would appear that uniparental inheritance is character- istic of these ferns as well in O. xruggii. (Stein, 1985). A similar result has been found in a tree fern hybrid, Cnemidaria horrida (L.) K. Presl x Cyathea arborea (L.) Smith (Stein & Conant, un- published). Such studies, which examine chloro- plast DNAs from mature plants, do not rule out the possibility of chloroplast inheritance from both parents with subsequent sorting of the chloroplast types. However, in Phyllitis scolopendrium the mode of plastid transmission is apparently bipar- ental, yet both green and pale plastids were main- tained (not sorted) in the mature sporophytes (An- dersson-Kotto, 1931). Examination of very young hybrid individuals by Southern hybridization would address this question. Duckett & Bell (1971) re- ported that male gametes of archegoniate plants often do not transmit plastids to the egg. Thus, despite the one report of biparental transmission of chloroplasts in ferns (Gillham, 1978), uniparen- tal, maternal inheritance of chloroplasts in ferns seems more likely to be a common pattern. We also examined nuclear ribosomal DNA cut with eight restriction enzymes for evidence of the hybrid nature of these plants (Fig. 2). Seven of the eight endonucleases did not reveal informative re- striction fragment length polymorphisms. How- ever, the restriction enzyme Ара 1 cut the ribo- somal DNA in the two parental types to yield several similar large fragments (ranging from 14.7 kb to 4.0 kb) and also two small fragments of slightly different mobility, 1.6 kb in P. acrostichoides and 1.4 kb in P. braunii. These small fragments are both present in the hybrids, as expected for nuclear DNA. A similar additivity of different lengths of ribosomal DNA fragments was demonstrated in an intergeneric hybrid between Tolmiea menziesii and Tellima grandiflora (Saxifragaceae) by Doyle et al. (1985). It is likely that these small fragments come from the intergenic spacer region, which has been shown to be the most variable (Appels & Dvorak, 1982), and length variation in this region may be related to loss or gain of a subrepeat. In addition to the ribosomal DNA fragment additivity, starch-gel electrophoresis revealed that all four hy- brids summed isozyme bands fixed in each pro- genitor species at PGM-1 and SkDH (Barrington, unpublished) but did not show differences in fixe heterozygosity. Thus, while the isozymes and the ribosomal DNA data both verified that these ferns were hybrids, only the chloroplast DNA analysis demonstrated that they were undergoing recurrent formation. The morphology of the hybrids was examined by one of us (DSB), while only the other (DBS) was aware of the genetic makeup of the cytoplasm. The 26 morphometric characters evaluated to de- termine if the cytoplasm accounted for a shift in morphology toward that of the parent contributing the cytoplasm included 21 leaf size and shape char- acters, stomate length and width, sorus position, receptacle shape, and true-indusium diameter. No clearcut relationship between morphology and the cytoplasmic compositions was detected in a dis- criminant function analysis of the hybrids and their parents. The discovery of two chloroplast genomes in the population of hybrid P. Xx potteri suggests that examination of a number of hybrid individuals might be routinely carried out as a means of documenting parentage and demonstrating recurrent hybrid for- mation. For example, Stein & Conant (unpub- Volume 77, Number 2 1990 Stein 8 Barrington Recurring Hybrid Formation in Polystichum x potteri A123 4 B A FIGURE Chloroplast DNAs from Polystichum ac- rostichoides (A), P. ee s ), and four hybrid plants of P. x potteri (1, 2 4) digested with Hind III. The e by gel elec- l and 2 contain the P. brani chlo- fragments were de udo 0.7% agarose Hybrids 1 iybrids З and 4 have the P. acrosti- choides chloro wi genome. А = А DNA cut with Hind III. The bands in this lane correspond to (from the top) .0 kb. 23.1, 9.4, 6.6, 4.4, 2.3, and 2.0 lished), in a preliminary examination of the tree ferns, examined nine individuals of Cnemidaria horrida X Cyathea arborea for chloroplast DN composition. Eight contained the chloroplast DNA of Cnemidaria horrida; the ninth revealed the chloroplast genome of C. arborea. This documents the second parent; but more importantly it shows that hybridization has occurred more than once in these tree fern hy Chloroplast DNA has also been examined in studies of polyploid species in other evolutionary groups. Bryophyte allopolyploids have been shown to have had multiple origins. In a study of Pla- -1.6 -1.4 FIGURE 2. Ribosomal DNA fragment length poly- morphisms in P. acrostichoides (A), P. braunii (B), and four ка plants o x potteri (1, 2, 3, 4) digested with Apa I. The fragments were separat ›п 0.7% agarose, transferred to Zetabind by Southern blotting, and hybridized to a ??Р-!а beled е DNA repeat arp Polystichum acrostichoi and P. braunii contain kb and 1.4 kb fragments, mone nie the hybrids contain both fragment lengths. giomnium medium (Bruch, Schimper & Gumbel) Kop. (Wyatt et al., 1988), an analysis of isozymes and chloroplast DNA documented allopolyploidy and the maternal parent, respectively. Only one chloroplast DNA type was found, but the study does not state how many individuals were examined for their DNA composition. Multiple origin of allopolyploids has recently been documented in several angiosperm taxa. ‘al lace & R. Jansen (in prep.) have used chloroplast DNA to demonstrate recurrent hybrid formation in Microseris (Asteraceae). An allotetraploid, M. 338 Annals of the Missouri Botanical Garden ANCESTRAL DIPLOIDS © La UA Fi (STERILE) iit d doubling DERIVATIVE POLYPLOIDS (FERTILE) Diploidization (chromosomal & | genetic) -and- Divergence at the new level PALEOPOLYPLOID SPECIES GROUP PHYLOG INFERRED FRO CHLOROPLAST DNA | A hypothetical evolutionary history show- ing formation of pale oallopolyploids that esie ien ntly gave FIGURE 3. rise to three 1 o of the modern species would contain chloroplasts deriv ed from one chlo- roplast genome (e.g., the “a? „Chloroplast genome), where- as one of the modern species would have R derived from a different c Мотора genome (e.g., “p” ea genome). Thus, relationships as ed by clea A would suggest sal divergence, a chloroplas DNA would support a closer relationship be- en two of the three descendants. nodern species. In this case t heterocarpa (Nutt.) K. Chambers, has originated ice. During each origin, the chloroplasts were contributed by different subspecies of M. douglasii .) Schultz-Bip. (i.e., subsp. douglasii and subsp. tenella [Asa Gray] Chambers). No tetraploid so far examined has contained the chloroplast genome of М. lindleyi (DC.) A. Gray, the other progenitor of the tetraploid. Similarly, Doyle, Doyle, Brown, and Grace (pers. comm.) have provided evidence of multiple origin of allotetraploid Glycine taba- cina (Labill.) Benth., taking advantage of cpDNA variation within one of the progenitors. In a coDNA tudy of the Tragopogon polyploid complex, Soltis & Soltis (1989) found that the cpDNA genomes of the two progenitors (7. dubium Scop. and T. pratensis L.) are each represented in different populations of their allotetraploid T. miscellus Ownbey. 2, The fact that two chloroplast genomes exist in current populations of hybrids and allopolyploid species suggests that paleopolyploids (ancient poly- ploids whose progenitors are extinct) would have a high probability of having similar cryptic varia- tion. Thus, measuring relationships with cpDN comparisons could give differing patterns of rela- tionship if two species were descended from tet- raploids of similar ancestry but carrying different chloroplast genomes. Figure 3 illustrates a set of events that would yield such a spurious phylogeny. While such a scenario might be rare, it may be a useful model to consider when chromosome num- bers suggest ancient polyploidy, as for example in the fern genus Osmunda (Wagner & Wagner, 980) LITERATURE CITED m f I. 1931. The genetics of ferns. Bib- liogr. 269-294. APPELS, * & n Dvorak. 1982. Relative rates of di- DRE a я and gene sequences within the rDNA odii: in Piu pies eae: implications for e prada e of hom ud of a е gene family. Theor. Appl. G 361- BARRINGTON, D.S. 1986. The rpg and eyo of Polystichum X potteri, hybr. . (= P. acros da x P. и Rhodora. 88: 297-313. R . R. WERTH. 1989. rea acra and d an concepts in the ferns. Amer. Fern J. 79: 5 CONDE, M. F., D. R. Princ & C. S. Levines Ш. 1979, Maternal is of organelle DNA’s in Zea mye E perennis reciprocal crosses. J. Heredity Td 1. - UT .. DOYLE. 1988. Natural interspecific hybridization in eastern North American Claytonia. Bot. 75 _ . E. SoLris & P. S. SoLTIS. 1985. An in- tergeneric hybrid in the E evidence from Amer. J. Bot. 72: 1388- ribosomal RNA genes. Duckett, J. G. & P. К. BELL. 1971. Studies on fer- tilization in archegoniate plants 1. Changes structure of the spermatozoids of Pteridium aguili- ге (L.) Kuhn during entry into the archegonium. Cytobiologie 4: 421-436. GILLHAM, N. W. 1978. Organelle Heredity. Raven Press, New York. ы V. 1977. EN Evolution. W. H. Free- n, San Francisc cca HTEL, W., 1980. Maternal des gh ei . J. Heredity 71: NA in ve Oenothera specie 191-19 HAUFLER, С. H. & D. E. Souris. 1986. ( suggests that homosporous ferns "od high chromo- ad. U.S.A. loid. Proc. Natl. Ac some numbers are di 83: 4389-439; Ни, К. W. Identification of the “A” genome of finger millet using chloroplast DNA. Genetics 118: 163-167. JORGENSEN, R. A E. CuELLAR, W. К. THOMPSON & TA ‚В. CAVANAGH. 1987. Structure and variation Volume 77, Number 2 1990 Stein & Barrington 339 Recurring Hybrid Formation in Polystichum x potteri in ribosomal RNA genes of pea. Pl. Molec. Biol. 8: 3-12 Kirk, J. Т. О. & К. А.Е. peck е 1978. Тһе Plastids. Elsevier, Amsterda KLEKOWSKI, E. J., Jr. & Н. С. e 1966. Evolu- tionary significance: : к їп da Pteridophyta. PALMER, J. D. 1986. elias and "ЖҮГҮ n of chloroplast DNA. Meth. Enzymol. 118: -186. & TEIN. 1982. Chloroplast DNA from the fern Osmunda cinnamomea: physical organi- , C. В. : . J. 0 1983, Chloroplast DNA evolátion and the origin of amphidiploid Brassica species. Theor. Appl. Genet. . 1986. Plant organelle genetics through somatic hybridization. Pp. 96-121 in Oxford Sur- P. veys of Plant Molecular and Cell Biology, Volume 3. Oxford Univ Press, London. SMITH, D. M. & D. A. Levin. 1963 манаа study of reticulate evolution in the Appalachian As- plenium complex ‚ J. Bot. 50 952-958 Ѕогтіѕ, D. E. € P. S. Sorr. 1989. Allopolyploid — des in Tragopogon: insights from chloroplast . Amer. J. Bot. 76: 1119-1124. STEIN, D. B. 1085. Nucleic acid comparisons аз a too in understanding species qu e and phy- logeny. Proc. Roy. Soc. Edinburgh 86B: 283-288. Tryon, R. M., JR. 1940. An Osmunda а. Атег. 8 e. ER . Reticulate evolution in the Appalachian aspleniums. Evolution 8: 103-118 F. S. Wac Ag 1980. Polyploidy i in pteri- dophytes. Pp 14 in W. H. Lewis (editor), Polyploidy: шу Relevance. Plenum Press, New York. 1. N. MILLER, JR. & D. Н. WAGNER. 1978. New Guertin: on the royal fern hybrid Osmunda КА. Rhodora 80: 92-106. WERTH, С. GUTTMAN & . ESHBAUGH. 1985a. A et evidence of reticulate evo- lution in the Appalachian Asplenium complex. Sys Bot. 10: 184-192. & 85b. Recurring origins of allopolyploid species in dues nium. Science 228: Wis, $. б. 1983. Polypeptide composition of ru- bisco as an aid in studies of plant phylogeny. Pp. 182-190 in U. Jensen & D. E. Fairbrothers (edi- tors), Proteins and Nucleic a in Plant System- atics. Springer- e Berlir Worr, P. G., С. Н. HAUFLER & E. SHEFFIELD. 1987. Electrophoretic e чеин ice for genetic diploidy in the bracken fern (Pteridium aquilinum). Science 236: -949. — ‚ 1. J. ODRZYKOSKI, A. STONEBURNER, Н. W. Bass & С. A. GALAU. 1988. Allopolyploidy in bryophytes: multiple origins of Plagiomnium medi- Acad. U.S.A. 85: 5601-5604. YATSKIEVYCH, G., D. B. STEIN & С. J. Gastony. 1988. Chloroplast DNA evolution and systematics of Phan- ини (Dryopteridaceae) and related fern gen- a. Proc. Natl. Acad. U.S.A. 85: 2589-2593. FELIX QUI POTUIT RERUM COGNOSCERE CAUSAS (VIRGIL). (Photo taken by Walter H. Hodge.) TURNERACEAE: NOVEDADES PARA LA GUAYANA VENEZOLANA! María Mercedes Arbo? RESUMEN En Venezuela viven los dos géneros sudamericanos de Turneraceae: Turnera y Piriqueta. Las colecciones botánicas realizadas en los últimos años han proporcionado especime eta cistoides subsp. caroliniana (Walter) Arbo. Se gcn un neotipo para P. undulata nueva combinación, Pirique nes de siete nuevas especies de Turnera. Se propone una Urban, porque el sintipo sobre el cual fue basado se destruyó en el herbario de Berlin ABSTRACT The botanic z lw a made in the Venezuelan Guayana during recent years have provided specimens representing seven new species of Turnera, described paruana, and T steyermarkii. Arbo. A in the Berlin herbarium here: T. annectens, rgen One new combination is proposed: Piriqueta cistoides subsp. caroliniana (Walter) A neotype is аы for Piriqueta undulata Urban, because the syntype on which it was based was destroyed T a beri, Т. tea, T. castilloi, T. cicatricosa, Turnera annectens Arbo, sp. nov. TIPO: Vene- zuela. Territorio Federal Amazonas: Cerro Si- papo (Paráque), frequent in savanna at 1,400 m, 1-3 Dec. 1948, Maguire & Politi 27482 (holótipo, NY; isótipo, P). Figura 1. Fruticulus 50 cm altus, ramis шкы жм striatis, ramis а strigoso-pilosis. Folia eolata, ellipti ica vel obovata, glabra, nectariis m Flores verisimiliter nium фт абы liberi 10-30 mm longi; pedicelli nulli; prophylla cordata vel late ovata, acuta; mm longus; corolla lutea, petalis ad intus pilosis; filamenta basi tubo 1.5-2 mm tota А adnata, et inter sese + tubuli instar con Semin obovata, 3.2-3.5 mm longa, puberula, асн ele. vatim, transversim obsolete densissimeque striata calyx 25-21 Arbusto 50 cm de alto, ramas pardo-negruzcas, cilindricas, marcada e irregularmente estriadas, algo lustrosas, glabras, entrenudos cortísimos, cicatrices oliares marcadas por estrias transversales profun- das, lenticelas oblongo-lanceoladas marcadas por estrias transversales, ramas jóvenes rojizas con pu- bescencia estrigosa. Hojas con 1-2 yemas seriales, estipulas diminutas, cónicas o triangulares, 0.2- 0.5 mm de largo, rojizas, pilosas; peciolo 1.5-4 mm de largo, estrigoso; lámina lanceolada, elíptica u obovada, 30-62 mm de largo, 10-26 mm de ancho, coriácea, glabra o con algunos pelos simples en las venas, algo lustrosa, ligeramente maculada, a neada, margen subentero o serrulado, dien- tecillos glandulosos, ápice agudo, a veces obtuso, venas ligeramente salientes en la haz, prominentes en el envés, venas secundarias alternas o subo- puestas, curvadas, Angulo de divergencia 40—50°, venas terciarias f. 1d frecuencia un angulo de 90° con la vena media, venación menor conspi- cua en el envés. Alabastro recto, mucrones libres en el ápice. Flores solitarias, presumiblemente he- terostilas (los especimenes estudiados presentaban ores brevistilas); pedúnculo 10-30 mm de largo, ligeramente estriado, estrigoso; pedicelo nulo; pro- filos 2, opuestos, dispuestos en la base del cáliz, ovados o cordiformes, 9-17 mm de largo, 7-12 mm de ancho, subglabros, base truncada, margen serrulado, ápice agudo o brevemente acuminado, sin apéndices basales; cáliz 25-27 mm de largo, tubo calicino 7.5-9 mm de largo, cilíndrico, con algunos pelos simples sobre las venas en la cara externa, glabro por dentro, lóbulos triangular-lan- ceolados, prefloración quincuncial, 3-5 nervados, ' Este trabajo fue realizado en gran parte en el Missouri Botanical ezco a ж curadores de los herbarios BM, BR, CTES, DS, F, ЕТ, С, СОЕТ, HBG, K, LIL, , US, VEN y W el préstamo del material estudiado; a la Dra. ntificar los tipos; y as Dr. William T. Stearn su guía para interpretar el material linneano. Las figuras Victor МагипаК y las figuras 2 y 3 runo Manara. Jessie Smith Noyes. Agrade , MY, NY, P, RB, nim para ide 4 fueron preparadas por Garden, con el apoyo de una beca de la fundación Alicia Lourteig su por a de Botánica del Nordeste, C.C. 209, 3400 Corrientes, Argentina ANN. Missouni Вот. GARD. 77: 340-352. 1990. Volume 77, Number 2 1990 Arbo 341 Turneraceae: Novedades para la Guayana Venezolana l | Al. Turnera annectens (Maguire & Politi 27482 Р).— А. Rama florifera.—B. Profilo. —C. Porción del FIGUR cáliz, cara externa. — D. Flor brevistila, gineceo. — E. Porción del tubo calicino, cara interna, con pétalos y estambres adnatos. —F. Cápsula con profilos. — C. Semillas con arilo. mucronados (0.7-1.3 mm); corola amarilla, 2-3 mm más larga que el cáliz, pétalos con la uña soldada al tubo calicino, lámina obovada, base cu- neada vellosa en la cara interna, ápice redondeado o brevemente acuminado; filamentos estaminales soldados en la base 1.5-2 mm al tubo calicino, complanado-subulados, glabros, 13-16 mm de lar- go en flores brevistilas, soldados entre sí por sus bordes a distintas alturas, + hasta la mitad, for- mando un tubo desgarrado en 2-3 porciones; an- 342 Annals of the Missouri Botanical Garden teras triangular-subuladas, 2-2.8 mm de largo, 0.5 mm de ancho, base levemente emarginada, muy brevemente apiculadas, rectas; ovario ovoide o có- nico, 2-2.5 mm de largo, 1.2-1.5 mm de ancho, glabro o con algunos pelos en el ápice, estilos 3, filiformes, pilosos en la porción media, 5 mm de largo en flores brevistilas, estigmas penicilados, 1 mm de largo. Cápsula elipsoide, 5 mm diam., valvas ovado-elipticas, cara externa pardo-rojiza, tuber- culada, glabra, algo lustrosa. Semilla obovoide, 3.2- 3.5 mm de largo, 1.4-1.7 mm de ancho, ligera- mente curvada, castaña, con estrias longitudinales bien marcadas y estrias transversales muy tenues, pubérula, hilo amplio cónico, rafe linear poco mar- cada, cálaza saliente, deprimida en el centro; arilo membranáceo, unilateral, rasgado, llegando hasta la mitad de la semilla. Material adicional estudiado. | VENEZUELA. TERRI- TORIO FEDERAL AMAZONAS: pid yn (Paráque), oc- aag in A Grande, mpo Grande, 1,500 l Jan. 1949, Maguire n Politi 28688 (P); Cerro сав East Bas sin Savannas, 1,900 m, 26-28 Jan. 1949, Maguire & Politi 28672 (NY) Esta especie, propia de la sabana, pertenece a la serie Stenodictyae Urban por tener el pedünculo floral libre, carecer de pedicelo y presentar el epis- perma longitudinalmente estriado. Se distingue fá- cilmente por la ausencia de nectarios foliares y por sus profilos ovados de gran tamano. Es la ünica especie del género que presenta los filamentos es- taminales unidos entre si a distintas alturas; este carácter la acerca a la serie Annulares Urban, cuyas especies tienen flores con los filamentos es- taminales más o menos connados en la base. Su nombre, annectens (ligando, conectando), alude a la adhesión de los filamentos estaminales que la convierte en un nexo con la serie Annulares. Turnera argentea Arbo, sp. nov. TIPO: Venezue- la. Territorio Federal Amazonas: Dep. Ata- bapo, sabanas y bosques al E del Cano Perro de Agua, afluente derecho (oriental) del Rio Orinoco, a unos 30 km al SE de la confluencia Orinoco- Ventuari, aprox. 03?47'N, 67°00'W, ca. 100 m, 30 Nov.-1 Dec. 1978, Huber & Tillett 2809 (holótipo, VEN; isótipos, CTES, US). Figura 2A- EF. ex | m altus, pilis crispulis et aliis antrorsis densis e dm Folia linearia vel anguste elliptica, obtusa, lon- gitudinaliter plicata, sericeo-velutina, margine integro res verisimiliter heterostyli; pedunculi 1.5-2 mm longi, a eine pedicelli nulli; calyx 7-7.7 mm longus, extus ceo-velutinus; tl flava, intus ad medium usque pilo: sa; filamenta glabra subulata, basi tubo 0.5 mm adnata. Semina breviter obovata, 1.6-2 mm lat longa, vix curvata, brunnea, reticulata. Arbusto 1 m de alto, tallo erecto, ramificado; corteza pardo-negruzca con estrias longitudinales marcadas y estrías transversales muy tenues, pelos crespos y otros largos antrorsos suberectos, muy densos, cicatrices foliares no prominentes. Esti- pulas subuladas, pilosas, 0.4 mm de largo. Yemas seriales 2, la basal florifera, la apical vegetativa. Peciolo semicilindrico; nectarios 2 en la unión de peciolo y lamina, 0.3-0.4 mm diám., parte central cubierta con membrana rojiza, borde amarillento, pubérulo; lámina foliar verde-plateada, linear o an- gustielíptica, 10-27 mm de largo, 2-3 mm de ancho, longitudinalmente plegada, base atenuada, ápice obtuso, borde entero, haz y envés ѕегісео- velutinos, 6-7 pares de venas secundarias apenas visibles sobre el envés. Flores presumiblemente he- terostilas (los ejemplares estudiados presentan flo- res longistilas), solitarias, amarillas, agrupadas en los extremos de las ramas; pedúnculo 1.5-2 mm de largo, soldado al peciolo; profilos 2 en la base del receptáculo, subulados, 1.5-2.5 mm de largo, 0.3 mm de ancho, cara externa indumento como las hojas, cara interna glabra; pedicelo no desa- e largo, tubo calicino 3.5-3.7 mm de largo, cara externa sericeo-velu- tina, cara interna vellosa, lóbulos triangulares, bor- des internos membranáceos, cara interna glabra, cara externa con indumento más largo que en el tubo; pétalos con la una soldada al tubo calicino, lámina 4-5 mm de largo, 1-1.5 mm de ancho, obovada, cara superior con pelos largos en la base y sobre la vena media hasta la mitad; filamentos estaminales subulados, glabros, soldados en la base 0.5 mm al tubo calicino, 4-4.7 mm de largo en flores longistilas, anteras ovadas 0.7 mm de largo, dorsifijas, base emarginada, ápice apiculado; ovario mm de largo, densamente piloso, placentas 3-ovuladas, estilos 3, cilindricos, glabros, 3.5-4.2 mm de largo en flores longistilas, estigmas 1.2- 1.5 mm de largo, penicilados, ca. 12 ramas. Cap- sula 2.5-3 mm de largo, valvas lisas por fuera, con pelos simples crespos y otros largos antrorsos, rrollado; cáliz 7- cara interna lisa, lustrosa, amarillenta con man- chitas oscuras irregulares. Semilla 1.6-2 mm de largo, 1.2 mm de ancho, ligeramente curvada, hilo cónico breve, cálaza saliente, oscura, cóncava, rafe linear, episperma pardo-negruzco reticulado, aréo- las pequeñitas cuadrangulares o transrectangula- res, muros longitudinales más notables que los transversales, arilo unilateral más corto que la se- milla, membranáceo y blanquecino en seco. Especie perteneciente a la serie Leiocarpae Ur- ban por sus pedúnculos florales soldados a los pe- ciolos, pedicelos no desarrollados, frutos lisos y Volume 77, Number 2 Arbo 343 Turneraceae: Novedades para la Guayana 1990 Venezolana St N FIGURA 2. Hábito. — B. Flor longistila, gineceo. —C. Po E. Semilla con arilo. F-K. 7: Auberi (F, Huber 641 VEN; G-K, Williams 12988 US). —F. Hábito. —G. Porción del tubo calicino, cara atea con pétalos y estambre е — Н. Flor brevistila, gineceo. — I. Porción del cáliz, N).—L. Hábito. — M. cara externa. —J. Profilo.—K. Semilla con arilo. L-Q. Turnera paruana (Huber 4386 VE Porción del cáliz, cara externa. — №. Porción del tubo calicino, cara interna, con pétalos y estambre adnatos. — O. Flor longistila, gineceo. — P. Profilo. —Q. Semilla соп arilo. 344 Annals of the Missouri Botanical Garden semillas reticuladas. Es afin a T. elliptica Urban de Goiás, Brasil, que se diferencia por sus hojas elípticas sésiles o subsésiles, 4-5 veces más anchas y aproximadamente de la misma longitud. Turnera castilloi Arbo, sp. nov. TIPO: Venezuela Territorio Federal Amazonas: Dep. Atures, uerto Ayacucho, bosque húmedo del Rio Ca- taniapo sector puente, margen izquierdo, 06?25'N, 67?25'W, 37 m, 14 Feb. 1983, Castillo 1582 (holotipo, CTES). Figura ex 1-3 m altus, pube simplici brevissima et aliis pilis curvato-erectis adspersus. Stipulae 0.6-0.9 mm lon- gae. Folia elliptica vel ovata, subtus juxta marginem 14- 20 nectariis orbicularibus minutis praedita; petiolus nec- tariis amplis marginibus glabris instructus. Flores hete- rostyli; pedunculi liberi 2-12 mm longi; pedicelli nulli; calyx 16-23 mm longus; petala aurantiaca, extus ad basin pilosa; styli pilis erectis hirsuti; filamenta glabra, basi tubo 0.6-1 mm facie externa adnata. Semina obovata, 2.7-3 mm longa, puberula, longitrorsum elevatim, transversim obsolete densissimeque striata Arbusto 1-3 m de alto, a veces ramas trepa- doras, corteza castaria, tenuemente estriada, ramas cilindricas, entrenudos de longitud variable, por- ciones con entrenudos cortos alternando con por- ciones con entrenudos largos, cicatrices foliares prominentes, ramas del ano con indumento blan- quecino, pelos cortisimos erectos y pelos cortos antrorsos. Estipulas triangular-subuladas, pilosas, 0.6-0.9 mm de largo. Yemas seriales 2, la basal florifera. Hojas coriáceas, lustrosas; peciolo cilin- drico, 3.5-10 mm de largo, indumento como el tallo; 1-3 pares de nectarios sésiles, circulares o elipticos, opuestos o alternos, 0.7-2 mm de largo, con un reborde amarillento, glabro, lustroso, de ancho irregular, ubicados en la unión de peciolo y lámina, los basales orientados lateralmente y los distales hacia el envés; lámina eliptica u ovada, 1.5-15 cm de largo, 0.6-4 cm de ancho, haz de color pardo-rojizo en seco, con máculas irregulares más claras, glabra o con algunos pelos simples sobre las venas, envés de color castaño, también con manchas irregulares más claras, algunos pelos sim- ples sobre las venas, con nectarios sésiles, circu- lares, 0.4-0.5 mm diám., ubicados cerca del mar- gen entre las venas secundarias y las escotaduras del borde; base cuneada o angostada, ápice acu- minado o agudo, margen aserrado-crenado, venas prominentes en ambas caras, más en el envés, venas secundarias alternas u opuestas, ángulo de divergencia aprox. 75°, venas terciarias formando frecuentemente un ángulo de 90? con la vena me- dia. Alabastros rectos con mucrones libres en el ápice. Flores solitarias, heterostilas, hojas tectrices a veces reducidas (15-20 mm de largo, 10-14 mm de ancho); pedünculo cilindrico, 2-12 mm de largo, indumento como el tallo; pedicelo nulo; pro- filos 2, opuestos, persistentes, situados en la base del cáliz, subulados, 2- de largo, 0.3-0.4 mm de ancho, pilosos, con 1-3 pares de apéndices basales laciniados; cáliz 16-23 mm de largo, tubo calicino cilindrico, 6—10 mm de largo, cara externa con pelos cortos antrorsos, cara interna glabra, lóbulos triangulares o angustiovados, bordes inter- nos membranáceos, ápice mucronado, mucrón 0.6- 0.9 mm de largo, corola rojo-anaranjada, 1.5-2 m más larga que el cáliz, pétalos con la uña soldada al tubo calicino, lámina obovada, 13-18 mm de largo, 3.5-5 mm de ancho, glabra por dentro, por fuera pilosa en la base; filamentos es- taminales complanado-subulados, glabros, 5-5.5 mm de largo en flores longistilas, 12 mm de largo en flores brevistilas, soldados en la base 0.6-1 mm al tubo calicino, anteras 3-5.5 mm de largo, dor- sifijas, filamento inserto a 1 mm de la base emar- ginada, ápice brevemente Du recto después de la dehiscencia; ovario ovoide, mm de largo, densamente estrigoso, estilos ES cilíndricos, erectos, hirsutos en la porción basal y media, 6 mm de largo en flores brevistilas, 10-11 mm de largo en flores longistilas, estigmas 1-2 mm de largo, amarillos, penicilados. Cápsula 5 mm de largo, valvas ovadas, ápice agudo, cara externa parda tuberculada, con pelos simples cortos an- trorsos, cara interna glabra, lustrosa, castano-ama- rillenta. Semilla obovoide, 2.7-3 mm de largo, mm de ancho, ligeramente curvada o rec- ta, episperma con estrias longitudinales bien mar- cadas y estrias transversales tenues, con pelos cor- tisimos crespos, cálaza prominente, deprimida en el centro, rafe linear, hilo cónico breve; arilo mem- branáceo, envolvente, rasgado, más desarrollado sobre la rafe, casi tan largo como la semilla. Material adicional estudiado. VENEZUELA. TERRI- Э в Aures, Puerto A cho, bosque húmedo del же e las Pavas y la Refor 06%25'N, 18 Feb. 1983, Castillo 1618 (CTES, о Puerto Aya- cucho, Samariapo, a 100 km del puente sobre el Río Cataniapo, i derecha cerca Caño Carinagua, 06?25'N, 67°25'W, 25 July 1981, Castillo 1338 (CTES); Puerto Ada ei hümedo del Río Cataniapo, comunidad de Las Pavas, 06?25'N, 67?25'W, 37 m, 6 Au Я Castillo 1525 (CTES); Puerto Ayacucho, bosque húmedo del Rio ng entre Raudal aay y comunidad de Las , 06?25'N, 67°25'W, 90-100 m, 13 Aug 1986, Castillo 2194 (CTES, MO). Turnera castilloi pertenece a la serie Stenodyc- tiae Urban; las especies más afines son 7. auran- tiaca Benth. que se diferencia por sus pedúnculos Volume 77, Number 2 Arbo 345 1990 Turneraceae: Novedades para la Guayana Venezolana FIGURA 3. Turnera castilloi, T. steyermarkii. A-F. T. castilloi (A, Castillo 2194; B-F, Castillo 1582 CTES).— A. Rama florifera. — В. Flor longistila, gineceo. — C. Porción del tubo calicino, cara interna, con estambre y pétalos adnatos.— D. Porción de cáliz y corola, cara externa. —E. Profilo.—F. Semilla con arilo. G-K. T. steyermarkii (Bunting et al. 3671 MY).—G. Rama florifera.—H. Profilo. —I. Flor brevistila, estambre.—J. Flor brevistila, gineceo. — K. Porción del cáliz, cara interna, con pétalos adnatos. 346 Annals of th Missouri еа Garden florales 9-37 mm de largo y sus estilos general- mente glabros; T. acuta Willd. ex Schultes con profilos lanceolados u ovado-lanceolados 2-9 mm de ancho y estipulas 0.3-0.6 mm de largo y T. macrophylla Urban que posee nectarios de re- borde piloso y cáliz corto 7-10 mm de largo. De- dico esta especie al Prof. Anibal Castillo quien colectó los ejemplares estudiados. Turnera cicatricosa Arbo, sp. nov. TIPO: Ven- ezuela. Territorio Federal Amazonas: Rio Ori- elow San Fernando de Ata- m, 11 May 1954, Level 64 (helótipo, P; isótipo, NY). Figura 4. Frutex 0.5-3 m altus, pilis crispulis brevissimis et aliis curvato-erectis intermixtis. iorum 1.5-3 mm longae. Folia subtus supra basin ad marginem nectariis hee glabris praedita. Flores ан pedunculi liberi 1- mm longi; pedicelli nulli; calyx 8-14 mm fien db flavida; filamenta puberula; styli pilis erectis hirsuti. Se- mina obovata, 2.7-3.2 mm longa, glabra геа elevatim, transversim obsolete densissimeque в striata Arbusto apoyante, 0.5-3 т de alto, ramas cilíndricas, pardo-negruzcas o pardo-rojizas, mar- cada e irregularmente estriadas, lustrosas, gla- brescentes, lenticelas suborbiculares u oblongo-lan- ceoladas, entrenudos de longitud variable, l- mm de largo, cicatrices foliares salientes, ramas jóvenes levemente estriadas, rojizas, pubescentes a tomentosas, pelos cortos, crespos, entremezclados con otros más largos, simples, antrorsos. Hojas con 2 yemas seriales, la basal florifera, la apical a veces desarrollada en ramita florifera; estipulas persis- tentes, рее subuladas, rojizas о negruzcas, pilosas, mm de largo; peciolo canaliculado, жш a tomentoso, 1-7 mm de largo; lámina elíptica u oblongo-lanceolada, a veces obovada, 10-85 mm de largo, 6-30 mm de ancho, cartácea, haz pardo-rojiza, a veces maculada, levemente es- triada en seco, con pelos simples curvado-antror- sos, envés pubescente, más densamente en las ve- nas, a veces glabro, base atenuada o brevemente atenuada, a veces ligeramente auriculada, gene- ralmente con un de nectarios amarillentos, con borde glabro, situados sobre el envés, discoideos en hojas nuevas, margen simple o doblemente ase- rrado-crenado, ápice agudo, obtuso o brevemente acuminado, venas ligeramente hundidas en la haz, prominentes en el envés, venas secundarias alter- nas a subopuestas, ángulo de divergencia 35-500, venas terciarias curvadas, con frecuencia dispues- tas en angulo recto con la vena media. Alabastros rectos, mucrones libres en el ápice. Flores solita- rias, heterostilas; pedúnculo cilíndrico, 1-5 mm de largo, rojizo; pedicelo nulo; profilos 2, opuestos, dispuestos en la base del cáliz, lanceolados, 7-10 mm de largo, 1.5-3 mm de ancho, pilosos, margen dentado, dientes glandulosos, ápice agudo o acu- minado, provistos de apéndices a divididos en 2-3 lacinias subuladas, formando en conjunto un pseudocaliculo persistente; cáliz 8-14 mm de largo, tubo calicino 4-7 mm de largo, piloso por fuera excepto en la base, velloso por dentro en la porción apical, lóbulos angusti-triangulares, preflo- ración quincuncial, trinervados, pilosos en la cara externa y glabros en la interna, mucrón 0.5-0.7 mm de largo; corola amarilla, a veces blanquecina, 3-4. mm таз larga que el cáliz, pétalos con la uña soldada al tubo calicino, lámina obovada, 7-12 mm de largo, 3-4.5 mm de ancho, haz pilosa en la base, ápice redondeado o brevemente apiculado; filamentos estaminales soldados en la base 1-1.5 mm al tubo calicino, complanado-subulados, 7-11 mm de largo en flores brevistilas, 6-7 mm de largo en flores longistilas, pubérulos, anteras dorsifijas, eliptico-lanceoladas, 3-4 mm de largo, 0.7 mm de ancho, base emarginada, brevemente apiculadas, rectas o curvadas después de la dehiscencia; ovario ovoide, 1.5-1.8 mm de largo, 1-1.3 mm de ancho, densamente estrigoso, ca. 20-ovulado, estilos 3, filiformes, hirsutos, 12 mm de largo en flores lon- gistilas, 3-5 mm de largo en flores brevistilas, estigmas penicilados 1-1.5 mm de largo. Cápsula ovoide o globosa, 3.5-5.5 mm de largo, 3-4.5 mm de ancho, valvas ovadas, cara externa pardo-ama- rillenta, tuberculada, pilosa, cara interna glabra, lustrosa, amarillenta. Semilla obovoide, ‚2 mm de largo, 1.2— tana, con estrias longitudinales bien marcadas y estrias transversales muy tenues, glabra, hilo có- nico, rafe linear apenas marcada, cálaza saliente deprimida en el centro, arilo membranáceo, uni- lateral, rasgado, tanto o más largo que la semilla. mm de ancho, curvada, cas- Material adicional estudiado. VENEZUELA. BOLÍ- VAR: Río Caroní, cerca de la boca de Tirika (Guayana), 400 ш, May 1945, Cardona 1169 (US); en playas de a Maripa, Alto Caroní, 370 m, eb. 1953, Car- nd 2830 pau) zx l1 Cerro Kurün-tepui, aprox. ri эк а 4'N, 62°43'W, 1,050 m, 31 . 1983, Huber ie Е d (CTES); Distr. Piar, vi- id of ош. снн Rio Acanán, affluent of Rio Carrao, W of Cerro Los Hermanos, 05%56'N, 62°17'W, vm 3 Ma 1986, Stevermar et al. 131899 (MO). GUYANA. ESSEQUIBO: Pakar aaa qe Im- pere im aee upper Mazaruni Biver r, 550 m, 21 Oct. 1 кле 32173 (NY, P). Б Esta especie ha sido coleccionada en Venezuela y Guyana, en selvas marginales y sabanas. Es muy afin a Т. acuta Willd. ex Schultes, que se diferencia por los profilos anchos y las estipulas cortas 0.3- 0.6 mm de largo, y a Т. macrophylla Urban que Volume 77, Number 2 1990 Arbo 347 Turneraceae: Novedades para la Guayana Venezolana С ro * - OM TEC, e. e aes: Sst oe 301704 pr f Turnera cicatricosa. —(A, Maguire 32173 P; B-H, Level 64 P).— А. Rama florifera. — B. Detalle Profil IGURA 4. del envés foliar mostrando los nectarios.— C. Profilo. — D. Porción del cáliz, cara externa. — E. Porción del tubo calicino, cara interna, con estambre y pétalos adnatos. — С. Cápsula con profilos. — gineceo. — К. H. Semilla con arilo. se distingue por sus nectarios foliares con borde piloso y sus frutos grandes. T. castilloi, tambien afin, presenta 14-20 nectarios diminutos en el envés, cerca del margen foliar. Turnera huberi Arbo, sp. nov. TIPO: Venezuela. Territorio Federal Amazonas: Estación de Pis- cicultura de Puerto Ayacucho, entre la ca- rretera hacia Samariapo y la pista de aterrizaje de Puerto Ayacucho, aprox. 'N 67°36'W, aprox. 75 m, 16 Apr. 1977, Huber 641 (holótipo, VEN). Figura 2F-K. Herba vel fruticulus 13-40 cm altus, pilis crispulis brevibus et aliis curvato-erectis indutus. Folia linearia vel anguste elliptica, 8-23-plo longiora quam latiora, acuta, margine plano denticulato. Flores heterostyli; pedunculi 3-6 mm longi, 0.5-4 mm petiolo adnati; pedicelli nulli; calyx 7- mm longus, extus densissimeque pilosus; co- rolla flava; filamenta glabra, subulata, basi tubo 0.5-0.6 Flor brevistila, mm tota facie adnata; styli glabri. Semina breviter obo- vata, - mm longa, vix curvata, brunnea, reticulata. Hierba o sufrütice 13-40 cm de alto, base le- nosa, con xilopodio, tallos 1 a varios, 0.8-1 mm diám., ramas viejas con corteza rojiza, longitudi- nalmente estriada, ramas nuevas cilindricas, con pelos simples crespos y otros más largos, antrorsos, muy densos hacia los ápices. Hojas con 2-3 pares de estipulas 0.1—0.3 mm de largo, cónicas, rojizas; yemas seriales 2, la basal florifera; peciolo 1.5-3 mm de largo, semicilindrico, indumento como el tallo; nectarios 2 en la unión de peciolo y lámina o en la base de la lámina foliar sobre el envés, circulares o elipticos, 0.5-0.7 mm de largo, la parte central con una membrana provista de un poro, el borde pubérulo o en parte glabro; lámina foliar linear o angustieliptica, base cuneada o ate- nuada, 17-57 mm de largo, 2-5 mm de ancho, relación largo : ancho = 8-23: 1, ápice agudo, 348 Annals of the Missouri Botanical Garden borde serrulado excepto en la base, haz punteada con algunos pelos simples especialmente sobre las venas y cerca del borde, vena principal prominente, 6-7 pares de venas laterales a veces ligeramente salientes, ángulo de divergencia muy agudo, envés con venas más prominentes, pelos simples y otros glandulares capitados sésiles; venación menor in- conspicua. Flores solitarias, heterostilas; pedúnculo floral 3-6 mm de largo, soldado al peciolo, porción apical libre; profilos 2, subulados, 2-6 mm de largo, 0.4-0.6 mm de ancho, borde entero o se a pilosos en ambas caras; pedicelo no de- arrollado; cáliz 7—10 mm de largo, tubo calicino -5 mm de largo, cara interna con pelos blandos, cortos, por fuera con pelos cortos crespos, lóbulos triangulares o angustiovados, mucrón 0.5 mm de largo, cara interna glabra, cara externa con pelos simples crespos y pelos muy largos sobre las venas, — bordes internos membranáceos; pétalos con la una soldada al tubo calicino, lámina obovada angosta, 5-7 mm de largo, 1.5-2.3 mm de ancho, glabra o con algunos pelos en la base y sobre la vena media; filamentos estaminales soldados en la base 0.5-0.6 mm al tubo calicino, 5-9.7 mm de largo en flores brevistilas, 4.5 mm de largo en flores longistilas, anteras angustiovadas o elípticas, 0.7— 1.2 mm de largo, dorsifijas; ovario cónico 1-1.5 mm de largo, ovuladas, estilos 3, cilindricos, glabros, 3-5 mm densamente piloso, placentas 3-4 de largo en flores brevistilas, 5 mm de largo en flores longistilas, estigmas 1.7-2 mm de largo, penicilados, ca. 12 ramas. Cápsulas 2-4 mm diám., valvas ovadas, por fuera lisas, pilosas, castañas o amarillentas, por dentro lustrosas, jaspeadas. Se- millas obovoides, 1.8-2.2 mm de largo, 1-1.2 mm de ancho, ligeramente curvadas, hilo cónico breve, cálaza pigmentada, ligeramente saliente, general- nte cóncava, rafe linear, episperma reticulado, pardo, aréolas cuadrangulares o rectangulares; ari- lo unilateral casi tan largo como la semilla o lige- ramente mayor, blanquecino en seco. Material adicional Nr VENEZUELA. TERRI- TORIO FEDERAL : Estación de des ultura de Puerto Ayacucho, entre E carretera hac a pista de pisi de Puerto Ayacucho, aprox. О 7°36'W, aprox. 75 m, 19 Apr. 1977, Huber 693 (VEN); бечет е саш po Florida, ep са каа hacia El Si- papo carretera Puerto abd — 22 Mar. 1979, Trujillo & id 15152 ( ); Puerto Ayacucho, 100 m, 18 May 1940, Williams 12988 (US), 12089 (VEN). En esta especie, perteneciente a la serie Leio- carpae Urban, la heterostilia está poco diferencia- da, pues especialmente en los ejemplares brevisti- los, anteras y estigmas están parcialmente en contacto. La diferencia en longitud hallada entre androceo y gineceo es de 1-2 mm en flores lon- gistilas y de 0.7—2 mm en flores brevistilas. Especie afin a Т. lanceolata Cambess. del centro de Brasil, que se diferencia por su tallo aristado y por sus hojas con peciolos cortos y venación conspicua en ambas caras. También está emparentada con 7. argentea, que se diferencia por sus hojas sericeas, obtusas, y de borde entero. Dedico esta especie al Dr. Otto Huber quien coleccionó el ejemplar tipo y numerosos especimenes de Turneráceas. Turnera paruana Arbo, sp. nov. TIPO: Venezue- la. Territorio Federal Amazonas: Dep. Ata- bapo, Serranía del Parú (Aroko), sobre colinas onduladas en el sector centro- nororiental de la serranía, al S del Rio Parú, aprox. 04°33'N, 65°31'W, 710 m, 5-6 Oct. 1979, Huber 4386 (holótipo, VEN). Figura 2L-0 sabanas Frutex 50 cm altus, pilis crispulis brevibus et aliis curvato-erectis praeditus .2-2.5 mm longi aati pedicelli nulli; calyx 7.5 mm longus, extus glaber; corolla flava; filamenta glabra, PH d basi it >. 3 mm tota facie adnata; semina brev obo- vata, 1.8 mm longa, vix curvata, brunnea, кш = Arbusto 50 cm de alto, ramificado, tallo cilin- drico, parte basal con corteza rojiza con grietas longitudinales, cicatrices foliares no salientes, pelos crespos cortos у pelos largos antrorsos о erectos. Hojas obscuras en seco, en vivo haz verde mediano, enves verde claro; estipulas 0.5 mm de largo, su- buladas; yemas seriales 2, la basal florifera, la apical vegetativa; peciolo semicilindrico 2-4 mm de largo; lamina elíptica 15-30 mm de largo, 6- ancho = 2.5- 3: 1, base brevemente atenuada, ápice agudo, bor- de piloso plano, serrulado, dientes agudos termi- nados en una emergencia subulada, haz lustrosa, pardo-rojiza, glabra excepto algunos pelos sobre la vena media y cerca del borde, 6-9 pares de venas 11 mm de ancho, relación largo : laterales apenas salientes, envés castaño, vena prin- cipal y secundarias prominentes, con pelos largos, simples, antrorsos, ángulo de divergencia 45—50°, venas terciarias visibles, no prominentes, venación menor inconspicua; nectarios 2, situados en la base de la lámina, sobre el envés, elípticos, 0.6-0.9 mm de largo, parte central cubierta con una membrana blanquecina, borde rojizo glabro. Flores presumi- blemente heterostilas (el ejemplar estudiado presen- ta flores longistilas), agrupadas en racimos apicales hojosos, flores basales alejadas; hojas floriferas api- cales bracteiformes, obovadas и romboidales, 7- Volume 77, Number 2 1990 Arbo 349 Turneraceae: Novedades para la Guayana Venezolana 12 mm de largo, 3-5 mm de ancho; pedúnculo floral totalmente soldado al peciolo, 1.2-2.5 mm de largo; profilos 2.5-3.5 mm de largo, 0.3-0.4 mm de ancho, subulados, rojizos, uninervados, con pelos largos y simples en los bordes y en la cara externa, insertos en la base del receptáculo; pe- dicelo ausente; cáliz 7.5 mm de largo, tubo calicino mm de largo, glabro por fuera en la base, con algunos pelos cortos sobre las venas en la parte superior, por dentro velloso en la porción media, lóbulos triangulares, glabros por dentro, por fuera con pelos simples largos especialmente. sobre las venas; corola amarilla, pétalos con la una soldada al tubo calicino, lámina con algunos pelos en la base, en la cara interna; filamentos estaminales subulados, glabros, soldados en la base 0.3 mm al tubo calicino, 5 mm de largo en flores longistilas; anteras ovadas, 1 mm de largo, 0.6 mm de ancho, dorsifijas, base emarginada, ápice agudo; ovario cónico, 0.7 mm de largo, densamente piloso, pla- centas 3—4 ovuladas, estilos 3, cilíndricos, glabros, 4.5 mm de largo en flores longistilas; estigma 1.7 mm de largo, penicilado, ca. 10 ramas. Cápsula ovoide, 2-3 mm diám., valvas ovadas, lisas, pilosas por fuera, por dentro lustrosas, castaño claro, vena placentaria prominente. Semilla elipsoide, ligera- mente curvada, 1.8 mm de largo, 1.2 mm de ancho, hilo cónico, pequeño, rafe linear, cálaza no prominente, ligeramente cóncava, episperma re- ticulado, aréolas + hexagonales; arilo unilateral, ovado, casi tan largo como la semilla, borde eroso, membranáceo y blanquecino en seco. Esta especie pertenece a la serie Leiocarpae Urban por tener pedúnculo floral soldado al pecíolo, pedicelo no desarrollado, estambres soldados en la base al tubo calicino, y frutos lisos. La diferencia en longitud observada entre androceo y gineceo es e 1.7 mm. Es afin a T. lineata Urban que se diferencia por tener hojas floriferas no bracteifor- mes, nectarios con borde piloso, y tubo calicino completamente piloso. También es afin a lochioides Cambess. var. arenaria Urban que se diferencia por sus ramas estigmáticas pilosas en la base y por sus semillas más grandes y curvadas. . те- Turnera steyermarkii Arbo, sp. nov. TIPO: Ven- ezuela. Territorio Federal Amazonas: Dep. Ca- siquiare, Río Temi, a lo largo de la orilla de- recha del río, aprox. 1-2 horas abajo de Yavita, en sabanita, 100-140 m, 6-19 July 1969, Bunting, Akkermans & Van Rooden 3671 (holotipo, MY). Figura 3G-K. Frutex ramosus 1-1.5 m altus. Folia elliptica vel an- guste elliptica. dun: vel кыш. supra glabra, subtus strigillosa, кане obsolete remoteque crenulato, nerv medio su mpresso strigoso, lateralibus ascendentibus sub Rn 38- 60? abeuntibus. Flores verisimiliter hete- rostyli, pedunculi subnulli, pedicelli nulli, prophylla sub- spathulata, calyx 2.1-3 mm lo “р filamenta pilosa, an- therae breves conectivo to lingulato-apiculato pi- loso. Semina longitudinaliter striata, dense pilosa. Arbusto muy ramificado 1-1.5 m de alto, ramas con corteza pardo-negruzca, obscuramente plega- do-estriadas, porciones con entrenudos muy cortos alternando con porciones con entrenudos bien de- sarrollados, cicatrices foliares no salientes; ramas del año estrigosas con pelos muy cortos, simples, adpresos. Hojas con estipulas 0.3 mm de largo, triangulares; yemas seriales no observadas; peciolo de sección aprox. triangular, aplanado en la cara adaxial, 2-4 mm de largo, estrigoso; lámina co- riácea, elíptica o angustielíptica, 40-78 mm de largo, 8-19 mm de ancho, relación largo : ancho = 3-6: l, base cuneada o brevemente atenuada, ápice obtuso o subagudo, borde plano o ligeramente revoluto, obscuramente crenulado, con escotaduras remotas, inconspicuas, con 2-3 pares de nectarios diminutos ubicados en las escotaduras basales, haz glabra, verde oscura en vivo, pardo-grisácea en seco, vena media hundida y estrigosa, 8-11 pares de venas laterales apenas visibles o salientes, ángulo de divergencia más frecuente 52? (3 venas terciarias a veces un poco prominentes, envés ver- de claro en vivo, pardo-rojizo en seco, estrigiloso, pelos cortos estrictamente antrorsos, vena media muy prominente, estrigosa, venas laterales salien- tes, venación menor conspicua o no. Flores pre- sumihlemente heterostilas (los ejemplares estudia- ) hrevistilas деп esnigas hikes 10-20 floras, axilares o terminales; brác- teas tectrices coriáceas, peciolo 0.7-1 mm de lar- go, estrigoso, lámina eliptica, 1.2-2 mm de largo, m de ancho, ápice agudo, borde entero o denticulado, haz lustrosa, parda, estrigilosa, envés densamente dorado-estrigoso con vena media sa- liente; pedúnculo nulo o subnulo; profilos 2, insertos en la base del cáliz, subespatulados, 1.5 mm de largo, 0.5 mm de ancho, ápice agudo, cara interna glabra en la base, estrigilosa en el ápice, envés densamente dorado-estrigoso; cáliz 2.1-3 mm de largo, tubo calicino 1 mm de largo, dorado-estri- goso por fuera, cara interna glabra o con algunos pelos en la garganta; corola igual que el cáliz o .5 mm más larga, pétalos con la uña soldada al tubo calicino, lámina obovada, 1.2-1.5 mm de largo, 0.5 mm de ancho, base cuneada, pilosa por dentro, ápice agudo; filamentos estaminales 1.6- 1.8 mm de largo en flores brevistilas, subulados, pilosos en ambas caras, anteras 0.5 mm de largo, 350 Annals of the Missouri Botanical Garden curvadas a la dehiscencia, base emarginada pilosa, conectivo terminado en un apiculo subulado 0.3 mm de largo, piloso; ovario ovoide, piloso, 0.7- 0.8 mm de largo, placentas uniovuladas, estilos 3, cilíndricos, 0.6-0.8 mm de largo en flores brevisti- las, hirsutos, terminados en estigmas apenas divi- didos. Capsula elipsoide 4 mm de largo, valvas angustielipticas, tuberculadas y estrigosas por fue- ra, por dentro glabras, rugosas, jaspeadas. Semilla obovoide, 3 mm de largo, 1.8 mm de ancho, lon- gitudinalmente estriada, densamente pilosa, calaza saliente, concava, hilo conico, rafe linear; arilo unilateral membranaceo, mas corto que la semilla. Material adic a estudiado. BRASIL. AMAZONAS: Rio ca г egro, acima da cachoeira, praia baixa perto de Boca d Capanary, 26 Feb. 1936, Ducke ). Turnera steyermarkii es afin а Т. venosa Ur- ban, que se diferencia por tener hojas agudas y venas laterales con ángulo de divergencia ca. 70. Urban ubicó esta última especie en la serie Ste- nodictyae con dudas, encontrándola afin también a T. glaziovii Urban. Turnera steyermarkii y T. venosa presentan filamentos estaminales pubérulos y anteras pilosas como 7. serrata Vell.; flores sésiles reunidas en inflorescencias axilares o terminales, y tienen semillas pilosas como 7. bra- siliensis Willd. ex Schultes y T. glaziovii Urban. omo todas estas especies pertenecen a la serie Salicifoliae Urban, considero que 7. venosa у Т. steyermarkii deben ubicarse en dicha serie y no en la serie Stenodictyae Urban. Dedico esta especie al Dr. Julian A. Steyermark, quien estudio y anoto el ejemplar tipo, marcando sus diferencias con la especie afin, 7. venosa Ur- an. tienen Piriqueta cistoides (L.) Griseb., Fl. Brit. W. I. 298. 1860 Piriqueta cistoides var. macrantha Urban, Jahrb. Ko- Gart. Berlin 2 . 1883. Tipo: Brasil. : Martius s.n. (holótipo Ten ои а уаг. exasperata Urban, Jahrb. Konigl. Bot. Gart. Berlin 2: 73. 1883. TIPO: Cuba: Wright 2609, plantas 2, 4 y 5 (lectótipo (aquí designado), 5). Urban (1883) describió P. cistoides y P. car- oliniana con los siguientes caracteres: P. cistoides es anual, con flores homostilas, cáliz 4-9 mm de largo, y corola igual o hasta el doble más larga que el cáliz; P. caroliniana es perenne, con flores he- terostilas, cáliz 6-12 mm de largo, y corola el oble o más larga que el cáliz. El aspecto vegetativo de ambos táxones es muy similar, con la misma variación en tamaño, forma e indumento de hojas, porte y ramificación; su distribución geográfica cu- bre prácticamente toda el área del género, desde el sur de E.E. U.U. hasta el norte de Argentina. Los conjuntos de caracteres usados por Urban se encuentran netamente definidos en los extremos del área que ocupan ambos táxones, así en los ejemplares de E.E. U.U. se reconocen los carac- teres de P. caroliniana, y en los ejemplares de Paraguay y Argentina se reconocen los caracteres de P. cistoides. Sin embargo, en el resto del área, desde el Caribe hasta Mato Grosso (Brasil) se en- cuentran ejemplares de una u otra entidad, y otros cuyos caracteres son intermedios. Ya Urban tro- pezó con estos ejemplares y describió numerosas variedades; en el ejemplar Wright 2609 determinó las distintas plantas como P. cistoides “var. ge- nuina,” P. caroliniana var. glabra y P. carolin- iana var. exasperata. Hay ejemplares de P. car- oliniana var. integrifolia y de P. cistoides var. genuina que son idénticos excepto en la longitud de androceo y gineceo, y lo mismo ocurre entre P. caroliniana var. caroliniana y P. cistoides var. latifolia. Después de estudiar todo el material disponible (470 ejemplares de numerosos herbarios) y de me- dir los caracteres cuantitativos, he llegado a la conclusión de que se trata de una sola especie con dos subespecies que muy probablemente se hibri- dizan en las áreas donde conviven. La mayor parte del material estudiado puede identificarse a subes- pecie, pero hay ejemplares dudosos, ya sea por su variabilidad o porque presentan caracteres inter- medios. Dichos ejemplares, que incluyen algunos tipos, se citan después de las subespecies; los si- nónimos citados más arriba son los no identificados a nivel de subespecie. combinación do qs cistoides Meyer ex St ШО. Nomencl. Bot. 4. 1824, usada рог Urban (1883) está basada en Turnera cistoides Hort. (non L.), nomen nudum. Piriqueta cistoides (L.) Griseb. subsp. cistoi- es. Turnera cistoides L., Sp. Pl. ed. 2, 1: 387. 1762. P. cistoides var. genuina Urban, Bur Konigl. Bot. Gart. Berlin 2: 74. 1883. TIPO: (holótipo, Linn. Herb. 384. 6). 7 villosa Aublet, Hist. pl. Guiane 1: 298, t. 117. 1775. о var. latifolia Urban, Jahrb. Ko- nigl. ce t. Berlin 2: 74 3. TIPO: Caienne, (holótipo, BM. isótipo, Herb. J. J. Rousseau, Turnera aspera Poiret in Lam., Encycl. 8: 141. TIPO: Caienne, Poiret herb. (holótipo , P). Turnera hirta Willd. ex Schultes in Roemer et Schultes, Syst. veg. 6: 678. 1820. TIPO: Herb. Willdenow (holótipo, B n.v., microficha estudiada). Volume 77, Number 2 1990 Arbo 351 Turneraceae: Novedades para la Guayana Venezolana iios tomentosa Kunth in H. B. & K., Nov. gen. . 6. 1823. TIPO: Missiones del Orinoco, inter Atu- et Maypures, May, Humboldt & Bonpland s.n. (hold tipo, P). Turnera Кабо P г Cambess. in Saint Hilaire А., £e 2 сэр : 9,0 £ Ф un B p Paran nahyba, Saint Hilaire s.n. (holótipo, Р; isótipo, Piriqueta n) Benth., Hook. J. Bot. 4: 117. 1842. Guyana: banks of the Rupunoony, Schom- rans 137 (биро К; isotipos, BM, FI, Р). a lorr Garcke, Linnaea 27: 63. 1849. Piri- cistoides var. foliosa ( [тые Urban, Jahrb. Kön igl. Bot. Gart. Berlin 2 . TIPO: Suri- name: ad vias prope plant. б. Kegel 110 (ho- o, СОЕТ; isótipo, Р). Pirque citoides var. Ld Urban, Jahrb. Kö- . Bot. Gart. Berlin 2: 74. 1883. TIPO: Suriname: Pine 265 (lectótipo d designado), K; iso- ctótipos, FI, P, W). Piriqueta cistoides var. ao Urban, Jahrb. Kó- . Gart. Berlin 2: 74. 3. TIPO: Jamaica: Wu lls "o egel 844 (lectótipo pus designado), M; isolectótipos, GOET, W). Todos los ejemplares estudiados de México y Centroamérica y del centro-sureste de Brasil (Mi- nas Gerais, Rio de Janeiro, Sao Paulo, y Parana), Paraguay, y norte de Argentina (Formosa) corre- sponden a esta subespecie. Vive ademas en nu- merosas islas del Caribe, Venezuela, Trinidad, las Guayanas, Brasil (Amazonas hasta Mato Grosso do Sul), y Bolivia. Son plantas generalmente anuales, con raices típicas cortas y ramificadas; flores de 4-11 mm de largo, relación en longitud corola : cáliz = 1.5: 1; son homostilas, pero en muchos casos el andro- ceo puede ser hasta 1.5 mm más largo que el gineceo o, a la inversa, el gineceo puede ser hasta 1 mm más largo que el androceo. Se la ha coleccionado a alturas diversas, desde el nivel del mar hasta los 1,100 m en Cuba Material ma examina MÉXICO. TA- 839, Dm 62 (MO, P, 1929, Lundell 535 (DS, F, К, O). NICARAGUA. ZELAYA: Puerto 5m ‚3 July 1980, Stepes 17799 (CTES, MO). poing ISLA DE PINOS: near Nueva ; May 1904, Curtiss 496 (F, MO, P, US). Dist. Bayamón, E o, Bo. Sabana, 9 Mar. 1938, Otero 521 (F, M ). COLOMBIA. VALLE: Piani del Valle, entre Palmira a y Cali, 25 Apr. 1963, López Figueiras ж (US). VENEZUELA. BOLÍVAR: Dist. Piar, Paviche, NW de El Manteco, 215 m, márgenes del lago de Guri, July 1978, Delascio & Liesner 7165 (MO, VEN). TRINIDAD: St. Augustine, 23 June 1926, Broad- YT way 6316 (F, MO, S, US). SURINAME: Kwakoegron, 19 Oct. 1944, Maguire 25006 (F, MO, NY). BRASIL. MINAS GERAIS: na margem do Paranahyba, July 1892, Ule 163 (НВС, P, RB). BOLIVIA. BENI: Prov. Vaca Diez, Guaya- ramerin, 65%22'W, 10?49'S, 18 Apr. 1979, Krapovickas & Schinini 35096 (CTES). PARAGUAY. AMAMBAY: Sierra de Amambay in campis humidis Esperanza, Nov. 1907, Hassler & Rojas 10710 (BM, C, K, LIL, LY, NY, P, S, W). ARGENTINA. FORMOSA: Clorinda a Pilcomayo, km 6, ruta 11, 21 Oct. 1946, Morel 1496 (CTES, LIL). Piriqueta cistoides subsp. caroliniana (Wal- г) Arbo, comb. nov. Waltheria caroliniana Walter, Fl. Carol. 175. 1788. Turnera car- oliniana (Walter) S. Watson, Smithsonian Misc. Collect. 15: 391. 1878. Piriqueta car- oliniana (Walter) Urban, Jahrb. Konigl. Bot. Gart. Berlin 2: 1883. TIPO: Е.Е. U.U. South Carolina: ай: ripas fluvii Santee, n.v. Turnera TO Willd. ex Schultes., Syst. veg. 6: 679. 0. Piriqueta caroliniana var. integrifolia t ex Schultes) Urban, Jahrb. Koónigl. Bot. Gart. Berlin 2: 72. 3. TIPO: Herb. Willdenow 6089 (holótipo, В n.v., microficha estudiada). Turnera glabra DC, Pro dr. 3: 347. 1828. Piriqueta glabra (DC) Griseb., Cat. pl. Cub. 285. 1866, non Chapman. TIPO: Santo Domingo, Bertero s.n. (ho- lótipo, DC herb. n. v.; isótipos, M, MO, W). phu. fulva Chapman, Fl. South. U.S. 146. 1860. TIPO: E.E. U.U. Florida: South Florida, Chapman 1. (holótipo, NY; isótipos, MO, Р). ía glabra Cha apman, Fl. South. U.S. 147. 1860. TIPO: E.E. U.U. Florida: South Florida, Chapman 1. (holótipo, NY). T MERE tomentosa Alph. Wood, Brun. Bot. Fl. 129. 1874. Tipo: Е.Е. U.U. Florida Piriqueta t tomentosa auct. non Kunth. Piriqueta caroliniana var. — Urban, Jahrb. Ko- Gart. Berlin 2: 73. 1883. TIPO: Brasil. Bahía: Serra Jacobina, Blanchet 2708 (lectótipo aquí designado), P; isolectótipos, BR, NY, Piriqueta cistoides var. ramosissima nigl. Bot. jak Berlin 2: 74. 1883. Goiás: between Funil et Sao Joao, Burchell 9 (lectotipo, Р. ео К, NY, Р). е жылу и Small, Fl. s.e . 794, 1335. 1903. ›: Е.К. U.U. чы : vicinity of Eustis, Lake Co., h 830a (holótipo, NY). Piriqueta tracyi Gand., ull. Soc. Bot. France 45: 27. 1918. TIPO: E.E. U. U. Florida: Sanibel Island, 17 May 1901, Tracy 7459 (holotipo, LY; isotipo, F). E 16-31 May 1894, Todos los ejemplares de E.E. U.U. corresponden a esta subespecie: son plantas generalmente pe- ennes, con órganos subterráneos ligeramente en- grosados, a veces retorcidos en tirabuzón; las flores son grandes, 10-20 mm de largo, corola : cáliz = 2: 1; en flores brevistilas el andro- ceo es 2-5 mm más largo que el gineceo, y en flores longistilas el gineceo es 1.5-5 mm mayor; las anteras son ligeramente más grandes que en la subsp. cistoides, y los estigmas están mejor de- y la relación Annals of the Missouri Botanical Garden sarrollados. Vive además en Venezuela, Colombia, Brasil (Piaui y Bahia hasta Mato Grosso do Sul), Bolivia, y las islas del Caribe. Se la ha coleccionado en lugares bajos, a nivel del mar, y hasta los 800 m en Goiás, Brasil. l баеса А examinado. Е.Е. .U. SOUTH CAROLINA: Bluffton, Beaufort Dist., 1884, Melli- champ s.n. (F, K, MO, fà "US ). ys Pinar del Río city, at Laguna del Hun 30 Oct. 1923, Ekman 17860 » a Re к DOMINICA ANA. т between Bayagua and Guerra, 8 Nov. 1 , Howard 9902 (S, US). COLOMBIA. ES San uen del Guaviare, 240 m, 11 Nov. 1939, Cuatrecasas 7645 (F). VENEZUELA. BOLÍVAR: Bajo Caura, sabana Guayapo, 1 , 2 May 1939, Williams 12025 (F, S, VEN) BRASIL. BAHÍA: basin on Sao Francisco river, ca. 26 km of Bom Jesus da Lapa the road to Calderao о, ca. 500 m, 43°13'W, 1309'S, 17 Apr. 1980, ш et al. 21443 (CTES). MATO Grosso: W of km 229, vantina- Ca- chimbo road, 20 Dec. 1967, Philcox et aL 3628 (K, MO, P, RB) MATERIAL DUDOSO ejemplares dudosos de Cuba, Repüblica Do- minicana, Venezuela, Guayanas, Perü, y Brasil (Amazonas a Mato Grosso do Sul). Este material representa el 10% del total estudiado: Material representativo examinado. CUBA. PINAR DEL RÍO: San Gabriel to Pinal de La Catalina, 18 Jan. 1912, Shafer 1844 (F, MO). REPÚBLICA DOMINICANA. Pacificador: Pimentel, 20-25 Jan. 1921, Abbott 678 (US). VENEZUELA. TERRITORIO FEDERAL AMAZONAS: be- tween o and Esmeralda Ridge, in Esmeralda savana, r Orinoco kv 150 a 21 ns 1944, Schomburgk 189 (BR, F, K, P). SURINAME: ле 8 May 1961, Hekking 794 (BR, С, MO, NY, Р, VEN). PERO. SAN MARTÍN: San Roque, 9 May 1925, Melin 124 (5). BRASIL. AMAZONAS: Rio Negro, Manaus, Feb. 1901, Ule 5403 (HBG, К). MATO Grosso: 45 km SW de Poconé (Transpantaneira), 56%50'W, 16%25'S, 16 Dec. 1976, Krapovickas 29936 (CTES, MO). Piriqueta undulata Urban. Repert. Spec. Nov. Regni Veg. 13: 154. 1914. síNTIPOS: кр Botellas, Passarg Selwyn 325, 360a, 379, 431 (B, vins db NEÓTIPO (aqui designado): Venezuela. Bolivar: Dist. Cedeño, alrededores de Caicara de Ori- посо, 200 m, 66?11'W, 07°38'N, 27 June 1981, Rutkis 354 (VEN; isoneótipo, CTES). Selecciono esta colección como neótipo porque se encuadra muy bien en la descripción original y procede de un lugar próximo a la localidad típica. El ejemplar de VEN es más completo, con flores y cápsulas bien desarrolladas. n la ee original la localidad tipo figura como “Las Botillas.” He corregido la ortografía de acuerdo con la pee de Passarge (1933). LITERATURA CITADA PASSARGE, S. 1933. Wissenschafliche Ergebnisse einer eise in Gebiet des сва Caura und Cuchivero turwiss. 12: 1-281 STEUDEL, E. G. Naat “Wanenal: bot. ed. 2, 2: 344. URBAN, I. 1883. Monographie der familie ier н гасееп. Jahrb. Kónigl. Bot. Gart. Berlin 2: THREE NEW ANDEAN SPECIES OF AULONEMIA (POACEAE-BAMBUSOIDEAE) Lynn G. Clark? and Ximena Londoño? ABSTRACT Three new species of Aulonemia are described and illustrated, and comments on the generic limits of Aulonemia are included. Aulonemia longiaristata, an upper-montane forest species from Ecuador, has large, prominently aristate spikelets. Aulonemia robusta, of u Aulonemia pumila is a diminutive páramo species from Colombia, distinguished by its aristate glum upper-montane forests and paramos in Venezuela and Colo mbia, has robust culms and large, ovate foliage leaves with well-developed fimbriae on the sheath apex and margin. The distribution of each species is mapped, and additional comments on related or similar species are included. Aulonemia Goudot is a widespread but poorly known bamboo genus of perhaps 30 species, allied to the Arthrostylidiinae (Soderstrom & Ellis, 1987; Soderstrom et al., 1988). It is distributed from southern Mexico through Central America and along the Andes to Peru and Bolivia, with several species known from the Guayana Highlands and central and southeastern Brazil, including the states of Goias, Minas Gerais, and Espirito Santo south to Rio Grande do Sul. The Andean and Guayanan species usually occur at elevations above 2,000 m, whereas the Brazilian == tend to occur at some- what lower elevation Soderstrom (1988) noted that the vegetative branching of A. queko Goudot (the type species of the genus) and 4. hirtula (Pilger) McClure differs in some respects from that found in other members of Aulonemia. Most species of Aulonemia exhibit one branch per node, at least on mid-culm and higher nodes, with all the internodes more or less equal in length. In contrast, A. queko and А. hirtula usually have a series of several short in- ternodes alternating with the long internodes, and a branch complement derived from the basal re- branching of the dominant primary branch. Soder- strom (1988) suggested that these two groups were possibly not congeneric, and that the species with one branch per node were perhaps better accom- modated in Matudacalamus Maekawa, which McClure (1973) submerged into Aulonemia. Many of the Matudacalamus-type species do usually produce only one branch per culm, but we have observed that at least A. trianae (Munro) McClure and 4. patula (Pilger) McClure can pro- duce Aulonemia-type branch complements at their lower nodes while producing only one branch per node at the upper nodes. The bulky branch com- plements of the lower nodes are not usually rep- resented in herbaria, so specimens tend to be mis- leading as to the potential extent of branch development. complete revision of Aulonemia is necessary to resolve these morphological and taxonomic ques- tions; for the present, we regard Matudacalamus as falling within the limits of Aulonemia. The species described in the present paper all appear to be of the Matudacalamus type, but we have not verified this in the field for A. longiaristata. During the preparation of these descriptions, we examined material of a number of the Andean species of Aulonemia. The morphology and dis- tribution of the fimbriae on the foliage leaf sheaths were especially valuable as vegetative characters in distinguishing specie Aulonemia longiaristata L. G. Clark & Lon- ono, sp. nov. TYPE: Ecuador. Azuay: Nudo de Portete, pass between headwaters of the rios Tarqui (Atlantic) and Girón (Pacific), 9,000 ft., 10 Mar. 1945 (fl), Camp E-2177 (holo- dwork was supported by the National Science Foundation (DEB-8102766) ede a Foreign Travel Grant ' Fie from Iowa State University (Clark), the National Geographic Society (Clark, Londono), I de Investigaciones Cientificas) of Cali, Colombia (Londono), a f COL, for access to the herbarium material. We are grateful to Gerrit Davidse, 1 the curators o ISC, a and US Emmet Judziewicz, and Joh yer for h ? Department of Botany, ped State Universit 3 Museo de Ciencias Naturales de Cali F. Carlos Lehmann V., А.А. CIVA (Instituto Vallecaucano nd COLCIENCIAS of E (Londono). We thank 5660, Cali, Colombia. ANN. MISSOURI Bor. Garb. 77: 353-358. 1990. 354 Annals of the Missouri Botanical Garden е EL 7] LL e 22 () EL a d 7, Ze / Ce Nem № FIGURE 1. Aulonemia longiaristata. —А. (жу leaf and sheath apex Res auricle and apical fimbriae. —B. ikelet. — С. Base of fertile lemma showing the sms callus. (А based оп iP 128, B and C based on Camp -2177.) type, US 2 sheets; isotype, NY). Figures 1, Culmi usque ad 1.8 cm diametro, 2-4 m alti, erecti. pu KP а. Inflorescentia culos fertiles continentes; lemmata fertilia 10-20.4 mm longa arista exclusa, 9-nervata, arista (4-)9.5-20 mm longa; callus 3-5 mm longus, tumidus, nitidus, flavidus. Culms to 1.8 cm diam., 2-4 m tall, erect to slightly arching at the tip. Internodes to 31 cm long, terete, hollow, smooth. Culm leaf sheaths 10.5-13 cm long, 1.9—3 times as long as the blade, apically long-fimbriate, auriculate on one or both sides of the apex; blades 4.4—5.5 cm long, reflexed, pseudopetiolate; girdle 3-7 mm wide, smooth. Nodes somewhat swollen; supranodal ridge prom- inent; branch one per node. Foliage leaf sheaths keeled, glabrous, apically auriculate on the side of the overlapping margin (rarely both sides auricu- late); fimbriae restricted to the sheath apex, 1.5- 4 cm long, + erect, their basal portions straight, scabrid, their apical portions о to curly; blades (11.5-)14-29 cm long, (1.9-)3.5-10 cm wide, narrowly ovate to ovate, reflexed, adaxially gla- brous, abaxially glabrous to scabrid, not tessellate, the apex acuminate, the base asymmetrical, one side rounded to rounded-cuneate, the other atten- uate to rounded-attenuate; pseudopetiole 0.4-1.2 cm long, pulvinate; outer ligule 0.3-1 mm long; inner ligule ong. Inflorescence a + open panicle 14-20 cm long; rachis + complanate, glabrous; branches angular, the edges scabrous, otherwise glabrous, loosely appressed; pedicels 1- 5 cm long, angular, scabrous-pubescent. Spikelets 2.4-3.4 cm long excluding awns, pubescent; glume I 6.8-9.5 mm long (including awn), subulate to awned, 3-nerved, the awn to 4 mm long; glume II 10.4-11.2 mm long (including awn), 5- or 7-nerved, the awn 3.7-4.8 mm long; sterile lemma (8.4—) 11-13 mm long (excluding awn), 7- or 9-nerved, sometimes enclosing a rudimentary palea, the awn 4.5-5.5 mm long; fertile florets 4—6; disarticula- tion apparently above the glumes and between the florets, but the florets not separating readily from each other; fertile lemma 10-20.4 mm long (ex- cluding awn), 9-nerved, the awn (4—)9.5-20 mm long, straight to slightly curved, the callus 3-5 mm long, rounded, swollen, shiny, yellowish, prominent in florets at least before anthesis; palea 7.5-11.2 mm long, 2-keeled, acuminate-apiculate, pubes- cent, 4-nerved, the sulcus broad; rachilla inter- nodes 2.5-5 mm long, finely pubescent, flattened; lodicules not seen; stamens 3, the anthers 5-5.4 mm long; fruit unknown; terminal 1—2 florets ru- dimentary, 4.8-6 mm long, if two florets then the lower one often awned, and the second awnlike. Eus Н examined. ECUADOR. BOLÍ- R: 42 km W of Guaranda on the old road to Babahoyo, Wie 128 (AAU, MO, QCA, US). CHIMBORAZO: Canon of the Rio Chanchan, about 5 km N of Huigra, Camp E-3393 (NY, US). EL ORO: between Curtincapa and Gua- gra Uma, 8 mi of Curtincapa on SW slopes leading to Chapel, mail 53924 (US). IMBABURA: El Tam- bor, areas de Buenos Aires, Cordillera ns Acosta- Solis 17218 (US). LOJA: Cariamanga(?), S. Loja, 7 May 1946 (fl), Espinosa 309 (US). This species, an Ecuadorian endemic, is named for the long, slender awns of the fertile lemmas. Aulonemia longiaristata is also characterized by Volume 77, Number 2 1990 Clark 8 Londoño New Andean Species of Aulonemia 355 FIGURE 2. leaves. — B. In E. A. robus escence. ана pumila and A. robusta. A-C. A. pu —C. Spikelet. (A based on Londoño et al. 379, В and С bas umila. — A. Habit and branching of culm, with Ev ed on Londoño et al. 3 ll Foliage leaf and sheath apex showing marginal and деч fimbriae. —E. Spikelet. (D based 7.) D, on Clark & Cavelier 295, E based on Garcia-Barriga & Jaramillo M. 1 its auriculate foliage leaf sheaths with fimbriae re- stricted to the sheath apex, large foliage leaf blades, pubescent spikelets 2.4-3.4 awns), and the swollen, rounded callus present at the base of each fertile lemma. According to labe data, clumps of 4. longiaristata form extensive stands in upper montane forest and on ridge tops at elevations of 2,000 to 2,700 m. The plant is cm long (excluding called zadilla or carrizo, and Camp (£-2177) noted that its culms were used along with Arundo donax L. in the manufacture of baskets and other items. Aulonemia longiaristata has been confused with another high-elevation member of the genus, patula. (Aulonemia sodiroana (Hackel) McClure is probably synonymous with 4. patula.) The two 356 Annals of the Missouri Botanical Garden species are rather similar vegetatively, both having large foliage leaf blades, but 4. patula lacks the sheath auricles, and in addition to its apical fimbriae also exhibits prominent marginal fimbriae that ex- tend about halfway down the overlapping margin of the leaf sheath. The spikelets of A. patula may reach 3 cm in length, but usually have seven or eight fertile florets, and the fertile lemmas have short awns usually no more than 4 mm long. A rounded callus is present at the base of each fertile lemma; it is not swollen and prominent as in A longiaristata. Aulonemia pumila L. G. Clark & Londono, sp. nov. TYPE: Colombia. Putumayo: km 34 from Pasto on the Pasto-Sibundoy road, Páramo San Antonio del Bordoncillo, 3,210 m, 5 Feb. 1988 (fl), Londoño & Clark 382 (holotype, COL; isotypes, ISC, MO, NY, TULV, US, WIS). Figures 2A-C, 3. ulmi 2-3 mm diametro, 0.5-1 m ali erecti a procumbentes. Vaginae foliorum parum carinatae glabra vel pubescentes, nonauriculatae, fimbriatae tantum Ad apicem; fimbriae 5-6 mm lon nguste ovatae, d Inflorescentia paniculata, parum contracta, 5.5-11 cm longa. Spiculae 8.4-11.2(-12.6) mm longae, е 2-3(-4) flos- culos fertiles continentes; lemmata fertilia 5.5-9.2 mm longa, subulata vel subulata-aristata, 5- vel 7-nervata; callus ad 2 mm longus, planus, nitidus. Rhizomes pachymorph. Culms 2-3 mm diam., 0.5-1 m tall, erect to procumbent. /nternodes 8- 10(-18) cm long, terete, hollow but relatively thick- walled, pubescent. Culm leaf sheaths 5-5.5 cm long, ca. 7 times as long as the blade, occasionally a few apical fimbriae present, 2-6 mm long, au- ricles absent; blade 0.8 cm long, reflexed, shortly pseudopetiolate; girdle to 1 mm wide. Vodes slight- ly swollen; supranodal ridge prominent; branch one per node. Foliage leaf sheaths slightly keeled to- ward the apex, glabrous to pubescent, nonauricu- late; fimbriae restricted to the sheath apex, 5-6 mm long, erect, fine, smooth, wavy to curly; blades 3-7 cm long, 0.6-1.4 cm wide, narrowly ovate, reflexed, adaxially glabrous, abaxially glabrous to more commonly scabrid-pilose, not tessellate, the apex acuminate, the base rounded; pseudopetiole 1-1.5 mm long, pulvinate; outer ligule 0.2 mm long; inner ligule to 0.5 mm long. /nflorescence a somewhat contracted panicle 5.5-11 cm long; гасыз + complanate, pubescent; branches angu- lar, pubescent, ascending to loosely appressed; ped- icels 2-9 mm long, angular, pubescent. Spikelets 8.4-11.2(-12.6) mm long, pubescent; glume I (4.3-)7.4-8.3 mm long (including awn), 1-nerved; glume II (5.5-)7.5-8.4 mm long (including awn), 3- or 5-nerved; sterile lemma absent; fertile florets 2—3(-4); disarticulation above the glumes and be- tween the florets, but the lowermost floret often not disarticulating; fertile lemma 5.5-9.2 mm long, subulate to subulate-aristate, 5- or 7-nerved, the callus to 2 mm long, flat, lighter, shiny; palea 4.3- 6.3 mm long, 2-keeled, acuminate, 4-nerved, the keels ciliate, the sulcus broad, pubescent; rachilla internodes 1.5-2 mm long, finely pubescent, slight- ly flattened; lodicules not seen; stamens not seen; fruit unknown; terminal floret rudimentary, 2.9- 4.3 mm long Additional Т examined. COLOMBIA. CAUCA: г ‚ Valle de Las Papas, alrededores de Valencia, 11 Se ш; Oct. 1958 (9), Idrobo et al. 3820, 3830 (US). PUTUMAYO: alta cuenca del Rio Putumayo, e de la do entre El Encano y Sibundoy, Páramo San Antonio del Bordoncillo, 4 Jan. 1941 (fl), Cua- А HIT 77 (COL, US); lado sur de la Laguna de la mo de Santa Lucia (nacimiento del Río Ali- “ocha, Рага sales), 9 on 1941 (fl), Cuatrecasas 11874 (COL, US); m : f " ‚3 “Маг 987 PUL V, US); 32-33 ln from Pasto on the Pasto- oy road, Páramo San Antonio del Bordoncillo, Lon- dono et i 379 (COL, ISC, MO, NY, TULV, US Aulonemia pumila is one of the smallest mem- bers of the genus, hence its specific epithet. It is characterized by delicate culms 2-3 mm in di- ameter and 0.5-1 m tall; nonauriculate foliage leaf sheaths with a few, erect fimbriae 5-6 mm long restricted to the sheath apos foliage leaf blades 3-7 cm long and 0.6-1.4 cm wide; + contracted panicles 5.5-11 cm a and pubescent spikelets 8.4-11.2(-12.6) mm long with 2-3(-4) fertile flo- rets and distinctly aristate glumes. This species is known only from three marshy páramo areas in southern Colombia at elevations of 2,900 to 3,250 m. Aulonemia pumila is similar and perhaps re- lated to A. trianae, a more widespread Colombian species. Aulonemia trianae is distinguished from A. pumila by its culms 1-2.5(-6) m tall, apical fimbriae ca. 1 cm long, foliage leaf blades 6-15 cm long and l-2 cm wide, more open panicles, spikelets 10-17 mm long with 5-6(-8) fertile flo- rets, and acuminate, not aristate, glumes. Aulonemia robusta L. G. Clark & Londoño, sp. nov. TYPE: Venezuela. Mérida: Dtto. Sucre, via Estanquez-Las Coloradas-El Molino-Ca- naguá, Páramo Las Coloradas (Páramo La Laguna), 2,800 m, 14 June 1989 (fl), Clark et al. 533 (holotype, VEN; isotypes, ISC, MO, Volume 77, Number 2 1990 Clark 8 Londoño New Andean Species of Aulonemia 357 NY, US, and Facultad de Ciencias, Univer- sidad de Los Andes, Merida). Figures 2D-E, 3. Culmi 1-2.5 cm diametro, 2-4(-5) m alti, erecti. Va- ginae foliorum rotundatae, glabrae, interdum farinosae, т учать fimbriatae ad apicem еї ad marginem in ; fimbriae tenues, glabrae, crispatae, api- cales (1- )3- 4 сш 1опрае, maruinales 1-3 cm longae, pectinatae; laminae foliorum (5-7.5)19-32 cm longae, (1-1.5)5-13.5 cm latae, ovatae, reflexae. Inflorescentia paniculata, aperta, 35-50 cm longa. Spiculae 1.1-1.6 cm longae, pubescentes, 4-7 flosculos fertiles continentes; lemmata fertilia 5.2-6.7 mm longa, acuminato-apiculata, 7- vel 9-nervata; callus ad 2 mm longus, planus, nitidus. Culms 1-2.5 cm diam., 2-4(-5) m tall, erect. Internodes 30-40 cm long, terete, hollow, some- times farinose. Culm leaves often not clearly dis- tinguishable from the foliage leaves especially on nonbranching culms; sheaths 12 cm long, .9-2 times as long as the blade, fimbriae as for foliage leaves, the apical ones 2-3 cm long, the marginal ones 2.5-3 cm long, auricles absent; blades 6-8.5 cm long, reflexed, pseudopetiolate; girdle 3- 5 mm wide, smooth. Nodes slightly swollen; su- pranodal ridge prominent; branching restricted to culm apices, usually one branch per node. Foliage leaf sheaths rounded on the back, not at all keeled, glabrous, frequently farinose, nonauriculate; fim- briae fine, smooth, their basal portions straight, their apical portions wavy to curly, the apical fim- briae (1-)3-4 cm long and + erect, the marginal fimbriae 1-3 cm long, pectinate, extending about halfway down the overlapping margin; blades (5- 7.5)19-32 cm long, (1-1.5)5-13.5 cm wide, ovate, reflexed, adaxially glabrous, abaxially retrorsely scabrid, not tessellate, the apex acuminate, the base rounded; pseudopetiole 5—7 mm long, + pulvinate; outer ligule 1 mm long; inner ligule 2-8 mm long. Inflorescence an open panicle 35-50 cm long; rachis + angular, smooth, mottled; branches an- gular, smooth, the primary branches basally pul- vinate and diverging from the rachis; pedicels 2- 10 mm long, angular, smooth, often somewhat sinuous. Spikelets 1.1-1.6 cm long, pube ume I brous, 3-nerved; glume II minate, abaxially with a few scattered hairs toward the apex, 5- or 7-nerved; sterile lemma absent; fertile florets 4—7; disarticulation above the glumes and between the florets, often the lower 1-2 florets tardily or not disarticulating; fertile lemma 5.2- 6.7 mm long, acuminate-apiculate, abaxially pu- bescent, 7- or 9-nerved, the callus to 2 mm long, flat, distinct, lighter, shiny; palea 5.3-6.4 mm long, 2-keeled, acuminate-apiculate, 2-nerved, the sul- cus broad, pubescent; rachilla internodes 1.5-2 mm long, glabrous to pubescent, slightly flattened; a A. longiaristata | 4 A. pumila $n € A. robusta | FIGURE 3. Distributions of Aulonemia longiaristata, A. pres and 4. robusta lodicules 3, apically short ciliate, the posterior to 0.9 mm long, the anterior pair 1.1-1.2 mm long, fleshy at the base; stamens not seen; fruit unknown; terminal floret rudimentary, to 2.5 mm long. Additional specimens examined. COLOMBIA. NORTE DE SANTANDER: Bucaramanga-Pamplona road, between km 108 & 109, Clark & Cavelier 295 (COL, ISC, MO, US). NORTE DE SANTANDER/ CÉSAR: 20 km al sur de Abre- go, Las Jurisdicciones (Cerro de Oroque), 19/21 May 1969 (fl), García-Barriga & Jaramillo M. 19797 (US). SANTANDER: Mpio. Floridablanca/Piedecuesta, via Buca- гашапва- Вегіп- Pamplona, km 41 between Bucaraman- ga and Berlin, Londoño & Clark 477 (COL, ISC, MO, NY, TULV, UIS, US); Mpio. Guaca, via Berlin- Baraya, descending toward Baraya, Londoño & Clark 481 (COL, ISC, MO, TULV, UIS, US); Mpio. Tona, Santa Rita por la carretera de Bucaramanga a Pamplona antes del Pára- Murillo & Jaramillo 1217 region noreste Los Arbolitos, Briceño et al. 2329 (Fa- cultad de Ciencias, ULA, Mérida). MÉRIDA: Dtto. Liber- tador, Páramo El Portachuelo, Briceño & Adamo 1956 (ISC. Facultad de Ciencias, ULA, Mérida); Dtto. Andres Bello, Páramo El Tambor, 25 Oct. 1985 (fl), Briceño et al. 1443 (ISC, Facultad de Ciencias, ULA, Merida); Es- eV Жане нра Quinimari, 20 km al sur de San Vice 16 Jan. 1968 (fl), aid et al. 101052 (COL, F, NY, US, VEN). Aulonemia robusta is named for its overall ro- bust appearance, including the stout culms and large leaves. This species is characterized by the 358 Annals of the Missouri Botanical Garden culms 1-2.5 cm diam. and 2-4(-5) m tall with branching restricted to the apices; nonauriculate foliage leaf sheaths with fine, smooth fimbriae 1— 4 cm long distributed at the sheath apex and half- y down the overlapping margin; ovate foliage leaf blades 19-32 cm long and 5-13.5 cm wide; and pubescent spikelets 1.1-1.6 cm long with 4— 7 fertile florets. Aulonemia robusta occurs at el- evations of 2,500-3,200 m in upper montane for- est, subpáramo, and paramo vegetation in south- western Venezuela and northeastern Colombia. Vegetatively, A. robusta resembles А. subpec- tinata (O. Kuntze) McClure, a species known only from the Coastal Range of Venezuela. Although comparable in culm height and diameter to 4. robusta, А. subpectinata has narrowly ovate fo- liage leaf blades 9.3-22 cm long and 2-5.7 cm wide. Aulonemia subpectinata also exhibits both apical and marginal fimbriae, but they only reach 3 cm in length, and the basal portion of the fimbriae is flattened and thickened, while the apical portion is narrow, threadlike, and curly. In addition, the spikelets of A. subpectinata are (1.2-)2.4-3.6(- 4.6) cm long with (5-)6-8(9- 11) florets per spike- let. The spikelets of 4. robusta are very similar to those of A. trianae, but А. robusta has much larger inflorescences and is distinguished by its large, ovate foliage leaves and the well-developed apical and marginal fimbriae on the foliage leaf sheaths. The foliage leaves produced on the apical branches of culms of A. robusta are often much smaller than the foliage leaves of the main culms, and may be easily confused with those of A. trianae. However, the pattern of fimbriae development in 4. robusta is much different from that of А. trianae, which usually has only sparse, apical fimbriae 1 cm long. LITERATURE CITED MCCLURE, К. A. 1973. Genera of bamboos native to the New World (Gramineae: Bambusoideae). Smith- sonian Contr. Bot. 9: i-xii, 1-148. [Edited by T. R. Soderstrom.] ви Т.К. 1988. Aulonemia fulgor (Poaceae soideae), a new species from Mexico. Brittonia 40: 2231. ELLIS. 1987. The position of bamboo E and allies in a system of grass classification. 25-238 in T. R. Soderstrom et al. (editors), Grass oo ie and Evolution. Smithsonian Insti- of the tution Press, Washington, D.C. [Proceedings о zolini & R. E. Heyer (editors), de Janeiro. [Proceedings of a tropical Distribution. Rio de Janeiro, 12-16 January 1987.] DEVIA XEROMORPHA, A NEW GENUS AND SPECIES OF IRIDACEAE-IXIOIDEAE FROM THE CAPE PROVINCE, SOUTH AFRICA! Peter Goldblatt? and John C. Manning? ABSTRACT Devia xeromorpha, a new genus and species of Iridaceae- Ixioideae, is a local endemic of the Roggeveld Escarpment of Devia is of the bas or Tritoniinae is x — 11 у e seed coat еми some of the differences between Devia and (persistent tunics, proteranthous nk, actinomorphic flowers with helically rotated anthers; and mal cells are leni thickened, and submarginal sclerenchyma is absent; the bundle caps are strongly dev Hal gen at reach to t type for Iridaceae in its brown, microreticulate outer surface. Basic Chasmanthe and some species of Tritonia have n = 10. ape et with this pattern, but it differs from Crocosmia which has the basic number for the subtribe. Crocosmia reflect xeromorphic adaptations in the former and peculiar fibrotic leaves), the genus appears to have followed an independent the e epidermis; and stomata are restricted to the laminar Devia, evolutionary pathway, becoming specialized in the structure of the leaf and in orientation of the stamens and style. A new species of Iridaceae subfamily Ixioideae, discovered by the first author in 1981 in fruit on the Roggeveld Escarpment in the western Karoo of the Cape Province, South Africa, was collected in full flower in January 1989. The ample material that is now available makes it clear that the plant is not only a new species but also that it does not accord with any genus of Iridaceae so far described. The species is assigned to a new genus, Devia, named in honor of Dr. М. Р. de Vos, of Stellen- bosch, South Africa, in recognition of her extensive systematic, anatomical, and embryological re- search in Iridaceae and other families of southern African plants. Devia (Fig. 1) is characterized by large, persis- tent corms with tough long-lived fibrous corm tu- nics; a branched spike of numerous, small acti- nomorphic flowers; small membranous bracts; a style with short, apically notched branches; heli- cally rotated stamens; and globose capsules with one or two large seeds per locule. The leaves are unusual in both their structure and phenology. They are heavily fibrous and have two large grooves running the length of each surface, between the central (pseudomidrib) and secondary veins. The leaves are proteranthous and are quite dry by an- thesis, and their persistent bases accumulate for years in a thick fibrous mass around the base of the plants. Vegetative increase in the number of corms results in a clumped habit, and the plants sometimes grow in dense concentrations among low shrubs or on open slopes. LEAF ANATOMY The leaves of Devia are unusual in Ixioideae, although they conform to the basic isobilateral (equitant) type for the family. The central veins are heavily thickened, and deep grooves run the length of both surfaces between the pseudomidrib and secondary veins (Figs. 1K, 2A). There is also a pair of shallow grooves between the margin and secondary veins. A formal anatomical description ' Support for this study by grants DEB 81-19292 and BSR 85-00148 from the U.S. National Science Foundation is age acknowledged. We thank Dee Snijman and Graha В. А am Duncan for hel elp. rukoff Curator of African Botany, Missouri Botanical Garden, Р.О. Box 299, St. Louis, Missouri 63166, U.S.A 3 National Botanic Gardens of South Africa, Kirstenbosch, Private Bag X7, Claremont 7735, South Africa. ANN. Missouri Вот. GARD. 77: 359-364. 1990. 360 Annals of the Missouri Botanical Garden | FIGURE 1. Habit, reproductive morphology, and leaf anatomy of Devia xeromorpha (Snijman & Manning 1194). — A. Habit, x0.5. — B. Flowering spike, full size. — C. Corm, x0.67.— D. Side and top view of flower, x 2. — Volume 77, Number 2 1990 Goldblatt & Manning Devia xeromorpha, a New Genus and Species of Iridaceae-Ixioideae 361 Ficu desd lean 1 1194). ues Lal transection, x25.— (double arrow) x60. surface, without stomata, showing elongate, epapillate epidermal cells. LEAF TRANSVERSE SECTION Blade isobilateral, monofacial, more or less ob- long, with large paired, opposed grooves (sinuses) between the median and submarginal vein pairs (Fig. 2A-C), the sinuses with low median ridges, and with a pair of smaller shallow sinuses (Fig. 1K) without median ridges between the submarginal and marginal veins. Cuticle thick and domed over the cells, but thin within the sinuses. Epidermis cells wider than high, outer walls thickened and minutely denticulate in those along the ribs; cells 2-4 times longer than wide in surface view, without papillae (Fig. 2D); marginal cells palisade, heavily thickened (Fig. 2C), especially on the anticlinal walls; long papillae on the cells on the lips of the sinuses and on the ridges within the larger sinuses (Fig. 2B). Stomata restricted to the sinuses, sunken; guard cells with a slight outer cuticular lip, much smaller than the neighboring epidermal cells, which over- arch them. Vascular bundles in two opposite rows, ves of Devia xeromorpha (Snijman & Manning Laminar г groove ж-да йя papillate ridge (single arrow) and vascular girders — C. Leaf margin showing thickened marginal palisade epidermis (arrowed), x 10 0.—D. Leaf highly variable in size: primary vascular bundles median (forming the pseudomidrib); bundles submarginal; tertiary bundles single below each margin and also in two opposed pairs, between the primary and secondary bundles (below the sinus ridges); 2-6 quarternary or minor bundles alter- nating with the others, and one centripetal to each marginal tertiary bundle. Xylem pole oriented to the interior except in the marginal tertiary bundle, which is bicollateral and oriented at right angles to the rest. The xylem and phloem poles in the pri- mary and secondary bundles separated by two lay- ers of parenchymatous cells. Outer sheath of bun- dle sheaths a continuous layer of sclerenchyma in primary and secondary bundles, extending as T-shaped girders to the epidermis, the crossbars three or four cells thick, reaching to the edges of the ridges; outer sheath also extending as a scle- renchymatous girder to the epidermis from the sub- marginal tertiary bundles (Fig. 2B); inner sheath sclerenchymatous as a complete sheath in primary secondary — Br иелер х i. x 20. — К. Dissected flower with ovary and style, x3.—G. Anthers, x4.—H. Ape — |. Capsules, full size. — Ј. Seeds, x5.— К. Leaf transection (small submarginal sinuses arrowed), ex of style and style Annals of the Missouri Botanical Garden and secondary bundles, forming bundle caps only in tertiary bundles. Chlorenchyma forming broad bands surrounding the sinuses, about eight rows deep, cells at most slightly elongated. Central ground tissue of large loosely packed parenchy- matous cells, separated from chlorenchyma by an incomplete band of tanniniferous cells one or two cells wide; tanniferous cells also present sporad- ically in the chlorenchyma and outer sheath of marginal tertiary or quarternary vascular bundles. CHROMOSOME CYTOLOGY A diploid chromosome number of 2n = 20 was determined from mitosis in root tips harvested from sprouting corms. The root tips were pretreated in hydroxyquinoline and squashed after hydrolysis in FLP orcein following a technique outlined else- where (Goldblatt, 1980, 1981). The chromosomes are relatively small, 2-3 um long, and range from acro- to submetacentric. Two slightly longer chro- mosome pairs can be distinguished, and a pair of small satellites is located on the short arm of one of the short chromosome pairs. The chromosomes are comparable in size to those of most genera of Ixioideae (Goldblatt, 1971). The karyotype match- es closely those described for /xia and Dierama (x = 10) in number and appearance of the chro- mosomes and with Tritonia and Crocosmia (x = 11) in general appearance (Goldblatt, 1971; de Vos, 1982a) RELATIONSHIPS The affinities of Devia are most likely with Tri- tonia, Crocosmia, and Chasmanthe (Ixieae- Tri- toniinae sensu Goldblatt, 1971). Devia shares with most species of these genera a leaf structure in which the marginal epidermis is heavily thickened and not associated with submarginal sclerenchyma, while the laminar bundles have thick sclerenchyma caps often extending to the epidermis as girders (cf. de Vos, 1982b, 1984). This type of leaf anat- omy is uncommon in [Ixioideae but is also found in Sparaxis, Synnotia, Freesia, Anomatheca, Tri- toniopsis, and Anapalina (unpublished). Devia shares with Crocosmia and Chasmanthe rounded, hard-walled capsules containing few seeds, up to two per locule in Devia, and short, apically stigmatic and sometimes bifurcate style branches; and with Crocosmia alone, a persistent corm. Tri- tonia has many-seeded capsules, smaller and usu- ally angular seeds, and typically obovoid capsules. Devia differs from both Crocosmia and Chas- manthe in several significant features. The leaves of Devia are unique in Tritoniinae in having paired longitudinal grooves, although this is a common xeromorphic feature of several genera of other subtribes of Ixioideae (e.g., Gladiolus spp., Geis- sorhiza spp., and most notably, Romulea). The flowers are actinomorphic but with spirally rotated stamens and an eccentric style. Species of Cro- cosmia and Chasmanthe (de Vos, 1984, 1985) are mesomorphic and have broad, soft-textured leaves, either plane or plicate, and with the ex- ception of Crocosmia aurea Planchon, zygomor- phic flowers. In Crocosmia and Chasmanthe but not Tritonia the pseudomidrib consists of several pairs of vascular bundles (de Vos, 1984, 1985). Perianth color in Crocosmia and Chasmanthe is orange to red (yellow in one variant of Chasmanthe floribunda (Salisb.) N.E.Br.), which also contrasts with the dusty pink perianth of Devia. The stamens in Crocosmia and Chasmanthe are never spirally rotated, but are either unilateral and arched below the upper tepal in zygomorphic-flowered species or straight and surrounding the central style in acti- nomorphic-flowered Crocosmia aurea. The diploid chromosome number in Devia, 2n = 20, corresponds with that of Chasmanthe (x = 10) but not with Crocosmia (x = 11) (Goldblatt, 1971; de Vos, 1984) Devia is probably most closely allied to Cro- cosmia, yet in addition to the differences discussed above, the seeds of Crocosmia have a thick hy- drophilic coat (possibly an adaptation for bird dis- persal) that becomes loose on dehydration and then can be abraded easily (de Vos, 1982b). The seed coat of Devia appears to be of the basic type for Iridaceae, having a brown, microreticulate outer surface. Although some of the differences between Devia and Crocosmia reflect xeromorphic adaptations in the former (persistent tunics, proteranthous and peculiar fibrotic leaves), the genus appears to have followed an independent evolutionary pathway and become specialized in the structure of the leaf and orientation of the stamens and style. SYSTEMATICS Devia xeromorpha Goldbl. & Manning, gen. et sp. nov. TYPE: South Africa. Cape: Roggeveld Escarpment, farm Vierfontein, NW of Suth- erland, rocky loam in renosterveld, Snijman & Manning 1194 (holotype, NBG; isotypes, К, MO, PRE, S, STE, WAG). Figures 1-3. Plantae 50-70 cm altae, cormo globoso 2-2.7 cm diam., persistentis, tunicis fibrosis, foliis siccis sub anthesi linearibus 35-45 cm longis 2.5 mm latis, inflorescentia Volume 77, Number 2 1990 Goldblatt 8 Manning 363 Devia xeromorpha, a New Genus and Species of lridaceae—Ixioideae 18 19 20 1 1 1 Signs m A ps ees 1: = CARITA ELL LL. A КЕДЩ FER ZZ 9 Y tity FIGURE 3. spica 18-26 florum secunda, floribus actinomorphibus roseo-bubalinis, tubo anguste infundibuliformis ca. 9 mm longo, tepalis ovatis ca. 6 mm longis, staminibus helice rotatis, styl tricis, ramis ad api fi tis, capsulis globoso-trigonis, seminibus (0)1-2 per loculis, 2-3 an- gulatis, 3.5-4.5 mm longis. Plants 50-70 cm high, growing in dense tufts. Corm depressed-globose, 5-6 internodes long, 2- 2.7 cm diam., surface and inner tissue bright or- ange, persisting for several years, thus several corms lying above one another; tunics coarsely fibrous, persisting for several years and forming a thick neck around the base. Cataphylls 2, cartilaginous, pale or becoming dry and brown, especially above the ground, the inner largest and reaching 2-4 cm above the ground, sheathing the leaves initially, later decaying. Leaves dry at anthesis, usually 6, linear, 35-45 cm long, 2.5 mm wide, tapering gradually to a pungent apex, narrowly oval in sec- tion with the central vein area heavily thickened, and with 2 narrow grooves running the length of each surface. Stem (0)1-3-branched, sheathed en- tirely by 5-6 imbricate bracts, these dry by an- thesis. Inflorescence a spike, the main axis bearing Southwest portion of southern Africa showing the distribution of Devia xeromorpha. 18-26 flowers, the lateral spikes with 5-10 flow- ers, secund, axes ascending; bracts paired and opposed, dry-membranous, 2.5-4 mm long, trans- lucent at the edges, minutely brown-speckled to- ward the center and base, the outer bracts enclosing the inner ones and acute, the inner bracts about as long as the outer ones but forked apically. Flow- ers actinomorphic, dusty dull pink, odorless, the style eccentric; perianth tube narrowly funnel- shaped, ca. 9 mm long, the lower cylindric part 5-6 mm long, straight or curving upward; tepals ascending, ovate, ca. 6 mm long, ca. 3.5 mm wide. Filaments inserted at the base of the upper part of the tube, rotated counterclockwise the width of one tepal, ca. 9 mm long, thus reaching to the tepal apices; anthers opposite the inner tepals, longitudinally dehiscent, sub-basifixed, thecae sep- arate in the lower quarter, linear, ca. 4 mm long, yellow. Ovary obovoid, ca. 1.8 mm long, style ca. 16 mm long, reaching to about mid anther level, branches ca. 0.5 mm long, bifid, stigmatic only at the apices. Capsules globose-trigonous, 5-6 mm high, 6-8 mm wide, cartilaginous, hard when dry, pale straw-colored, lightly verrucose toward the 364 Annals of the Missouri Botanical Garden apex; seeds (0)1-2 per locule, dark brown, 2-3- angled, surface microreticulate, 3.5-4.5 mm long, ca. 3 mm at the widest, funicle whitish, often persisting, lightly adhering to the raphe. mosome number 2n = 20 Flowering time. December and January. Distribution and habitat. Devia xeromorpha is restricted to the highest parts of the Roggeveld Escarpment in the western Karoo (Fig. 3). Exten- sive populations occur northwest of Sutherland in the vicinity of Sneeukrans on several farms that extend along the steep escarpment edge. The el- evation in this part of the escarpment is 4,800- 5,600 ft. (1,100-1,300 m), and the area receives slightly higher precipitation than the surrounding, somewhat lower country. The escarpment extends for a considerable distance to the north, almost as far as Calvinia, and it seems likely that Devia may occur elsewhere along the escarpment in areas of higher elevation and precipitation, but it has yet to be recorded except close to Sneeukrans. The leaves emerge in March in early autumn and grow throughout the winter and spring, the wettest pe- riod in the Roggeveld. They begin to dry out in November and are usually dead at flowering time, which occurs in summer, normally the dry season here. Several individuals of a small species of long- tongued fly in the Bombyliidae were observed vis- iting and probing the flowers of Devia in the mid- morning, and may be the pollinator. Additional specimens examined. SOUTH AFRICA. CAPE: 32.20 (Sutherland) Uitkyk farm, Roggeveld NW of Sutherland (AD), Goldblatt 6357 (MO); Sneeukrans, south of Voelfontein, ca. 4,500 ft., Goldblatt 6343 (MO). LITERATURE CITED DE Vos, М.Р. 1982a. The African genus Tritonia p Gawler (Iridaceae): Part 1. J. S. African Bot. 05-163. 1982b. Die bou en ontwikkeling van die unifa- sale blaar van Tritonia en verwante genera. J. S. а Bot. 48: 23-37. The African к, Crocosmia Plan- African Bot. 50: 463-502. . Revision of s pues joe genus Lemaire S. African J. Bot. 51: GOLDBLATT, P. 1971. Cytological and or teorica studies in e southern African Iridaceae. J. S. Af- rican Bot. 37: 317-460. 1900, "Redetnition of Homeria and Moraea оова)“ in the light of biosystematic data, with Rheome gen. nov. Bot. Not. 133: 85-95. 19 otes on the о and distribution of Anapalina, Tritoniopsis, and Sparaxis, Cape Iridaceae. Ann. Missouri Bot. Gard. 68: 562-564. hom is 198: LEAF AND CORM Peter Goldblatt? and John C. Manning? TUNIC STRUCTURE IN LAPEIROUSIA (IRIDACEAE-IXIOIDEAE) IN RELATION TO I I CLASSIFICATION! ABSTRACT The tropical and southern. African Lapeirousia, compri sing 35 species, divides into two subgenera on the basi sis o of lineages in the genus. This forms the basis for our infrageneric classification that recognizes two subgenera, each with two sections The 35 species of Lapeirousia (Iridaceae—Ixioi- deae) occur primarily in the winter rainfall zone of southern Africa and in semiarid southwest tropical Africa, but a few species are widespread across sub-Saharan Africa. Based on its forked style branches and the absence of flavone O-glycosides, Lapeirousia was assigned to Watsonieae, one o three tribes of Ixioideae (Goldblatt, 1989, 1990a); Watsonieae also include Watsonia (52 species), Thereianthus (6 species), Micranthus (3 species), and Savannosiphon (1 species). All except the last are restricted to southern Africa. Until now rela- tionships among the species of Lapeirousia have been obscure, and there has been no phylogenetic analysis of the genus nor any modern attempt to provide an infrageneric classification. Baker's (1892, 1896) recognition of three subgenera no longer has any utility, since his subg. 4nomatheca is now regarded as a genus of tribe Ixieae. His subgenus Sophronia included at most four species and merits at best sectional rank. In this paper we investigate the major differences among the species of Lapeirousia, namely the nature of the corm tunics and leaf structure, and use these features to establish a new infrageneric classification. This ad has been made in conjunction with a system- revision of Lapeirousia in tropical Africa (Goldblatt, 1990b). Leaf structure and corm tunic structure vary in an apparently consistent pattern across species in tropical and southern Africa. That these two in- dependent features are correlated suggests a major division in the genus into two groups, for which we propose subgeneric rank. The anatomical basis for the leaf variation is explored here, and we propose a phylogeny for the major species groups in Lapei- rousia based on several independently varying characters (Table 2). We recognize subg. Pani- ! Support for this study by grant BSR 85-00148 from the U.S. National Science Foundation and grant 3749-88 from the National Geographic Society is gratefully acknowledged. 2 B. A. Krukoff Curator of African Botany, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. 3 National Botanic Gardens, Kirstenbosch, Private Bag X7, Claremont 7735, South Africa. ANN. Missouri Вот. GARD. 77: 365-374. 1990. 366 Annals of the Missouri Botanical Garden TABLE 1. Leaf anatomical characters for the species studied and voucher data. (G & M = Goldblatt & Manning.) All vouchers are housed at MO and the derived states by 1, 2 Pseudomidrib: О absent, 1 present Rib profile: О rounded, 1 truncate Vascular bundles Marginal bundle: O absent, 1 ANS subepidermal marginal sclerenchyma: 0 Bundle position: O all subepide . Leaf margins: 0 rounded, 1 flanged ec : 0 opposite, 1 alternate nt Blade profile transverse section: O plane, 1 zigzag, 2 elliptic, 3 elliptic and hollow Sclerenchyma bundle sheath: o phloem уи 1 pud Coding for each character is as follows with the primitive condition indicated by 0 , l prese rmal, 1 major bundles ie and minor ones embedded Number of character state Taxon 123 45 678 9 Voucher data Pillansia templemannii L. Bolus 200000000 Powrie s.n. Watsonia aletroides (Burm. f.) Ker 01 001011 0 Sidey 1754 laccata (Jacq.) Ker 01 00001 1 0 Rycroft 2531 Micranthus alopecuroides (L.) Rothm. 3.0001 101 € Grant 3489 Thereianthus minutus (Klatt) G. Lewis O01 00 1 1 0 1 I Bolus s.n. spicatus (L.) G. Lewis 0 1 0 1] 10 1] I Grant 5027 Lapeirousia subg. Paniculata Section Paniculata avasmontana Dinter 1 101 1001 0 G & M 8798 bainesii Baker 01 001 00 1 0 G & M 8808 coerulea Schinz 1 101 1001 0 С «€ M 8811A gracilis Vaupel 010010010 Seydel 3419 otaviensis R. Foster 01 001 00 1 0 G & M 8837 schimperi (Aschers & Klatt) Milne-Redh. 01 001 001 0 Grant 4507 Section Fastigiata corymbosa (L.) Ker 01 00 1 1 01 0 Gillett 4503 Lapeirousia subg. Lapeirousia Section Ad anceps (L.f.) 101 11001 0 Drége pots odoratissima Bake 101 1 1001 0 С& M8 pyramidalis (Lam. ) Goldbl. 101 1 1001 0 Mauve & p 233 Section Lapeirousia divaricata Baker 01111 00 1 0 Goldblatt 2754 Savannosiphon euryphylla (Harms) Goldbl. & Marais 000010000 Pawek 10753 culata for species with plane leaves having a dis- tinct central vein and with corm tunics composed of densely compacted fibers; subg. Lapeirousia, by contrast, is made up of species with corrugate (shallowly plicate) leaves lacking a distinct central vein and having corm tunics of uniformly woody texture. Inflorescence structure and bract morphology appear to coincide with this division to some extent but the patterns of variation in these two features are complex and sometimes contradictory, indi- cating a degree of convergence that obscures the major patterns of variation. Our conclusions are presented in the form of a cladogram for which Volume 77, Number 2 Goldblatt & Manning 367 1990 Leaf and Corm Tunic Structure in Lapeirousia TABLE 2. Characters used in the cladogram (Fig. 3). tions, were collected from living plants in the field The derived (apomorphic) states are listed first followed and fixed in FAA. Leaves taken from herbarium by the presumed ancestral = phic) conditions. specimens were rehydrated in aerosol OT. One leaf taken Gold Apomorphies for Savannosiphon are from Gold- was examined for each of the species studied. blatt (1989). Other possible Synapomorphies for t. Fas- Voücher informatióm is cited in Table 1. tigiata are discussed in the . Anatomical ме The cladogram (Fig. 3), based on the data matrix (Table 3), was constructed without the use of a computer. tions for L. corymbosa, the ae species of sect. Fastigiata known anatomically, are not included in the cladogram . Corm bases flat—corm bases rounded . Corm tunics woody—corms tunics of compacted fi- bers — CHARACTER ANALYSIS ho CORM TUNICS 3. Margins of corm toothed/spiny—margins of corm not elaborate There are substantive differences between the 4. Leaves corrugate—leaves plane and with a pseu- corm tunics of the two subgenera of Lapeirousia. domidrib Members of subg. Lapeirousia have corms with 5. Major veins alternate — major veins opposite : : tunics composed of concentric hard, woody layers 6. Ribs truncate — ribs r e : ‘ 7 with smooth surfaces. With age these layers gen- Stems angular —stems terete . . j rally fragment irregularly into smaller pieces but 8. Inflorescence a pseudopanicle — inflorescence a spike . 9. Flower zygomorphic—flower actinomorphic rarely become fibrous or cancellate (Fig. 1A, B). 10. Bracts small —bracts relatively large In contrast, the tunics of subg. Paniculata are 11. Bracts + membranous and dry above at anthesis — composed of densely compacted layers, which bracts herbaceous sometimes have a nearly woody texture and decay 12. Capsules coriaceous— capsules woody a different manner. They fray at the lower edges, 13. Plants short and inflorescence congested — plants tall жар into regular vertical strips that become and inflorescences not notably congested increasingly fibrous with age, and sometimes be- come distinctly netted above or entirely (Fig. 1 E). The anatomical basis for the differences is un- known. Within the two subgenera there are dif- ferences of lesser significance. In several species of subg. Lapeirousia the lower margins of the tunics are ornamented with small teeth, whereas in other species the edges are simply Gold blatt, 1972). In subg. Paniculata the way in which Leaf sections were prepared by following a stan- the tunics age is often characteristic of species. dard graded ethanol dehydration and wax embed- ome have tunics that form more or less regular ding technique. Serial sections were mounted on vertical strips at least at the base, but that do not slides and stained in saffranin, sometimes counter- become fibrous, whereas the development of a fi- stained in fast green, and permanently mounted in brous texture characterizes others. Based on com- Canada balsam. Leaves studied, with a few excep- parison with the genera closely allied to Lapeirou- the characters are either discussed here or taken from completed work (Goldblatt, 1989. 1990b, c) MATERIALS AND METHODS TABLE З. Data matrix for Savannosiphon and the major generic clusters of Lapeirousia, numbered as іп Table 3. Presence of the specialized condition is denoted by x; absence Character number Genus 1 2 3 4 5 6 7 8 9 10 11 12 13 Savannosiphon = = = = = = x = x = = x - Lapeirousia Subgenus Lapeirousia Section Lapeirousia x x x x x Е х = x = = х = Section Sophronia x x = x х х х = x = = х - Subgenus Paniculata ection Paniculata x — = = 2 x x x = x x x = Section Fastigiata x — = a = x x x = x = х х 368 Annals of the Missouri Botanical Garden FIGURE 1. Representative corm tunics in Lapeirousia. Subgenus е —А. Goldblatt 6052).—B. L. ас (Goldblatt & Manning 8808).— (Variously magnified. »lomitica Dinter (Norde T sia, we conclude that tunics composed of compacted fibers are the primitive condition. Similar, though not identical, tunics occur in 7hereianthus and Micranthus, corms of which can often only be distinguished from those of Lapeirousia by being rounded (vs. flat) at the base. Woody tunics are unknown in other genera of Watsonieae, but are present in a few genera of Ixieae (notably Hes- perantha, Geissorhiza, and Romulea) where they are most likely also a derived condition. LEAF ANATOMY (Table 1) Like all Iridaceae, the leaves of Lapeirousia are ensiform and equitant, and consist of a sheathing base and a blade that is typically lanceolate to linear and either plane or occasionally terete (subg. Paniculata), or ridged to corrugate (subg. Lapei- rousia). In transverse section the blades are iso- bilateral and monofacial, oblong to linear, and var- iously plane-sided, ridged or zigzag. The marginal subepidermal cells are unlignified apart from the phloem cap that is present in most species, the basic condition for Watsonieae and presumably for the whole family. The intercostal epidermal cells are brick-shaped and thin-walled, although the out- er periclinal walls may sometimes be slightly thick- ened. There is a notable correlation between leaf anatomy and subgeneric limits in the structure of the costal epidermis (Table 1). In subg. Lapeirou- sia the costal cells are abruptly enlarged and often heavily thickened in the outer periclinal walls, and the ridge is square and truncate (Fig. 2B). In subg. Paniculata the epidermis cells gradually diminish in size and usually lack a heavily thickened outer periclinal wall, and the ridges are smoothly rounded (Fig. 2A). The vascular bundles are in two rows with larger nstam & Lundgren 17 15). Sub — D. L. coerulea (Seydel 2609).— L. pee (Jacq.) Diels us Paniculat ainesii —E. L. шол (Klatt) Baker (Williams 875). (major) bundles separated by one or two smaller (minor) bundles; major bundles occupy the ridges and minor ones the intervening troughs. The xylem poles face the center of the leaf. The two subgenera differ somewhat in the arrangement of the bundles. Although the major bundles are, with one exception (Table 1), alternate in subg. Lapeirousia (Fig. 2B), they are mostly, but not always, opposite in subg. Paniculata (Fig. 2A). This is an obvious deter- minant of the zigzag leaves in subg. Lapeirousia and plane leaves in subg. Paniculata. (In species with largely alternate bundles the central bundles are often opposite, but the pairs become more alternate distally.) The zigzag leaf form is a spe- cialization in subg. Lapeirousia, and the accom- panying alternation of the major bundles is prob- ably so, given that all other genera of Watsonieae have opposed bundles. A pseudomidrib formed by larger central bundles is not evident in subg. Lapei- rousia even though the central bundle in L. di- varicata is sufficiently large to form a pseudomid- rib. This species resembles subg. Paniculata in having opposite bundles, but the ridges are those of subg. Lapeirousia, and the corm tunics place it firmly in the latter. The opposed bundles in L. divaricata, a fairly specialized species, are pre- sumably secondary. A pseudomidrib is present in all plane-leaved species of subg. Paniculata, even those in which the vascular bundles are alternate (L. coerulea and L. avasmontana), as well as in nearly all species of the tribe, and this is undoubt- edly the basic condition. The alternation of bundles in L. coerulea and L. avasmontana, thought to be the most primitive species in subg. Paniculata, is puzzling. On the basis of outgroup comparison, this must be viewed as a specialization for the two species. Volume 77, Number 2 1990 Goldblatt & Manning Leaf and Corm Tunic Structure in Lapeirousia 369 Transverse sections of leaf blades and vascular bundles in Lapeirousia. — A. The major bundles reach the epidermis or are separated from it by a single layer of parenchyma, while the minor bundles are a number of cell layers below the epidermis. In species with opposite bun- dles, some vascular bundles may fuse at the xylem poles. An outer bundle sheath is present as a single layer of parenchymatous cells distinct from the surrounding mesophyll cells. An inner sheath is present as a phloem cap in at least the central major bundles, and usually all major bundles; and the phloem cap may also occur in some minor bundles. (In L. corymbosa alone there is a more or less complete sclerenchyma sheath, which is more tenuous at the xylem pole and is restricted to the central bundles.) Even when the phloem cap is absent from some minor bundles, the marginal bundles are still usually capped (but not in L. an- ceps, L. pyramidalis, and L. corymbosa). Crystals are present, mainly in the outer bundle sheaths but also in isolated mesophyll cells. Chlorenchyma may IGURE 2. Subgenus Paniculata: L. bainesii. — B. Subgenus Lapeirousia: L. odoratissima. Scale bars = 150 um. (Voucher data in Table 2.) be distinct from the central ground tissue or may grade into it. Pillansia, the only genus of Ixioideae—Pillan- sieae, and the other genera of Watsonieae (Sa- vannosiphon, Micranthus, Thereianthus, and Watsonia) each differ from Lapeirousia in some aspect of their leaf anatomy (Table 1). Pillansia is characterized largely by primitive traits (Gold- blatt, 1990a). The leaf is plane in transverse sec- tion, and lacks a distinct midrib and ridges, al- though it is thickened toward the center and the profile is undulate. The epidermal cells are uni- formly square without differentiation between cos- tal and intercostal, although stomata are restricted to the intercostae. The vascular bundles are in two opposite rows of alternating major and minor pairs, and all except one or two very minor bundles are subepidermal and reach the epidermis, although the outer sheath is not distinct from the surrounding chlorenchyma at the phloem pole. All bundles have 370 Annals of the Missouri Botanical Garden a sclerenchyma phloem cap. A marginal bundle is absent, and there is no subepidermal marginal scle- renchyma. In Savannosiphon the plane leaf also has all the vascular bundles subepidermal and of similar size. In Micranthus most of the bundles are sub- epidermal with some of the very minor bundles more deeply embedded in the leaf. All major bun- dles, including the marginal ones, have a complete sclerenchyma sheath well-developed at both poles. Thereianthus is likewise characterized by a com- plete sclerenchyma sheath around the major bun- dles, at least those in the center of the leaf, but the minor bundles have a phloem cap only and are embedded in the mesophyll. The marginal bundles lack sclerenchyma. We assume that a complete sclerenchyma sheath is a synapomorphy for Mi- cranthus and Thereianthus. The latter differs fur- ther from the other genera in having nonvascu- larized flanges on the leaf margins, but not all species of the genus have been examined for this character. Watsonia resembles most species of Lapeirousia subg. Paniculata in the distribution and nature of the vascular bundles: two opposed rows of large subepidermal bundles separated by embedded minor bundles with sclerenchyma largely restricted to the phloem cap. It is distinct in the tribe in the extensive development of subepidermal sclerenchyma in a V- or U-shaped region at the leaf margin, associated with and extending well beyond the marginal phloem caps. There is a sim- ilar development of sclerenchyma in some genera of Ixioideae, and in Iridoideae—Irideae, e.g., Dietes (Rudall, 1983), and the shrubby genera of Niveni- oideae (Rudall & Burns, 1989). In the last-men- tioned two groups the sclerenchyma is not asso- ciated with the marginal bundles. There is little doubt that the presence of subepidermal scleren- chyma on the leaf margins is a specialized condition and a synapomorphy for Watsonia in addition to those бо by "Goldblatt ( (1989). DISCUSSION The five genera of Wat phyletic line united by the derived deeply divided style branches and the absence of flavone O-gly- cosides (Goldblatt, 1989, 1990a). The sister tribe, Ixieae, differs by having a putatively more spe- cialized type of corm ontogeny and by having fla- vones and flavone O-glycosides. Within Watsonie- ae, Watsonia constitutes one clade and the four remaining genera a second. Several synapomor- phies unite the species of Watsonia (Goldblatt, 1989), and now the presence of subepidermal mar- ginal sclerenchyma can be added. The clade formed by Thereianthus, Micranthus, Lapeirousia, and Savannosiphon share three synapomorphies: small corms; corm tunics formed only from the cataphylls (the foliage leaves are inserted on the flowering stem and do not contribute to the tunics); and corm tunics composed of densely compacted fibers (pre- viously defined as hard or woody by Goldblatt, 1989). Only subg. Lapeirousia has truly woody tunics of a uniform texture and smooth surface. Thereianthus and Micranthus ne clade, defined by their fusiform seeds (Goldblatt, 1989). The presence of complete sclerenchymatous bundle sheaths in the leaves is a possible additional syn- apomorphy (only three of the ten species in the two genera have been examined for this character), although this xeromorphic feature is a fairly com- mon condition in Iridaceae. Savannosiphon, which is monotypic, and Lape- trousia are weakly related (Goldblatt, 1989). They both have angular to winged stems and coriaceous capsules (polarization of the latter character is un- certain). Leaf anatomy makes no contribution to our perception of this clade. The unusual flat-based corms of Lapeirousia, the major synapomorphy for the genus (Goldblatt, 1989), separate it from Savannosiphon whose in- cluded styles and stamens readily distinguish it from other Watsonieae. The position of the vascular bundles in Savannosiphon resembles that in Pil- lansia, but whether this represents a reversal to the primitive state or an indication that the genus is misplaced in our phylogeny (Fig. 3) is uncertain. Within Lapeirousia there appear to be two ma- jor groups (Fig. 3) defined by their corm tunics, leaf morphology, and anatomy. Species with woody tunics of uniform texture and a smooth surface, corrugate leaves with truncate ribs, and mostly alternate veins form one clade, which we recognize as subg. Lapeirousia. The remaining species are plesiomorphic for these characters (Table 1) but ave a presumably derived inflorescence structure (Goldblatt, 1990b), and we recognize the group as subg. Paniculata. In general, species of subg. Pa- niculata have highly branched inflorescences (pseudopanicles) (Fig. 4A, B), whereas those of subg. Lapeirousia generally have spikes (Fig. 4С), although these may also be branched to some de- gree. Some members of subg. Paniculata, notably L. abyssinica, have few-branched inflorescences that are indistinguishable from those of subg. Lapeirousia. A simple or few-branched spike is presumed to be the basic condition in Watsonieae and Ixieae (Goldblatt, 1990a, b). Species of subg. Lapeirousia have herbaceous bracts that are often Goldblatt & Mannin 371 Volume 77, Number 2 g 1990 Leaf and Corm Tunic Structure in Lapeirousia х E Ў К ri [4 $ subgenus Paniculata subgenus Lapeirousia section Fastigiata 13 plant short, infl. dense 10 bracts small 9 flower zygomorphic | = 8 infl. a pseudopanicle stamens/style included flower white 12 capsules coriaceous 7 stems angular section Paniculata 11 bracts + dry above section section Sophronia Lapeirvusia 3 corm margins toothed 9 flower zygomorphic 6 ribs truncate 5 veins alternate 4 leaves corrugate 2 tunics woody 1 corm bases flat FIGURE З. Hypothetical phylogeny of Lapeirousia showing the major infrageneric lineages and the apomorphic characters that define them. Characters used in the cladogram are listed in Table 3. enlarged and sometimes ridged, keeled, or toothed. In subg. Paniculata the bracts are generally small and either entirely herbaceous or partly membra- nous and are often dry at anthesis. We assume that small bracts constitute a second, weak, syn- apomorphy for the subgenus (Table 2), as шее, firm herbaceous bracts are the presumed basic condition for the tribe. Thus subg. Lapeirousia has diverged in the nature of its vegetative organs whereas subg. Paniculata has to some extent in the floral axis. The occurrence of actinomorphic flowers in three species of subg. Paniculata, L. corymbosa (from the southwest Cape) and L. coerulea and L. avas- montana (from Namibia) remains puzzling. These species do not appear to be closely related but they may well be primitive in the subgenus, and there seems no reason to believe that the actinomorphic flower is a derived trait in a genus of largely zy- gomorphic-flowered species. We accept here that actinomorphy is the primitive condition for the genus and that zygomorphy must have evolved iru in each subgenus of Lapeirousia as we in Savannosiphon. There is considerable floral convergence in Ixioideae, and there are single instances of presumably primitive actinomorphy in the otherwise zygomorphic-flowered genera Wat. sonia (Goldblatt, 1989) and Thereianthus (Lewis, 1941). By contrast, in subg. Lapeirousia the zygo- morphy is thought to be the basic state and acti- nomorphy derived in the few species in which it occurs (Goldblatt, 1972), These actinomorphic species are all acaulescent, and we assume that actinomorphy is adaptive for these low-growing plants whose flowers are borne close to the ground — vertical presentation with consequent actinomor- phy enhances floral display and access to long- tongued pollinators. A further refinement of the infrageneric clas- sification of Lapeirousia is suggested by discon- tinuous patterns of variation in both subgenera (Fig. 3). All the Cape species of subg. Paniculata have dark corm tunics, a trait only occasionally found in the tropical African species; fairly short, falcate leaves; rather congested inflorescences; and ee low stature (Fig. ese > also have basic chromosome number o Go ldblatt, 1971, 1972), whereas the tropic ee species range from n = 8 ton = 3 (Goldblatt, 1990c). The bracts of the tropical species are characteristically dry at anthesis (there are a few exceptions), which we 372 Annals of the Missouri Botanical Garden FIGURE 4. Representative species of Lapeirousia showing the morphology of the inflorescence. — A. L. bainesii (sect. Paniculata). — В. L. corymbosa subsp. fimbriata (sect. Fastigiata).—C. L. dolomitica subsp. lewisiana (sect. Lapeirousia). —D. L. odoratissima (sect. Sophronia). Scale: x0.5. Volume 77, Number 2 1990 Goldblatt 8 Manning 373 Leaf and Corm Tunic Structure in Lapeirousia assume is a derived state (Table 2). We suggest that the tropical and Cape members of subg. Lapei- rousia constitute two monophyletic lines for which we propose sectional rank: sect. Fastigiata for the Cape species and sect. Paniculata for the tropical species. In subg. Lapeirousia those species with the specialization of elaborate, toothed to spiny lower corm tunic margins probably constitute a mono- phyletic line and also merit sectional segregation. Whether the remaining species constitute one or more lines remains to be determined. We find no reason to favor a particular hypothesis here but suggest they be segregated in one section, which incidentally contains, among others, all the acau- lescent species (Baker’s subg. Sophronia). SYSTEMATICS ке Pourret, Mem. Acad. Sci. Toulouse 2. 1788. TYPE: Lapeir pressa Pourret (= L. fabricii (de la Roche) Ker). ousia com- — Subgenus Lapeirousia Plants with hard, woody corm tunics of uniform texture, the surface smooth and often glossy, de- caying irregularly into unequal fragments, rarely becoming fibrous, the basal margins sometimes pro- duced into teeth or spines. Stem aerial or entirely subterranean. Leaves corrugate (shallowly plicate), without a prominent midrib and all major veins + equal. Inflorescence a simple or branched spike, or the internodes very short and forming a tuft at ground level; bracts herbaceous, short to long, the margins and keel sometimes undulate, crisped or toothed, the inner shorter than the outer. Flowers actinomorphic (only in acaulescent species) or zy- gomorphic, short- to long-tubed, the upper tepal usually larger than the others, and the lower three with contrasting markings; stamens symmetrically disposed or unilateral and arcuate. Section Lapeirousia Plants with woody corm tunics with the basal margins toothed or spiny. Stem aerial and usually branched. Inflorescence a simple or branched spike. Flowers always zygomorphic with the stamens uni- lateral and arcuate. Eight species occurring in the southwest Cape, Namaqualand and the Karoo, South Africa, and in southern Namibia. Section Sophronia (Lichst. ex Roemer & Schultes) enn & Manning, stat. nov. Lapeirousia Sophronia (Lichst. ex Roemer Schultes) Baker, Handbk. Irideae 174. 1892. Sophronia Lichst. ex Roemer & Schultes, Syst. 1: 482. 1817. ТҮРЕ: Lapeirousia pli- cata (Jacq.) Diels. Plants with woody corm tunics with the bases entire, not produced into teeth or spines. Stem aerial and usually branched or not produced above the ground. Inflorescence a simple or branched spike, or the internodes very short and forming a tuft at ground level. Flowers actinomorphic in the acaulescent species or zygomorphic with the sta- mens unilateral and arcuate. Ten species mostly in southern Africa, although L. odoratissima widespread across south tropical Africa and Namibia. 2. Subgenus Paniculata Goldbl. & Manning, ubgenus nov. TYPE: Lapeirousia erythran- 5 tha (Klatt) Baker. Plantae cormi tunicis duris, fibrosis compactis com- positis, foliis usitate тя costatis, raro teretibus, inflo- rescentia usitate moso et + paniculato, bracteis bre- vibus das ado siccis supra Plants with hard, persistent corm tunics com- posed of densely compacted fibers (sometimes + woody in texture), decaying into vertical strips or other florescence usually highly ramified and + panicle- like (a pseudopanicle) or sometimes a branched spike, the ultimate branches bearing 1-8 flowers, and those below the terminal flower always sessile; bracts short, herbaceous, sometimes dry apically or for their entire length at anthesis, the inner bracts about as long as the outer ones. Seventeen species occurring in the southwest Cape, South Africa, and in tropical Africa from central Namibia to Ethiopia and Nigeria. Section Paniculata. TYPE: as for the subgenus. Plants usually very -o Corm tunics pale straw to blackish, decaying into regular vertical fibrous and то, Leaves gested at the end of the branches; bracts herba- ceous to membranous, usually becoming dry in the 374 Annals of the Missouri Botanical Garden upper half by anthesis. Flowers actinomorphic or zygomorphic, the tepals subequal or unequal (then with the upper tepal largest and reflexed or arched over the stamens); stamens symmetrically disposed or unilateral and arcuate. Fourteen species extending from central Na- mibia, Botswana, and the Transvaal across central Africa to Ethiopia in the north and Nigeria in the west. Section Fastigiata Goldbl., Contrib. Bolus Herb. 1972. TYPE: Lapeirousia corymbosa (L.) Ker. Plants usually very branched. Corm tunics dark brown to black, decaying into regular vertical strips. Leaves plane, the midvein sometimes raised, fal- cate or straight. Inflorescence a pseudopanicle or a spike, usually congested; bracts herbaceous, the apices often reddish or purple. Flowers actino- morphic or zygomorphic, the tepals subequal; sta- mens symmetrically disposed or unilateral and ar- cuate or declinate. Three species restricted to the southwest Cape, South Africa. LITERATURE CITED BAKER, * G. 1892. Handbook of the Irideae. George Bell & a "id ae. In: W. T. Thiselton- Dyer (ed- op) Ad санаа 6: 88-89. Lovell Reeve & Co., London. GOLDBLATT, P. 1971. Cytological and morphological studies in the southern African Iridaceae. J. S. Af- rican ee 37: 317-460. 972. А revision of the genera Lapeirousia a and Anomatheca Ker in the winter irm region of South Africa. Contrib. Bolus Her 111. 1989. The southern African m. Watsonia. Ann. Kirstenbos -148. ch Bot. Gard. 19 1990a. Phylogeny үле veris of Iri- daceae. Ann. Missouri Bot. . 77: (in press) —————. 1990b. Systematics of Lapeirousia (Irida ceae- »- [xioideae) in tropical Africa. Ann. Missouri Bot. Gard. an d press). ———. Oc “жа variability in the African кеа Lapeirousia (Iridaceae-Ixioideae). Ann. Mis- t. Gard. 77: 375-38 Pen rà J. 1941. Iridaceae. New genera and species and miscellaneous notes. J. 5. African Bot. 7: 19- RupaALL, P. 1983. Leaf anatomy and ces of Dietes (Iridaceae). Nordic J. Bot. 3: 471- & P. Burns. 1989. Leaf anatomy of a woody South African Iridaceae. Kew Bull. 44: 525-532. CYTOLOGICAL VARIABILITY IN THE AFRICAN GENUS LAPEIROUSIA (IRIDACEAE-IXIOIDEAE)! Peter Goldblatt? ABSTRACT The African genus Lapeirousia (Iridaceae-Ixioideae) comprises two subgenera each with two sections. The basic chromosome number for th nine muc Paniculata Goldbl. & Ma species, but chromosome number ranges from n = 10 to = 10 and asymmetric 3 less strongly bimodal чырр ae Sa Pan ith Species of the section with the highest chromosome num has species with n = 8, , 5, 4, and 3 karyotypes with one lon с Perle ana pair. Total chromosome length, a = 6 and approximately twice the total chromosome length compared with all involved in = (€ of only these two species. D nsible for the variation in chrom i i in all except two species, which hav other species examined. Polyploidy appears to have been reduction is thought to have been re e genus is postulated to be x = ch smaller pairs. This karyotype occurs in at least some species of three sections and is exclusive in subg. anning sect. Fastigiata Goldbl. 2 subg. Lapeirousia the bimodalit 0 in a strongly bimodal karyotype with one long and ity is preserved in all . Genera most closely allied to Lapeirousia also have x niculata, which is entirely tropical in distribution, e bimoda a crude measure of genome size, is similar from irap AE morphology suggests that descending ашы pate repeatedly in the genus and that low number and 3, were achieved in separate lineages Lapeirousia Pourret, a genus of Iridaceae- Ixioideae (cf. Goldblatt, 1990a), comprises some 35 species (Goldblatt, 1972, 1990b; Goldblatt & Manning, 1990) distributed in two subgenera each with two sections. The genus is widespread in Africa south of the Sahara, with centers in the winter- rainfall zone of the southern African west coast and in the drier parts of tropical Africa, particularly Namibia. This pattern is unusual for Iridaceae, in which most African genera are either restricted to the Cape region of South Africa or extend into the wetter parts of eastern southern Africa, some as far north as Ethiopia. Only Gladiolus (Ixioideae), Moraea (Iridoideae), and Aristea (Nivenioideae) have ranges comparable to Lapeirousia, but they are absent or poorly represented in areas of tropical Africa where Lapeirousia is best developed (Gold- blatt, 1990b) Chromosome cytology of pa in aed ern África in moderately well nine species counted, about half "s total (Goldblatt, 1971, 1972), but until now there have been no counts for any tropical African species. The karyo- types of 13 species in tropical Africa (of a total of 16) and an additional seven in southern Africa are described here. The cytology of only five species remains unknown. Data indicate that Lapeirousia is unusually variable cytologically. Haploid num- bers of n = 10, 9, 8, 7, 6, 5, 4, and 3 have now been recorded in the genus. This contrasts with the majority of Ixioideae, which are cytologically uniform (Goldblatt, 1971) and typically have only one base number and relatively little polyploidy. Only Romulea and Crocus have until now been exceptions to this pattern in the subfamily (De Vos, 1972; Brighton, 1976a, b, 1977). Variation in chromosome number in Lapeirou- = is accompanied by major differences in karyo- . Strong bimodality appears to be the rule (Goldblatt 1972), with one long chromosome pair and a variable number of much smaller pairs (Gold- blatt, 1971, 1972). The bimodality encountered in all southern African species examined also oc- curs in some of the tropical species. The patterns ! Support for this study by grant BSR 85-00148 from the U.S. National Science Foundation and grant 3749-88 from the National Geographic Society is gratefully "gc ag I thank the following for their help in obtaining live material of species of Lapeirousia: Jean Pawe мч la Croix, Gairloch, Scotland; Sylvester Chisumpa, , San Jose, California; Georges Delpierre, Durbanville, South Africa; J. W. Loubser, Strand, South Africa; Maurice oe Verdun, France Kitwe, ; Jan de Koning, Maputo, Mozambique; Zambia; and W. Giess, Windhoek, Na mibia. rukoff Curator of African Botany, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A ANN. MISSOURI Вот. GARD. 77: 375-382. 1990. 376 Amnals of the Missouri Botanical Garden of variation in form and overall size of the chro- mosome complements in Lapeirousia and in closely allied genera suggest that the direction of chro- mosome change has been from high to low numbers and that polyploidy has played only a minor role in the evolution of the genus. MATERIALS AND METHODS Wild-collected seeds or corms (Table 1) were sprouted in the greenhouse. Root tips, harvested when 1-2 cm long, were pretreated either in 0.003 M hydroxyquinoline for 6-8 hours at refrig- erator temperature or in saturated aqueous 1 -bro- monaphthalene at room temperature for 3 hours, and then fixed in 3: 1 absolute ethanol- glacial acetic acid for 1-2 minutes. Tips were then stored in 70% ethanol or immediately macerated in 10% HCI at 60°C for 6 minutes, washed in tap water, and later squashed in lactopropionic orcein (Dyer, 1963) or FLP orcein (Jackson, 1972). This method differs from the paraffin-section technique I pre viously used for Lapeirousia (Goldblatt, 1971, 1972) and has yielded more satisfactory results. Difficulties, noted previously, in growing corms were still encountered and an adequate number of root tips for study could be obtained only with difficulty in several species. Seeds, however, germinate easily and provide ample material for examination. Total length of the chromosome complement was determined in selected species (Table 2) by linear measurement of camera-lucida-drawn chro- mosomes in karyotypes of estimated comparable degree of contraction and figured at the same mag- nification. Error in this type of estimation of ge- nome size is considerable, but the results appear internally consistent so that some confidence can be attached to the results. OBSERVATIONS SUBGENUS PANICULATA SECTION PANICULATA Restricted to tropical Africa and previously un- counted, sect. Paniculata is remarkably variable cytologically. The two putatively most primitive members of the section and the only ones with actinomorphic flowers, L. avasmontana and coerulea, have 2n = 16 and 8, respectively. The karyotype of L. avasmontana (Fig. 1A) consists of one long acrocentric chromosome pair ca. 8 um long and seven short acrocentric to submetacentric pairs ca. 3 um long. No satellites were noted. Lapeirousia coerulea, for which there are counts for seven populations, has a very different karyo- type (Fig. 1B). In all populations there are two pairs of longer acrocentric pairs 5-6 um long (up to 7 um in some preparations where the chromo- somes are less contracted), and two pairs of ac- rocentric to submetacentric chromosomes 2.3-3 um long. When satellites are visible (three of the seven populations examined) they are small and situated on the end of a short arm of one of the small chromosome pairs. In one population, Lavra- nos & Pehlemann 21101 (Table 1), small con- strictions are present near the ends of the long arms of a long chromosome pair, and no satellites were noted in these plants. The widespread Lapeirousia erythrantha has 2n — 12 in all eight populations examined, these covering a substantial part of the range of the species from southern Malawi to northern Zambia. The karyotypes are similar in all plants examined, with the exception of those with B chromosomes, and consist of three longer acrocentric pairs 6-7 um long and three shorter acrocentric pairs 3.3- 4 um long (Fig. 1C). Size differences are not as sharp as in the preceeding two species. Small sat- ellites are present on the ends of the short arms of one of the longer pairs. One to three B chro- mosomes were noted in plants from Mufulira, Zam- bia (Goldblatt 7575). The B chromosomes varied in number in different plants and are identified by their particularly small size, ca. 1 um smaller than the next smallest chromosomes. A collection from eastern Zambia (Faden et al. 74/83), which has = 14, is probably best interpreted as having 2n = 12 + 2B. There are four small pairs in this collection and the smallest of these are probably B chromosomes. Lapeirousia rivularis, evidently closely related to L. erythrantha, also has 2n = 12 (Fig. 1D), although one seedling examined had 2n = 18 and is apparently triploid. The karyotype of the 2n = 12 plants closely resembles that of L. erythrantha. Total chromosome length in these two species (Ta- ble 2) is close to twice that of all the other species studied, suggesting their possible polyploid origin. Two species apparently closely related to Lapei- rousia erythrantha, L. abyssinica and L. setifo- lia, have 2n = 8. They have similar karyotypes with two long acrocentric pairs 5.5-6.5 um long (to 9 um in preparations of less contracted chro- mosomes) and two much shorter pairs ca. 3 um long. Small satellites are located on the end of the long arm of a long pair in L. abyssinica (Fig. 1E) and on the end of the short arm of a long pair in L. setifolia (Fig. 1F). In the latter, one of the small chromosome pairs is metacentric. Total chromo- some length in L. abyssinica and L. setifolia (Table 2) is similar and about two-thirds that of L. ery- Volume 77, Number 2 1990 Goldblatt 377 Cytological Variability in Lapeirousia | J FIGURE 1. coerulea. —C. L. erythrantha. —D. L. rivularis. > ~ Mitotic metaphase in Lapeirousia subg. Paniculata sect. E —F. №? - bn K — А. L. avasmontana E- га ssinica setifolia. —G. L. otaviensis. — schimperi. —1. L. bainesii.—J. L. gracilis. —K. L. sandersonii. dans are given in Table 1. Scale bar, t um. thrantha (and L. rivularis). Other species closely allied to L. erythrantha are unknown cytologically, and there are no counts for L. erythrantha from Mozambique, Zimbabwe, or Zaire. Counts for this species, however, include both major variants, the small- and crimson-flowered typical form from the southeast part of its range and the blue-flowered form (corresponding to the type of L. briartii) from northern Zambia. The lack of any obvious struc- tural variation among the populations examined is notable in view of the morphological variability and wide distribution of L. erythrantha. Three of the six long-tubed species of sect. Pani- culata, Lapeirousia otaviensis, L. bainesii, and L. schimperi, have similar basic karyotypes with 2n — 10. Their karyotypes consist of one pair of long submetacentric (almost metacentric) chro- mosomes 8-10 um long (depending on the degree of contraction), and four shorter pairs 3.5-6 um long that show no sharp size discontinuities (Fig. 1G, H). The longest of the latter four pairs is nearly metacentric, and the others are acrocentric. Sat- ellites are present on the ends of the short arms of one of the shorter acrocentric pairs in L. ota- viensis and L. schimperi. Satellites were not seen in populations of L. bainesii with this karyotype. Two populations of L. bainesii (Table 1) have 2n — 6 (Fig. 1I) and nearly equal metacentric chro- mosome pairs 7.5-8 um long. There are secondary constrictions close to the centromere in one pair and in the midpart of an arm on another. Two other long-tubed species, Lapeirousia gracilis and L. sandersonii, have 2n — 12 and 10, respectively, but otherwise identical bimodal karyotypes with one long acrocentric pair ca. 8 um long and either five or four much shorter acro- 378 Annals of the Missouri Botanical Garden TABLE 1. Chromosome numbers in Lapeirousia. Original counts are marked with an asterisk. The taxonomy used here is based on the revisions of Goldblatt (1972, 1990b). Previous counts were reported by Goldblatt (1971, 1972). Acronyms (abbreviated according to Holmgren et al., 1981) following the collection data refer to the herbaria in which the vouchers are housed. Presence of a bimodal karyotype indicated by Y, absence by N. Species Diploid number Karyo- Collection data Subgenus Paniculata Section d L. abyssinica . Rich.) = Baker L. avasmontana Dinter L. bainesii Baker L. coerulea Schinz L. erythrantha (Klotzsch ex Klatt) Baker L. gracilis Vaupel L. otaviensis R. Foster ivularis Wanntorp L. sandersonii Baker L. schimperi (Asch. & Klatt) Baker L. setifolia Harms Section Fastigiata L. corymbosa (L.) Ker subsp. corymbosa subsp. fastigiata Goldbl. ~ Lam.) L. falcata L. micrantha (Klatt) Baker Subgenus Lapeirousia Section Sophronia L. anceps (L.f.) Ker L. exilis Goldbl. 12* 10* 20 20 20 4 aA uve << Ethiopia, Muger valley, Edwards et al. 97 (MO) Namibia, near Windhoek, Goldblatt & Manning 8798 (MO) Namibia, ENE of Otjiwarongo, Lavranos & Pehlemann 21031 ) Namibia, farm Norabis, Goldblatt & Manning 8826 (MO); Steinhausen road E of Windhoek, Goldblatt & Manning 8808 (MO) Namibia, Ameib, Giess 15284 (WIND); Ameib, Goldblatt & Manning 8811A (MO); Grootfontein, Valhal, Goldblatt & Manning no voucher; NW of Omaruru, Wanntorp 805 (S); Etosha Pan, Giess 15283 (MO); N of Tsumeb, Lavranos & Pehlemann 21101 (WIND); W of Otavi, J. Lavranos & 1. Pehlemann no voucher Malawi, University of Malawi, Zomba, Goldblatt 7521 (MO); Thondwe W of Zomba, Goldblatt 7575 (MO); Old Naisi Road, Zomba, Goldblatt 7514 (MO); near Ncheu, Goldblatt 7534 (MO); near Chileka, Goldblatt 7524 (MO); Chongoni Forest, La Croix 2698 (МО), Zambia, Chati Reserve, Gold- blatt 7567 (MO); Mufulira, Goldblatt 7575 (MO). Zambia, Sanje Hill, Faden et al. 74/83 (MO) Namibia, Asab, Goldblatt & Manning 8870 (MO) Namibia, Auros farm, Goldblatt & Manning 8837 (MO) Zambia, Lusaka, Goldblatt 7537 (MO) South Africa, Transvaal, E of Pretoria, Anon sub Goldblatt 5490 ( Namibia, farm Vaalwater, Goldblatt & Manning 8831 (MO); Zambia, Mufulira, cultivated Missouri Botanical Garden, Goldblatt s.n. (MO); Zimbabwe, Victoria Falls, G. McNeil no voucher; Malawi, Nyika, /. La Croix no voucher Malawi, Nyika Plateau, La Croix 4321 (MO); Nyika, Pawek 6674 (MO) Goldblatt (1971, 1972) Goldblatt (1971) Goldblatt (1972) Goldblatt (1971, 1972) South Africa, Cape, Cedarberg, near Algeria, Goldblatt 5150 (MO) Goldblatt (1971) South Africa, Cape, S of Piekeniers Kloof, Goldblatt 3026 ; Koeberg, Goldblatt 5105 (MO) South Africa, Cape, N of Springbok, Goldblatt 2649 (MO) Volume 77, Number 2 Goldblatt 379 1990 Cytological Variability in Lapeirousia TABLE 1. Continued. Diploid Karyo Species number type Collection data L. jacquinii N.E. Br. ca. 20 Y Goldblatt (1971) 18* Y South Africa, Cape, near Trawal, Goldblatt (no voucher) 18(-20)* Y South Africa, Cape, near Klawer, Goldblatt 2266 (MO) L. littoralis Baker subsp. littoralis 16* Y South Africa, Cape, E of Springbok, Wisura s.n. (NBG) subsp. caudata (Schinz) 16* Y Mozambique, near Maputo, Goldblatt 6585 (MO) Goldbl. L. odoratissima Baker 16* Y Namibia, Gobabis district, Tolken s.n. (BOL); ЕМЕ of Otji- ongo, Lavranos & Pehlemann 21059 (WIND); E of Windhoek, pong & Manning 8803 (MO) 18* Y Namibia, Palmflache, Merxmuller & Giess 30147 (M); Ma- lawi, Pawek 8166 (MO) L. oreogena Goldbl. ca. 18; 16- Y (Goldblatt, 1971, 1972) 8 L. plicata (Jacq.) Diels subsp. plicata 16” Y South Africa, Cape, Matjesfontein, Goldblatt 6091 (MO) L. pyramidalis (Lam.) ca. 18 Y (Goldblatt, 1971) Goldbl. 20* Y South Africa, Cape, foot of Gifberg, Goldblatt 2196 (MO) L. silenoides (Jacq.) Ker — 20* Y South Africa, near Garies, Goldblatt 2767 (MO) L. verecunda Goldbl. 18* Y South Africa, Cape, Spektakel Pass, Goldblatt 5710 (MO) Section Lapeirousia L. arenicola Schltr. 16* Y South Africa, Cape, near Hondeklipbaai, Goldblatt 4242 (BOL) L. divaricata N.E. Br. subsp. divaricata ca. 20 Y De he 1972) * Y South Africa, Cape, origin unknown, NM Missouri Bo- x Garden, Goldblatt s.n. (M subsp. grandiflora Goldbl. 16* Y South Africa, Cape, Richtersveld, a 5716 (MO) L. dolomitica Dinter subsp. dolomitica 16-18 Y (Goldblatt, 1972) 16* Y South Africa, Cape, Richtersveld, G. Delpierre 380 (no vouch- er); Williamson 3606 (NBG) subsp. lewisiana (B. 16* Y d P Cape, near Komkans, Nordenstam & Lundgren Nord.) Goldbl. L. fabricii (de la Roche) Ker 16 + 1B* Y L. violacea Gold 16* Y 5 (S); Ka niesberg, Goldblatt 3980 (MO) ы eid Cape, near Alpha, Goldblatt no voucher South Africa, Cape, Botterkloof, Goldblatt no voucher unted species: L. angolensis Goldbl.; teretifolia (Geerinck et al.) Goldbl. to metacentric pairs 2.5-3 um long (Fig. 1J, K). In both species satellites are located on the end of a short arm of a short acrocentric chromosome pair. Total chromosome length in these two species is similar (Table 2) and in the same range as that for most species of Lapeirousia, including those with low numbers such as 2n — SUBGENUS PANICULATA SECTION FASTIGIATA The bimodal karyotype (Goldblatt, 1971, 1972) already described for sect. Fastigiata is confirmed. Lapeirousia micrantha (Fig. 2A), counted here from a third population, has 2n — 20 and a karyo- L. barklyi Baker; L. masukuensis Vaupel; L. montana Klatt; L. type matching that already described for this and the two other species of sect. Fastigiata. There are one long acrocentric pair ca. 8 um long and nine short acro- to metacentric pairs 2-2.5 um long. Satellites, when observed, are always located on the short arm of a short acrocentric chromosome pair. SUBGENUS LAPEIROUSIA Sections Lapeirousia and Sophronia, including the two tropical African species of the latter, have similar karyotypes (Fig. 2B-H), closely resembling the karyotype described above for L. micrantha 380 Annals of the Missouri Botanical Garden TABLE 2. Total length of the chromosome comple- In the caulescent members of the section, 2n ment in selected Lapeirousia species. Measurements were made from illustrations of karyotypes of estimated com- parable degree of contraction, drawn at the same mag- nification. Species are grouped in subgenus and section by chromosome number to facilitate comparison. Total Diploid chromosome Species number length (um) Subgenus Paniculata Section Paniculata |. avasmontana 16 40.0 L. erythrantha 12 57.3 L. rivularis 12 59.5 L. gracilis 12 34.1 L. otaviensis 10 46.8 L. sandersonii 10 38.2 L. schimperi 10 49.1 L. abyssinica 8 40.5 L. coerulea 8 40.7 L. setifolia 8 38.2 bainesii 6 45.5 Section Fastigiata |. micrantha 20 41.2 Subgenus Lapeirousia Section ene L. littor subsp. poe 16 43.6 subsp. caudata 16 40.6 L. odoratissima 16 39.5 L. anceps 20 43.5 Section Lapeirousia L. arenicola 16 38.8 L. dolomitica 16 34.1 except for variation in the number of small chro- mosome pairs (Table 1). In sect. Sophronia there are new counts for nine species, six of which are the first records for these species. Numbers range from 2n — 16 to O. Lapeirousia littoralis subsp. littoralis (Fig. 2B) and the tropical African L. littoralis subsp. caudata have 2n — 16. The same number is also found in two acaulescent species, the southern Af- rican L. plicata subsp. plicata, and in two pop- ulations of the tropical African L. odoratissima (Fig. 2C). A third has 2n — 18. It is also likely that 2n = 16 is the correct number for L. oreo- na, previously reported as ca. 18 and 16-18 (Goldblatt, 1972), an acaulescent species closely related to L. plicata. I have been unable to obtain sufficient root tip material of this species to confirm its chromosome number. — 20 is confirmed for Lapeirousia anceps (Fig. 2D) and L. pyramidalis (Fig. 2E) and reported for the first time in L. silenoides. A diploid number of 2n — 18 was found in L. exilis and in two separate populations of L. jacquinii, previously reported as 2n — ca. 20 ts here for sect. Lapeirousia include the first records for three species and additional counts for three more, leaving only one species uncounted. А diploid number of 2n = 16 is the most common and the only one recorded for L. arenicola; L. divaricata subsp. grandiflora; both subspecies of L. dolomitica (Fig. 2F, G), previously reported as 2n = 16-18 (Goldblatt, 1972); and L. violacea (Fig. 2H). The single count for L. fabricii is 2n = 16 + 1B. Only L. divaricata subsp. divaricata differs in the section with two counts of 2n — It is possible that 2n — 16 is basic for sect. Lapeirousia and L. divaricata and that the ad- ditional two pairs in subsp. divaricata represent B chromosomes. DISCUSSION The extensive variation in chromosome number and karyotype in Lapeirousia is puzzling. The pat- terns of variation are particularly difficult to assess in the tropical sect. Paniculata, which is so chro- mosomally diverse yet appears to comprise a mono- phyletic assemblage (Goldblatt & Manning, 1990). There is no reason to doubt that the only two actinomorphic-flowered species of the section are similar to the ancestral type of the section. Yet at the chromosomal level these two species, L. avas- montana and L. coerulea, differ surprisingly. The former has a bimodal karyotype with 2n = 16, and the latter has 2n — 8 and weak bimodality. No direct polyploid relationship appears to link them and polyploidy seems an unlikely explanation for the numerical difference. Cells of L. avasmon- tana have about the same amount of chromosome material as L. coerulea, based on measurement of total chromosome length (Table 2). More likely, L. coerulea is dysploid and its lower chromosome number is the result of chromosome fusion by un- equal translocation and loss of centromeres and other nonessential genetic material (Jones, 1974, 1977; Goldblatt, 1979). I suggest that the basic karyotype in Lapeirou- sia is the strongly bimodal type that occurs in all sections, and exclusively in all but sect. Panicu- lata. Basic number d be x — 10, 9, or 8, most ikely 10. Base numbers in other genera in Wat- sonieae (Goldblatt, 1971, 1989. Goldblatt & Ма- p= Volume 77, Number 2 1990 Goldblatt Cytological Variability in Lapeirousia 381 FIGURE 2. sect. Sophronia (B-D) a odoratissima. —D. sect. Lapeirousia (F-H). — L. anceps. —E. L. . L. micrantha. — ramidalis. ms L. dolomitica vw dolomitica. —G. L. dolomitica subsp. um. d to, 8 A D ng Mitotic он іп Lapeirousia subg. Ni perius sect. Fastigiata (A) and in subg. Lapeirousia B. ds L. littoralis subsp. littoralis. — lewisiana. —H. L. violacea. Vouchers as given in Table 1. Scale bar, rais, 1976) are x = 10 (Thereianthus and ds cranthus), x = 9 (Watsonia), and = (Savannosiphon). The two first eet m a bimodal karyotype with one long and nine shorter pairs, and Watsonia has two long and seven shorter pairs (Goldblatt, 1971). In all three genera the degree of bimodality is less pronounced than in Lapeirousia. Savannosiphon (Goldblatt & Marais, 1976) does not have a bimodal karyotype. It seems reasonable to suggest that the ancestral base num- ber and karyotype of Lapeirousia were similar to those in Thereianthus and Micranthus, which probably have the basic karyotype for Watsonieae. The karyotype in Watsonia is also probably derived from the Thereianthus type (Goldblatt, 1971, 1989). Structural change appears to have proceeded in both subgenera of Lapeirousia toward reduction in number by fusion. The situation in subg. Lapei- rousia is somewhat confused by the occurrence of two numbers in some species, and this may be the result of the presence of B chromosomes or other types of supernumeraries, which are not uncom- mon in strongly bimodal karyotypes. The karyotype appears to be particularly un- stable in sect. Paniculata. Rapid structural change and accompanying decrease in chromosome num- ber, such as postulated above in Lapeirousia avas- montana and L. coerulea, appears to be frequent. It is presumed to have occurred within L. bainesii, in which x = 5 is probably basic. This base number and a similar karyotype are shared with the allied L. otaviensis and L. schimperi. The low n = 3 in two populations of L. bainesii can best be explained by so-called Robertsonian fusion of small acrocen- tric chromosomes to form correspondingly larger metacentrics. Significantly, the measures of total chromosome length (Table 2) are similar in the dysploid L. bainesii, and in L. otaviensis and L. schimperi, which have the presumed basic karyo- type for L. bainesii. Strongly bimodal karyotypes are present in La- peirousia gracilis, n — 6, and L. sandersonii, n — 5, both with long-tubed flowers. At least the rmer appears on morphological grounds (Gold- blatt, 1990b) to be closely related to L. bainesii and L. otaviensis, and this suggests the possibility that the bimodal karyotype may also be ancestral to the rather different basic karyotype in L. baine- sii, L. otaviensis, and L. schimperi. Their dis- tinctive karyotype, with its long submetacentric S chromosome pair, is presumably the result of ex- 382 Annals of the Missouri Botanical Garden tensive chromosomal rearrangement and fusion. There seems little doubt that the karyotype shared by these three last-mentioned species is a synap- omorphy indicating common ancestry. In Lapeirousia rivularis and the species of the L. erythrantha group the basic number is difficult to determine. These two species have 2n = 12 and a total chromosome length half again as much as in L. abyssinica and L. setifolia, which have 2n — 8. It seems reasonable to assume that the higher number, n — 6, in L. rivularis and L. erythrantha is polyploid (either dysploid from ancestors with 2n = 16 or derived from dysploid ancestors with 2n — 6) and that n — 4 in L. setifolia and L. abys- sinica is basic for the alliance. Possibly, п = 4 in the latter two species indicates a common ancestry with the actinomorphic-flowered L. coerulea, which also has this number. However, there are significant differences in the karyotypes of L. setifolia, L. abyssinica, and L. coerulea so that if this hy- pothesis is correct, then their karyotypes have di- verged appreciably. This model depends very much on crudely determined estimates of genome size (i.e., total linear measurement of illustrated chro- mosomes; Table 2), and this requires further in- vestigation. Critical measurement of genome size would be particularly useful in evaluating the re- lationships of the species of sect. Paniculata. The available data suggest that the unusually low numbers for Iridaceae, n = 4 and 3, may have been achieved independently in at least two sep- arate lines in subg. Paniculata (n — 4 in L. coe- rulea, L. setifolia, and L. abyssinica; n = 3 ina population of L. bainesii) and possibly more than twice if the hypothesis that L. abyssinica and L. setifolia are directly related to L. coerulea is not correct. Such parallels in dysploid reduction are also known in Iridaceae in Moraea, in which n = 10 is basic and n = 6 (or 5) has evolved in at least four lines independently (Goldblatt, 1976, 1986, unpublished data) and in three with the loss of all intermediate number e possession of a bimodal karyotype in Lapei- rousia may be related to its success in particularly arid habitats, among the most extreme in Iridaceae. The frequency of bimodal karyotypes is highest in plants of arid habitats (C. G. Vosa, pers. comm.), prominent examples being Agavaceae and the suc- culent Asphodelaceae- Alooideae. Numerical insta- ility in bimodal karyotypes is also known, although the reasons are not. The repeated patterns of more than one number in some species of Lapeirousia and in all sections except sect. Fastigiata may be another example of this phenomenon. In short, the variation in chromosome number and relative size in Lapeirousia are remarkable in lridaceae and for plants in general and merit further investigation for they may lead to a better understanding of the ways in which dysploidy occurs and its adaptive significance. More counts from new populations and methods such as photometric measurement of genome size and Giemsa banding may be useful in understanding better the cytological evolution in the genus. LITERATURE CITED BRIGHTON, C. 1976a. Cro Em Wu l. Crocus vernus aggrega Bull. 31: 32-46 1976b. Cytology of i ig us oliveri and its Kew Bull. 31: 209-2 1977. Cytology of p sativus and its allies. PI. Syst. Evol. 128: 157. De Vos, M. P. 1972. The fuc Ronnies in South Africa. J. 5. African Bot. Suppl. 9 Dyer, A. The use of lacto-propionic orcein in rapid squash methods. Stain Technol. 38: 85-90. GOLDBLATT, 1971. Cytological and Hae: studies in the southern African Iridaceae. J. S. Af- rican Bot. 37: 317-460. 1972. “A revision of the genera Lapeirousia Pourret and Anomatheca Ker in the winter rainfall region of South Africa. Contrib. Bolus Herb. 4: 1- 111. Cytological problems in the genus te. Kew allies. Evolution, cytology, and subgeneric Танайка. in Moraea (ЇгїЧасеае). Ann. Missouri Bot. ens 63: 1-23. —— ——. 1979. Chromosome cytology and variis pelis in Galaxia (Iridaceae). Pl. Syst. Evol. 133 61-69. 986. Cytology and systematics of the Mo- raea a fugax complex (Iridaceae). Ann. Missouri Bot. Gard. 73: 140- AA . The southern African genus Watsonia. Ann. Kirstenbosch Bot. Gard. 19: 1-148. 1990a. Phylogeny and classification of Iri- diced: Ann. Missouri Bot. Gard. 77 (in press). 1990b. Systematics of p rousia (lrida- ceae- -- [xioideae) i in tropical Africa. Ann. Missouri Bot. Gard. 77 (in press). a - MANNING. 1990. Leaf and corm struc- ture 1 L to phylogeny and A loc classification. Ann. Mis- souri Bot. Gar 65-374 & W. Marais. 1976. Savannosiphon gen. a fed of Lapeirousia (lridaceae- Ixioi- nn. Missouri Bot. 6: 8 0 ‚ М. KEUKEN & nov., deae). A HOLMGREN, P. E. K. SCHOFIELD. 19 bis Helium Part 1, ed. 7. Regnum Vegetabile 106. JACKSON, R. 1972 Chromosomal evolution in Haplo- pappus gra acilis: a centric transposition race. Evo- lution 27: 243-256. кы К. Chromosome evolution by Robertson- n translocation i м Gibasis (Commelinaceae). Chro mosoma (Berl.) 4 53-308. 197 ho role of Robertsonian change ir raid a ap РЕН in higher plants. eins Today 6: -129, BIOLOGIA FLORAL DE UNA Nelson Ramirez, Celia Gil, COMUNIDAD ARBUSTIVA a Alberto Seres, TROPICAL EN LA GUAYANA VENEZOLANA! RESUMEN El presente estudio analiza la biología floral de 55 especies de plantas de un arbustal en la Alta Guayana Venezolana. inflorescencia, o ambas). El diámetro de las flores varía significativamente de acuerdo al color. La longevidad floral es aproximadamente 12 horas para 35 (63.6%) especies. La orientación de las flores es horizontal en 28 (50.9%) especies, hacia arriba en el 40.0% (N= 22), y hacia abajo en el 9.1% (N = 5). Las flores con anteras de dehiscencia longitudinal (N = 36, 69.2%) están orientadas hacia arriba u horizontalmente. Las flores con anteras poricidas (N = 16, 30.89%) se disponen horizontalmente o colgantes. De 49 especies con polinización biótica la recompensa principal 5 de similaridad. А 90% de similaridad hay varios pares o grupos de especies que coinciden en su periodo de floración, Р utilizando las especies de polinizadores como caracteres son de poco valor predictivo para agrupar las especies de plantas de acuerdo a caracteres florales similares. ABSTRACT This study investigates the floral biology of 55 plant species from a shrub community in the Venezuelan Guayana dia 2 9.1% (N = 5) downward. Flowers with longitudinally — m = 36, 2%) are pee upward or horizontally, whereas species with poricidal anthers (N = 16, 8%) are oriented жы tally ог downward. From 9 plant species with biotic pollination, the main reward is nectar sti 20, 40.8%); however, pollen is given as the i % (N lengths of insect na However, the length/diameter ratio of flowers correlates with average size of insect аттанан (N = 49; г = 0.53; Р < 0.001). Cluster analysis оп floral characteristics of 55 plant species showed буе artificial groups at 37% similarity. At 90% similarity, there are some pairs or groups of species whose flowering eriods overlap. Some pollinators can be shared while in other cases the pollinators differ. Plant groups could not be established when they were clustered using pollinator species as characters Una de las preguntas que ha intrigado a los zación de una comunidad puede aportar información ecológos y evolucionistas, es ¿Cómo en comuni- a esta pregunta. De este modo, el desfasaje de dades de plantas se mantiene el flujo genético in- períodos de floración y fructificación entre especies traespecifico? El estudio de la biología de polini- es considerado como uno de los mecanismos que Agradecemos los valiosos comentarios y sugerencias a S. Renner, P. Berry, M. Condon, A. Herrera, W. S. Ga С. Sobrevila, у N. Xena. A las siguientes autoridades de sistematica vegetal por sus tada 'identifi- caciones del sur botánico: J. uin (Rubiaceae y varios grupos); J. Wurdack (Melastomataceae); C. Sastre (Orchidaceae); A. P. J. Maas y L. Cobb (Gentianaceae); J. Luteyn (Ericaceae); V. Badillo (Compositae); G. Carnevali (Orchidaceae); G. Agostini (Cybianthus); B. Garofalo, H. Debrot, y E. Cotton (Gramineae y Cyperaceae). También damos gracias a J. E. Ramirez por su colaboración en la M аш н de análisis de agrupamiento, a а López рог el trabajo mecanográfico, y а los Laboratorios Smith, Kline & French por el financiamiento de separa ? Universidad Central de Venezuela, Facultad de Ciencias, Escuela de Biología, Dpto. de Botánica, Aptdo. 20513, Caracas, Venezuela. ANN. MISSOURI Bor. GARD. 77: 383-397. 1990. 384 Annals of the Missouri Botanical Garden evita la interferencia entre especies que comparten polinizadores y dispersores comunes (Snow, 1965; Mosquin, 197 1; Gentry, 1974; Frankie et al., 1983; Stiles, 1975; Mori et al., 1978). La morfología floral también puede incrementar la selectividad de los polinizadores (Macior, 1971; Leppik, 1968, 1977; Orians et al., 1977; Feinsinger, 1978; Har- der, 1985). La divergencia en los tipos florales ha sido asociada con el desarrollo sensorial de los po- linizadores, particularmente en la capacidad de dis- tinguir y recordar ciertos tipos florales (Leppik, 1968). El tamaño floral y el de los polinizadores han permitido establecer algunas relaciones de in- terdependencia. La hipótesis de repartición del re- curso en función del largo de la flor y el largo de la proboscis del insecto destaca un acoplamiento entre ambas estructuras (Ranta & Vepsalainen, 1980; Harder, 1985). Sin embargo, la especifici- dad de los polinizadores ha sido sobrestimado y en casos de plantas con floración simultánea se ha encontrado un comportamiento oportunista (Fein- singer, 1978; Frankie et al., 1983; Kephart, 1983). Estudios de la biología floral en base a las ca- racteristicas morfológicas florales y al comporta- miento de las plantas en su ambiente contribuyen a entender la biología de polinización. En este con- texto, se han estudiado algunos parámetros como parte activa del proceso de polinización. La orga- nización de la inflorescencia, disposición de las flo- res, dehiscencia de las anteras, y la textura de las flores son caracteristicas florales que sumados a los colores, longevidad, y recompensa floral, contri- buyen a interpretar la biologia de polinización a nivel comunitario. Aunque estudios previos han analizado la dinámica reproductiva de varias co- munidades (Percival, 1974; Pojar, 1974; Simp- son, 1977; Arroyo et al., 1982; Bawa et al., 1985), las interrelaciones entre los atributos morfológicos florales y los tipos de polinización necesitan mayor número de estudios que permitan establecer las caracteristicas particulares de las comunidades. AREA DE ESTUDIO Esta investigación fue realizada en el Parque Nacional Canaima, sector Gran Sabana, ubicado al sureste del Estado Bolivar, Venezuela. Dicha area forma parte de la cuenca del Rio Caroni. El área de estudio seleccionada está situada aproxi- аи а 50 km al oeste del Fuerte Luepa, re Kavanayen y el campamento del Rio Parupa us N, 61?43'O) a una altitud de 1,350 m. Lo- calmente es llamada en lengua Pemon como **Gua- mu-pe" (Loma Guamu) y “El Jardin," nombre que alude a la abundancia de flores durante todo el ano. El sustrato edáfico esta formado en su mayor parte por areniscas rosadas, rojas o blancas del grupo Roraima, con algunos afloramientos de rocas igneas (Schubert, 1984). En el área seleccionada, el sustrato edáfico superficial está formado por arenas blancas que se tornan grises en algunos sitios por la materia órganica en descomposición. A poca profundidad del suelo arenoso destaca un enramado de raices delgadas y gruesas. Frecuen- temente afloran grupos de rocas, de tamano vari- able y sueltas entre si, entre las cuales crecen arbustos y hierbas. La temperatura media mensual de Kavanayen, situada a unos 15 km al este del area de estudio, presenta poca variabilidad a lo largo del ano (máx- ima 21.4°С y minima 19.9°C). La prm media anual es de 2,428 mm (rango: 1,8 ,400 m. mm), con una precipitacion mayor de 100 mm entre abril y diciembre. Los niveles de evaporación superan la precipitación desde enero a marzo, lo que sugiere un balance hidrico negativo. El área seleccionada ocupa una extensión de aproximadamente 10 hectáreas con una ligera pen- diente de aproximadamente 15 grados. La vege- tación es achaparrada, 1.0-1.5 m alto, y solo Clusia grandiflora destaca por alcanzar hasta 6 m (Ramirez et al., . Esta vegetación se in- tegra con las sabanas circundantes a traves de colonias de Heliamphora heterodoxa y Stegolepis angustata hacia las partes más bajas de la pen- diente o por un notorio esparcimiento de arbustos rodeados por Gramineae y Cyperaceae. Estudios previos sobre la estructura de la ve- getación y diversidad floristica permiten considerar al área de estudio como un arbustal siempreverde mesotérmico (Huber, 1986), desarrollado sobre suelos arenosos ері i plantas vasculares (91), dom mataceae (11.8%), Orchidaceae (10.5%), Com- positae (7.9%), y Cyperaceae (6.6%). Los periodos de floración y fructificación son mayores de un mes en la mayoría de las especies (Ramirez et al., 1988). METODOS Para un total de 55 especies de plantas se realizó un análisis morfológico basado en los tipos florales de Leppik (1977), en la secuencia evolutiva floral (haplomórfico, actinomórfico, pleomórfico, estereo- mórfico, y zigomórfico), y en la habilidad sensorial de sus polinizadores. Los colores de las flores, in- cluyendo el color principal y adicionales (guías de néctar, brácteas, estambres llamativos) fueron es- tablecidos utilizando el atlas de colores de Küppers Volume 77, Number 2 1990 Ramirez et al. 385 Comunidad Arbustiva Tropical en la Guayana Venezolana (1979). Posteriormente se establecieron seis ca- tegorías de colores: (1) blanco, considerando flores blancas y blancas amarillentas; (2) Rojo-rosado, considerando flores rojas, rosadas, y fucsia; (3) pardo, considerando flores marrones; (4) moradas, considerando flores morado claro e intenso; (5) amarillo; y (6) verde. Las dimensiones fueron re- gistradas para un total de 10 flores por especie. El largo de flores tubulares representa la longitud floral. En el caso de flores no tubulares, el largo floral representa la medida entre el receptáculo y la altura de pétalos o estambres. El diámetro es la medida entre ápices opuestos de los pétalos. En este caso se realizaron dos medidas perpendicula- res. Las unidades de polinización fueron caracteri- zadas de acuerdo a la organización de la inflores- cencia y al comportamiento de los polinizadores. Cuando las inflorescencias forman grupos densos, uniformes, en forma de seudantos y cuyas flores tienen una antesis relativamente sincronizada, en- tonces la inflorescencia es considerada la unidad de polinización. En este caso los polinizadores vi- sitan muchas flores simultáneamente, sin moverse de una inflorescencia a otra. Este tipo de organi- zación es considerado como *'colectivista" (Burtt, 1961). En el caso opuesto, las unidades de poli- nización son las flores individuales caracterizadas por inflorescencias laxas y/o sincrónicas en la an- tesis de las flores. Los polinizadores solo vistan una flor por turno; la visita de otra flor implica su vuelo. Una tercera categoría puede ser considerada como casos intermedios, caracterizado por la organiza- ción de las flores parcialmente densa, las cuales están dispuestas en un plano horizontal o ligera- mente diferente. La antesis es sincrónica o par- cialmente sincrónica. En este caso, los visitantes de gran tamaño se desplazan por la inflorescencia polinizando muchas flores y los visitantes de menor tamaño pueden visitar las flores en forma indivi- dualista. La biologia floral fue caracterizada en base a: (1) La orientación de las flores en relación al sus- trato, destacando flores orientadas verticalmente hacia arriba y hacia abajo. ores dispuestas horizontalmente se denotaron más precisamente de acuerdo a la variación angular hacia abajo y hacia arriba con respecto a un plano horizontal. (2) La longevidad floral fue estimada por observaciones directas en el campo para un mínimo de 10 flores por especies. En cada flor marcada en estado de yema se hicieron anotaciones sobre la receptividad estigmática, presencia de polen, y turgencia de los étalos a intervalos de seis horas durante dos o más dias. (3) El néctar y el polen como recompensa a los polinizadores fueron determinados por obser- vaciones en un minimo de 10 flores por especie, aisladas en bolsas de pelon. Esta información fue completada con información obtenida del análisis de la morfología floral. Los agentes polinizadores fueron estudiados para un total de 49 especies de plantas con sindromes bióticos y seis especies anemófilas (cuatro Grami- neae y dos Cyperaceae). La proporción de especies anemófilas incluidas en el total de 55 especies co- rresponde aproximadamente a la proporción natu- ral del arbustal (11 de 91). La actividad de los agentes visitantes se midió mediante períodos de observaciones continuas de 5 a 15 minutos por especie de planta. En algunos casos el período de observación fue prolongado a 30 minutos. La fre- cuencia de estos periodos de observación fue de cinco veces al día en cada especie. Esta metodología fue repetida durante tres a cinco días consecutivos o alternos en cada periodo de trabajo de campo (febrero—marzo; junio-julio y septiembre-octubre). teriormente fueron medidos en la longitud del cuer- po. Además se determinó el número de cargas de polen (representado por los distintos tipos de granos de polen) y la posición sobre el cuerpo del animal. La designación de polinizador fue establecida de acuerdo al comportamiento observado y corrobo- rado cuando la posición de la carga de polen sobre el cuerpo del animal hacia contacto con el estigma de las flores. Las variaciones de los tamaños florales relativos al color, simetría, tipo floral de Leppik (1977), y organización de la inflorescencia fueron estableci- das utilizando análisis de varianza de una sola vía. Previo a este análisis se realizó la prueba de ho- mogeneidad de varianza de Bartlett (Sokal & Rohlf, 1969). Las especies de plantas fueron agrupadas de acuerdo a las características florales (morfología y biología floral) y en base a las especies de poli- nizadores y visitantes previamente reportados (Ra- mirez, 1989) utilizando un análisis de agrupamiento (SAS, 1982). Este procedimiento está basado en el agrupamiento jerárquico de una matriz de datos multivariados donde los caracteres tienen el mismo matizada en forma de dendrogramas (SAS, 1982). RESULTADOS MORFOLOGÍA FLORAL Las caracteristicas morfológicas florales mues- tran gran variación en cuanto a la simetría, ta- maño, color, y orientación (Tabla 1). La simetría y el tipo floral (Leppik, 1977) están claramente Annals of the 386 Missouri Botanical Garden HIII гет Є`#Ї Id — N c-I I (Ds u Чишс 7) “y шпујиоХітә wmu122904. LI III esl 0°06 Id = М 1 14 (Ds ul "y ‘Yoox, 11421mquoyos biodoj0N 91 III Scl £'0c ld m: N e" 1 (8) 5 е "Я Y Н 020/8 vuvfeg SI ILIMIY IIFIFI 0'€ 0v oT + о4-№ 1 14 (D d I “Т 240yru2204 DIA yl әвәэв[1Аг) III I L'O сс oT + xS © A (DH 1 qunw ю0}р]]рхә vsodsoyoucyy El ПГП 90 6% oT + — c-€ A (DH 1 yuny (sasy) snpiziu sndapoouo8v'] ZI oeoov1odÁ^?) IDI #(€°€) 60 9v oT + N 1 ш (Ds ) O[peg siswaspayoqg ютиошәд 11 Ш *(Е`@) 90 0% ол + М 1 1q (Ys D ‘За »vyopisopisd оирууү 01 xo?punA 1e amey suaosaqnd стел IH gc V'61 o + М 2-1 19 (ws ul 'quioqQog ‘ао риртшрујиәд sidajojÁ duos 6 ГІ (876) @`1 vs oT + М 1 ш QDs 9 ds nuanjowosy) 8 IIFIFI #(9°SZ) EZ GS oT + о4-№ 1 d QD v 0) SP E 1d dio beds Demy BE aeysoduro”) П oS ү ST + od N I 1q (Н) 5 ! unus qp] PARADO Sao: Q әғРә?ецәшо:гЯ Ш ГҰР 009 oT -H N © 1 (8) 5 ч `Чшәң Dyofinn) onjduosiq С әвәәвтиои2т{ "ung wnsosns IIFIFI gl 91 oT H d OT-OI Id (9) Z y ‘ea "Koi wnupiupid UOLPUIPOY] Y IPIDPIY ПГП GG €I 01 + N-94 1 Id (н) a 1 sslay Xo TY Dpsnja4 хә} € avaoryoyinby IIFIFI Vee 6825 oT -H N 1 е (89) 5 ч WS “УЯ (9d V) 'шрузиәд vjpaspuvgy с IIFIFI 399 0°99 oT +H N © (WS ul uospoo y DUDIYÍAMQUOYIS DIO0YAOJ9D]DY 1 әғәоеи4оойұ epeaJosqo одәше(] ойлет еюиәэѕ [егор esuad (setp) 10[0) (eugaurs) uoroez satdadsq погәюло[ J ччәр X uorej -шоэәу рер [елор оар -tuod seue J бшш) pug. onem ep odı, -ueug) -IAQSUOT ap рери] "[e1snque [op sejue[d ap sardadsa сс әр [егор uoroejueuo А *esuedurooe: 'seordo[ojiour seysen) | УПЧУ], 387 Ramirez et al. Volume 77, Number 2 1990 Comunidad Arbustiva Tropical en la Guayana Venezolana ITI ха! ctl Id +H od 1 1 (Da ul eueu[ (pug) $иәпи n2020] OF IIFIFI 026 rl ld +H od I ul (Da ul UMOIg “Y `N Du42;D4f nun[onoqi] SE ITINI 6'6 97 Id +H od 1 1 (Da ul цоввә[с) (uag) vofipro9 DsayUDYdIg HE I LS Ж. 14 +H od I I (gd ul "DA (YPN) 020112 DUO EE ПГІГІ ТА £8 14 +H od €-Z ul (8g) d ul PUBLI] (Poey) 9744 do19]98 DUDILIY] ZE HIFIFI РОТ 0`1 Id +H od I ul (Da ul эа (TH 2$) vyofixn) 0132220. 1€ 1 30% ve Id +H od I ш (gd ul "pug отојагоа оәпоэруг 0E Ш UZI ve 14 +H od Z-I 19 (Dd ul "gueg 0] AydAyood рәпрәруү б 2B3IB]BULOISP|IA IIFIFI 0'8 S'I 01 + о4:№ 1 Id (WS - әлп8еу{ (1983) DIUNPD DSNAYIYA BZ IBIIRYIUBIO] Ш 901 S'I oT + °d-N I 1q S - TH 3s CIqny) »42fumsqoq тюїпшпң 17 ovoorLmung III I єс L9 01 +Н о4 1 [d (wa ul Ástoy;) (шет) snusofiuno] onusid 9g ШІ Les 08 oT +H od-ey 2-1 Iq (wd ч отпа -dsqns Asis D]psnd ют] SZ ПГП с'с9 Є`8 01 +H od-Y 6-1 19 (ч) а ul ‘щас ютлоуйритл@ msn ұс 9elogruin^) Sean] PIEM III I S'0 9'I oT + — €-Z A (DH 1 xe'g Э 'H) snsownjd иодоаХуэрір €z Ш 80 67 oT + m © A (DH 1 g RH unis unppdspq zc HINI L'O 6'I 01 + — “ А (DH 1 "ше wumnsoa42u UNIMDA IZ IFI c'o 0`1 oT + — €-z A (DH 1 ә$ецг) (1104) охәрит DuanjoulyI] Ос әвәшше1е) ulow (S 1ә y хә IIFIFI vo s'8 oT —H °d-N 1 ш (Da ul ЇР Д\) Psosowau DIYIDQ]A] 61 әвәэеиециә‹) uÁajn] x» `шыәЛәзс “Sey (USSISA хә Ту) UNID NUILIQNS ‘лел ШІ E'S OT 14 — oq-N I 19 (Ds ul 1eusste[N хә `Җ штулләдпа umm2204 91 ерелгәѕцо одәшец ойлет еюиәәѕ peop esuad (setp) Jojo (еіцәшіѕ) цор? saradsq цогәвло[ J -qəp uone -Ul029y рер [лор ой -tuod seue y ' шш) кюреши әр ойу -ueug -LAQSU0T ap pepun "'openuguo) ‘Т VI8v] Ё Ф y © ©] © 2 E 2$ m erue 5 23 o co c= < > 388 IIFIFI 0790€ TTIE oT -H od © e (Qs ul "KƏS DxOposajay DOY Ad UWMDJIH ZC IBIIRIUIIBIIPS IIFIFI € 1c VS 01 + № 2-1 Id (Ds ul "yula y $әрто1]08 vaundig 1с “hays (`Хәзс W PULIS) оурзәдоиәјѕ IIFIFI с'8 c9 oJ + N 1 Id (WS uy ‘dsqns "queg рлоріѕиәр mihuy Qc IIFIEI Cc 9€ oT + Od-N I ш Qs I ‘ÁIS (PULIS) sapioX4408. sisdoampsng 6Y nso2imaf `тел IIFIFI СФ 9% 01 + № I 19 (Ds 1 "ez ix CPIM) 250217n4/ nixnaqgooqq gt "ANS (MH) 2172427 сел ШІ sv 9% oT T N I Id (ws I эа Са Y Y) oimndos dDiuas0g Ly aeaoetqny ILI GPS 9'zI Id +H od I e (Н) d ul иоѕрә[‹) n;DjsnZun sidajosaig QF aesreaJedey IIFIFI 8'66 8 "Ip oT -H N I 14 (4)7 uy Три wnajsvip 2101405 GH П S'OI 0°06 ол -H N I 1 (9) 7 ul ушәмцәс `7) 03211) DIUOBO JS HF ПІ 8'v3 el -H N £-6 ed (4)7 ul [pur Dq071q sisdoug gp IIFIET 8801 с OT -H N I ш 92 YF 19qNH тәўәпр wniydaisidy zy әзиәп@рф1 `dsqns III U8 76 —H N I 1 (9) 7 ul MAH asuensnq1 wnipuapidy її ПГП Uve 070€ —H N £- FIGURA 1. en la Tabla 1. PORCENTAJE DE SIMILARIDAD m a | -0 | — NEN | — Análisis de agrupamiento para 55 especies de plantas en base a ocho caracteres np ic descritos Nümeros basales indican especies en la Tabla 1. Nümeros romanos indican grupos 394 Annals of the Missouri Botanical Garden PORCENTAJE DE SIMILARIDAD O M го : i 1 1 ? L 2319344 $ 3 © А! Análisis de agrupamiento para 55 especies de plantas en base a la presencia de polinizadores. Los agentes sin cargas tienen un valor de 1 y los polinizadores efectivos tienen valores de 2. Números basales indican especies de plantas en la Tabla 1. Números romanos indican grupos artificiales de afinidad. de polinización para insectos grandes). Esto evi- sino también por la selección de los polinizadores dencia que en la comunidad arbustiva los sistemas y la organización de la inflorescencia como unidad de polinización, no sólo pueden ser caracterizados de polinización. por la relación de tamano flor—tamano polinizador El polen y el néctar son recompensas básicas Volume 77, Number 2 1990 Ramirez et al. 395 Comunidad Arbustiva Tropical en la Guayana Venezolana que ofrecen las flores a los polinizadores en el arbustal. El porcentaje de especies que producen sólo néctar y en combinación con polen es similar al reportado por Percival (1974) para arbustos costeros de Jamaica. El elevado porcentaje de es- pecies productoras de néctar ha sido relacionado con la preponderancia de visitantes consumidores de néctar (Percival, 1974). Ciertamente en el ar- bustal estudiado hay una alta proporción de es- pecies ornitofilas y melitófilas productoras de néc- tar. Sin embargo, el 26.6% de las especies analizadas no producen néctar. La abundancia de especies anemofilas y de grupos taxonómicos adaptados a la polinización por vibración concuerdan con las proporciones encontradas. Además del polen y néc- tar, la resina ponente critico de ciertas comunidades tropicales (Armbruster, 1984). En el arbustal hay dos es- pecies del género Clusia cuyas flores femeninas floral es considerada como un com- producen resina como recompensa. La escasez de néctar en Bonnetia sessilis y la similitud morfo- lógica con Clusia especies podrian ser interpre- tadas como un caso de evolución convergente. La utilización indiscriminada de los polinizadores co- munes, probablemente de como resultado una es- trategia que reduce el costo de recompensa en B. sessilis, y mantiene a las poblaciones de poliniza- dores asiduos a estas especies. La ventaja de es- pecies con caracteristicas florales convergentes y polinizadores comunes ha sido señalada para plan- tas polinizadas por colibries (Brown & Kodric- Brown, 1979; Schemske, 1981; Kress, 1983). El mecanismo de polinización engañosa, en la cual los polinizadores son atraídos a las flores sin recompensa, aparece muy frecuentemente en Or- chidaceae No 1980, 1983; Dafni, 1983) y entre los arboles y lianas tropicales (Frankie et al., 1983; Haber, 1984). Este mecanismo ha sido aso- ciado a abejas jovenes inexpertas (Nilsson, 1980) y a los polinizadores con pobre capacidad discri- minativa (Dafni, 1983). La abundancia de especies con flores grandes y llamativas, y de floración si- multánea sugiere que la polinización engañosa pue- de ser una estrategia de polinización en el arbustal. Esto es reforzado por la marcada escasez en la producción de néctar, que fue más acentuada en rchidaceae entomofilas (Sobralia y Epistephium) y ornitófilas (Pogonia). Estas plantas fueron es- porádicamente visitadas por los polinizadores co- munes a otras especies del arbustal. Las flores adaptadas a la polinización por abejas tienen una longevidad de un día en la mayoria de las especies arbóreas de un bosque seco de Costa Rica (Frankie et al., 1983). En el arbustal, una alta proporción de las flores viven un dia. Percival (1974) encontró una alta proporción de flores res- tringidas a uno o medio periodo diurno, las cuales fueron asociadas a flores que reciben visitas por animales. La extensión en la longevidad floral po- dría estar asociada con la textura de las flores. Si bien todas las especies con un periodo diurno de actividad floral son zoofilas, en el arbustal hay especies melitófilas cuya longevidad es mayor, lo cual podría incrementar la probabilidad de polini- zación. Esta estrategia puede ser particularmente importante en especies con polinizadores poco pre- decibles. El caso más evidente ocurre en plantas anemófilas. El tamano de las flores es considerado como una función general del tamano del polinizador (Opler, 1980). Lindsey & Bell (1985) senalaron que el largo del tubo y del estilo puede ser correlacionado con el tamaño del insecto. Kodric-Brown et al. (1984) destacaron la misma situación entre flores y colibries. La longitud de la proboscis ha sido considerada una característica determinante en la selección de flores por parte de polinizadores (Ma- cior, 1979; Ranta & Vepsäläinen, 1980; Real, 1981; Pyke, 1982) . En contraste, las flores de tubo corto permiten el fácil acceso a gran variedad de polinizadores (Kephart, 1983). Sin embargo, existen excepciones que no apoyan la hipótesis de repartición del recurso (Macior, 1974; Bauer, 1983; Roubi uchmann, En el arbustal estudiado, la веса de las flores en relación al diámetro y el largo del tubo de la corola no apoyan la correlación entre tamaño flor y tamaño insecto, pero la combinación del tamaño floral expresado por la relación largo/diametro si se aproxima a reflejar el tamaño promedio de los insectos polinizadores. Esta relación destaca que solo la combinación de las dos medidas florales estan correlacionadas con la longitud promedio del con- junto de polinizadores. Flores con diámetros y lon- gitudes similares, son predominantemente polini- zadas por insectos pequeños, mientras que las flores cuya longitud es mayor que el diámetro son poli- nizadas por insector grandes. La validez de esta relación está sujeta a excepciones, entre las cuales destacan las flores poco profundas y con grandes diámetros (ej., Clusia spp.) y las flores pequeñas donde la inflorescencia es la unidad de polinización; ambasson polinizadas por insectos grandes y pe- queños. Las flores entomófilas del arbustal con dehis- cencia poricida son polinizadas por abejas vibrá- tiles, aunque Marcetia taxifolia es visitada por avispas y mariposas. La polinización en plantas con anteras poricidas pero visitadas por agentes no vibrátiles puede ser explicada parcialmente por el 396 Annals of the Missouri Botanical Garden hecho de que el viento (el cual es muy fuerte en el arbustal) puede facilitar la dispersión del polen (Tomlinson et al., 1979). Alternativamente, algu- nos insectos pueden morder las anteras y asi co- lectar el polen (Buchmann € Buchmann, 1981; Renner, 1983). Cuando los agentes visitantes son tanto vibrátiles como no vibrátiles, las visitas pre- vias por agentes vibrátiles depositan abundante po- len por toda la flor, lo cual podría contribuir a que otros agentes efectúen la polinización. La orientación de las flores respecto al sustrato puede ser considerada de gran utilidad en el estudio de los sistemas de polinización y del ambiente. Las flores polinizadas por agentes vibrátiles pueden es- tar orientadas en cualquier posición respecto al sustrato (Buchmann, 1978). El 9.1% de las flores del arbustal están orientadas hacia abajo; además, las corolas son urceoladas o subglobosas (constric- ción distal del tubo de la corola), o los pétalos están plegados a los estambres (Sauvagesia angustifo- lia). Esta organización floral conduce a que el polen en estas especies es depositado pasivamente sobre los pétalos y retenido principalmente en la cons- tricción de la corola. En este sentido, la colección de polen por agentes no vibrátiles (aves, dipteros, y abejas) puede ser efectiva. Además, esta orien- tación floral ha sido asociada con la Selec ridad por parte del polinizador (Tomlinson et al., 1979; Sullivan, 1984), complementariamente, podría re- ducir la interferencia de la lluvia, la cual es fre- cuente en el área de estudio. El comportamiento fenológico de las plantas con caracteristicas florales similares tiende a estar se- parado temporalmente para evitar la competencia por los polinizadores (Gentry, 1974; Frankie et al., 1983; Whitney, 19 teres morfológicos florales ligeramente uniformes pueden estar sometidas a la carencia de recono- 84). Las especies con carac- cimiento por parte de los polinizadores (ej., Sulli- van, 1984). Los sistemas de polinización raras ve- ces son idénticos (especies de polinizadores) en las especies de plantas del arbustal de la Gran Sabana (Ramirez , 1989). Algunas especies de plantas con caracteres morfológicos similares pueden tener po- linizadores comunes o diferentes. Esto sugiere que la similitud morfológica de los caracteres florales considerados en este trabajo no es evidencia di- rectas para suponer especies de polinizadores co- munes. Sin embargo, los grupos afines de especies de plantas, basados en los caracteres morfológicos, permiten destacar que hay mayor similitud en re- lación a los grandes grupos de polinizadores en comparación con la afinidad taxonómica. Clara- mente, en casos donde hay una alta especialización a un solo agente polinizador (ej., anemofilia), las plantas son categorizadas en grupos homogéneos de acuerdo a los caracteres morfológicos florales y a su posición sistemática (ej., Cyperales). La delimitación de grupos de plantas en base a los agentes polinizadores permite destacar que no existen grupos las especies eo Lu de polinizadores, sino таз bien un conjunto de plantas que superponen algunas caracteristicas morfológicas comunes, asociadas a ciertos grupos de visitantes: las especies polinizadas por pequenos insectos con flores pequenas y colores claros y las plantas polinizadas por abejas medianas y grandes con flores grandes y de colores llamativos. La va- riabilidad observada en este grupo de plantas podria estar asociada a la diversidad del conjunto de es- pecies visitantes. En contraste, las especies orni- tófilas y anemófilas están fuertemente relacionadas en sus caracteres morfológicos florales y agentes polinizadores. LITERATURA CITADA ARMBRUSTER, W. 1984. The role of resin in angio- sperm bo ecological and chemical consid- erations. Amer. J. Bot. 71: 1149-1160. ARROYO, М. Т. K., R. Primack & J. ARMESTO. 1982. керене, я studies in n Vias d in the high temperate Andes of Central Chil Pollination о апа а variation. Amer. J. Bot. 69: 82 ү Bauer, P. J. 1983. Bumblebee die veh relationships on the Beart Plateau tundra of Southern Mon- tana. r. J. Bot. 70: 134-144 Bawa, SH. Buttock, D. R. Perry, R. E. COVILLE & H. reum 1985. Reproductive biology of tro- pical lowland rain forest trees. II. Pollination systems. Amer. А Bot. 72 Brown, J. . Kopric-Brown. 1979. Conver- gence, оо and mimicry in a temperate com- munity of hummingbird-pollinated flowers. Ecology 1035. BUCHMANN, S. L. 1978. Abiophysical model m Erg pollination i in angiosperms. J. Theor. Biol. 557. . BUCHMANN. 1981. Anthecology of Mouriri myrtillioides (Melastomataceae: Memecy- leae), an oil flower in Panama. Biotropica 13 (sup- plement): 7-24. Bunrr, B. L. 1961. tional evolution. Trans. 39: 216-232 DAFNI, A. 1983. Pollination of Orchis caspia, a nec- tarless plant which deceives the pollinators of nec- tariferous species from other plant families. J. Ecol 4. Compositae and the study of func- & Proc. Bot. Soc. Edinburgh 71: 467-47 FEINSINGER, P. 1978. Ecological interactions betweer plants and Са in а successional ah community. Ecol. . 48: 287 FRANKIE, G. W., A. R P. А. a ER 8 К. 5. Bawa. 1983. ошын. and organization of the large bee pon ion system in the Costa Rican dry forest. Pp. 411-447 in C. E. Jones & R. J. Little (editors), Volume 77, Number 2 Ramirez et al. 397 1990 Comunidad Arbustiva Tropical en la Guayana Venezolana SIMPSON, J. C. ScHuLTZ & C. S. ToMOFF. 1977. Handbook of Experimental Pollination Biology. Van Nostrand Reinhold, New Yor GENTRY, А. Coevolutionary patterns in Central American Bignoniaceae. . Missouri Bot. Gard. 61: 728-759 HABER, W. A. 1984. Pollination by deceit in a mass- flowering tropical tree Plumeria rubra L. (Apocy- naceae). Biotropica 16: 269-275. HARDER, D. L. Mus ology as a predictor of flower choice by bumblebees. Ecology 66: 198- 2 HUBER, O. 1986. La vegetación de la cuenca del Río Caroni. Interciencia 11: 301-310. KEPHART, S. R The partitioning of pollinators among tree species of Asclepias. Ecology 64: 12 133 KopRIc-BROWN, A., J. Н. Brown, С. S. Byers & D. F. 984. Organization of tropical island com- munity of hummingbirds and flowers. Ecology 65: 368. 983. ав), barriers in neotropical Heliconia. Ann. Bot. (London) 52: 131-147. Atlas = los Colores. Editorial ume, Barcelona, Espan Hd E. E. 1968. Dire-tional trend of floral evolution. Acta Biotheor. 18: 87-102. 977. Floral evolution in relation to ara ecology. In: Verghese (editor), International Bros- To da ay & Tomorrow’s Printers 2d ico A Reproductive biology of Apia- ceae. 1. Floral visitors to Thaspium and Zizia and ni importance in pollination. Amer. J. Bot. 71: 87. . BELL. 1985. Reproductive biology of Apiaceae. II. Cryptic specialization and floral evo- lution in Thaspium and Zizia. Amer. J. Bot. 72: 231-247 Macior, L. W. Co-evolution of plants and ani- iplis ae к from plant-insect inter- actions. Taxon 20: 17-18. 197 Pollination ecology of the forest range of the Colorado Rocky Mountains. Melanderia 15: 1- . 1979. Bumblebees as pollinators of native and introduced plants. Proc. IV. Th. Frit. Symp. on Polli- А aryla ae Exp. Sta. Special Misc. Publ. 1. -44 Mont, 5 y C. & A. B. BoLTEN. 1978. Additional notes on the floral biology of neotropical Lecythidaceae. apum 3-130 MosqUIN, T. 197 мнн скы for pollinators as a stimulus pn de evolution of flowering time. Oikos 22: 398- NILSSON, L. A. 80. The pollination ecology of Dacty- orhiza sambucina (Orchidaceae). Bot. Not. 133: 367-385. . 1983. Anthecology e Orc ko mascula (Or- со Nordic J. Bot. 15 79, OPLER, Р. A. 198 Nectar о in a tropical желе m. In: В. Bentley & Т. Elias (editors), Bio- ogy of Nectaries. Columbia Univ. Press, New York ORIANS, С. i R. G. Cates, M. A. MARES, A. MOLDENKE, J. NEF D. F. Ruoapes, M. L. RosenzwEIG, В. B. 165-224 in G. H. ose’ (editors), Convergent Evo- Deserts. Dowden, Hutchinson & Resource utilization systems. Pp. Orians & O. lution in Warm Ross, Stroudsburg. — M. 1974. Floral ecology of coastal scrub in utheast Jamaica. Biotropica -129. Pun 1. 1974. Reproductive аад of four plant communities of southwetem British Columbia. Canad. J. Bot. 52: 1819-1834. Pyke, G. Н 82. Local geographic oe of umblebees near Crested Butte, Color tition and community structure. Ecology 63: 5995- 573. MEE N. 1989. Biologia de polinización en una unidad arbustiva Iopies al de la Alta Guayana (CRM керез а -3 . Git, М. Lop ‚ О. HokcHE Е Y. Brito. 1988. ‘Caracterización ro ener ural de una comunidad arbustiva en la Alta Guayana a (Gran Sabana, edo. Bolívar). Acta Ci. Venez. 39: 57-4 457- А Ranta, E. & K. VEPSÄLÄINEN. 1980. Why are there REaL, L. A. Nectar availability and bee- he on Ipomoea nm Reproductive Botany. Suppl. Biotropica 13: m Th RENNER, S. 1983. The f anther destruction by Trigona bees in Melastomataceae. Biotropica 15: -256. ROUBIK, W. & s T BUCHMANN. Nectar selection by Melipo and Apis А (Hymenop, tera: Apidae) and the ecology of nectar intake by colonies in tropical forest. Oecologi : 1-10 SAS. 1982. User’ s Guide: Stastics. SAS Institute, Cary. Floral convergence and polli- nator sharing in two bee-pollinated tropical herbs. Ecology 62: 946-954. SCHUBERT, С. 1984 үш geológicos de la Gran Saba n (editor), La Protección de la C А Simpson, В. 1977. Breeding systems of dominant pe- rennial plants of two disjunct warm desert ecosys- tems. me 27: 203-226. 1965 Snow, D. W. A possible selective factor in the evolution of fruiting seasons in tropical forest. Oikos 15: -281. SokaL, R. R J. RomLr. 1969. Biometry. W. Н. Freeman, San Francisco. STILES, F. G. 19 Ecology, flowering phenology and Rican Heli- SULLIVAN, J. R Pollination biology of de dins vicosa var. cinerascens (Solanaceae). Amer. 71: 815-820. TOMLINSON, Р. B., В. B. Primack & J. S. Bunt. 1979. Preliminary observations on floral biology in man- grove Rhizophoraceae. Biotropica 11: 256-277 WHITNEY, С. С. e reproductive biology of йер and me pollinator community structure. r. J. Bot. 71: 887-894. THE SYSTEMATICS OF SOLANUM SECTION ALLOPHYLLUM (SOLANACEAE) Lynn Bohs? ABSTRACT species of Solanum are included in a new section, Solanum sect. have been previously known; the third, 5. Allophyllum. Two of the species, S. dp is described as new. i -3-leaved sympodial units, simple leaves with decurrent bases, and tapered anthers, is unique to this group and de not agree well with any existing subgenus of Solanum. A group of three species first attracted my at- tention during monographic work on the solana- ceous genus Cyphomandra Sendtn. Solanum al- lophyllum (Miers) Standl., the most widespread and best-known species of the section, had been placed by various workers in Cyphomandra and in Solanum. Determination of the true affinities of this species entailed an examination of the validity of the characters separating the two genera. Ап account of the morphology of this species and its resultant exclusion from Cyphomandra is given in a separate paper (Bohs, 1989). Bitter (1914) first noted the similarities between S. allophyllum (his S. ellipsoideibaccatum Bitter) and S. mapiriense Bitter (his 5. phytolaccoides (Rusby) Bitter), but he did not create a forma taxonomic category to include them. Child (1984) erected sect. Allophylla to contain these two species and regarded the new section as belonging to Cy- phomandra rather than to Solanum. He also sur- mised that Cyphomandra chlorantha Rusby be- longed to sect. Allophylla, but Bohs (1986) confirmed that it belongs in Cyphomandra and that it has no close relationship to S. allophyllum or S. mapiriense. A recent analysis of S. allo- phyllum (Bohs, 1989) indicates that this species should be placed in Solanum instead of Cyphoman- dra. Accordingly, sect. Allophylla is here removed from Cyphomandra and placed in Solanum. All three species of Solanum sect. Allophyllum share the following combination of characters: (1) upright herbs or shrubs (up to 2 m tall); (2) 2-3- leaved sympodial units; (3) vegetative parts densely covered with “sand-punctae” (idioblasts containing crystal sand); (4) membranaceous leaf blades, usu- ally with decurrent bases and winged petioles; (5) unbranched, few-flowered, relatively short inflo- rescences; (6) flowers with usually white or greenish white membranaceous corollas; (7) stamens with glabrous filaments, distally tapered anthers, and very small terminal pores; (8) glabrous, filiform styles with small truncate stigmas; and (9) glabrous, globose to ellipsoidal fruits with obtuse apices. Ad- ditional characters may further distinguish the sec- tion, but these are as yet imperfectly known for all three species. For instance, 5. allophyllum and 5. mapiriense have light-colored corollas and an- thers both with darker spots at the base, but such spots have not been observed in S. morellifolium. Also, the fruits of 5. allophyllum have a distinctive color and shape, being white to light orange marked with dark green or purple longitudinal stripes and being ovoid to ellipsoidal in outline but flattened on opposite sides (Bohs, 1989). The color and shape of mature fruits of S. mapiriense and S. morel- ji are unknown of the ане states that define sect. Allophyllum are found in many other sections of Solanum and are probably plesiomorphic. These include (1) herbaceous habit; (2) exclusively simple, unbranched hairs; (3) absence of prickles; (4) pres- ence of sand-punctae; (5) cuneate and decurrent leaf bases; (6) glabrous filaments; (7) glabrous, fil- iform styles with small truncate stigmas; (8) gla- ! Į thank the herbaria listed for loans of specimens, M. Nee ‚ W. G. D'Arcy, and S. Knapp for helpful discussions and specimens, Julie Mitchell for the illustration of S. morellifolium, and C. McPherson for helping me search, albeit unsuccessfully, for 5. allophyllum in Panama 2D epartment of Biology, University of Utah, Salt Lake City, Utah 84112, U.S.A. ANN. MISSOURI Вот. Garp. 77: 398-409. 1990. Volume 77, Number 2 1990 Bohs Systematics of Solanum sect. Allophyllum brous, obtuse fruits; (9) stone granules in the fruits; and (10) small, flattened seeds. Probable derived character states manifested by members of the section include (1) tapered anthers with small ter- minal pores; (2) 2-3-leaved sympodial units; and (3) self-compatible breeding system (known only for S. allophyllum). The evolutionary significance of the laterally compressed fruits seen in S. allo- phyllum (and possibly present in the other two species) is uncertain, but compressed fruits occur in other subgroups of Solanum, such as the sub- genera Potatoe (С. Don) D’Arcy and Bassovia (Aubl.) Bitt. The distribution of flattened fruits among other solanaceous genera is unknown to me. Perhaps this character is relatively insignificant taxonomically and may be related to the enlarge- ment of the ovary wall and septum during fruit development. In summary, sect. Allophyllum seems to com- bine a host of primitive features with a few seem- ingly derived character states, most notably ta- pered anthers. PLACEMENT OF THE SECTION WITHIN SOLANUM Solanum is one of the largest genera of flowering plants, containing an estimated 1,500 species (D'Arcy, 1979; Hunziker, 1979). Substantial taxo- nomic problems still exist with regard to Solanum and related genera. At present, Solanum is loosely defined and consists of plants with poricidally de- hiscent anthers that lack the specialized features of associated segregate genera, such as enlarged anther connectives in Cyphomandra Sendtn., ster- ile anther beaks in Lycopersicon Mill., and dis- tinctive calyx morphology in Lycianthes (Dun.) Hass]. Although each of these segregate genera may be monophyletic, monophyly has not been established for Solanum. Poricidal anther dehis- cence may have evolved more than once in Sola- num and its allies, and may thus not be a synap- omorphy uniting these taxa. Certainly our understanding of phylogenetic relationships within Solanum is hampered by the lack of basic taxo- nomic and morphological information for many subgenera and sections within the genus. It seems inevitable that Solanum will be broken up into smaller monophyletic genera, but taking such a step now would be premature based on our current state of knowledge. The infrageneric classification of Solanum and the boundaries between Solanum and its related genera have not been firmly estab- lished, and the taxonomy of this group is likely to remain in a state of flux for many years to come. The placement of sect. Allophyllum in Solanum is enigmatic. No existing group in Solanum has the combination of tapered anthers, simple hairs, upright habit, simple leaves, and spinelessness seen in sect. Allophyllum. In fact, the anthers of this group resemble those of some species of Ly- cianthes, but sect. Allophyllum does not have the truncate calyx with subapical lobes, reduced inflo- rescences, and large stone cell aggregates in the fruits that distinguish Lycianthes (D'Arcy, 1986). 50500 Allophyllum may represent a new genus with Solanum and Lycianthes, but no synapomorphies are known that distinguish sect. Allophyllum from Solanum. Giv- en the possible polyphyly, or at least paraphyly, of Solanum, sect. Allophyllum, differentiated on the basis of a combination of characters but with no obvious synapomorphies, must be placed in Solanum. Further work on generic limits between SS and its allies will undoubtedly shed light the proper generic placement of this section. Additional problems surface when the affinities of sect. Allophyllum within Solanum are considered, because the infrageneric classification scheme in Solanum leaves much to be desired. Several attempts have been made to divide Sola- num into subgenera and sections (e.g., Dunal, 1852; Seithe, 1962; Gilli, 1970; Danert, 1970; D’Arcy, 1972). The most widely used recent scheme is that of D’Arcy (1972), who accepted seven subgenera of Solanum, in a classification with basic elements derived from earlier literature. I will refer exclu- sively to his classification in the discussion below. Although many well-defined sections can be dis- cerned within Solanum, the circumscription of sub- genera is more problematic. Most of the subgenera consist of one or several well-defined or well-studied sections along with a number of sections whose affinities are less certain. Thus, D’Arcy’s subgenera are often clearly defined with respect to one or a few component sections, but the boundaries be- tween the sugenera are less clear when all the associated sections are considered. The large number of species in Solanum and the relative paucity of adequate taxonomic treat- ments for many infrageneric groups hampers un- derstanding the genus. Because I lack the facilities for an exhaustive survey of morphology and vari- ation in the entire genus, it is possible that I have overlooked existing groups in Solanum that might be allied with sect. Allophyllum. Nevertheless, I will attempt to enumerate the possibilities for place- ment of sect. Allophyllum within the existing sub- genera of Solanum as 1 understand them. Section Allophyllum probably does not belong to the subgenera Archaesolanum Marz. (aneu- Annals of the Missouri Botanical Garden mapiriense morellifolium allophyllum FIGURE 1. Hypothesis of phylogenetic relationships in Solanum sect. Allophyllum. ploid-polyploid South Pacific species), Lyciosola- num Bitt. (restricted to a single South African species, 5. aggregatum Jacq., with sessile inflo- rescenses, short shoots, and very long filaments), or Brevantherum (Seithe) D'Arcy (exemplified by sect. Brevantherum Seithe, which contains species with entire leaves, elliptic to oblong anthers with blunt apices, and stellate hairs). Bitter (1913) and Morton (1944) postulated an alliance of S. allophyllum with subg. Leptoste- monum (Dun.) Bitt. because both groups have ta- pered anthers with small terminal pores. Such a relationship is doubtful, however, because all species of sect. Allophyllum lack the stellate hairs and prickles characteristic of subg. Leptostemonum. Members of sect. Allophyllum resemble vege- tatively those of sect. Solanum of subg. Solanum and have often been identified as members of sect. Solanum in herbaria. Both sections contain her- baceous plants with ovate to ellipsoidal leaf blades having decurrent bases; usually extra-axillary, rel- atively few-flowered inflorescences with an obvious peduncle; and often small aggregates of stone cells in the fruits. The lobed leaves of S. allophyllum also resemble those of S. dulcamara L., which D’Arcy placed in subg. Potatoe (С. Don) D’Arcy but which probably belongs to subg. Solanum. However, all species of subg. Solanum have oblong to ellipsoidal anthers that dehisce by large terminal pores opening into longitudinal slits rather than the tapered anthers with small pores that characterize sect. Allophyllum. Thus the resemblance of sect. Allophyllum to members of subg. Solanum ap- pears to be superficial and based only on vegetative characters. Section Allophyllum is not easily accommodat- ed in subg. Potatoe. Although some members of this subgenus have more or less tapered anthers, they are also dehiscent by large terminal pores that frequently develop into longitudinal slits. Many species of this group have twining stems and pin- nately compound leaves, and self-incompatibility is common. None of these attributes suggests a re- lationship with sect. Allophyllum e remaining subgenus, subg. Bassovia, also does not appear to be a good candidate. Most members of this group have stout anthers with large apical pores that later open into longitudinal slits, and inflorescences located in leaf axils. These char- acteristics are not seen in sect. Allophyllum. Ta- pered anthers occur in S. reptans Bunb. of the monotypic sect. Herposolanum Bitt., but they open by longitudinal slits, and this species has no further morphological resemblance to sect. Allophyllum. (The inclusion of Solanum reptans in subg. Bas- sovia is itself doubtful.) Laterally compressed fruits occur in some species of this subgenus, but so little is known of the distribution and development of this character that it is not possible to determine whether this is an important link between sect. Allophyllum and subg. Bassovia. The preceding considerations indicate that sect. Allophyllum does not fit well into any of the ex- isting subgenera of Solanum. The group certainly deserves sectional rank, but given the problems already noted for its placement at both the generic and infrageneric levels, I am hesitant to create a new subgenus for it now. INFRASECTIONAL RELATIONSHIPS Within the section, S. allophyllum is most likely the most advanced species, exhibiting a suite of d characters, such as pinnately lobed AMA inflorescences situated in branch forks, ro- tate-stellate corollas, and large, ellipsoidal fruits. Solanum mapiriense may be the most primitive element of the section, exhibiting none of the de- rived features mentioned above for 5. allophyllum. Solanum morellifolium shares with S. allophyllum few-flowered inflorescences, rotate-stellate corollas, and small ovate-elliptic leaf blades; on the other hand, S. morellifolium resembles S. mapiriense by having extra-axillary inflorescences, small glo- bose fruits, and by lacking pinnately lobed leaves. The presence of gland-tipped finger hairs (un- branched multicelled stalks with a unicelled glan- dular tip) on the stems and inflorescences of S mapiriense and some collections of 5. morellifol- ium may be an indication of the relatively primitive status of these species, for gland-tipped finger hairs are thought by Seithe (1979), based on her on- togenetic studies, to be the most primitive hair type in Solanum. Finger hairs (without glandular tips) and multicellular glands (unicelled stalks with mul- Volume 77, Number 2 1990 Bohs 401 Systematics of Solanum sect. Allophyllum ticelled glandular tips) are the hair types found on the mature foliage and stems of S. allophyllum. A preliminary hypothesis of the relationships among these three species is given in Figure 1. This should be interpreted only as a hypothesis for further examination and testing. Á rigorous cla- distic analysis can be attempted only when the proper placement of sect. Allophyllum within Solanum is known, when further information is gathered for many of the poorly characterized groups within the genus, and when appropriate hypotheses can be made concerning character po- larity within and among the various lineages of Solanum. TAXONOMIC TREATMENT Solanum section Allophyllum (Child) Bohs, comb. nov. Cyphomandra section Allophylla Child, Feddes Repert. 95: 292. 1984. Type species: Solanum allophyllum (Miers) Standl. Herbae vel frutices fere glabra vel pilis simplicibus, catum. Bacca glabra ovoidea vel ellipsoidea apice obtusa. Herbs or shrubs up to about 2 m tall. Branches glabrous to sparsely puberulent, abundantly sand- punctate. Leaves 2-3 per sympodial unit, the blades simple, lobed or unlobed, chartaceous to mem- branaceous, abundantly sand-punctate, the apex acuminate, the base subcordate to truncate or cu- neate, decurrent along petiole; the midrib and ma- 1 jor la teral glabrous to minutely puberulent, especially i in adax- ial channel. Inflorescence extra-axillary or located in a branch fork, unbranched, with up to 15 flow- ers, shorter than 10 cm long; pedicels 3-9 mm long, articulated at the base. Calyx membrana- ceous, veiny and sand-punctate. Corolla white or greenish (possibly blue?), usually with darker spots at the base, membranaceous, veiny and sand-punc- tate, glabrous to sparsely puberulent abaxially and adaxially, the margin glabrous to minutely tomen- e tiole > winged, the corolla base, the free part of the filaments shorter than 1 mm and much shorter than the anthers; anther thecae narrowly triangular, strong- ly tapered distally, sagittate at base, thickened abaxially in proximal half, the pores very small and directed distally. Ovary glabrous, ca. 1-3 mm long, 1-2 mm diam.; style glabrous, filiform, less than 0.5 mm diam., longer than the ovary; stigma truncate, the same diameter as style. Fruit gla- brous, obtuse at apex, ovoid or ellipsoidal, laterally compressed in S. allophyllum and possibly also in the two other species. KEY TO SPECIES OF Sor. 4NUM SECTION ALLOPHYLLUM la. pen iet unlobed, elliptic, 10-27 cm long, wide; inflorescences 8-15-flowered; corola stellate the lobes 6- y mm long and 3- PEPEN СГ Solanum с 1Ь. Leal “blades lobed or a the unlobed leave h blades ovate, ovate-elliptic, or il 2. 5- 17 cm long, 1.3-7.5 cm wide; inflores- cences 3- 7-flowered; corolla rotate-stellate, the lobes 2-6 mm long and 1.5-3 mm wide. 2a. Inflorescences 0.5-2.5 cm long; peduncles 0.3-1.5 cm long; corolla radius 4-5 mm long, the tube 1-2 mm long, the lobes 2- 4 mm long; anther thecae 2.5-4 mm long; fruits globose, ca. 1 cm long and 1 ст in diameter; eastern s and adjace ivia MEET ка кка p lanum morellifolium Inflorescences p ко = northwestern South ши о allophyllum 1. Solanum mapiriense Bitter, Repert. Spec. Nov. Regni Veg. 11: 16. 1912. TYPE: Bolivia. La Paz: San Ant ear Mapiri, 850 m, Dec. 1907, Buchtien 1434 (lectotype, US #1399273, here d ated; isolectotype, US #1175838). Bassovia е Rusby, Bull. М.Ү. Bot. Gard. 4: 317. 1907. Solanum А (Rusby egni Veg Dec 1740 (lectotype, NY, here rale i isolectotypes, A, BM, E, CH, NY, US). Figure 2. Herb or shrub ca. 1-2 m tall. Branches gla- brous to sparsely puberulent with gland-tipped fin- ger hairs. Leaves 2 per sympodial unit, the blades unlobed, elliptic, glabrous to sparsely puberulent or pubescent adaxially, nearly glabrous abaxially, (5-)10-27 cm long, (2-)4-12 cm wide, the length : width ratio ca. 2-3 : 1; major veins 5-11 on each side; petioles 1-5 cm long. /nflorescence extra- axillary, 8-15-flowered, 2-6 cm long; peduncle 402 Annals of the Missouri Botanical Garden PLANTE BOLIVIANA Solanum nutans, R. 4 P y No. 1740 - FIGURE 2. Isolectotype of Bassovia phytolaccoides (= Solanum mapiriense) (Bang 1740 BM). 1-3 cm long; rachis 1-4 cm long; pedicels 5-9 mm long, in fruit 7-9 mm long, spaced (1-)2-6 mm apart. Peduncle, rachis, and pedicels sparsely to moderately puberulent with curled gland-tipped finger hairs. Calyx glabrous to moderately glan- dular-puberulent, the radius 2 mm, the margin nearly entire with very shallow obtuse lobes. Co- rolla white to cream or greenish white with green spots at base (with deep blue center fide Rusby), stellate, the radius 8- 10 mm, the tube 1-3 mm long, the lobes 6-8 mm long, 3-4 mm wide, elliptic to ovate, the apex short-tomentose. Stamens ca. 6 mm long; anther thecae yellow or white with green spots at base, (darker at the base fide Rusby), narrowly triangular, ca. 4-6 mm long, 1.5-2 mm wide at base, ca. 0.5 mm wide at apex. Style ca. 6 mm long, 0.2-0.5 mm diam.; stigma 0.2-0.5 mm diam. Fruit globose, 5-9 mm long, 4-9 mm diam. (when immature?), the color when ripe un- known; presence of stone cell aggregates in me- socarp unknown (not present according to Bitter, 1912); seeds unknown. Distribution. Known only from moist forest of western Bolivia in the Province of La Paz, 850— 1,700 m (Figure 3). Additional “үе е те BOLIVIA. LA PAZ: Mapiri region, San Carlos, 850 m, 12 Dec. 1926, Buch- tien 1259 (NY); Prov. Nor Yungas 10 km by road N and above Caranavi, са. 15?47'S, 67°32'W, 1,400 m, 1 Nov. 1984, Nee & Solomon 2 305 (NY, UT) Prov. Nor Yungas, 4.6 km below Yolosa, then 19.1 km on road up the Rio Huarinilla, 16°] 2'S, 67°53 W, 1,700 m, 12 Nov. 1982, Solomon 8757 (NY). This species differs from the others of the section by its large elliptic leaves and relatively large stel- late corollas with long lobes. The fruits are small Volume 77, Number 2 1990 Bohs Systematics of Solanum sect. Allophyllum о 200 400 600 800 1000кт LA ee il "TJ Q 100 200 300 400 500 600 miles FIGURE 3. Distribution of Solanum sect. Allophyllum. Dots = S. allophyllum. Triangles = S. morellifolium. GURE Stars = S. mapiriense. (Base map copyright 1979 by the University of Utrecht. and globose like those of S. morellifolium. Thus far S. mapiriense has only been collected from a restricted area of western Bolivia. Although it is not possible to determine fruit shape from the herbarium specimens available, Rusby’s description mentions that the fruits of S. mapiriense are slightly depressed and have two grooves. Laterally compressed fruits may therefore be characteristic of this species as well as of S. allophyllum. Solanum mapiriense is the correct name for this species, as the combination Solanum phyto- laccoides (Rusby) Bitter is a later homonym of S. phytolaccoides C. H. Wright. Rusby did not specify the holotype for his Bas- sovia phytolaccoides, so a specimen at NY bearing his annotation has been chosen as the lectotype. Bitter cited two syntypes for S. mapiriense, Bang 1740 and Buchtien 1434. He stated in the protologue that he examined these specimens in 404 Annals of the Missouri Botanical Garden "herb. Buchtien." Although most of Buchtien’s collections are at US, duplicates exist elsewhere, so "herb. Buchtien" does not precisely reveal the location of his types. The only specimens I have seen with a definite indication that Bitter examined them are Buchtien 1434 (US #1399273) and Bang 1740 (US 41175839) at US. The former sheet bears the annotation “Solanum mapiriense Bitter (det. Bitter)," but the handwriting does not appear to be Bitter's. Furthermore, the label data do not precisely match that of the protologue; on the label it says “ап Waldwegen," whereas the protologue reads “in viis silvaticis." The latter sheet has an annotation label in Bitter's writing in the packet. However, this label reads “Solanum phy- tolaccoides (Rusby) Bitt.," with Bassovia phyto- laccoides Rusby and Solanum mapiriense Bitter given as synonyms. Аз Bitter was not aware of Rusby's name when he described S. mapiriense in 1912, he must have examined this specimen at a later date, and it cannot be considered the type. In the absence of any better alternative candidates for the type of S. mapiriense Bitter, I have des- ignated Buchtien 1434 at US as the lectotype. 2. Solanum morellifolium Bohs, sp. nov. TYPE: Peru. Ucayali: Prov. Coronel Portillo, Bosque von Humboldt Experimental Station, carre- tera marginal, km 86 toward Puerto Bermu- dez, 75%05'W, 8?45'S, ca. 330 m, 18 Apr. 1982, D. Smith, Angulo & Lynch 1337 (ho- lotype, MO; isotype, NY). Figure 4. Herba fere glabra. Laminae foliorum non lobatae ova- = basi decurrentes. Inflorescentiae extra xillares breves 7 floribus praeditae. Corolla rotato- ea iridi-alba, s -4 mm longi. Thecae antherarum 2.5-4 mm longae. Stylus circa 3 mm longus. Bacca globosa circa 1 cm diametro. Herb or shrub 0.5-1 m tall. Branches glabrous to very sparsely puberulent, rarely moderately pu- berulent with gland-tipped finger hairs. Leaves 2- 3 per sympodial unit, the blades unlobed, ovate to ovate-elliptic or + triangular, glabrous to sparsely piti adaxially and abaxially, 2.5—17 cm long, 1.3-7 cm wide, the length : width ratio ca. 2: 1; major veins 4-8 on each side; petioles 0.5-7 cm long. Inflorescence extra-axillary, 3-7-flowered, 0.5-2.5 cm long; peduncle 3-15 mm long; rachis 3-10 mm long; pedicels 3-7 mm long, in fruit 5— 8 mm long, spaced 1-4 mm apart. Peduncle, ra- chis, and pedicels glabrous to very sparsely and minutely puberulent, rarely moderately puberulent with gland-tipped finger hairs. Calyx glabrous, the radius 1-2 mm, the margin nearly entire except for 5 very short obtuse lobes or with deltate lobes ca. 0.5-1 mm long and 1 mm wide. Corolla white or greenish white (blue fide Smith et al. 1337), rotate-stellate, the radius 4—5 mm, the tube 1-2 mm long, the lobes 2-4 mm long, 1.5-2 mm wide, triangular, the apex short-tomentose. Stamens 3.5— 4 mm long; anther thecae yellow or white, 2.5-4 mm long, 1-1.5 mm wide at base, 0.3-0.5 mm wide at apex, triangular. Style ca. 3 mm long, ca. 0.2-0.5 mm diam. stigma ca. 0.2-0.5 mm diam. Fruit globose, 7-12 mm long, 6-12 mm diam. (when immature?), green or red (Kayap 750) or green with purple lines (Smith et al. 1337; Knapp & Mallet 6642); mesocarp with small aggregates of stone cells; seeds ca. 2 mm long, 1.5 mm wide. Distribution. Forests or forest clearings of eastern Andean slopes and adjacent lowlands in Peru and Bolivia, 100-1,300 m (Fi . Chuagkáteme (Kayap Vernacular names. Additional (p e examined. PERU. AMAZONAS: 12-15 km N of mpami, 1,200 m, 2 Oct. 1972, Hern 156 (MO); Quebrada Huampami, Rio Cenepa, hss y 1973, Kayap 750 (MO). AYACUCHO: Rio Apurir b. near Kimpitiriki, 400 m, 10 May 1929, Killi & Smith 22972 (NY, US). cuzco: Prov. La Convencion, Rio Chalpimayo above Pacchar, 3,850 ft., 1975, Plowman & Davis 4858 (GH). HUANUCO: near Tingo María, ca. 600 m, 1 June 1977, Hart 601 Dist. H | de Abeja, 300-400 т. 1067, Bohunke 17 d dp oo Bridge; near La Merced, 800-1,300 m, 1-3 June 1929, Killip & Smith 24015 (NY, US). LORETO: Prov Mayas trail from Indiana on Rio Amazonas to Rio Na 200 m, 24 May 1978, ad et al. 22172 (MO). near Yurimaguas, 180-200 m, Nov. 1982, Ochoa & Hooker 14885 (US). MADRE DE DIOS: Prov. Parque Na- cional Manú, Cocha Cashu Station, 71°23’W, 11°53’S, 350 m, 12 Sep. 1986, Foster 11363 (NY); same locality, 71?0'W, 11?45'S, 400 m, 15 Aug. 1986, Núñez 5724 (NY). PASCO: Oxapampa, km 15 of Palcazu Road (km 73 Villa Mairo) along Ri 1 ‚ 380 m, 17-18 Aug. 1984, Knapp & eh 6642 NT US); Río Palcazu Valley, Iscozacin, 984, Whalen & Salick 654 (NY). BOLIVIA. PANDO N on o pat © LL Aug. 1985, Nee 31505 (NY); Prov. Manuripi, along Rio Madre de Dios, З km W of Humaita, 12%1'S, 68°18'W, 150 m, 30 Aug. 1985, Nee 31658 (UT). This species closely resembles S. allophyllum, but 5. morellifolium has much smaller inflores- cences, flowers, and fruits. The lobed leaves often seen in S. allophyllum apparently do not occur in this species. Although most collections are nearly glabrous, a few have glandular-puberulent inflo- rescence axes like those of 5. mapiriense. The specific epithet refers to the great similarity Volume 77, Number 2 1990 Bohs 405 Systematics of Solanum sect. Allophyllum B | || RE 4. рат торек ань — А. Crown branch. — B. Stem and peduncle. — C. Stamens (left, abaxial view; right, adaxial view). — D. Flower. Smith et al. 1337 MO; F based on Nee 31505 NY. of the leaves of this species to those of Solanum sect. Solanum, which was formerly known as Mo- rella or Maurella until the name was changed to sect. Solanum in accordance of the rules of bo- tanical nomenclature (Seithe, 1962). Data from Knapp & Mallet 6642 indicate that Ithomiine butterflies of the genus Thyridia oviposit on S. morellifolium. With this exception, all the known larval food plants for Thyridia belong to Cyphomandra (Drummond, 1986). If S. morel- lifolium indeed turns out to be a food source for these butterfly larvae, it may be indicative of a . Gynoecium — Е. Corolla opened to show insertion of stamens. (A-E based on ) closer relationship of sect. Allophylla to Cypho- mandra than can be inferred on the basis of mor- phological characters (Knapp, pers. comm.). 3. Solanum allophyllum (Miers) Standl., J. Wash. Acad. Sci. 17: 16. 1927. Pionandra allophylla Miers in Seem. Bot. Voy. Herald, 174. 1854. Cyphomandra allophylla (Miers) Hemsl., Biol. Cent.-Amer., Bot. 2: 417. 1882. TYPE: Panama: in waste places, Seemann 169 (lectotype, BM, here designated; isolectotype, K). Figure 5. 406 Annals of the Missouri Botanical Garden Aeunabo flere ty do (Coes el fora et her 24 ; phoma FIGURE 5. Solanum ellipsoideibaccatum Bitter, Repert. Spec. Nov. egni Veg. 11: 486. 1913. Bassovia ellipsoidei- aa Биш Cat. Fl. Venez. 2: 350. 1947. erect, branching, t shady and somewhat damp places below 1,000 ft., corolla pale greenish, 12 Nov ‚Н.Н. Smith 1153 (distributed as кы, chenopodioides Lam. ?) (holotype, B, destroyed; isotypes, A, BM, E, O, NY, P, US 15 боолот о уаг. 2 e spud uid pert. Spec. Nov. Regni Veg. 13: 173. 1914. Isolectotype of Pionandra allophylla (= Solanum allophyllum) ~ Seemann 169 K). Panama. Chiriquí: near San Félix, 0-120 m, Pittier 5237 (holotype, US #715441; isotype, US) Herb or shrub 0.4-1.5 m tall. Branches gla- brous or sparsely puberulent, especially when young. Leaves 3 per sympodial unit, the blades unlobed or pinnately 2-5-lobed, sparsely eglandular-pu- bescent adaxially, glabrous abaxially; if unlobed, the blade ovate, 3-14 cm long, 1.5-7.5 cm wide, the length : width ratio ca. 2: 1; if lobed, the blade Volume 77, Number 2 1990 Bohs Systematics of Solanum sect. Allophyllum 3-14 cm long, 2-12 cm wide, the terminal lobe elliptic to obovate, 2.5-12 cm long, 1-5.5 cm wide, the lateral lobes ascending, 1-10 cm long, 0.5-4 cm wide, the blade divided nearly to midrib, the sinuses rounded, acute; major veins 4-6 on each side; petioles 1-11 cm long. Inflorescence extra-axillary or located in a branch fork, 4-6- flowered, 1.5-7 cm long; peduncle 1-4.5 cm long; rachis 0.5-2.5 cm long; pedicels 4-6 mm long, in fruit 6-9 mm long, spaced 3-13 mm apart. Pe- duncle, rachis, and pedicels glabrous, or rarely sparsely puberulent. Calyx glabrous, the radius 2- 3 mm, the lobes deltate, acute or obtuse, apiculate, 1-2 mm long, 1.5-2 mm wide. Corolla white or greenish, rotate-stellate, the radius 7-10 mm, the tube 3-4 mm long, the lobes 4-6 mm long, 3 mm wide, triangular. Stamens 6-6.5 mm long; anther thecae yellowish or white with a greenish patch near the base, narrowly triangular, 5-6 mm long, 1.5-2 mm wide at base, ca. 0.5 mm wide at apex. Style 4-5 mm long, 0.3-0.5 mm diam.; stigma 0.3-0.5 mm diam. Fruit ovoid or ellipsoidal, lat- erally compressed, 2.5-4 cm long, 1.5-2.5 cm wide, 1-1.8 cm thick, white or light yellow to orange, often with dark green or purple stripes; mesocarp with 3-6 stone cell aggregates ca. 2 mm diam.; seeds 2 mm long, 1.5 mm wide, rugose- verrucate. Distribution. Forest and disturbed areas, es- pecially in drier sites, Honduras to Panama, Co- lombia, and Venezuela, 0-900 m (Fig. 3). All col- lections from Panama are from Tropical Moist Forest, sensu Holdridge et al. (1971). Flowering and fruiting. April through Jan- uar Vernacular names. Bleo de gallinazo (Ro- mero 6321) (probably a misprint for bledo, a Span- ish term for herbs that can be used as food (Hun- ziker, pers. comm.), cumapan (Delascio & Liesner 6988), hierba de gallinazo (Standley 28138), hierba gallota (Pittier 6788), yerba de gallote (Bro. Paul 154), zopilote (Moreno 22030, Nee 28133). Uses. Leaves used in salads, broths, and chopped meat dishes; ripe fruits used in stews (Co- lombia, from Romero-Castaneda, 1965). Additional specimens examined. HONDURAS. CHOLUTECA: vicinity of Pespire, 160-200 m, 18-25 Oct. 1950, Standley 27108 (BM, F, US); same locality, 18- 27 Oct. 1950, Standley 27237 (F). NICARAGUA. BOACO: ae Grande, 12%25'N, 85°45'W, 200 m, 30 Sep. 980, Moreno 3286 (МО); San Lorenzo, Sierra El Espino, UN 85°39'W, 500-600 m, 11 Nov. 1982, Moreno 18544 (MO); Santa Cruz, 12°24'N, 85°49’W, 160-200 m, 15 Nov. 1982, Moreno 18608 (MO). CHONTALES: La Asunción, km 120 carretera Juigalpa, 12?9'N, 85°31'W, О m, 19 Oct. 1980, Moreno 3705 (МО); Hda. San Martin, near confluence of Rio El Jordan and Río La Pradera, 12°17'N, 85?15'W, са. 390 m, 30 July 1984, Stevens 22977 (MO, NY). GRANADA: P ra de Apoyo, “Babilonia,” 11%55'N, 86?4'W, , 30 May 1981, Moreno & Henrich 8898 MO» camino de Casa Tejas, 11?46'N, 85%54'W, ca. 40-60 m, 21 June 1982, Moreno 16658 (MO). LEÓN: Isla de Momotombito, Lago de Managua, ca. 200 m, 21 Oct. 1979, Araquistain 362 (MO); same locality, ca. 150 m, 22 Oct. 1979, Ara- quistain 386 (MO). MANAGUA: El Carrizo, carretera a San Francisco, 12*23'N, 86°7'W, са. 70-80 m, 10 Dec. 1980, Moreno 5080 (MO). MASAYA: A. de Apoyo, 11*56'N, 86°2’W, 100-140 m, 20 Sep. 1981, Moreno 11149 (MO). MATAGALPA: carretera a Jinotega, km 134, 800-900 m, July 1982, Bustos 44 (HNMN). RIVAS: Isla de Ometepe, 11?29'N, 85?29'W, 40-55 m, 1983, Moreno 22030 (MO); above Balgue on facing slopes of Volcán Maderas, Isla de Ometepe, 11?28'N, 85*31'W, 600 m, 14 Sep. 1983, Nee & Téllez 28019 (MO, NY); same locality, 11°29’N, 83%31"W, 50-100 m, 15 Sep. 1983, Nee 28133 (MO, NY); Isla de Ometepe, Volcán Maderas, “La Palma,” 11?27-29'N, 85°28-30'W, 100-200 т, 21 Sep. 1984, Robleto 1221 (МО); Isla de Ometepe, entre Cuatro Esquinas y San Fernando, 11°30- 32'N, 85°33-34'W, 17 July 1981, Sandino 1017 (MO). Costa RICA. GUANACASTE: La Pacifica, Nov. 1976, Haber 57 (F, MO); between Las Cañas and Liberia, Pan Amer- ican Highway, 100 ft., 12 Nov. 1953, Heiser 3719(US); Santa Rosa National Park, 25 Sep. 1975, Lie 10221 (MO); same locality, ca. 10%50'N, 85°37'W, 0-320 m, l July 1981, Janzen 12083 (МО); along Rio Higuerón near agricultural experimentation area near Taboga, 10°20'N, 85°12'W, 0-100 m, 29-30 June 1977, Lies- ner et al. 2724 (МО); Finca La Pacifica, Cañas, я Farm Road, 22 Oct. 1971, Opler 472 (MO). PANA CANAL ZONE: Barro Colorado Island, 1931, Aviles ae (F); vicinity Cerro Viejo on K16C, 13 Oct. 1965, Blum 1263 (MO); Farfan Beach, from Thatcher Hwy. to Palo Seco, 27 Dec. 1966, Burch et al. 1410(MO, NY); Gatún, Nov. 1859, Hayes 563 (NY); vicinity of Madden Dam, 50 ft., 3 Dec. 1966, Lewis et al. 27 (M, MO); Madden Forest Preserve, along Las Cruces Trail and highway, 8 969, Lewis et al. 5315 (MO); Sosa Hill, Balboa, с. 1923, Standley 25289 (US); Balboa, Nov. 1923-Jan. 1924, 2 25449 (US), 26109 (US); vicinity of Summit, 7 Jan. 1924, Standley 30142 (US); Barro Colorado Island, El Cermeno, 1 Dec. 1942, Zetek 5040 (F, ‚20 m, 12-14 July 1937, P. H. Allen y MO): ее of DN = 20 m, 6 Oct. 1938, P. H. Allen 939 (F, MO, NY, US). PANAMA: Bayano, 24 Nov. 1973, Alvarado 42 is Cerro Cam- pana, 12 Nov. 1975, D'Arcy 9613 (MO); ca. 6 mi. E of Chepo on Pan Am Highway, 28 Sep. 1961, Duke 4091 (MO); same locality and date, Duke 4092 (MO); vicinity of El Llano, 14-19 Oct . 1962, Duke 5798 (MO); near Madden Lake, 3 Au ; Camino de Las Saban irio 256 (US); Juan Díaz. 10 Nor кзз Herrera 42 408 Annals of the Missouri Botanical Garden (MO); same locality, 25 m, 30 Sep. 1917, td 3100 (MO); along Tapia River, 75 m, 21 Oct. 1917, Killip 3154 (MO); 1 km S of Madden Dam, 80 m, 20 Dec. 1973, Nee 8900 (F — photos, Му 8901 (МО); Sabanas, E of Panama City, Oct. 1932, Bro. Paul 154 (US); same locality, Nov. 1932, Bro. Paul 175 5 (GH); Chepo, 60 m, Oct. 1911, Pittier 4689 (BM, US); E experiment station at Mat нт илас 10 Sep. 1914, Pittier 6788 (US); Тарона bind: Dec. 1923, Standley 27089 (BM, US); Río Tapia, 7 Dec. 1923-11 Jan. 1924, Standley 28138 (US); Rio Tecumen, З Jan. 1924, Stand- ley E (US). SAN BLAS: c on mainland in front of Ustupo, 9 Nov. 1975, D'Arcy 9472 (МО); cult. at Missouri Bot. gin di seed of Diarey 9472, кай D'Arcy 9472a MO); on mainland in front o tupo, 9 Nov. 1975, D'Arcy 9529 (MO, NY); Mligandi mis along trail from ocean to waterfall on river, 0-200 m, 7 Oct. 1978, Hammel & D'Arcy 5002 (MO). AER BOLÍVAR: La Popa, near Cartagena, 50-175 m 1 & Smith 14080 (GH, NY, US); Тогай; near Turbaco vai = 150-300 m, 7-19 Nov. 1926, Killip th 14647 (F, NY). CESAR: Rincón Hondo, 11 A 24, C. Allen 373 (MO); same locality, 18 Aug. 1924, С. Allen 42 (MO); same locality, 20 Aug. nn C. Allen 451 (MO); Poponte, 23 Sep. 1924, C. Allen 737 (MO) Becerril, ca. 100 m, 15 Sep. 1943, Haught 3674 (US). CHOCÓ alrededores de Tilupo, 21 Jun , Romero 6321 (Е, , NY). GUAJIRA: Сеггејоп, ca. Am . 1949, Haught 6637 (US). NORTE DE SANTANDER: inar del Rio Peralonso en los alrededores de Santiago, 120 m, 21 D 1948, Molina & Barkley 86 (MO, US). Secu Barranca Bermeja, Magdalena Valley, between Sogamoso and Carare rivers, on Aguas Blancas Creek ca. 25 of El Centro, 150 m, 20 Nov. 1936, Haught 2084 (F, GH). VENEZUELA. BOLÍVAR: Dist. Piar, La Camilera, 40 km al W de El Manteco, 250-260 m, July 1978, De- lascio & Liesner 6988 (MO, NY). GUÁRICO: ca. = SSW of Calabozo on Hato Masaguaral, 100 m, 8%56'N, 67°60'W, 9 Nov. 1982, Ronde pau 123 (US). zu LIA: Dtto. Bolívar, via entre la carretera Lara-Zulia y Piedras Blan cas (desviando en km 70 de la Lara-Zulia al SE del puente sobre el lago), entre km 1-3, May 1979, Bunting & Fucci 7651 (MO); San Martin, on Rio del Palmar, Urdeneta, 15 Oct. 1922, Pittier 10530 (US); vicinity of Mene Grande, 31 Oct. 1922, Pittier 10610 (GH, NY, US) Solanum allophyllum differs from the other species of the section by its large, ellipsoidal fruits and frequently pinnately lobed leaves. The u leaves are very similar in size and shape to those of S. morellifolium. The flowers of S. allophyllum, though rather large like those of 5. mapiriense, are rotate-stellate with short lobes like those of S. morellifolium. Details of the architecture, branch- ing pattern, compatibility, and cytology of S. al- lophyllum may be found in Bohs (1989). e herbarium sheet of Seemann 169 at the British Museum (BM) bearing Miers's annotation has been chosen as the lectotype of Pionandra allophylla Miers. Two collections are included here that show morphological features more characteristic of 5. morellifolium than of S. allophyllum. I have iden- tified them as S. allophyllum because of their occurrence in northwestern South America (Fig. 3). Haught's collection 2084 bears only flowers. The vegetative features conform to those of S. allophyllum, but the inflorescence is axillary, few- flowered, and has very small flowers, all charac- teristics more typical of S. morellifolium than of S. allophyllum. The collection Rondeau 123 also resembles S. allophyllum in vegetative features, except for denser puberulence than is usual in that species. The fruits on this collection are globose and less than 1 cm in diameter. Although it is not known if these fruits are mature, their size and shape are more typical of S. morellifolium than of S. allophyllum. No flowers are present on this collection. LITERATURE CITED BITTER, G. 1912. Solana nova vel minus cognita. II. е Pow Nov. Regni Veg. 11: 1-18. ———. Selena nova vel minus cognita. VII. epen. Se Nov. Regni Veg. 11: 481-491. — ——,. 1914 “olana nova vel minus cognita. XV. Re eper ert. Spec. Nov. Regni Veg. 13: 169-173. Bons, L. 1986. The biology and сеа of Cy- phomandra (Solanaceae). Ph.D. Dissertation. Har- vard Univ., Cambridge, Massachusetts. Solanum allophyllum (Miers) Standl. and the generic delimitation of Cyphomandra and Solanum (Solanaceae). Ann. Missouri Bot. Gard. 76: 1129-1140. CHILD, A. 1984. Studies in Solanum L. (and related genera) 3. A provisional conspectus of the genus Cyphomandra Mart. ex Sendtner. Feddes Repert. 95: 283-298. DANERT, S. 1970. л. к, дег Gattung Solanum L. Kulturpflanze 18: 25 D'Arcy, W.G. 1972. ads den de ji typification of subdivisions of Solanum. Ann. Missouri Bot. Gard. 59: 262-278 The classification of Solanaceae. Pp. J.G. Hawkes, R. N. Lester & A. D. Skelding pen jd Biology and Taxonomy y the Solan- aceae. Academic Press, London. 1986. The calyx in Lycianthes and some de genera. Ann. Missouri Bot. Gard. 73: 117- 27. DRUMMOND, В. A., П. 1986. Coevolution 7 Ithomiine butterflies and solanaceous plants. Pp. 307-327 in . G. D'Arcy (editor), Solanaceae: Biology and Sys- tematics. Columbia Univ. Press, New York DunaL, M. F. 1852. Solanaceae. Pp. 1-690 in A. P. DeCandolle, Prodromus Systematis Naturalis Regni . Bestimmungsschlüssel der Subgenera und ET e der Gattung Solanum. Feddes Repert. 81: | иш E R., W. C. GRENKE, W. H. HATHEWAY, T. ШАМС & J. A. Тоз, JR. 1971. Forest Environ- ments in Tropical Life Zones: A Pilot Study. Per- gamon Press, New Yor Hine A. T. 1979. South American Solanaceae: a synoptic survey. Pp. 49-85 in J. G. Hawkes, R. Volume 77, Number 2 0 Bohs 409 Systematics of Solanum sect. Allophyllum N. Lester & A. D. Skelding (editors), The Biology and Taxonomy of the Solanaceae. Academic Press, London. Morton, С. V. 1944. Some South American species of Solanum. Contr. U.S. Natl. Herb. 29: 41-72 ROMERO-CASTAÑEDA, R. 1965. Flora del Centro de Bolivar, Volume 1. Universidad Nacional de Colom- ota bia, Bog 1962. Die Haararten der Gattung Solanum . und ihre taxonomische Verwertung. Bot. Jahrb zm 81: 261-336. 1979. Hair types as taxonomic characters in Solanum. Pp. 307-319 in J. С. Hawkes, R. М. Lester & A. D. Skelding (editors), The Biology and Taxonomy of the Solanaceae. Academic Press, Lon- don. SEITHE, À. NOTES TWO NEW SPECIES AND A NEW COMBINATION IN VISMIA (GUTTIFERAE- HYPERICOIDEAE) Work on the Guttiferae subfamily Hypericoi- deae for the Flora of the Venezuelan Guayana has revealed three species of Hypericum and 14 species of Vismia in the area of the Flora. Two of the Vismia species, one lode рү and the other d below. Another was previously se as a variety and is here elevated to specific status. endemic, are new an Vismia schultesii N. Robson, sp. nov. TYPE: Co- lombia. Amazonas-Vaupés: Rio Apaporis, Raudal Jirijirimo (below mouth of Kananari), 0°05'N, 70%40'W, са. 270 m, 21 Jan. 1952, Schultes & Cabrera 14946 (holotype, BM; isotypes, NY, Arbor 3-13.5 m alta, latice rubro ramulis dense fusco- chocolatino-tomentellis. a a 10-20 mm longo; lamina (100-)140-280 x -145 mm, late ovata vel ovato-oblonga vel oblonga ау гаге lanceolata, apice api- ¡ profunde cordata vel que nullis, coriacea, s a ferrugineo- nl haud glabrescens. Inflorescentiae terminales et interdum е же duobus superioribus ortae, rotundato- amulis dense fusco-chocolatino- vel ferr ru- , alabastris к vel late оу Flores homostyli; sepala 6-8 mm oe о е vel subaequalia exteriora elliptico- aie oblonga, paginis in alabastro expositis dense l fi i-i n longa, obovata; d fasciculi diis ovarium со um scis hirsutis. Fruc tus immaturus vir 10 mm longus, ovoideo- о haud vel seb Alle -punctatus e 3-13.5 m tall, latex red, young branches densely dark-chocolate-tomentellous. Leaves with petiole 10-20 mm long; lamina (100-)140-280 х 50-145 mm, broadly ovate or ovate-oblong to oblong or rarely lanceolate, apex apiculate to short- ly acuminate, base deeply cordate to rounded (or rarely broadly cuneate), main lateral veins 13-17 ANN. pairs, subsidiary laterals usually absent, coriaceous, dark green above, ferrugineous-tomentellous be- neath, not glabrescent. Inflorescences terminal and sometimes axillary, rounded pyramidal, axes dense- ly dark-chocolate- to ferrugineous-tomentellous, buds globose to broadly ovoid. Flowers homosty- lous; sepals 6-8 mm long, unequal or subequal, outer elliptic-oblong or all narrowly oblong, exposed surfaces (in bud) densely appressed-dark-choco- late- to ferrugineous-tomentellous, inner margins at least distally ciliate, spreading in fruit; petals white to yellow with glands linear, reddish, 8—10 mm long, obovate; stamen fascicles deciduous; ovary subglabrous. Fruits (immature) green, 10 mm long, ovoid-globose, not or very sparsely gland-dotted. Vismia schultesii is a lowland species of gallery forest and white sand forest and savanna in the extreme south of Venezuela and elsewhere in the Amazonian basin from Para to Rondónia (Brazil), southern Colombia, Peru (Loreto), and adjacent ¿cuador. Additional specimens examined. VENEZUELA. TERRITORIO FEDERAL AMAZONAS: Dept. Rio N Gentry & Stein 46392 (BM, MO). vicHaDa: Parque Nacional Natural “El Tuparro,” ca. 20 km E of El Tapór, 5%13'N, 69°05'W, 20 Mar. 1985, Zarucchi & Barbosa 3792 (BM, MO). vaurfs: Mitu, 22 Oct. 1976, Davis 110 (U). AMAZONAS-VAUPÉs: Rio Apoporis, entre os ríos Kananari y Pacoa, 250 m, 21 Jan. 1951, Schultes & Cabrera 12728 (NY, U); Río Apaporis, Soratama, 1- 15 Dec. 1951, Garcia-Barriga есеи ery AMAZONAS: 18-22 km N of Nee near Los Alpes, 19 Nov. 1974, Gentry 12762 (BM, МО); Leticia bs vicinity, 21 Jan. 1968, Stout 20 (NY). PUTUMAYO: Mocoa, Pueblo Viejo, 580-600 m, 28 Dec. 1940, Cuatrecasas 11. 385 (F, US). ECUADOR. PASTAZA: Ri А 48'W, 200 m, 4 Jun 31497 (AAU, mouth of Rio ~ ppe ÓN 1, 26 June 1929, атк 9076 (Е), upper Rio Nanay, Manfinfa, 30 June 1929, Williams 1144, (Е); Missouni Вот. GARD. 77: 410-411. 1990. Volume 77, Number 2 1990 Notes ре pe ын 17 Oct. 1973, Ayala 468 (BM, MO). B : Rio Solimóes, mun. бао Paulo de Olivença, jos Belem, 26 Oct.-11 Dec. 1936, Krukoff 8728 (BM, K, F, NY, P, U); Manaus area and lower Rio Negro, Chagas 5887 (NY), Chagas MG 21209 (NY), 9 (NY), Lowe 4235 da Silva 4508 (BM, NY); Rio Purús, Coa, 3 Ferreira 33-55 (NY). RONDONIA: basin of Rio Made na, 18 Nov. 1968, Prance et al. 8593 of Rio Madeira, Mutumparana, 5 Jul al. 5553 (BM, К, NY); Porto Velho-Cuiabá M Santa Barbara, 12 Aug. 1968, Prance & Ramos 6896 (BM, NY); Porto Velho-Cuiabá highway, near Ariquemes, i 17 Aug. 1968, Forero & Wrigley 7136 (BM, NY). P Santarém, Palhao to Igarape do Pilao road, 18 "e i 1969, Silva & Sousa 2311 (BM, NY). Vismia schultesii is closely related to the wide- spread Amazonian V. japurensis Reichardt but differs in having leaves usually broadly ovate to oblong rather than broadly ovate to lanceolate with the base sometimes more deeply cordate and the lower surface ferrugineous, not tawny to mustard- brown; sepals shorter, chocolate-tomentellous, the outer ones elliptic-oblong to narrowly oblong, not ovate; petals shorter. Some specimens of V. schultesii have been iden- tified by Ewan as V. tomentosa Ruiz & Pavon and V. catachrysa sp. ined. Vismia steyermarkii N. Robson, sp. nov. TYPE: Venezuela. Bolivar: alto Rio Cuyuni, Rio Uiri- yuk, La Escalera, 500-600 m, 20 Aug. 1962, В. & C. Maguire & Steyermark 46799 (ho- lotype, BM; isotype, NY). 3-5 m alta, colore laticis ignota, ramulis dense chocolatino-tomentellis. Folia petiolo 8-15 mm longo; lamina (95-)110-180 x (35-)40-65 mm, elliptica vel rarius elliptico-lanceolata, apice apiculata ve breviter acuminata, basi rotundata vel cuneata, venis lateralibus principalibus 10-14 jugis, venis lateralibus subsidiaris E ramulis dense ala 5-7 mm ү же, рөк шз anguste о lance la ta, iti e 5 md Р un e = m = о er С t £g Га ES Р " O ep E [99] 2 E. a © 5 n @ 5 = p < & ч lanceolata; staminorum fasciculi demum decidui; ovarium glabra, stylis glabris. Fructus immaturus virens, 8 mm longus, globosus, brunneo-glanduloso-punctatus. Tree 3-5 m tall, latex color unknown, young branches densely chocolate-tomentellous. Leaves with petiole 8-15 mm long; lamina (95-)110-180 x (35-)40-65 mm, elliptic or more rarely elliptic- lanceolate, apex apiculate to shortly acuminate, base rounded to cuneate, main lateral veins 10- 14 pairs, subsidiary laterals usually absent, sub- coriaceous, dark green above, chestnut- to cin- namon-tomentellous beneath, not glabrescent. In- florescence terminal, shortly rounded-pyramidal, axes densely appressed-chocolate-tomentellous, buds broadly ellipsoid. Flowers probably heteros- tylous; sepals 5-7 mm long, subequal, narrowly oblong to lanceolate, exposed surfaces (in bud) densely appressed-tomentellous, inner margins cil- iate, spreading in fruit; petals greenish white to cream with glands sparsely punctiform, 7-8(-9?) mm long, oblanceolate; stamen fascicles eventually deciduous (?); ovary glabrous. Fruits (immature) green, 8 mm long, globose, brown-gland-dotted. Vismia steyermarkii, a species of montane woodland at 500-1,040(-1,300?) m, is known only from southeastern Venezuela, from the upper Rio Cuyuni south of El Dorado into La Gran Sa- bana. Additional шел examined. VENEZUELA. BOLÍVAR: Cerro La Danta, NW of Cerro кеч head- waters of Rio Venamo, T 040- 1,060 m, 13 A 1960, Steyermark & Nilsson 17 (NY); Roraima Plateau, Rio Cuyuni 132.5 km S of El Dorado, 865-1,3 26- 27 July 1970, Steyermark & Dunsterville DA (NY); Donde road on Gran Sabana, 114 km lena, 1,040 m, 27 July 1982, Croat 54313 (BM, MO); е El Dorado-Santa Elena de Uairén, km 107, 560 m, 13 Aug. 1957, Trujillo 3517 (MY, U). Vismia steyermarkii appears to be a higher- altitude relative to V. schultesii, differing from it usually in stature and in having smaller, usually elliptic leaves with a narrower base, chestnut- to cinnamon-colored (not ferrugineous) beneath; flow- er buds broadly ellipsoid; flowers probably heter- ostylous, the petals greenish white to cream, short- er, and with glands punctate only; glabrous styles; and (immature) fruits smaller and gland-dotted. Vismia tenuinervia (E. v.d. Berg) N. Robson, comb. et stat. nov. Vismia cayennensis var. tenuinervia E. v.d. Berg, Acta Amazonica 4: 16, f. 19. 1974, I am grateful for the loan of material of Vismia to the curators of DUKE, F, MO, NY, and U. —Norman К. В. Robson, British Museum (Nat- ural History), Cromwell Road, London SW? 5BD, England, U.K DOS ESPECIES NUEVAS DEL GENERO SCHWENCKIA (SOLANACEAE) DE VENEZUELA Durante un estudio sobre la tribu Schwenckieae (Solanaceae) para la flora de Venezuela, he encon- trado dos nuevas especies del género Schwenckia: S. huberi Benitez, del Territorio Federal Amazonas, Departamento Atures, y S. trujilloi Benitez, del Parque Nacional Henri Pittier, Estado Aragua. La primera pertenece a la sección Schwenckia y la segunda a la sección Cestranthus Bent Schwenckia huberi Benítez, sp. nov. TIPO: Ven- ezuela. Territorio Federal Amazonas: Depto. Atures, en sabanas arenosas al noreste de Ga- lipero, vía Puerto Ayacucho ca. 80 m, 27 Feb. 1986, C. E. Benítez de Rojas & F. Rojas 3397 (holotipo, MY; isotipos, MO, MYF, NY, VEN). Figuras 1, 2 Herba repens radice axonomorpha instructa, caule pri- mario brevissimo in ramos elongatos repentesque argen- uius erectis, ne к\н vel acuta, basi cuneata ‘vel late glabro, rubescenti- nitido. Flores н sed ee decidui pauci fruc- tificantes; в im longis, glabris. Calyx tu- А losus 4 mm lors extus laxe ее 5-lobulatus, obulis acutis 1 mm longis. Corolla губа glabra, obscura, o vel badio suffusa, 1.1-1.2 cm longa, 5 lobulis mediis claviformibus minus 1 mm b lobulis lateralibus utrunque emarginatis, papillosis, n attin pus Stamina filamentis deii basim ve c ris 1 mm longis. Fructus ie patie globosus mm diam., seminibus 10-20, obscure castaneis. Hierba rastrera, de raiz axonomorfa, tallo prin- cipal muy corto ramificandose en largas ramas rastreras, éstas plateado-verdosas y densamente velloso-sericeas, los pelos erectos. Hojas firme-pa- piráceas, con peciolos 3-6 mm largo, velloso-se- riceos, con pelos erectos; lámina elíptica u ovada, a veces angosto elíptica o angosto ovada o mas frecuentemente ancho ovada, (10-)13-20(-40) mm largo y (4-)6-13 mm ancho, velloso sericea en la haz y en el envés, los pelos largos, erectos y en ambas caras entremezclados; ápice obtuso o agudo, base cuneada o ancho cuneada, la nerva- dura principal sobresaliente con 3-5 pares de venas secundarias. Inflorescencia terminal, erguida, hasta 95 cm alto, en panicula muy laxa, desnuda, con eje glabro, rojizo y lustroso. Flores muchas, en general caedizas, fructificando pocas; pedicelos 2- 4 mm largo, glabros. Cáliz tubular 4 mm largo, laxo-piloso por fuera, 5-lobulado, lóbulos agudos, 1 mm largo. Corola erecta, glabra, oscura con tonos violaceos o castaños, 1.1-1.2 cm largo, con 5 lóbulos medios claviformes menores de 1 mm de largo, lóbulos laterales contiguos unidos formando entre si emarginadura, papilosos, más cortos que los lóbulos medios. Estambres fértiles 2, filamentos laminares, hacia la base pubescentes, anteras 1 mm largo, oblongas. Fruto parcialmente incluído en el cáliz, globoso 2-2.5 mm diám.; semillas 10— 20, de color castano oscuro. Habitat y distribución. En sabanas arbusti- vas sobre suelo bien drenado, localizada en el Te- rritorio Federal Amazonas, Departamento Atures, Venezuela. sta especie es dedicada al Dr. Otto Huber, biólogo especialista en ecologia tropical, quien ha contribuido al conocimiento de la flora y vegetación de la Guayana venezolana, con sus fecundas ex- ploraciones botánicas. dir i ыл eus JELA. TERRITORIO FEDERAL AMAZO- NAS: Depto. res: sabanas y a en la región de Rincones de и unos 30 ro 30 abr. 1988 (fl, fr), ы de Rojas et ^ 3959 (F, K, MO, MY, PORT, TFAV, VEN). Schwenckia huberi, perteneciente a la sección Schwenckia, es muy peculiar por el indumento velloso-sericeo, tanto en los tallos como en las hojas. Su hábito de crecimiento rastrero con ramas largas, plateado-verdosas y sus hojas firme papiráceas, la diferencian de S. americana, su especie más re- lacionada. De S. elegans difiere porque en esta ültima, las hojas son fasciculadas y con el margen ligeramente ondeado. ANN. MISSOURI Bor. GARD. 77: 412-417. 1990. Volume 77, Number 2 Notes 1990 Y O y P A Vi f AN 25 2 0 AÑ DM er ARTA SS ИЕ, "eet УУМ АЗ ANO У ДДУ ) 2 Pr a r IES 17 } ИЕ ЖД ADIA aa- d 2%, EIA $5 d e Y, Wb Yes FIGURA 1. Schwenckia huberi (Benitez de Rojas et al. 3959 MY).— Habito. Annals of the 414 Missouri Botanical Garden gy a з. EX wa 2 1 RASO jer Supe b NS YE К Ке 2 - PY maru Nm IY me. АЛАУЫ AN JaN > — ago ES X ыс ASCENSO "AN NS 147 ZZ : AS PAS е ARTS SN OS SED NON dd 24) «2 ola UA irs irs E O INT 2 Е ISP Vel 74 ys TTT x MEANS mm Schwenckia huberi (Huber 5736 MY).— a. Cáliz y corola. —b. Vista interna de la corola expandida. FIGURA 2. c. Fruto. Volume 77, Number 2 1990 Notes 415 FIGURA 3. hojas e inflorescencia. Schwenckia trujilloi Benitez, sp. nov. TIPO: Venezuela. Aragua: Dtto. Girardot, en ba- rrancas de la carretera a Guamitas, Parque Nacional Henri Pittier + 700 m, 17 Ag. 1963, Baltasar Trujillo 5745 (holotipo, MY; isotipo, VEN). Figuras 3, 4. а. Mata 4 Schwenckia trujilloi (C. Benítez de Rojas, F. Rojas £ V. Badillo 3839 MY). Rama mostrando Frutex pluricaulis, caulibus badio-olivaceis, 2-3 mm iam., usque 4 т longis, primum erectis et aphyllis, dein flexuosis paulo ramosis ac foliosis. Folia membranacea, petiolis 9-25 mm longis pubescentibus; pagina late ovata, 3-7 cm longa, 1.5-4 cm lata, supra pilis ascendentibus conspersa, subtus pilis praesertim secus nervos, apice attenuata vel acuta, basi late subcordata vel cordata. 416 Annals of the Missouri Botanical Garden А. Mato НМ. FIGURA 4. superior. —b. Sección longitudinal de la mism Infloresc axillares vel terminales, laxae, expansae; Pedicelli 3 6 mm longis pilosis. Calyx tubu- losus 3- evibus marginatis 3-4 mm lor ongis. Stamina fertilia duo, exserta; filamentis parte su- M Schwenckia trujilloi а аде, МҮ). —а. Vista de frente de flor abierta en su extremo . Fru periori laminatis m i aie 1 mm longis, oblongis, extrorsis. Stylus exsertus. Fructus globosus, 3- .9 mm diam., calyce sub ^ Dus deua seminibus 6-14, obscure castaneis rubescentibus Arbusto hasta 4 m largo, de varios tallos de 2- 3 mm diám., marrón oliváceo, partiendo de una base leñosa común, al comienzo erguidos y sin Volume 77, Number 2 1990 Notes hojas, luego flexuosos poco ramificados y foliosos, con pelos antrorsos especialmente en los tallos jó- venes. Hojas membranáceas, con peciolos 9-25 mm largo, pubescentes; lámina ancho-ovada, 3-7 cm largo y 1.5-4 cm ancho, haz con pelos espar- cidos, ascendentes, envés con pelos especialmente en las nervaduras, ápice atenuado o agudo, base ancho subcordada o cordada. Inflorescencia en pa- nicula axilar o terminal, laxa; pedicelos 3-6 mm largo, pilosos, Cáliz tubular 3-3.5 mm largo, piloso por fuera y dentro, 5-lobulado, lóbulos 1-1.5 mm largo. Corola erecta, glabra, verde violácea, 1.5- 2 cm largo, tubo cilíndrico, con la parte superior un tanto ensanchada, con 5 lóbulos medios den- tiformes, curvos, minimos, lóbulos laterales conti- uos unidos formando una porción + triangular, glabros, 3-4 mm largo. Estambres fértiles 2, ex- sertos; filamentos laminares en su parte superior, alli pubescentes, anteras 1 mm largo, oblongas, extrorsas. Estilo exserto. Fruto globoso 3-3.5 mm diám., con el cáliz constricto abajo de los lóbulos; semillas 6-14, de color castaño oscuro o rojizo. abitat y distribución. Crece en arbustales sobre taludes, en suelos deleznables, localizada sólo en el Parque Nacional Henri Pittier, Estado Ara- gua, Venezuela. Dedicamos esta especie al destacado botánico Profesor Baltasar Trujillo, quien hizo especiales observaciones de campo a la muestra que se ha escogido como holotipo. atipos. VENEZUELA. ARAGUA: Parque Nacional F. Rojas 3756 (MY, PORT, US, (fl), Benitez de Rojas et al. 3838 au 29 dic. 1987 (fl, fr), C. Benítez de Rojas, F. Rojas & V. Badillo 3839 (F, MO, MY, NY, P, US). Esta especie, de la seccion Cestranthus, muestra afinidad con S. grandiflora Benth., diferencián- dose por ser una planta perenne de tallos alargados hasta 4 m de largo, flexuosos pero no trepadores; con hojas menores de 7 cm de largo y de 4 cm de ancho, siendo las hojas de S. grandiflora Benth. mayores de 7 cm de largo y de 4 cm de ancho. Las flores miden hasta 2 cm de largo en S. trujilloi Benitez y más de 2 cm en S. grandiflora Benth. La autora agradece a Bruno Manara, quien ela- boró la descripción latina de las especies y a Aris- tides Mata, quien realizó las Figuras. —Carmen E. Benitez de Rojas, Instituto Botánica Agrícola, Facultad de Agronomía, Universidad Central de Venezuela, Apartado 4579, Maracay 2101-A, Aragua, Venezuela. A NEW SPECIES OF TRIGONOSPER MUM (COMPOSITAE, HELIANTHEAE) FROM CENTRAL AMERICA In preparing treatments of Compositae for the Flora de Nicaragua project, an unusual specimen was discovered in the genus Trigonospermum that could not be placed with existing revisionary lit- erature (McVaugh & Laskowski, 1972; Turner, 1978). Examination of material of the genus from other institutions revealed additional specimens and led us to treat these as a new taxon at the specific level. Trigonospermum stevensii Sundberg et Stues- sy, Sp. nov. TYPE: Nicaragua. Depto. Madriz: Cerro Quisuca, lower S and SW slopes, ca. 13°30'N, 86?31'W, 800-— 1,100 m, dry forest on rocky slopes, 23 Nov. 1979, W. D. Stevens & A. Grijalva 16156 (holotype, OS; isotypes, MO, OS, OSH). Figure 1. gh & Laskowski et T. melampo- paleis exteriores glabris, apice m longis, m longis, A T. annuo McVau dioide DC. simile, ed ciliatis, floribus radii 3-4, ramis styli 0 ae ligulis 4.5-7 mm longis, achaemis 2.5-3 ibe disci 18-25, lobis corollae sparsim ыз. vel glabris, antheris 1.1-1.3 mm longis. Annual herb, ca. 1 m tall. Stems hispidulous, stalked-glandular. Leaves opposite; blades rhom- bic-ovate, 5-15.5 cm long and 3-11 cm wide, hispidulous and occasionally glandular above, stri- gose and glandular below, acute at the apex, with margin serrulate; petiole winged, to 5.5 mm long. Peduncle slender, hispid, stalked-glandular. Capitu- la numerous; involucre 2.8-4 mm tall; outer phyl- laries linear-elliptic, 2.4-4 mm long, 0.8-1.6 mm wide, hirtellous, glandular, grading into middle phyllaries; middle phyllaries rhombic to obovate, slightly keeled, 2.2-3.2 mm long, apex acute to apiculate, fimbriate-ciliate; inner phyllaries n ping around achenes, obovate, scarious, white, 3 mm long, 10-nerved, glabrous except at the rounded, fimbriate-ciliate apex. Ray florets 3—4; corollas yellow; ligules 4.5-7 mm long, 3-6 mm wide, gland-dotted both sides, deeply trilobed, with lobes rounded, 2-4 mm long; tube pubescent, 0.3- 0.5 mm long; style branches 0.5-1.4 mm long; achenes broadly elliptic to obovate, sometimes 3- ANN. angled, broadest above the middle, 2.5-3.2 mm long, with terminal scar elevated 0.2 mm. Disk florets 18-25, with abortive ovaries; corollas 2.1— 2.7 mm long; lobes 5, 0.1—0.3 mm long, glabrous to sparsely pubescent; styles fused (not bifid); an- thers 1.1-1.3 mm long, exserted half their length at anthesis. Pales filiform-subulate to oblanceolate, fimbriate at apex, 1.8-2.6 mm long. Iditional ens examined. GUATEMALA. DEPTO. os o: Chimaltenango HEP Station, О m, 5 Nov. ‚ A. Molina R. & A. R. Molina и РЕРТ. HUENUETENANGO: ca. 5 mi. WNW - по 23 Oct 76, Т. К. Stuessy & К. . Gardner X); DEPT. ЕЧ slopes of a cán de N S of Santa e Jesús, 1,800-2,100 m, 10 Dec. 1938, P.C. ати 59378 (MICH). This new taxon is morphologically intermediate between Trigonospermum annuum McVaug Laskowski and 7. melampodioides DC. (Table 1). Most importantly, it differs from 7. annuum in ray corolla length, disk floret number, anther length, and achene length. It differs from 7. melampo- dioides in pubescence of lobes of the disk corollas and outer pales. The morphologically intermediate nature of Т. stevensii could be due to interspecific hybridization, but two observations argue against this. First, pol- len stainability of the specimens is high (99% of 500 grains in cotton-blue in lactophenol), and the pollen grains appear normal. Second, the possible parents, 7. melampodioides and T. annuum, occur in Mexico reaching only as far south as Oaxaca, and therefore do not occur within the range of T. stevensii. This new species, in fact, is allopatric to all other taxa in the genus. These considerations urther suggest specific recognition for T. stevensii. Additional fieldwork could alter this viewpoint, how- ever, as well as concepts of other poorly collected taxa in the rest of the genus (e.g., T. auriculatum Turner). The discovery of Trigonospermum stevensil in Nicaragua extends the range of the genus south- ward into Central America, previously known only from Mexico and Guatemala. The new species is named in honor of Dr. Warren Douglas Stevens, Missouni Bor. GARD. 77: 418-420. 1990. Volume 77, Number 2 1990 Notes 419 Trigonospermum stevensii, showin who is coordinating the Flora de Nicaragua proj- ect. Thanks are expressed to The Ohio State Uni- versity Graduate School for a University Postdoc- toral Fellowship to the senior author 1986-1987, FIGURE 1. g habit, capitulum (with facing ray floret removed), and outer phyllary and achene (same scale). All Stevens & Grijalva 16156 (holotype). when this paper was written; NSF Grant INT- 402888 for support of manuscript preparation; MICH and TEX for loan of specimens; curators of GH, MO, NY, and TEX for courtesy during visits; B. L. Turner for critical comments on the manuscript; and James Zech for drawing Figure 1. 420 Annals of the Missouri Botanical Garden TABLE 1. Comparison of Trigonospermum stevensii with T. annuum and Т. melampodioides. Character Т. annuum T. stevensii T. melampodioides Duration annual annual perennial Inner phyllary Vestiture labrous labrous strigose argin apically ciliate apically fimbriate-ciliate apically erose Ray floret umb 3-4 3-4 (4-)5(-10) Style branch length (mm) 0.4-1 0.5-1.4 0.9-1.5 Ligule length (mm) 1.5-3.5 4.5-7 5-8.4 Achene length (mm) 2.8-3.8 2.5-3.2 1.8-3 Disk floret Number 8-13 18-25 19-50 Corolla lobe vestiture sparsely pubescent sparsely pubescent or glabrous densely pubescent Anther length (mm) 0.5-1 - LITERATURE CITED McVaucH, R. & C. W. Laskowski. 1972. The genus Trigonospermum Less. (Compositae, Heliantheae). Contr. Univ. Michigan Herb. 9: 495-506. TURNER, B. L. 1978. New species dnd combinations in the genera Sigesbeckia and Trigonospermum (Com- positae: Melampodiinae). Brittonia 30: 64-68. — Scott D. Sundberg and Tod Е. Stuessy, De- partment of Botany, The Ohio State University, Columbus, Ohio 43210, U.S.A. (Present address of senior author: Department of Botany, Univer- sity yl id Seattle, Washington 98195, U.S.A TWO NEW SPECIES OF MONOTAGMA (MARANTACEAE) FROM THE VENEZUELAN GUAYANA Two new species are described. One is from eastern Bolívar, Venezuela, and is related to Mono- tagma spicatum (Aubl.) Macbr. The other species is related to Monotagma laxum (Poepp. & Endl.) chum. but is confined to white sand areas. Inflorescence terminology follows Andersson (1976). oe ovatum Hagberg, sp. nov. TYPE: uyana: Pakaraima Mts., Mt. embaru (5T N- —60°33- 34 W), 600 m, 12 Nov. 1979, Maas & Westra 4329 (holotype, U; isotypes, BRG not seen, CB, K, NY). Fig- ure l Herba rosulata 0.5-0.7 m alta. Folia basalia 1-4; ultimum folium cum lamina ovata, ad basin obtusa, ad к. bo a, 22-32 x 9-15 apicem acuminata ad c cm, adaxiali pagina supra tota superficiem hirtella pr ae pagina glabra supra totam superficiem ad apicem hirtella; ER glabro, 1.2-2.6 longo, annulato; pe tiolo gla -45 cm longo: va pyracea ad char nga subtenta, raro frondoso, 3-6 nodis t 6-17 florescentiis, peduncu o 17-55 cm longo. Flo- m longa, -— 17 SK pedunculo 1.1-1.5 сш kn infima spatha 2.5-3.4 o ae m. -2 mm latus, ain glaber vel raro parce pilosus, in intra Doe parte basali excepto, lobi virelli, X 2-4 mm; staminodium ex- terius aurantiacum, 4 j x 2-3 mm; staminodium cal- losum 4-6 x 3-4 mm, cum parte о callosa, sine parte заар ovarium ик (1-)2-3 mm longum Capsula 9-12 x ca. 3 mm; semen 8-12 x x 2-3 mm (arillo incluso); arillus cum duobus lobis, 2-3 mm longus. Herb 0.5-0.7 m tall. Rhizomes branched to form clusters of a few shoots, each shoot with 1- 4 basal leaves. Ultimate basal leaf with sheath 14- 30 cm long, papyraceous to chartaceous, glabrous or slightly appressed-pilose along the margins; pet- iole 6-45 cm long, flattened laterally, glabrous; a prominent annulus present between petiole and pul- vinus; pulvinus 1.2-2.6 cm long, flattened later- ally, glabrous; blade 22-32 x 9-15 cm, ovate, chartaceous, broadly obtuse at base, acuminate to caudate at apex, oblique (apex displacement 1.8- 3.7 cm), adaxial surface not papillose, hirtellous ANN. MISSOURI Вот. GARD. throughout or at least at apex, midrib glabrous, abaxial surface not papillose, glabrous or rarely appressed-pilose at apex. Inflorescence a 3-6-no- ate synflorescence with 6-17 florescences, low- ermost node with a bladeless sheath 3.6-5.4 cm long or rarely with a leaf, subsequent nodes with bladeless sheaths, first internode 2.5-11.5 cm long, peduncle 17-55 cm long, glabrous or sparsely hirtellous to hirsute, hairs sometimes appressed. Main florescence 6-11 cm long with 5-17 disti- chous to unilaterally arranged spathes, first inter- node 0.8-1.4 cm long, peduncle 1.1-1.5 cm long; lowermost spathe 2.5-3.4 x 0.8-1.2 cm, yellow- ish green, chartaceous, abaxially glabrous or hir- tellous at apex, or rarely hirtellous throughout. Florescence component with 4—5 one-flowered cy- mules; first prophyll 12-17 mm long, abaxially glabrous or sparsely hirtellous throughout or only at apex or at base, ecarinate; interphylls and brac- teoles absent. Pedicel 1-2 mm long; sepals 11- 14 x 1-3 mm; corolla tube 20-27 mm long and 1-2 mm wide, yellowish, glabrous to rarely slightly hirsute outside, inside hirsute except for a glabrous 9- -4 mm, greenish; e 6-7 x 2-3 mm, orangish with entire apex; fertile stamen 3-5.5 mm long, petaloid appendage absent, theca ca. 1 mm long; inner staminodes reddish, at least toward base; callose stamin -4 mm, with a shelf-shaped callus 2-4 mm from apex, without petaloid ap- pendage; cucullate staminode 3-6 mm long, lateral appendage ca. 1 mm long; ovary (1-)2-3 mm long, glabrous; style 2-5 mm long. Capsule 9-12 х ca. 3 mm; seed 8-12 x 2-3 mm, with a 2-3 mm long two-lobed aril. Distribution and habitat. Known only from the Pakaraima Mountains in Guyana and from adjacent parts of Venezuela. It has been found in crevices and at bases of rocky escarpments and on moss-covered boulders; in Guyana, associated with the fern Pterozonium (Maas, pers. comm.), 900-1,300 m Monotagma ovatum is closely related to Mono- tagma spicatum (Aubl.) Macbr. but differs by hav- 77: 421-424. 1990. 422 Annals of the Missouri Botanical Garden 5cm FIGURE 1. 4432 (U).—A. Inflorescence. — B. Outline of ultimate basal leaf. Monotagma ovatum, Maas & Westra ing ovate (vs. elliptic) leaves, less than 10 cm (vs. over 11 cm) from the base to the broadest part of the blade, midrib both adaxially and abaxially gla- brous (vs. pilose), pulvinus glabrous (vs. puberulent throughout or at least adaxially), main florescence peduncle 1.1-1.5 cm (vs. 0.4-0.9(1.2) em), co- rolla tube 20-27 mm (vs. 17-23 mm) and outer staminode orangish (vs. bluish, fide coll.). GUYANA: Paka- 5°59’N- Additional specimens examined. raima Mts., t. Korak, Mazuruni River, 60°37'W, 600 m, Nov. 1979, Maas & Westra 4432 (BRG not а СВ, 0). VENEZUELA. BOLÍVAR: Cerro Vena- mo, Jan. , Steyermark et al. и (US, VEN); El ы о 1957, Couret 1; 78 (US); El ЖТ Sta. Elena km 105, Dec. 1956, AA 2680 (F, US, VEN, e coll., specimen in MY = M. spicatum); El D Sta. Elena km 107, Aug. 1957, Trujillo 3520 (MY, U); El Dorado-Sta. Elena km 130, Feb. 1968, Bunting 3017 (MY) Monotagma vaginatum Hagberg, sp. nov. TYPE: Brazil. Amazonas: about 5 km N of Presidente Figueiredo on the Manaus-Caracarai road (BR- 174), 8 Mar. 1986, Andersson & Hagberg 1750 um INPA; isotypes, GB, K, NY, S). Figu Herba па 0.5-1.7 т alta. Rhizoma + erecta, in parte supraterranea. Folia basalia 5-25, spirodistichae; н folium cum lamina anguste ovata, ad basin cu- neata obtusa, ad apicem acuminata ad caudata, cen- trica ad bone 20-51 x 2.4-13.5(-16.5) cm, adaxiali pagina hirtella ad puberula secus costam, nervos laterales majores, margines et ad apicem, vel solum secus costam et ad apicem, epidermide papillosa vel non papillo- sa, abaxiali pagina glabra supra totam superficiem vel ad apicem hirtella, epidermide nonpapillosa; pulvino supra puberulus, 0.7-2.8(-4) em longo; «шер hirsuto vel deest; petiolo destituto; vagina chartacea ad coriacea, 18-65 cm longa. nor phyllomate frondoso nt 2-6 nodis et (3-)5-20(-2 5) forescentis pedun cu / mm; staminodium 3-7 mm, cum parte specifica callosa, sine parte petaloidea; ovarium vertice е um, 1-3 mm longum. Capsula (9-)13- э х 3-4.5 mm; semen (8.5-)12.5-19 x 2-4.5 mm (аго incluso); АЕА сит duobus lobis, 2-4.5 mm longus. Herb 0.5-1.7 m tall. Rhizomes + erect, partly aboveground, subtended by “511” roots, branched to form clusters of 3—5 shoots, each shoot with 5- 25 spirodistichous basal leaves. Ultimate basal leaf with sheath 18-65 cm long (6.5-37.5 cm in cau- line leaf), chartaceous to coriaceous, + glabrous or appressed-hirsute along margin, distally auricu- late and furnished with a hair tuft; petiole absent; a hirtellous annulus usually present between sheath and pulvinus; pulvinus 0.7—2.8(—4) cm long, light green, laterally + flattened, puberulent to hirtellous adaxially, hairs sometimes appressed; blade 20-51 x 2.4-13.5(-16.5) cm [(12.5-)16.5-41(-54.5) x 2.3-12.3(-14) cm in cauline leaf], length from base to broadest part (2.9-)6-13.5(-18) cm, very narrowly to narrowly ovate, chartaceous to cori- aceous, markedly scalloped, cuneate to obtuse at base, acuminate to caudate at apex, centric to slightly excentric (apex displacement rarely to 3.5 cm), adaxial surface usually not papillose, dark green, hirtellous to puberulent on midrib, major lateral veins, margins, and apex (or at least on midrib and apex), abaxial surface not papillose, Volume 77, Number 2 1990 Notes 423 FIGURE 2. Monotagma vaginatum. — А. Habit (from the type population, Andersson & E 1750).— Terminal part of a florescence with spathes and a flower (scale — 2 cm, Hagberg & Medin 361 glossy green, glabrous or somewhat hirtellous at apex. Inflorescence a 2-6-nodate synflorescence with (3- )5-20(-25) florescences, lowermost node (sometimes also second and third node) with a cau- line leaf, subsequent nodes with bladeless sheaths, first internode 2.7-12.5 cm long, peduncle (10-) 13.5-50 cm long, appressed-pilose or rarely + glabrous. Main florescence 10-25 cm long with 8-24 distichous spathes, first internode 1-2 cm long, peduncle 1.2-2.4(-3.3) cm long; lowermost spathe 2.6-3.6(-3.9) x 0.7-1.1 cm, greenish to yellowish green, chartaceous, glabrous or rarely puberulent abaxially at apex. Florescence com- ponent with 3-5 one-flowered cymules; first pro- phyll 13-19(-21) mm long, abaxially glabrous, ecarinate; Zm rarely present; bracteoles ab- sent. cel 1-2 mm long; sepals 6-12 x 1-2(- 3) mm, ai white; corolla tube 21-30 mm long and 1-2 mm wide, white, outside glabrous or rarely hirsute above tip of sepals, inside hirsute except for a glabrous basal portion, lobes 4-8(- 10) x 2-5 mm, white to pale green; outer stam- inode 6-13 x 4—7 mm, white, spathulate with entire apex; fertile stamen 3-8 mm long, petaloid appendage absent, theca 1-2 mm long; callose staminode 4— 3-7 mm, white, sometimes green toward apex, with a shelf-shaped callus 2-4 mm from apex, without petaloid appendage; cucullate staminode 3-6 mm long, white, lateral appendage 1-2 mm long; ovary 1-3 mm long, puberulent apically; style 3-6 mm long, back of style turning red when tripped. Capsule (9-)13-20 x 2-4.5 mm; seed brownish, (8.5-)12.5-19 x 2-4.5 mm, with a whitish, 2-4.5 mm long two-lobed aril. Distribution and habitat. Monotagma va- ginatum occurs in the white sand forests (Ama- zonian caatingas/campinas) around Iquitos, Peru; in southern Venezuela and adjacent parts of north- western Brazil; around Manaus, Brazil; in scattered 424 Annals of the Missouri Botanical Garden campina areas in eastern Acre, Brazil; and in Рага, Brazil. It has also been found in the coastal forest in the basin of the Oyapock River and on Piton Rocheux in French Guiana. Monotagma vaginatum is related to Monotag- ma laxum (Poepp. & Endl.) К. Schum. from which it differs by having rhizomes partly aboveground (vs. on ground level), basal leaves spirodistichous (vs. distichous or shoot apex Wr twisted), mature shoots with 5-25 (vs. 1 15)) basal leaves, which are comparatively narrower [length/ width (3-)3.7-8.5 vs. 2.75-4.5(-5.5) cm], petiole absent (vs. present) and sepals 6-12 (vs. 9-17 mm) long. Additional specimens e СЕ ^d. BRAZIL. ACRE: 11 km S of Rio Branco, Oct. 0, Lowrie et al. 628 INPA); 12 km S of Rio dos on roa 1980, Cid & Souza 3030 (INPA); Madureira, Oct. 1968, Prance et al. 7789(F, GH, INPA. K, MG, NY, S, U). amazonas: 12 km up on aes Cuieiras, Feb. 1969, Keune ly 112 (DUKE, GB); 50 km up on Rio егар, Арг. 1974, Ongley & аай P21782 m S of Serra Central da Serra Aracá, Feb. 1984, Атага! 1674 А km S of Кыны i 1986, Andersson & Hagberg 1754 (GB, INPA. S); ane Rio Tupana and vastanha, on road Manaus~Porto Velho, July 1972, M. F. Silva et n гез Ана: "o reserve, 45 km N of Manaus on , Dec. 1977, 4. В. Anderson 321, 339 (INPA); Manas 187 20 rt 430 (P); Aug. 1957, Rodrigues 480 (INPA); Jan. 1963, E. Santos 1487 (GB); Feb. 1960, de la eon 2459 (LIL); . 1901, Ule 5415 (G, HBG, MG); errs 5 km N o як амы оп road to Hacostiara, Mar. 1986, 4ndersson & Hagberg 1773 (GB, INPA); Manaus, Igarapé do Pensador, Oct. 1955, оше 2133 (1 tal. 11683 (GB, GH, INPA qu K, MO, U); near ye Dec. 1964, Vogel anaus-Caracarai (B 74) km 130, Dec. 974, Gentry 12966 (INPA, МО); Ne 1973, Berg et Ру Р19525 (GB, INPA); Feb. 1974, Steward et al. 20294 (GB, INPA); des tal km 131, Mar. 974, и et al. s.n. (I ne naus- Caracaraí km 60, INPA reserve, ^u aE Oct. 1976, Kirkpatrick 118 (INPA Caracarai km 10, Oct. 1966, Prance et al. 2 F, K, INPA, MG, NY, 5, U, UBC, UC); ML Rio Negro, Apr. 1947, Pires 476 (IAN); Porto Camanaus, Oct. 1978, Madison et al. 389 (INPA); Rio Xeriuini, Apr. 1974, Pires 13997 (INPA, МС); Sào Gabriel da Cachoeira, Mar. 1975, M. R. Cordeiro 435 (IAN); Sào Gabriel, airport, Oct. 1978, ein et al. 506 (INPA, , US); Tunui, Igana, Oct. 1947, Pires 711 (IAN, NY); above Santa Isabel do Rio Negro, Oct. 1971, Prance et al. 15381 (GB, INPA, NY, U); junction of Rio Cuieras and Branquinho, Apr. 1974, Campbell et al. P21926 (INPA, NY); mouth of Rio s Sep. 1952, Fróes & Addison 28648 (IAN). PARA: 17 km S of Ligação do Pará on road Belém- Brasilia, e 1980, Plowman 9398 (MG); 40 km SW of Marabá, on road PA 150, Dec. 1981, же a 2. 1693 (GB); Paragominas, Tingi do Рага, Dec. Mac "pe al. 466 (MG); Tucuruí, Oct. 1977, A. E Sus. et al. 173 (K, MG, MO, NY, 5); km 324 on road Belém- B Aug. 1960, Oliveira 1013 (IAN); km 174 on road Belém- Brasilia, May 1960, Oliv- eira 664 (IAN). COLOMBIA. GUAINIA: San Felipe, Nov. 1952, Humbert 27441 (P). FRENCH GUIANA: Crique Ga- baret, basin of Oyapock River, Apr. 1988, Cremers 9929 (GB); Piton Rocheux, Crique Armontabo, Feb. 1981, Cremers 7044 (GB, P). PERU. LORETO: Brillo Nuevo, Rio Yaguasyacu, Apr. 1977, Plowman et al. 6806 (F, GH); Lago Llanchama, near Rio Nanay, Aug. 1972, Croat 18714 (MO); Mishana on Rio Nanay, Jan. 1976, Gentry et al. 15837 (F, MO); Feb. 1987, Hagberg & Medin 36 1 (AMAZ, Е, GB), 365 (AMAZ, F, GB, U); July 1984, Vasquez et al. 5298 (GB); Nauta, June 1984, Vasquez & Jaramillo 5090 (СВ); Pto. Almendras, SW of Iquitos, May den Andersson 15 (GB); Quistococha, near Iqui- tos, Nov. Asplund 14662 (S), 14663 (S); Dec. 1979, binas & Jones 9685 (GB); May 1978, Gentry & Jaramillo 22304 (GB). Mp QUE AMAZONAS: Cerro Aracamuni, Quebrada Camp, Oct. 7, Liesner & Car- nevali 22327 (GB); Oct. 1987, ueni d df 22249 GB); Cerro e _ Neblina, ESE slope above Rio Mawa- 984, Thomas 3248 (GB); IVIC study site, i , Liesner 6157 (GB, MO, VEN); Mamurividi, Rio Packman, June 1984, Davidse a Miller 267 14 (GB e lower Río Baria, June 1984, Davidse & nn 26746 (GB); Rio Мечата, 3-5 Кт о го de La Neblina base camp, Mar. 1984, Liesner 16 344 (GB); Rio Paci- moni, Apr. 1 ao ermark & Bunting 102457 (MY, NY, VEN — © E egro, near airport, Apr. 1970, Steyermark e Bunting 102767 (GB, US, VEN); 0-0.5 km NE of San Carlos de Rio Negro, Nov. 1 Liesner 3683 (VEN), O ee (MO); 5 km S of San Carlos de Rio Negro, 979, Liesner 6495 ( VEN); Solano, Brazo тү June 1984, Davidse & Miller 26662 (GB). I thank Lennart Andersson, who proposed Monotagma as a subject for my doctoral thesis and who critically read the manuscript. The field studies were supported by grants from the Royal Swedish Academy of Sciences (A. F. Regnells Bo- taniska Gávomedal, K. O. E. Stenstróms Fond, and Р. К. Wahlbergs Minnesfond), W. & M. Lund- grens Vetenskapsfond, Kungl. och Hvitfeldska Sti- pendieinrattningen (Overskottsfonden) and Anna Ahrenbergs Fond. LITERATURE CITED ANDERSSON, L. 1976. The synflorescence of Maranta- ceae. Organization and descriptive terminology. Bot. Not. 129: 39-48 —Mats Hagberg, Department of Systematic Bot- any, Gothenburg University, Carl Skottsbergs Gata 22, 41319 Goteborg, Sweden BOOK REVIEW Oostendorp, Cora. The Bryophytes of the Palaeo- zoic and the Mesozoic. 1987. Bryophytorum Bibliotheca, Band 34. 112 pp., 49 pls. J. Cramer, Stuttgart. ISBN 3-443-62006-X. Retail price: 120 DM This comprehensive catalog of Paleozoic and Mesozoic fossil bryophytes collates all published literature up to 1980. Many of the original illus- trations are reproduced photographically in the plates at the end of the book. Only gametophytic material or sporophytes attached to gametophores, described or annotated recently as bryophytic, have been included. A few incertae sedis, questionable bryophyte-like fossils, are also included. Russian and Polish but not the original Latin or French diagnoses have been translated into English. For each form species (not assigned to an extant family) in the alphabetized list, Oostendorp lists the no- menclatural authority and a reference to the illus- trations reproduced in the back of the book. This is followed by a list of synonyms with publication references according to the International List of Periodical World Abbreviations (1968). Each form species is then treated further under four sub- headings: (1) Icones, a list of publications with original or reproduced figures; (2) Type, including the location of the collection in which the holotype is deposited and the field locality, stratigraphical position, and age of the type collection; (3) Di- agnosis, sometimes presented as Original Di- agnosis and an additional Description, or Di- agnosis and Diagnosis emended, with reference to the original publication; and (4) Systematic Position, to order if known. Form genera are treated similarly to form species with the subhead- ings Type-species, Diagnosis, and Systematic position. After a brief section on problematic bryophyte- like fossils, treated similarly to the form species, there follows an exhaustive reference list and index, the latter differentiating among accepted names (130), synonyms, and new combinations (four new combinations made by C. Oostendorp). Preceding the lists described above are brief introductory chapters on “History and Classifica- tion” and on “The Fossil Record.” The first chap- ter discusses the development of Paleozoic and Mesozoic paleobryology and defines the structural characteristics of the form genera. It also classifies form genera with well-preserved remains within the present classification of bryophytic orders, follow- ing the International Code of Botanical Nomen- clature. Only one order, the Protosphagnales, does not have any extant members. Table 3 lists chrono- logically all described species with their synonymy of the form genera Thallitis, Hepaticites, Jun- germannitis, Metzgeriites, Marchantites, and Muscites. The chapter summarizing the fossil record re- views chronologically the major form species for each geological period from the Devonian to Cre- taceous. In addition, it lists all well-preserved form genera and species of each period for each order in a systematic review. This is the only part of the book where an attempt is made to evaluate some records in the light of bryophyte phylogeny and classification. Conclusions derived from this anal- ysis are brief and general, without any deviation from commonly accepted concepts suggested in earlier review papers. database I have been unable to find additional records published before 1980 that were not collated in this book. The number of typo- graphical errors is very small. The larger tables are not well differentiated from the main text, and headings are frequently separated by page breaks from the following text. The gray-toned photo- graphic reproductions of the figures are a major improvement over the photocopies of the less ac- cessible publications I have often had to study. The selection of figures reproduced for the plates is sufficient and representative. With such a large amount of disparate material to catalog and ana- lyze, C. Oostendorp was not able to evaluate crit- ically every record. However, this book will be invaluable as a starting point for the study of the relationship between fossil bryophytes and phylo- genetic research on living mosses and liverworts.— Jan A. Janssens, Department of Ecology, Evolu- tion, and Behavior, 318 Church Street, Univer- sity of Minnesota, Minneapolis, Minnesota 55455, U.S.A ANN. MISSOURI Bot. Garp. 77: 425. 1990. Volume 77, Number 1, pp. 1-224, of the ANNALS OF THE MISSOURI BOTANICAL GARDEN was published on February 16, 1990. о ө Н НЕЧ ЧӨӨ ө ө ө — AA AAA A New Monograph in Systematic Botany from the Missouri Botanical Garden Number 29 ADVANCES IN LEGUME BIOLOGY Edited by С. Н. Stirton and J. L. Zarucchi Advances in Legume Biology contains the proceedings of the Second International Legume Con- ference, in June 1986. The Conference was sponsored jointly by the Missouri Botanical Garden and the Royal Botanic Gardens, Kew, and it brought together анна 230 people from 26 countries to discuss current advances іп our understanding of biology. This Pp dec volume contains 35 papers addressing a wide range of tops. Soft cover. Price: $10 To order, send a deat dne, ^ the form below, with Pii gs to Department Eleven, Missouri Botanical Garden, P.O. Box 299, St. Louis, MO 63166-0299, U.S.A tt ИНЫЕ E i NOTE: A cumulative index for Cal Dodson's Icones Plantarum Tropicarum— Series | and 1! is now available from the Missouri Botanical Garden. This index contains a substantial number of name Changes since the original publication of the plates. It is included with the purchase of any fascicle in IPT Series Il, and it can be ordered separately by sending $2.50 (for postage and handling) to De- partment Eleven at the Missouri Botanical Ga rden. 19 Place an order, send a photocopy of this form. Please send me сору(іеѕ) ot X5 Bi $100.00 each, plus postage. A losed К< Note: E .00 invoicing fee, if payment not enciosed. & US. shipments: add 1.50 for | book non-U.S. shipn nts: check or money order in US. funds, payable to U S bank, add $2.50 for one book, $.50 for each k. Orders should b Prepaid; a $1.00 fee will be added t to orders requiring invoices. No shipments are made until payment is received. Май form with your check or t Eleven Missouri Botanical Garden P.O. Box 299 St. Louis, MO 63166-0299, U.S.A. : ease send Advances in Legume Biology to: ; A с | | М a Name э е none ен: | Ж ЫЫ, 4 added to total). — Postal Code ^ Country | pee 77(2) P Aj Al P TO ORANGE VYHFIDO I IU rve NOTES. CONTENTS Introduction: A Festschrift in Honor of Alice Faber Tryon and Rolla Milton Tryon, Jr. avid S. Barrington, Organizer : William Jackson Hooker and the Generic Classification of Ferns Cathy A. Paris & David S. Barrington 228 Studies of Neotropical Isoetes L. I. Euphyllum, a New Subgenus К. James Hickey Phy Two New Species of Cnemidaria (Cyatheaceae) from Panama Robbin C. | 25. Pteridophytes of the Venezuelan Guayana: New Species Alan К. Smith... 249 f Observations on Ctenitis (Dryopteridaceae) and Allied Genera in America Robert RRE G. Stolze 274 Defense Strategies in Bracken, Pteridium aquilinum (L.) Kuhn Gillian A. Cooper- Driver | MS Pityrogramma calomelanos (L. ) Link (Adiantaceae) on Layers of Volcanic Ash in Los À s, State of Veracruz, Mexico Ramon Riba & Irma Reyes J. ............. 281 Observations on the FADES Biology of Alsophila Species and Hybrids (Cyathea- КЫ ceae) _ David S. Conant ..... MAT. : Hybridization and ba: in (ена) American Polystichum: q... and — | ju. Isozyme Documentation David S. Barrington Eri 291 v р | Electrophoretic Evidence for Allotetraploidy with Segregating Heterozygosity in 1 South e African Pellaea rufa А. F. Tryon (Adiantaceae) Gerald J. Gastony o... _ Biosystematic Analysis of the Cystopteris tennesseensis (Dryopteridaceae) Сотрех 2 Christopher H. Haufler, Michael D. Windham & Thomas A. Ranker n- ME. x The American Paradox in the Distribution of Fern Taxa Above the Rank of Species ГАУ : PED Kho uo c oL A клу све — 99 А Judas Hybrid F wires ina a Population of Palystithum X potteri: Evidence from; С . Chloroplast DN arisons Diana B. Stein & David S. Barrington ... (34 Tumersceae Novedades para T3 eres Venezolana. María Mercedes Arbo dor ine Three New Andean Species of Aulonemi (P Bambusoideae) - Lynn С. Clark & ep S peer L DOE ^ M Pa ae | Devia xeromorpha, a New бони d Species of Iridaceae-Ixioideae fron the Ci Province fe RUE. Peter Goldblatt & John C. presen out SEE > Leaf and Corm Duk Structure in Lapeirousia (Iridaceae-Ixioideae) in Relation to Phylogen | and Infrageneric Classification Peter Goldblatt & John C. Manning ci Ww Cytological Variability і in the African Genus s Lay irousia (Irid Ixioideae) | P ее = Goldblatt кое. н" є. Biología Floral den una Conti Abi Tropical en la байа Venezolana - ; Nelson Кати ez, Celia Gil, Omaira Hokche, Alberto Seres & ney Brito —À The Systematics of Solanum section ела оймен) ime Bohs ~ EA Two New Species a And a New Combination: in Vismia a Gai iin 5З n. Norman: К. B. Robson .......... а асарны Dos Especies Nuevas del Genero Schwenckia (Solanaceae) de > Venezuela is Benitez de Rojas ... 2 pm = А New Species of Trigonospermun Compose Holas) rom Cental Amer | : E ^ D. Sundberg & Tod F. Stuessy - VE tees Ie M is Two New Species of Monotagma Mireia) É from the Venen Guayana. 2 E Неа. на te ae ene Volume 77 Number 3 Volume 77, Number 3 Annals of the Summer 1990 Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, con- tributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be accepted. Authors should write the Editor for information concerning arrangements for publishing in the ANNALS. Instructions to Authors are printed in the back of the last issue of each volume. Editorial Committee Marshall R. Crosby | Gerrit Davidse Editor, _ Missouri Botanical Garden Missouri Botanical Garden John D. Dwyer | Amy Scheuler : Missouri Botanical Garden & Managing Editor, Saint Louis University Missouri Botanical Garden | Peter Goldblatt Glenda Nau | Missouri Botanical Garden. Magdalen Lampe E Publications э Dale E. Johnson Missouri Botanical Cardó Henk van der Werff Missouri um сит ; For subscription information contact Department - . "The ANNALS OF exe MISSOURI BorANICAL CARDEN Eleven, P.O. Box 299, St. Louis, MO 63166. Sub- (ISSN 0026-6493) is published quarterly by '* scription price is $75 per volume U.S., $80 Canada - Missouri Botanical Garden, 2345 Tower Grove AV- and Mexico, $100 all other countries. Four r issues ~ enue, St. Louis, MO 63110. Second class postage per volume. — —— : ae. paid at St. Louis, MO and additional mailing a M uu uda rie POSTMASTER: Send address changes to ANNALS ; THE Missouri BOTANICAL GARDEN, Departmen Eleven, P.O. Box 299, St. Louis, MO 63166. .. € Missouri Botanical Garden 1990 — — — THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER. — | Volume 77 Number 3 1990 Annals of the Missouri Botanical Garden NZ THE PUBLICATION OF THE FLORA OF CHINA WILL BE A GREAT CONTRIBUTION TO THE SCIENTIFIC CIRCLES OF THE WORLD Opening speech by Professor Wu, Zheng-Yi' at the second editorial meeting for the joint Flora of China project, Guangzhou, Guangdong, China, August 1989 The first book of the Flora Reipublicae Po- pularis Sinicae (FRPS) (volume 2) was published as a gift to the tenth anniversary of the People's Republic of China (Ching, 1959). Thirty years have passed since then. Although we lost seven years during the Cultural Revolution, Chinese botanists have worked hard under various difficult condi- tions. Up to now, more than half of the FRPS has been published or is in press. Most of the remaining volumes have been completed, and only a few are in the middle stages of writing. The complete work is planned to be published within ten years; thus we shall give this monumental work to the world in the year 2000. While the FRPS is being pre- pared, botanists in every province or district are very actively preparing floras of their own regions. For example, the floras of Hainan, Jiangsu, Hubei, Xizang, and the herbaceous plants of northeast China have been published. Flora of Mount Hen- duan has been submitted to the publisher. Floras of the remaining provinces or districts are presently being written or are partially published. These flo- ras have added or will add new information to the FRPS. For various reasons very few plant speci- mens from Taiwan and Hongkong have been avail- able as we compile the FRPS. We have had to use information in the literature in cases when her- barium materials are inadequate. Problems that still could not be solved have had to be set aside until additional information is available. We are very glad that the Flora of Taiwan has been completed and now is being revised, and that the Flora of Hongkong is being compiled under the direction of Dr. Shiu-ying Hu at the Chinese University of Hongkong. We would like to thank members of the biology department of the Chinese University of Hongkong and Dr. Shiu-ying Hu for giving many excellent plant specimens to the herbarium of the Institute of Botany at Beijing. We hope that in the near future our colleagues in Taiwan can give us Taiwanese plant specimens or can exchange spec- imens with us. The Flora of China (FOC) is a joint project by Chinese and American botanists and is based on the FRPS. The FOC will be another monumental work, in English and different from the FRPS. FOC ! Director Emeritus, Kunming Institute of Botany, Academia Sinica, Kunming, Yunnan, People's Republic of China. ANN. Missouri Bor. GARD. 77: 427-429. 1990. 428 Annals of the Missouri Botanical Garden will have more or less complete information on specimens from Hongkong and Taiwan. The Chinese and American botanists have experience writing floras of large areas, so we are confident that the joint project can be carried out as planned and that the Flora of China can be published. It will be useful to persons worldwide who are concerned with Chinese plants. We thank the Chinese gov- ernment's open policy, which has made this joint project possible, has given us the opportunity to make friends with our foreign colleagues, and has allowed scientific exchange with our Hongkong and Taiwan colleagues. Our academic exchange and cooperation with colleagues in America and other countries has promoted and will continue to pro- mote the progress of science in China. When we are together with our American and Hongkong colleagues today, we are very pleased that the FRPS is near completion, floras of various prov- inces and districts in China are coming up like bamboo shoots after a spring rain, the Flora of Taiwan is under revision for a second edition, and the Flora of Hongkong is being written. The achievement in plant systematics in China has also promoted the development of plant floristic geography. China has a rich flora that not only has an ancient origin, but also has a very compli- cated history of migration and merging. The Chinese flora has attracted the attention of botanists all over the world. Many foreign scholars have come to China to investigate and study the flora of China during the last century. Since the 1920s, the Chinese botanists Hsen-Hsu Hu, Tcheng-ngo Liou, and Hui-lin Li have published many books and articles on the Chinese flora, probing into the origin of the flora of certain regions in China and the relationships between the flora of East Asia and North America (Hu, 1936, 1948; Li, 1944, 1950, 1952, 1953a-c, 1957; Liou, 1934, 1944, 1955). These botanists made great contributions, but be- cause the floras of many regions were not known in great detail, the conclusions they made were unavoidably simple and sketchy. With the large- scaled field expeditions after 1949, we have col- lected a large number of plant specimens, which have made possible the publication of many mono- graphs on plant genera and families. Compiling the FRPS and floras of various provinces and districts has laid the foundations for studying the flora of China. I have preliminarily summarized previous work on the flora of China and described the floristic characteristics of the distributional types and sub- types of Chinese seed plants, and their integrating relationships (Wu, 1963, 1965; Wu & Wang, 1983). Professor Zhang, Hong-da of Sun-yat-sen University also has published an article on the origin and development of the Chinese flora in which he differs from my opinions (Zhang, 1980). In order to make a thorough study of the Chinese flora, I proposed a research project—Study on the Chinese Flora —and applied to the Natural Science Foundation of China (NSFC). The committee of the NSFC has attached importance to this project and is about to approve it." Most participants in this project are the main authors of the I believe that our colleagues from Hongkong and Taiwan will also be interested in this project. Al- though they cannot participate in it for the time being, they can observe and comment, and they have academic exchanges with us. We hope that in the near future we can build another milestone in our rich and beautiful motherland and make even greater contributions to scientific knowledge. ur American friends have given us a lot of help in promoting the development of botany in China. We have exchanged visiting scholars, ex- changed plant specimens, and cooperated with each other very well. The joint Sino-American project of Flora of China is a typical example. In October of last year under the Metasequoia trees at the Missouri Botanical Garden, Dr. Peter Raven and I cosigned an agreement for this joint project. The preparation has proceeded smoothly for almost one year. For example, the project has been approved by the National Science Foundation (NSF) of the United States, and the publication problem has been resolved. Cooperation between the American and Chinese members of the joint Editorial Com- mittee, especially between Dr. William Tai (the project coordinator) and Professors Dai, Lun-kai (acting member of the joint Editorial Committee) and Cui, Hong-bin (deputy editor-in-chief of the FRPS and a member of the joint Editorial Com- mittee), has been very successful. They have made great efforts to reach a mutual understanding by overcoming various difficulties caused by the dif- ferent customs and systems in our two countries. They have arranged for volume 17 to be a test treatment followed by volumes 15 and 16. They also selected the Chinese and American authors for the remaining volumes and helped these authors contact one another. The authors of the seven- teenth volume: Professors Chen, Shou-liang (Ver- benaceae), Li, Xi-wen (Laminaceae), and Lu, An- min (Solanaceae) are due gratitude. The first two ? Subsequent to this meeting, NSFC approved the project. Volume 77, Number 3 1990 Wu 429 The Flora of China's Contribution to Science authors have finished their manuscripts. The first product of our joint project is to come out at the end of next year or in the beginning of the year after next. Professors Chen, Shou-liang and Li, Xi-wen have had questions when preparing the treatments. This is the exact purpose of the test treatment. We can discuss these problems at this meeting and modify the author's guidelines ac- cordingly. We realize that it will be a very complicated process to accomplish this monumental work. It will involve many people and many related projects. Every plant family has its own characteristics, and every author will have questions. According to our experiences, the authors’ guidelines need to be modified as the writing continues. We need overall regulations on how to cooperate with each other and how to translate and edit the Flora, and both sides should abide by them. At the same time, we should be flexible and try to solve specific problems with specific methods under the direction of the joint Editorial Committee. We should also have regulations on how to regulate loaning and col- lecting specimens and gathering references. These questions should be discussed at this meeting. The Editorial Committee of the FRPS has an office in Beijing. This office is also responsible for the work of the Chinese side of the joint Editorial Committee. It is a pleasure to meet Professor Paul Butt of the Chinese University of Hongkong at this meet- ing. It is unfortunate that our colleagues from Taiwan have not been able to meet with us in Guangzhou. But we welcome our colleagues to come to the mainland for visiting or research at any time. We also hope that we can visit Taiwan someday. I have been conducting floristic studies for many years and have visited every region in China except Taiwan. It will be a pity if I cannot visit there in my lifetime. We would like to thank the host—the South China Institute of Botany at Ghangzhou, particu- larly Professor Lin, Yeou-ruen, who arranged everything necessary for the meeting in a very short time. Although the meeting could not be held in Hongkong as planned, we give our sincere ap- preciation to Professor Paul Butt and Professor hang Shu-ting, his department head, for their excellent preparations for the meeting. LITERATURE CITED е aa 1959. Flora Reipublicae Popularis icae, Volume 2: 1-406. Science Press, Beijing. Hu, жеңе HSU. 1936. The distribution of the Chinese conifers and taxads. Acta Bot. Sin. 2: 767-784 . 1948. sr Td of China. Thoughts and Time Monthly 52: Li, Hur. LIN. 1944. The ОО] ne of i i e. Proc. Floristic — and problems of eastern Asia. Taiwania 1: 1—5. 19 Floristic relationships between eastern Asia and eastern North America. Trans. Amer. Phi- los. Soc. п. ser. 42: 371-429, maps 1-56. 1953a. Endemism in ligneous flora of eastern Asia. Proc. 7th Pacific Sci. Congr. (New Zealand) ME 212-216, 1953b. Floristic interchanges between For- mosa and the Philippines. Pacific Sci. 7: 179-186. 1953c. Present distribution and habitats of the conifers and taxads. Evolution 7: 245-261. : The genetic affinities of the Formosan flora. Proc. Sth Pacific T Congr. 4: 189-195. ыо NGO. the phytogeography of rth and West China. бй. Inst. Bot. Natl. Acad. Pepin 2: 423-45 944. ola of Yunnan. Pp. 37 in Collected Papers for Celebration of Mr. Shi- Zeng Li’s 60th Birthda 55. The distribution of plants in = China. Ш. Man. Woody Pl. Wu, Zn YI. 1963. On de pus s Chinese floristic eas. Proc. Symposium 30th Anniv. Chin. Bot. Soc 1965. On the quen affinity of the Chinese flora Sci. News Januar HE-SHEN WANG. 1983. In: The Editorial Committee of the Natural Geography of China (ed- itors), The Natural Geography of China, Volume 1. Phytogeography. Science ав Beijing. ZHANG, HONG-DA. origin and development of Cathaya flora. Ácta Sei. Nat. Univ. Sunyatsen 1: 0. SYSTEMATICS OF Peter Goldblatt? LAPEIROUSIA (IRIDACEAE-IXIOIDEAE) IN TROPICAL AFRICA! ABSTRACT The genus Lapeirousia, a member of the predominantly African subfamily Ixioideae of Iridaceae, is one of five genera of tribe Watsonieae and is the only one that is widespread i in southern and tropical Africa. In this revision of the tropical members of the genus, 16 species are recognized, 14 assigned to the largely tropical sect. Paniculata (s ubg. Paniculata) and two to the largely temperate southern African sect. Sophronia (subg. Lapeirousia). One new species, L. angolensis, is described and L. teretifolia is raised from subspecies to species rank. The center for tropical African Lapeirousia is northern Namibia, but species occur across south tropical Africa to Mozambique and north to Ethiopia, Sudan, and Nigeria. Lapeirousia is one о anh A represented in both tropical Africa and the winter-rainfall region of temperate southern Africa, к it is unique in its wide distribution in drier parts of Africa rather than the well-watered eastern highland areas of = pana Variation among the species is largely floral, and flowers range from completely actinomorphic to medianly биши and from short-tubed to extremely long- omosomal variation occurs in sect. Paniculata with numbers ranging from n = 8, о З and karyotypes from strongly bimodal to relatively uniform. Chromosome cytology correlates to some ii dad with patterns of morphological variation and provides жш support for the phylogeny of the tropical specie Lapeirousia, a member of the largely African subfamily Ixioideae of the Iridaceae, is one of the few genera in the family well represented in tropical Africa and in the southern African winter-rainfall zone of the western Cape Province of South Africa and southwestern Namibia. Lapeirousia comprises two subgenera (Goldblatt & Manning, 1990), subg. Lapeirousia (17 species—Goldblatt, 1972), large- ly temperate, and subg. Paniculata with sect. Fas- tigiata (5 species—Goldblatt & Manning, in prep.) restricted to the SW Cape of South Africa, and sect. Paniculata (15 species) mostly tropical Af- rican (Table 1) but extending well into the Trans- vaal, and with a disjunct species in the SW Cape (Goldblatt & Manning, in prep.). Although centered along the west coast of South Africa, subg. Lapei- rousia includes two species, L. odoratissima and L. littoralis, having wide ranges in tropical Africa (Table 1). Of the approximately 38 species in the genus some 2] are restricted to the southern Af- rican winter-rainfall area (Goldblatt, 1972), and 16 are largely tropical. Only L. littoralis (= L. caudata) is widely shared between the two areas. The tropical African species are concentrated in south central Africa, particularly in Namibia, An- gola, and Zambia with a decreasing representation in East Africa and Mozambique. Two species reach Ethiopia and one extends to Nigeria. The taxonomy of Lapeirousia in tropical Africa has long been considered problematic, and there has been no complete treatment of the genus out- side southern Africa since Baker's study in Flora of Tropical Africa (1898). Although completed at a time when tropical Africa was incompletely ex- plored botanically, this revision admitted 14 species of true Lapeirousia and four more that are now treated as the genus Anomatheca. Since 1898, 21 tropical species have been described, making a total of some 35 for the area. The local floristic treatments published over the past 20 years for Flora of West Tropical Africa (Hepper, 1968), Namibia (Sólch, 1969) and Zaire (Geerinck et al., 1972), and a new species described by Wanntorp (1971) represent the only significant contributions ' Support for this study by grant BSR 85-00148 from the U.S. National Science Foundation and grant 3749-88 from the National Geographic Society 18 gratefully acknowledged. I thank Sylvester Chisumpa, Kitw owledged with gratitude. I also extend my appreciation to John esa and Margo Branch for the illustrations e add pets to the value of this study. Krukoff Curator of African Botany, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, USA. ANN. Missour! Bor. GARD. 77: 430-484. 1990. Volume 77, Number 3 1990 Goldblatt 431 Systematics of Lapeirousia in Tropical Africa TABLE 1. summary of their distributions. The species of Lapeirousia in tropical Africa alphabetically arranged within their sections, including a SUBGENUS PANICULATA SECTION PANICULATA L. abyssinica eastern Ango western and northern Namibia L. angolensis L. avasmontana sii L. baine erythrantha northern Ethiopia, eastern Sudan gola central and northern Namibia, southern Angola, Botswana and northwestern Transvaal central and northern Namibia, northwestern Botswana eastern Angola, Zambia, southern Zaire, Zimbabwe, Malawi, western and southern Tanza- nia, Mozambique, northeastern Botswana L. gracilis L. masukuensis L. otaviensis central Namibia L. rivularis L. sandersonii L. schimperi Ethiopia, Su L. setifolia L. teretifolia SUBGENUS LAPEIROUSIA L. littoralis subsp. littoralis Transvaal central Mozambique, eastern Transvaal, southeastern Zimbabwe Namibia, southern Angola and Zambia central and western Transvaal, eastern Botswana northern Namibia, southern Angola, Zambia, Zimbabwe, Tanzania, northern Kenya, northern Mala, p Tanzania, Zimbabwe western Zambia, northeastern Angola, southern Zaire southwestern Angola, western and southern Namibia, Botswana, northern Cape, western caudata northern Namibia, Zambia, Zimbabwe, southern Mozambique L. odoratissima nia northern Namibia, southern Angola, Zambia, southern Zaire, Zimbabwe, Malawi, Tanza- to our knowledge of Lapeirousia outside DE Africa. The species of the winter-rainfall z southern África were revised in 1972 (Goldblatt, 1972) when Anomatheca, treated by Baker (1892, 1896, 1898) as a subgenus, was removed from Lapeirousia and restored to generic rank. The present treatment, summarized in Table 2, represents a complete revision of Lapeirousia in tropical Africa. This treatment differs substantially in the delimitation of L. erythrantha (Geerinck et al., 1972) and includes several changes in the circumscription of species in Namibia (cf. Solch, 1969). In Namibia I recognize L. avasmontana (previously included in L. coerulea) and L. ota- viensis (included by Sólch in L. bainesii), whereas I consider L. bainesti and L. vaupeliana conspe- cific, although both were recognized by Solch. The widespread L. caudata has an earlier synonym, L. littoralis, but I still consider the species to comprise two subspecies, one largely tropical and the other largely temperate (cf. Goldblatt, 1972) Treatment of the Lapeirousia erythrantha complex is difficult and I doubt that any solution rhodesiana and L recognize as distinct L. sandersonii, kuensis, L. teretifolia, L. setifolia, and L. ango- including L. lensis (the last-mentioned a new species). Geerinck et al. (1972) treated some of the foregoing species as varieties of L. erythrantha. GENERIC DEFINITION AND RELATIONSHIPS Lapeirousia is a member of the Old World and largely African subfamily Ixioideae, one of the four subfamilies of Iridaceae (Goldblatt, 1990a). It is currently assigned to tribe Watsonieae (5 genera, ca. 95 species), one of three tribes of Ixioideae (Goldblatt, 1989, 1990a), the others being the monotypic Pillansieae and the large Ixieae, with some 25 genera and over 700 species, nearly half the total for the family. Distinguishing character- istics of Watsonieae are deeply forked style branch- es, a derived condition (Goldblatt, 1989), and corms produced entirely from a bud near the base of the flowering stem and attached laterally to the flow- ering axis (Goldblatt, 1990a). The latter is thought to be ancestral to corm development in Ixieae in which the corm is formed, in part, from the base of the flowering stem, and the flowering stem is attached to the corm near the corm apex (Goldblatt, 1989). Lapeirousia is readily defined by its character- istic corm and corm tunics. The corms of all species are more or less bell-shaped with a flat base or side 432 Annals of the Missouri Botanical Garden (Fig. 1C-H). The tunics are typically hard and cartilaginous to woody and consist of concentric layers of either densely compacted fibers or woody material of uniform texture. Corms of this shape are not found in other Watsonieae but occur in a few genera of Ixioideae where they are found in only a few specialized species of Romulea and Hesperantha (which coincidentally also have woody corm tunics). The genus probably most closely re- lated to Lapeirousia is the monotypic south trop- ical African Savannosiphon Goldbl. & Marais. The two genera share two unusual features in Watsoni- eae, angular to winged stems and membranous walled capsules. Outgroup comparison suggests that both characters are specialized. Lapeirousia and Savannosiphon are probably most closely allied to Thereianthus and Micranthus, two small south- western Cape genera (Goldblatt, 1989). These four genera form a clade, united by having the foliage leaves inserted on the flowering stem rather than on the corm (Fig. 1B, C); thus the leaves do not contribute to the formation of the corm tunics, as they do in Watsonia (the remaining genus of Wat- sonieae) and in the primitive monotypic Pillansieae. The evolutionary relationships of Lapeirousia are discussed further in the section below dealing with phylogeny. GEOGRAPHY Lapeirousia is one of few widespread African genera of Iridaceae that have centers in the winter- rainfall zone of the southern African west coast and in tropical Africa (Fig. 2). This pattern is unusual for Iridaceae, in which most African gen- era are either restricted to the Cape region of South Africa or occur only in the summer-rainfall parts of eastern southern Africa, sometimes as far north as Ethiopia. Only Gladiolus, Hesperantha, and Romulea (Ixioideae), Moraea (Iridoideae), and Aristea (Nivenioideae) have ranges comparable to that of Lapeirousia. However, all of these genera favor mesic habitats in tropical Africa and occur either in montane and high-plateau areas or in zones of particularly high rainfall. Lapeirousia alone has radiated into dry parts of Africa and is well represented in Namibia, adjacent southern Angola, Botswana, and Zambia. Elsewhere in tropical Af- rica the number of species of Lapeirousia de- creases rapidly; five species occur in Zimbabwe, three in Mozambique and Tanzania, two in Ethio- pia, and one in Sudan and Nigeria. The area of greatest species concentration is the northern half of Namibia (Fig. 2A). Here the en- demic Lapeirousia avasmontana occurs locally in the Windhoek area, and L. gracilis extends along the length of the country from the Fish River Canyon in the south to the Kaokoveld in the north. Lapeirousia otaviensis is nearly endemic, extend- ing from the Erongo Mountains in the central west to southern Angola, and L. coerulea, widespread and relatively common in central and northern Namibia, also occurs in the northwest of Botswana. Lapeirousia bainesii is particularly common in northern Namibia, but it extends into southern Angola and across Botswana to the northwestern Transvaal. In addition, L. schimperi, L. rivularis, L. littoralis, and L. odoratissima occur widely in Namibia as well as elsewhere in tropical Africa. In temperate southern Namibia three more species of Lapeirousia, L. barklyi, L. dolomitica, and L. plicata, are predominantly southern African and have a winter-growing and spring-flowering phe- nology. A second, minor center for Lapeirousia in trop- ical Africa (Fig. 2B) is western Zambia, southern Zaire, and eastern Angola where two species are endemic, L. teretifolia in the north and L. an- golensis in the south. Also widely occurring in this area are L. erythrantha, L. rivularis, and L. lit- toralis. Other relatively localized species are L. abyssinica, restricted to northern Ethiopia; L. ma- sukuensis, central and southern Mozambique, the eastern Transvaal, i and L. vaal and eastern Botswana. The tropical African species of Lapeirousia are, in general, relatively widely distributed, the most prominent example being L. schimperi. and southeastern Zimbabwe; sandersonii, the central and western Trans- This species extends across south tropical Africa from northern Namibia to Zimbabwe and has a series of disjunct populations in northern Tanzania, northeastern Kenya and southern Ethiopia, northern Ethiopia, and western Sudan. Lapeirousia erythrantha has a comparable distribution, being common across south central Africa from eastern Angola to the Mozambique coast, and it has a series of populations in northern Nigeria. Unlike L. schimperi, L. erythrantha con- sists of a number of distinctive regional populations across its range. e only ы of subg. Lapeirousia that ос- cur in tropical Africa, L. littoralis and L. odora- tissima, also have wide distributions, the latter extending from western Angola and Namibia to central Tanzania. This contrasts with the southern African members of the subgenus that have narrow ranges, and in some cases are known from only one or two localities , L. montana, L. oreo- gena, L. verecunda a 1972)). Such disjunctions as encountered in L. schim- Volume 77, Number 3 Goldblatt 433 1990 Systematics of Lapeirousia in Tropical Africa ( == 2 = LES 22 2222 FIGURE 1. Growth forms and main flower and corm types іп Lapeirousia. — А. L. bainesii, divaricately branched pseudopanicle with zygomorphic flowers (subg. Paniculata sect. Paniculata).—B. L. corymbosa, congested pseu- dopanicle with actinomorphic flowers and plane leaves with midrib (subg. Paniculata sect. Fastigiata).—C. L. dolomitica, branched spike, zygomorphic flowers, and corrugate leaf (subg. Lapeirousia sect. Lapeirousia).—D. L. plicata (subg. Lapeirousia sect. Sophronia).—E. L. divaricata (subg. Lapeirousia sect. Lapeirousia).—F. L. micrantha (subg. Paniculata sect. Fastigiata). —G. L. bainesii. —H. L. coerulea (subg. Paniculata sect. Paniculata). -C x 0.5, E-H full size. (Drawn by М. L. Branch & J. C. Manning.) 434 Annals of the Missouri Botanical Garden a 101 ~ | t hk же, 7, ^ Aq J т 3 0 UA || "X 10 Jd AJ > acere | WY t | | 2 0 * | + | | 10 | + P dX 3 | | 1 1 20 T ni B ү ү ^ o 1091400 #00 800 1000 км A д, - Lo tl * ap » | £ FIGURE 2. Geography of Lapeirousia. — А. Distribution of subg. Paniculata. —B. Distribution of subg. Lapei- rousia. Figures represent the number of species recorded in each 2°30'-degree square grid. peri and L. erythrantha are uncommon in Irida- ceae, although they are known in a few species in several of the widespread African genera of the family (e.g., Moraea schimperi, Gladiolus da- lenii, and Hesperantha petitiana [Goldblatt, 1977, 1986 ]). However, the pattern of narrow distribu- tions for many species in southern Africa vs. wide ranges for most tropical species is common in Af- rica and is consistent with patterns for many genera in different families. MORPHOLOGY ROOTSTOCK The most distinctive feature of Lapeirousia, and the one that defines it, is the flat-based corm (or more correctly flat-sided corm since the flat portion is oriented obliquely in the ground—Fig. 1). The corms are bell-shaped in all species when not dis- torted by growing conditions. The corm coverings or tunics reflect the corm shape and consist of a separate basal disc and bell-shaped upper part. Other Ixioideae usually have rounded corms, al- though a few species of Romulea (de Vos, 1972) and Hesperantha (Goldblatt, 1984) have a flat side and are bell-shaped. The nature of the tunics varies considerably and has substantial phylogenetic and taxonomic signif- icance. In subg. Lapeirousia the tunics are dark ‘he basal margin un e Goldblatt, 1972) in sect. Lapeirousia (Fig. 1С, E), probably a specialized condition (Gold- blatt & Manning, 1990), whereas the margins in sect. Sophronia are entire or lightly lobed (Fig. 1D) In subg. Paniculata the tunics range from blackish to pale straw in color and vary in texture from more or less woody to coriaceous. They ap- pear to differ fundamentally from those of subg. Lapeirousia in consisting of densely packed fibers Volume 77, Number 3 1990 Goldblatt 435 Systematics of Lapeirousia in Tropical Africa instead of having a uniformly woody texture. Sim- ilar textured corm tunics are characteristic of the related genera Thereianthus and Micranthus, and tunics of densely compacted fibers appear, on the basis of outgroup comparison, to be the basic type for Lapeirousia (Goldblatt & Manning, 1990). The way in which the outer layers of the tunics decay in subg. Paniculata is often diagnostic. In the five Cape species of subg. Paniculata sect. Fastigiata (Fig. 1F) the hard, blackish layers fragment rather distinctly into vertical strips that separate from the base (Goldblatt, 1972). A similar pattern is evident in several members of sect. Paniculata, such as L. erythrantha, L. rivularis, and their allies in tropical Africa. The softer-textured tunics of L. bainesii, L. otaviensis, L. gracilis, L. schimperi, and L. coerulea are assumed to be a specialized condition. These tunics decay with age to form a coarse to sometimes fine reticulum. Lapeirousia avasmontana, considered conspecific with L. coe- rulea by Solch (1969), has distinctive dark brown to blackish tunics that break with age into brittle membranous pieces. The tunics of L. sandersonii closely resemble those of L. avasmontana, and while the two species have similar divaricately branched panicles with only 1—2 flowers per branch, they nevertheless have such different flowers that it is difficult to believe they are closely related. Their similar corm tunics and inflorescence branch- ing are regarded here as convergent (Fig. 3). The woody tunics of subg. Lapeirousia are considered to be a specialized condition in the genus. A few other taxa in Ixioideae have similar woody tunics (Romulea, Syringodea, Hesperantha, Geissorhi- za), but the corms in these genera are rarely bell- shaped and these genera are generally considered to be only distantly related to Lapeirousia (Lewis, ). 1954; Goldblatt, 1971, 1990a CATAPHYLLS The first foliar organs produced by the sprouting corm, the cataphylls, are entirely sheathing and submembranous. They surround the base of the stem and reach only a few centimeters above the ground. In Lapeirousia usually two are produced, and they are often dry and dead at flowering time. LEAVES As in most Iridaceae, the foliage leaves of Lapei- rousia are ensiform and equitant. They are at- tached to the stem near the base, usually close to ground level (e.g., Fig. 1C), but unlike most genera of Ixioideae, the sheathing leaf bases do not con- tribute to the corm tunics. This is probably a de- rived condition (Goldblatt, 1989) shared, in Wat- with Thereianthus, Micranthus, and Savannosiphon, and is important in separating this sonieae, line from Watsonia. The leaves are plane with at least a discrete central vein (pseudomidrib) in subg. Paniculata (e.g., Fig. 1B), but are corrugate in subg. Lapeirousia (e.g., Fig. 1C). Anatomical dif- ferences such as opposed vs. alternate veins and truncate ribs (Table 3) accompany the external differences (Goldblatt & Manning, 1990). In sect. Paniculata the leaves are typically narrowly lan- ceolate, but sometimes are linear (forms of L. erythrantha, L. setifolia, L. sandersonii) or terete (L. teretifolia). The leaves of L. sandersonii are distinctive in being particularly rigid and fibrotic and in having the several thickened veins set closely together. STEMS Stems are usually aerial and branched, and in- variably angular to winged, a synapomorphy shared with Savannosiphon. While there are subtle dif- ferences in the degree of angularity of the stems in different species, the character is not useful taxonomically. In some species of subg. Lapeirou- sia sect. Sophronia in southern Africa (Goldblatt, 1972) and in the tropical African L. odoratissima (see Fig. 17) the stem is not produced above the ground, and the whole inflorescence is congested into a tufted, rosettelike structure borne at ground level. INFLORESCENCES A spicate inflorescence is characteristic of most Ixioideae and is a synapomorphy uniting Watsoni- and Ixieae, two of the three tribes of the subfam- - (Coldblatt, 1990a). Most members of subg. Lap- eirousia have spikes, but in a few species the whole aerial axis is contracted into a cushionlike tuft borne at or near ground level. Among the tropical species of subg. Lapeirousia, L. littoralis has spikes f 4-12 flowers, these somewhat congested and fewer-flowered in subsp. littoralis, and rather lax and with more flowers in subsp. caudata. In L. odoratissima the aerial stem is contracted and the plant has a tufted appearance (Fig. 17), a habit shared with and perhaps independently evolved in the southern African L. oreogena, L. montana, and L. plicata (Goldblatt & Manning, 1990). An extensively ramified flowering axis is char- acteristic of subg. Paniculata (Fig. 1A, B), and, although variously called a panicle or corymb, this pseudopaniculate structure is probably a highly ramified spike. Despite sometimes highly developed branching, the ends of the main branches usually carry at least two flowers, and those below the 436 Annals of the Missouri Botanical Garden terminal flower are always sessile and thus arranged exactly like those in the spike of subg. Lapetrousta and other Ixioideae with spikes. The pseudopanicle of subg. Paniculata is considered a specialized condition in the genus and is the only major syn- apomorphy for the subgenus (Goldblatt & Man- ning, 1990). The degree of inflorescence branching in subg. Paniculata is often typical of a species. Lapei- rousia abyssinica has few-branched stems, and the inflorescence is virtually a branched spike, hardly or not at all different from the branched spikes found elsewhere in Ixioideae. I assume that this is a reversal from the pseudopaniculate con- dition. In L. erythrantha the main branches of the rather lax panicle have 3—5 flowers that tend to be crowded terminally, which impart a corymblike appearance to the sometimes massive inflores- cence. The related L. teretifolia and L. angolensis have only one or two flowers per major inflores- cence branch, a feature shared with L. sander- sonii. Lapeirousia masukuensis, which resembles L. sandersonii in its long-tubed flowers, has in contrast 5-8 flowers on the terminal branches. Among the long-tubed species of the arid southwest, L. bainesii typically has 1(ог 2) flowers per main inflorescence branch, whereas L. otaviensis and L. gracilis, otherwise easily confused with L. bainesii, have 3-5 flowers per main branch. Sim- ilarly, an important distinction between the easily confused L. avasmontana and L. coerulea is the number of flowers on the main terminal inflores- cence branches, usually 2-4 in L. avasmontana and seldom more than one in L. coerula. In subg. Paniculata the floral bracts vary little among the species except in size, and they provide limited taxonomic information. However, the bracts are membranous and dry above in sect. Panicu- lata, which outgroup comparison suggests is spe- cialized (Goldblatt & Manning, 1990). In sect. Fastigiata the bracts are herbaceous. The bracts are more or less equal in length, but the inner (adaxial) bract is always smaller and is often api- cally forked or notched. In subg. Lapeirousia the bracts are herbaceous and usually have a firm = texture, and the inner bracts are shorter than the outer ones. Bract morphology is variable in this predominantly southern African alliance (Goldblatt, 1972), and species can sometimes be recognized by their bracts alone. FLOWERS Floral variation is extensive in Lapeirousia, as much in tropical as in southern África. Two species, L. coerulea and L. avasmontana, have actino- morphic flowers and short perianth tubes. Their flowers closely resemble those of the SW Cape L. corymbosa (sect. Fastigiata) and are, І assume, the basic flower type in the genus. The other species of sect. Paniculata have zygomorphic flowers with unilateral stamens. The acaulescent L. odoratis- sima also has actinomorphic flowers but has a long perianth tube. There is considerable variation in perianth tube length in the genus, a feature closely related to the pollination systems of the species and not always indicative of a close phylogenetic relationship. Long tubes of ca. 12-15 cm are found in L. schimperi and L. odoratissima. A long perianth tube also characterizes L. littoralis, in which the degree of variation in length is unusual: subsp. littoralis has a tube ca. 28-35 mm long, whereas in subsp. caudata the length ranges from 25 to 30 mm in some northern Zambian populations to 50-70 mm in the populations from southern Mozambique. Lapeirousia bainesii, L. otaviensis, and L. gra- cilis, centered in Namibia, each have similar and relatively long perianth tubes, usually 25-35 mm long. In the L. erythrantha complex, L. masu- kuensis and L. sandersonii have tubes 15-25 mm long compared with 8-12 mm in other members of the complex, including L. erythrantha and L. setifolia in which the tube is more or less twice as long as the bracts, a condition that I assume is basal for the alliance. The two tropical species of sect. Paniculata with actinomorphic flowers, L. coerulea and L. avasmontana, have very short perianth tubes, less than 3 mm long, comparable to those in the southern African L. corymbosa complex. The flowers of Lapeirousia coerulea and L. avasmontana resemble relatively closely those of the southern African actinomorphic-flowered species, and it seems likely in the absence of con- trary evidence that the actinomorphic, short-tubed flower with a blue perianth having white markings is basal for the genus (Goldblatt & Manning, 1990). If this is correct, it follows that zygomorphic flowers have evolved independently in subg. Paniculata in both the Cape sect. Fastigiata and the tropical sect. Paniculata, as well as in subg. Lapeirousia. Floral actinomorphy in Lapeirousia odoratis- sima may be secondary, and related to the low stature and tufted habit of this species (Goldblatt & Manning, 1990). I suggested a similar reversal of floral zygomorphy to actinomorphy in the south- ern African tufted species (Goldblatt, 1972). The most common perianth form in the zygo- morphic species is for the lower three tepals to be Volume 77, Number 3 1990 Goldblatt 437 Systematics of Lapeirousia in Tropical Africa joined for a short distance and displayed horizon- tally, while the upper tepal is reflexed and held more or less in the same plane as the lower tepals. This is unusual in Ixioideae in which species with zygomorphic flowers usually have the upper tepal held erect or slightly arched over the stamens. The latter type of flower is found in L. rivularis, while reflexed tepals are characteristic of all the members of the L. erythrantha alliance, including the long- tubed L. sandersonii and L. masukuensis. The reflexed upper tepal is regarded here as the basal condition, and the erect to arched tepal is consid- ered derived. he stamens are either erect with filaments con- tiguous in the actinomorphic-flowered species or unilateral and erect to arcuate in zygomorphic- flowered species. The style follows the orientation of the stamens, and is thus central and erect in species with actinomorphic flowers, and arched be- hind the stamens in species with zygomorphic flow- ers. The style branches are divided and recurved for half their length in most species of Lapeirousia, and this is assumed to be the basic state for the genus and the tribe Watsonieae. However, several tropical African species of Lapeirousia and some forms of the southern African L. plicata have undivided style branches. The character is variable within some populations of L. plicata and L. bainesii that I have examined in the field. It is difficult to assess the significance of this variation, and I have not regarded undivided style branches as having any taxonomic significance. CAPSULES AND SEEDS Apart from differences in the size of the capsules and the seeds they contain, fruit characters are of limited use in the taxonomy of Lapeirousia. The capsules have firm-membranous walls and range from globose-trigonous to more or less oblong, al- most always with the outline of the seeds distorting the outer walls. The seeds are globose to м ly oblong, tapering slightly toward the funicle. The raphal ridge is the only feature distorting the oth- erwise uniformly microreticulate surface. The seeds are sometimes lightly distorted by pressure. Seed diameter ranges from 2.5 mm in L. odoratissima to 1.3-1.8 mm in L. setifolia. CHROMOSOME CYTOLOGY Chromosome number is remarkably variable in tropical African Lapeirousia (Goldblatt, 1990b). In sect. Paniculata haploid numbers range from n = 8 (L. avasmontana) to n = 3 (one population of L. bainesii) (Table 2). Numbers for subg. Lapei- TABLE 2. Chromosome numbers in tropical African Lapeirousia (from Goldblatt, 1990b). Species Diploid number 2n SUBGENUS PANICULATA SECTION PANICULATA L. angolensis unknown L. avasmontana 16 L. bainesii 10, 6 L. coerulea 8 L. erythrantha 12 L cilis 12 L. masukuensis unknown L. otaviensi L. rivularis 12 L. sandersonii 10 L. schimperi 10 L. setifolia L. teretifolia unknown SUBGENUS LAPEIROUSIA L. littoralis 16 L. odoratissima 16, 18 rousia in southern Africa are л = 10, 9, and 8 (Goldblatt, 1972, 1990b). The karyotypes for all species of the subgenus are similar and strongly bimodal, comprising one long chromosome pair and six or seven small pairs less than half as long as the long chromosomes. Of the two tropical species of the subgenus, L. littoralis has n = 7 and L odoratissima, n = 8 and 7. The Cape species of subg. Paniculata have = 10 (Goldblatt, 1972) except Lapeirousia ne- glecta (Goldblatt & Manning, in prep.), which has n = 5, and a bimodal karyotype similar to that in subg. Lapeirousia. Among the tropical species, L. avasmontana has a karyotype most like the, Cape members of the subgenus, with one long and seven short pairs in a comparably bimodal karyotype. The species presumably most closely related to L. avasmontana, L. coerulea, has n = 4, now known from seven populations. The karyotype comprises two long and two medium-sized chromosome pairs. Lapeirousia erythrantha appears to be based on x = 6, a number recorded in several populations across its range from southern Malawi to northern Zambia. Lapeirousia rivularis also has n = 6, but the allied L. setifolia and L. abyssinica have n = The long-tubed species Lapeirousia sander- зот, L. bainesii, L. otaviensis, and L. schimperi appear to be based on x = 5. The karyotypes of the last-mentioned three species are similar to each 438 Annals of the Missouri Botanical Garden other, consisting of one particularly long metacen- tric pair and four shorter pairs, with one metacen- tric and the other three acrocentric. Lapeirousia sandersonii has a more bimodal karyotype with a long, acrocentric pair and four much shorter pairs. While this pattern seems fairly coherent, L. gra- cilis, closely allied to L. otaviensis, has n = 6 and a bimodal karyotype of one long and five much shorter pairs. Karyotypic variation occurs in L. bainesii, one population of the three studied having n = 3, with a karyotype of three metacentric chromosome pairs. This pattern suggests Robert- sonian fusion of the smaller acrocentric chromo- somes of the presumed basic x = 5 karyotype of this and related species. Variability in the karyotypes in Lapeirousia is puzzling and difficult to interpret. Elsewhere (Gold- blatt, 1990b) I have regarded the bimodal karyo- type with x = 10 as basic for the genus. It occurs in both subgenera of Lapeirousia and in the related Thereianthus and Micranthus, in which the karyo- type is weakly bimodal. This presumes that the lower numbers, n = 4 in L. coerulea, L. setifolia, and L. abyssinica, are derived. Judging from the chromosomal variability in L. bainesii alone, it seems that the chromosome constitution of Lapei- rousia is unusually unstable. Structural rearrange- ment and numerical change may form part of the adaptive strategy of a genus that is remarkable in the family for having radiated in semiarid habitats more severe than those that most genera of Iri- daceae favor. PHYLOGENY Within Watsonieae, Lapeirousia can be distin- guished by its specialized flat-based, bell-shaped corm, the primary synapomorphy for the genus (Goldblatt, 1989; Goldblatt & Manning, 1990) Perhaps most closely allied to Lapeirousia (Gold- blatt, 1989) is the monotypic Savannosiphon (Goldblatt & Marais, 1979), which shares with Lapeirousia compressed and angled to winged stems, also a derived condition, and coriaceous to membranous capsules. Except for Savannosiphon, which is tropical African, the remaining members of Watsonieae are southern African and are cen- tered in the southwestern Cape. Thereianthus and Micranthus are endemic to the SW Cape, and Watsonia is centered there, although nearly a third of the genus occurs in eastern southern Africa. Within the tribe, Watsonia forms one major clade and the Micranthus—Thereianthus and Lapeirou- sia~Savannosiphon line form the other (Goldblatt, ). This latter clade is defined by having the foliage leaves attached to the stem rather than to the corm, and in having thick, hard-textured tunics composed of compacted fibers. Thereianthus and Micranthus have an unusual type of seed, appear generally similar, and are almost certainly closely related. Savannosiphon, although presumed to be the genus most closely allied to Lapeirousia, seems rather different in its broad, soft-textured leaves, unbranched habit, and general appearance, and the possibility that it is misplaced in this scheme cannot be ignored. Within Lapeirousia there are two major species clusters, treated as subgenera by Goldblatt & Man- ning (1990). epa Lapeirousia has specialized corrugate leaves and some associated anatomical specializations (Goldblatt & Manning, 1990), woody corm tunics, and predominantly spicate inflores- cences (the latter a symplesiomorphy) (Table 3). Predominantly southern African, the phylogeny of subg. Lapeirousia is not considered further here. Subgenus Paniculata has unspecialized plane leaves with a central vein, tunics consisting of densely matted fibers that become ridged, cancel- late, or fibrous on aging, two symplesiomorphies, and pseudopaniculate inflorescences and small flo- ral bracts, both derived states. The SW Cape and tropical African members of subg. Paniculata ap- pear to comprise separate monophyletic lines, treated as sects. Fastigiata and Paniculata, re- spectively (Fig. 3). The largely tropical sect. Panic- ulata, although similar in many ways to sect. Fas- tigiata, can be distinguished by its generally more lax inflorescences and small bracts that are typi- cally dry and membranous above or entirely. Except for Lapeirousia coerulea and L. avas- montana, the species of sect. Paniculata have zygomorphic flowers, the synapomorphy that unites these 12 species (Fig. 3). In the section, four species form a clade, defined by their long-tubed flowers, pale perianths, and moderately fibrous corm tunics. Of these four species, all except L. gracilis have a similar and specialized basic karyotype with n = 5 and form a clade within which L. defined by two synapomorphies: divaricate branch- bainesii is ing and inflorescence branches with predominantly one flower each; L. schimperi is set apart by two synapomorphies: exceptionally long-tubed flowers and a white perianth. I have not identified any synapomorphy for the remaining L. otaviensis. The remaining eight tropical species of sect. Paniculata seem related, and they have flowers of similar form, color, and basic markings, yet there is no obvious synapomorphy uniting them. Lapei- rousia sandersonii and L. masukuensis share one synapomorphy, a moderately long perianth tube, Volume 77, Number 3 1990 Goldblatt 439 Systematics of Lapeirousia in Tropical Africa and L. sandersonii is further distinguished by its divaricately branched inflorescence with predom- inantly one- or two-flowered axes and peculiar, smooth corm tunics that fragment into vertical strips. Lapeirousia avasmontana shares these two last-mentioned synapomorphies, but its actino- morphic perianth and different karyotype suggest that its similarities with L. sandersonii are the result of convergence. Lapeirousia teretifolia and L. angolensis appear united by their terete leaves. he remaining species are undoubtedly closely related, but none share notable specializations that indicate that they are more closely allied to one another than to other species in the group. The most notable specialization is in L. rivularis in which the upper tepal is erect to hooded instead of being reflexed to lie in the same plane as the lower tepals. This species and L. erythrantha are evidently polyploid, п = б, but this may be an independent specialization in each and is so treated here. The possession of the derived chromosome number n — 4 in L. setifolia and L. abyssinica may indicate close relationship, but too few species in the alliance are known chromosomally for this character to be interpreted with confidence. No doubt relationships in this alliance will be better understood when the cytology is better known, but for the present the phylogeny cannot be further resolved objectively. HISTORY ОЕ LAPEIROUSIA The first tropical African species of Lapeirou- sia, L. abyssinica, was discovered in 1809-1810 by Henry Salt during his East African travels (Salt, 1814), and it had been thoroughly documented in herbaria by 1850 when it was described as a species of Geissorhiza (Richard, 1850). In 1878 С. abys- sinica was transferred to Lapeirousia, and the presence of the genus in tropical Africa was thus established (Baker, 1878a). Until then Lapeirou- sia was thought to be an exclusively southern Af- rican and largely Cape genus, although Ovieda erythrantha from Mozambique had been described in 1864 and transferred to Lapeirousia in 1878, at the same time as L. abyssinica. The early taxo- nomic history of Lapeirousia and its establishment as a genus thus largely concerns the nontropical species. Lapeirousia was based on L. compressa, de- scribed by Pourret in 1788 from a single specimen that was thought to have been collected in Mau- ritius by the French botanist Philibert Commerson (Goldblatt, 1972). Lapeirousia compressa is, how- ever, conspecific with the Cape and Namaqualand endemic, L. fabricii, which had been described some 22 years earlier as /xia fabricii by Daniel de la Roche in 1766. The first species of Lapei- rousia known to science was described 10 years prior to this, when Linnaeus named /xia corym- bosa, but the species was only recognized as a member of Lapeirousia in 1802. Species of Lapei- rousia described in the eighteenth century were also placed in Gladiolus and Galaxia. In 1802, Pourret's Lapeirousia was accepted by Ker who transferred the five species of the genus then known to it. Ker also accepted L. juncea, which he as- signed to the new Anomatheca in 1805. 1817 Sprengel described Ovieda, at first without any species, but in 1825 he placed all the known species of Lapeirousia, including by name L. compressa, in Ovieda, which has the same circumscription as Lapeirousia sensu Ker. Re- named Meristostigma by Dietrich in 1844 because it was a homonym for Ovieda L., Ovieda Sprengel nevertheless remained in use, for example by Klatt (1866), until 1876 when J. G. Baker revived Lapeirousia. Sophronia Licht. ex Roemer & Schult. was erected in 1817 for the acaulescent L. plicata, but the genus was reduced to a subgenus of Lapei- rousia by Baker (1892). By then subg. Sophronia included three southern African acaulescent species, all now considered conspecific (Goldblatt, 1972). Baker did not, however, regard the acaulescent /,. odoratissima from tropical Africa as a member of subg. Sophronia. Baker also transferred Anomatheca Ker (1805), a genus of five central and southern African species, to Lapeirousia as a third subgenus (Baker, 1892). Anomatheca is probably most closely related to the southern African Freesia (Goldblatt, 1971, 1982), and it has been removed from Lapeirousia (Goldblatt, 1972), although Geerinck et al. (1972) treated А. grandiflora as a species of Lapeirousia. Freesia and Anomatheca are currently regarded as members of Ixieae, and their deeply divided style branches, shared with Lapeirousia, are re- garded as a convergent development (Goldblatt, 19904). The knowledge of Lapeirousia in tropical Africa began to accumulate late in the nineteenth century when Welwitsch's Angolan lridaceae, collected during 1853-1861 (Rendle, 1899), were studied by Baker (1878b). At this time the widespread L. schimperi had already been discovered in Ethiopia by Schimper and referred to Tritonia (Klatt, 1866), although it does not have either round-based corms with finely fibrous tunics or the undivided style branches of the latter genus. The same species 440 Annals of the Missouri Botanical Garden LAPEIROUSIA subgenus Lapeirousia subgenus Paniculata CC ү section Paniculata 5 ^ E E: © $ 3 5 E: c c 2 = = o c E o = = 8 ©. E. 5 S 2 2 Е о © © 2 Е с © Кы © 5 S c e o G E Ф 2 " 2 = Ф 4 Ф ш 2 = Є 5 2 s E: 2 p 2 E 3 Е o oO 2 = E = Ф = o o E c c € E: a = ә о Ф = > a о о > c z o o © Ф S 3 Ё © Ei e 2 @ 5 5 © © о 2 5 5 o o & Ф © S = Ф x E D o o о a % $ $ $ -i 2 E Е 3 E E E E 2 zi E 2 3 15 18 m = 20 =Е25 = 19 3 9 flower zygomorphic 3 corm margins toothed $ = 13 plant short, infl. dense 11 bracts + dry above == 9 flower zygomorphic T 6 ribs truncate + 5 veins alternate + 4 leaves corrugate —+ 10 bracts small + 2 tunics woody -]- 8 infl. a pseudopanicle = flower white -]- stamens/style included T 9 flower zygomorphic 1 corm bases flat 12 capsules coriaceous 7 stems angular FIGURE 3. relationships for sect. Paniculat den rs use by double E lines and re consistency index including шы. C collected in Angola was referred by Baker (1876) to Anomatheca (А. monteiroi), and two more Ап- golan collections were described as L. fragrans and L. cyanescens, respectively (Baker, 1878b). Tritonia schimperi was transferred to Acidanthe- ra (now a synonym of Gladiolus) by Baker (187 8а), as A. unicolor, because its long perianth tube and поса pr ny of Mere ош the major infrageneric po and presumed species cladogram are listed in Table 3. Parallelisms are indicated sals by crossed pen The cladogram was eed manually: length — 44; — 0.64. white flower accorded well with Acidanthera. This one species of Lapeirousia was thus placed in four different genera over a period of 12 years. The tropical African species of Lapeirousia that occur in the northern Cape and Transvaal were discovered relatively late, with the exception of L. littoralis. Although based on an 1859 Welwitsch Volume 77, Number 3 1990 Goldblatt 441 Systematics of Lapeirousia in Tropical Africa TABLE Characters used in the cladogram (Fig. 3). The derived (apomorphic) states are listed first followed y the presumed ancestral (plesiomorphic) conditions. ¡e for Savannosiphon are taken from Gold- blatt (1989). Anatomical specializations for sect. Fasti- giata known only in one species, L. corymbosa, are not included in the cladogram. 1. Corm bases flat —corm bases rounded 2. Corm tunics woody — corm tunics of compacted fi- bers 3. Margins of corm toothed/spiny — margins of corm not elaborate 4. Leaves corrugate— leaves plane and with a pseu- domidri 5. Major veins alternate — major veins opposite 6. Ribs truncate — ribs rounded 7. Stems angular — stems terete 8. Inflorescence a к d a spike 9. Flower zygomorphic — flower actinomorphic 10. Bracts is at ge relatively large 11. Bracts + membranous and dry above — bracts her- baceous 12. Capsules coriaceous — capsules woody 13. Plants short and inflorescences congested — plants tall and inflorescences not notably congeste 14. Perianth pale-colored — perianth colored blue to pur- 15. Perianth white without ле pale or deeply colored, usually with mar 16. Upper tepal erect or йе ыш, a recurved and lying in the same plane as the lower 17. Perianth tube 15-40 mm long—tube rarely ex- ceeding 15 mm lon 18. Perianth tube 8-12 cm long— tube shorter than 8 cm 19. Branching pattern с + айег- nate with a main axis dominan 20. Terminal branches of the rc 1(-2)-flow- ered — terminal branches with more than 2 flowers 21. Corm tunics becoming fibrous and pale with age — corm tunics es relatively densely fibrous and dark-color 22. Corm tunic a dd tical strips — corm tunics rough, d 209 pis to uam fibrous 23. Leaf terete —leaf plane with midrib evident 24. Base number х = 5 paginis not bimodal)— karyo- type bimodal and n = 25. Base number x — 4— base abe higher 26. Base number x = 6 and polyploid—base number different and not polyploid collection from Angola, L. littoralis was actually first collected by William Burchell in 1812 in the northern Cape, but L. littoralis had already been described (Baker, 1878b) when Baker named Bur- chell’s plants L. burchellii in 1892. The identity of L. littoralis has only now been established, the species having previously been known as L. cau- data subsp. burchellii (Goldblatt, 1972). Lapei- rousia caudata, now L. littoralis subsp. caudata, was described in 1890 from specimens from north- ern Namibia. It was based on collections by the Finnish missionary Martti Rautanen and the Swiss botanist Hans Schinz. The largely Transvaal L. sandersonii was first collected by John Sanderson in 1852, but again was named much later by Baker (1892), while L. bainesii was first recorded from the remote Kobi Pan in western Botswana by the landscape artist Thomas Baines in 1863 and later found by Em Holub in 1876 and by Edward Lugard in 1887, also in Botswana. The wide distribution of this species in Namibia was discovered much later, and only in 1942 was L. bainesii documented in the Transvaal, where it is rare. The first east tropical African species of Lapei- rousia was collected by the German explorer Wil- helm Peters in Mozambique in the mid 1840s. This was the first record of the widespread and common L. erythrantha, described by Klatt (1864) as Ovie- da erythrantha. Later exploration in interior Mo- de and Malawi by John Kirk and David provided additional records of this e distrib mns Its wi led to L. bride de being given different names in Katanga (L. briartii de Wildeman, 1900), An gola (L. spicigera Vaupel, 1912), and Zimbabwe (L. rhodesiana N. E. Brown, 1911). Rudolf Schle- chter's Mozambican collections made in the 1890s yielded additional L. erythrantha, although Vaupel (1912) described two as separate species, L. gra- minea and L. plagiostoma. At this time Schlechter made the first collections of L. masukuensis, also described by Vaupel (1912). Knowledge of Lapeirousia in Namibia was par- and unusual variability ticularly slow to accumulate, owing to difficulties of travel until after the First World War, despite the rich development of the genus there. The com- mon L. coerulea as well as L. otaviensis were first collected by the Swedish traveler and explorer Ture Een in 1879. These collections were overlooked, and L. coerulea was described in 1892 based on later collections, while L. otaviensis was named in . C. Foster who based the species on a 1925 collection made by Kurt Dinter. The wide- spread L. littoralis subsp. caudata was first col- lected by Rautanen in 1885 and described as L. caudata by Schinz. Later collections of this dis- tinctive subspecies were given different epithets, L. lacinulata being described from a Zambian col- lection, and L. delagoensis from specimens from 442 Annals of the Missouri Botanical Garden southern Mozambique. In Namibia the common L. bainesii was first collected by Rautanen in 1892, ut was not then associated with the Botswanan species. When collected later by Kurt Dinter, it was described as L. vaupeliana. Dinter’s Namibian collections are particularly important as his several duplicates were distributed to many herbaria. His collections also formed the basis for L. dinteri (= L. schimperi) and L. stenoloba (= L. littoralis), described by Vaupel in 1912; L. avasmontana, L. uliginosa (= L. schimperi), L. juttae (= L. odora- tissima), and L. ramossisima (= L. littoralis), all described by Dinter. Although Namibia is now fairly well explored botanically, it was only in 1971 that Lapeirousia rivularis was described by the Swedish botanist H. E. Wanntorp. The species had been collected as early as 1900 by H. Baum in southern Angola, and by К. Н. Barnard in 1921 in northern Namibia but was confused with L. coerulea or other species. Western Zambia and the eastern half of Angola await full botanical exploration, and it is from here that the two new species in this treatment come. Recognized in 1972 as a variety of Lapeirousia L. teretifolia is raised to species rank in this treatment. The new erythrantha by Geerinck et al., L. angolensis is incompletely known and is based on only two collections. Further exploration in An- gola is expected to resolve any doubts about this species and should also make it possible to deter- mine the identity of L. welwitschii, which was described by Baker (1878b) but cannot be matched satisfactorily with any known species at present owing to the state of the type material. Lapeirousia bainesii and L. otaviensis are relatively common in northern Namibia but are known from only one or two collections in southern Angola, where they may be fairly common. The establishment of the complete ranges of these species awaits further exploration of this area. The current circumscription of Lapeirousia dates from 1972 (Goldblatt, 1972). In this revision of Lapeirousia in the winter-rainfall zone of southern Africa, rousia and subg. Sophronia was not recognized at all. I established two sections for the 19 species Anomatheca was excluded from Lapei- treated, sect. Lapeirousia including Sophronia and sect. Fastigiata for the Cape species of what is now recognized as subg. Paniculata (Goldblatt & . The infrageneric classification did now subg. Paniculata sect. Paniculata, but rec- ognized the apparently important distinction be- tween corrugate-leafed subg. Lapeirousia with woody corm tunics and plane-leafed subg. Panicu- lata with tunics of compressed fibers. The infra- generic classification used here (Goldblatt & Man- ning, 1990) is a refinement of my earlier classification. ETHNOBOTANY As early as 1912 Dinter reported on the use of the corms of Lapeirousia species as a food in Namibia. Dinter mentioned specifically that corms of L. littoralis (as L. coerulea, L. schimperi (as L. uliginosa), and L. odoratissima (as L. juttae) are a valued food eaten after roasting in hot ashes. Dinter provided vernacular names onduvi (ozonduvi pl.) (Herero) and garib (Khoi) for Lapeirousia. Dinter's observation has been confirmed repeatedly by plant collectors and eth- nobotanists, notably R. Story for the Kung Bush- men (L. coerulea, Story 6121, and L. littoralis, Story 6162). Rodin (1985) documented that corms of L. coe- rulea, L. bainesii (as L. vaupeliana), and L. schimperi (as L. cyanescens) are eaten both raw and roasted by the Kwanyama Ovambos in north- ern Namibia. The Kung Bushmen are also reported to eat the corms of L. odoratissima for their water content (Marshall, 1976). The Kung are reported to bake and eat corms of L. bainesii or to pound the cooked corms into a meal, then eat them as a gruel with water (Fox & Young, 1982). In addition, a collection of L. gracilis (Seydel 3419) docu- ments the edibility of this species. n light of the amply documented use of the corms of several Namibian species of Lapeirousia as food, Watt & Breyer-Brandwijk’s (1962) un- substantiated report that L. coerulea is poisonous is doubtful, particularly since their tests performed on a frog for cardiac glycoside action proved neg- caudata), L. ative. Lapeirousia appears to have little value to hu- man populations outside Namibia and presumably adjacent Angola and Botswana. There are, how- ever, isolated reports that corms of L. erythrantha are eaten in the Shire Highlands (southern Malawi) “in time of great famine” (Buchanan 426). Col- lection notes (Simpathu 60 from Victoria Falls, Zimbabwe) also indicate that corms of L. erythran- tha are eaten raw. SYSTEMATIC TREATMENT мрн os Mem. Acad. Sci. Toulouse . Ker, Konig & Sims, Ann. Bot. 1: s Mem Baker, J. Linn. Soc. Bot. Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa 443 l: 154-156. 1878; Handbook Irideae 167- 174. 1892; Fl. Capensis 6: 88-97. 1986; FI. Tropical Africa 7: 350-355. 1898, excluding subg. Anomatheca. Solch, Prod. Fl. Südwest- afrika 155: 6-10. 1969. Goldblatt, Ann. Bo- lus Herb. 4: 14-74. 1972. Geerinck, Bull. Soc. Roy. Bot. Belgique 105: 333-351. 1972. For the generic synonymy see Goldblatt (1972). Plants perennial geophytes, deciduous in the dry season. Rootstock a bell-shaped corm with a flat base, the tunics hard-textured, woody to coria- ceous, entire and concentric initially, becoming irregularly fragmented with age, or composed of compressed fibers and then becoming cancellate to loosely fibrous with age. Cataphylls usually 2, membranous, the inner one reaching shortly above the ground, the outer about half as long, pale to partly brown or uniformly dark brown. Leaves 2- several, often only l inserted at the base near ground level and this largest, sheathing below, blade isobilateral for the most part but often channeled above the sheath for a short distance or up to half its length, other leaves sometimes basal or more often inserted aboveground and P eue shorter above, linear to lanceolate with a pla surface and a central vein evident, sometimes te- rete to oval in section (subg. Paniculata), or cor- rugate and the central vein no more prominent than the other veins (subg. Lapeirousia). Stem erect, somewhat compressed and 2-3-angled, often more conspicuously so above, the angles often weakly winged. /nflorescence paniclelike or a sim- ple to branched spike, the panicles often somewhat corymbose; individual flowers always with a pair of opposed bracts at the base of the ovary; bracts herbaceous to membranous, the outer abaxial and often the largest, sometimes ridged or keeled, the keel sometimes crisped or toothed (subg. Lapei- гоиѕіа); the inner bract adaxial, usually with 2 veins, and frequently bifurcate. Flowers actino- morphic or zygomorphic, the perianth petaloid, often brightly colored, forming a short to extended tube, the tepals subequal or unequal, then usually with the upper tepal largest and the lower 3 smallest and forming a lip and provided with nectar guides of contrasting coloration. Stamens symmetrically disposed or unilateral and erect to arcuate; fila- ments filiform, inserted below the mouth of the tube; anthers oblong to linear, longitudinally de- hiscent, sub-basifixed to nearly centrifixed. Ovary globose to ovoid, concealed by the bracts; style filiform, dividing into 3 above, the branches usually forked for up to half their length, occasionally entire or barely bifid. Capsules membranous to coriaceous, globose to 3-lobed; seeds + globose to weakly angled by pressure, the surface rough. Haploid chromosome numbers n — 10, 9, 8, 7, 6, 5, 4, 3, the karyotypes often bimodal. TYPE SPECIES: Lapeirousia compressa Pourret (= L. fa- bricii (de la Roche) Ker). Named by French naturalist Abbé Pierre André Pourret (1754-1818) in honor of his colleague and contemporary, Philippe Picot de Lapeirouse, botanist and mineralogist at the University of Tou- louse. The first and only species to be so named was thought by Pourret to have come from Isle de France (Mauritius). Had Pourret known that it was a Cape species he probably would have placed it in Gladiolus or Ixia, genera to which other species of Lapeirousia were at that time assigned. KEv TO LAPEIROUSIA IN oe AFRICA INCLUDING BOTSWANA, NAMIBIA, AND TRANSVAAL, SOUTH AFR * Species marked with an asterisk are predominantly southern African and restricted to southern Namibia, largely to the southwestern corner that receives winter rainfall. They are treated fully by Goldblatt (1972). la. Axis contracted above ground level, bracts herbaceous and hardly different from t 2a. Floral bracts 6-15 cm long; perianth tube 10-14 cm vs m long; perianth tube ca. 3.5 cm Я pais including the inflorescence extending above the ud ad relatively lax; floral bracts аана to + membranous and dry but unlike the foliage leaves. 2b. Floral bracts 4-5 cm -— = the inflorescence congested and plants tufted in appearance; floral 16. L. odoratissima icata ч Perianth tube (2-)2.5-15 cm long (if less than 2.5 cm then the tepals narrower than 2 mm). Perianth tube 2 15 cm long; Rape 6-7 mm wi vide 4b. Perianth tube cm long; tepals 1 14. L. schimperi Tepals at a 15(-30) mm dee flowers shades of white to cream, rarely purple, with or without markings on the lower tepals 6a. Tepals 1.3-3 mm at the widest; flowers without markings on the lower xr PNE . L. littoralis 6b. Tepals 4-5 mm at the widest; flowers with violet markings on lower чи L price 444 Annals of th Missouri Bois Garden 5b. Tepals 8-14 mm long; flowers shades of blue to violet, ci white to cream, or greenish, but always with contrasting markings on the lower thre 7a. Inflorescence a divaricately branched panicle with e main branches l(or 2)-flowered. 8a. -~J = lants of Mozambique, Transvaal, and Zimbabwe 9b. Perianth white or pale blue with bluish markings ‹ on the lower ‘tepals; plants of 11. L. a Perianth white to pale pink; corm perianth tube 25-34(-40) mm long, rarely shor 8b. Perianth blue to violet; corm tunics dark strips; perianth tube 15-18(-20) mm ......... Inflorescence a panicle or spike with 9a. Perianth greenish or blue-violet with red to purple MR on the bs tepals; 9. mibia m tunics straw- 6 and + fibrous and e . L. bainesii brown, шо ra of — оа ж L. sandersonii at least some main branches 3-8-flow L. masukuensis gracilis 3b. Perianth tube 1-20(-25) mm long (if 20-25 mm then the tepals at least 2 mm wide). Perianth tube shorter than 2 mm; flower actinomor lla. rphi Flowers l(or 2) on terminal branches of the оа tepals 11-15.5 mm long; согт tunics dark brown to blackish and decaying irregularly into vertical strips .......... llb. Flowers (2-)3-5 on terminal branches of the ee tepals 7-9.5 mm i cor cs light brown and forming a fibrous netwo L uasmoniána m coerulea 10b. Perianth Wm 3-20(-25) mm long; flower medianly cygomorphie with arcuate unilateral £ stamens. ves corrugate and without an evident main vein; floral bracts herbaceous, usually longer 6 mm Namibia 14Ь. 15b. Perianth tube 8-12(-15) mm long. 16 16b. a. Perianth tube with the throat much wider than the lower part . 13b. Perianth tube slender and + uniform throughout 12b. ndm either plane and with an evident midvein (at least when living) or rounded to terete and the midrib then not SEDE but not corrugate; floral bracts herbaceou b c and the outer bract much longer than the inner; South Africa and southern _ L. dolomitica* L. bark lyi* s to membranous, and the outer bract about as long or shorter than the inner; a. ает tube 15-20(-25) mm long ain inflorescence bandes Lor 2)-flowered; leaves 1-3 mm wide; branching . L. sandersonii hin + divaricate ... Main inflorescence branches 5-8-flowered; leaves 3-6 mm iis branc ight 9. L. masukuensis Inflorescence a simple or branched spike with rarely more than a lateral m high; branches; bracts 6-8(-10) mm об! plants = exceeding 20 L nae plants of northern Ethiopia and eastern Su Inflorescence a panicle or several- branched ane (rarely with fewer than 3 lateral branches); bracts 5-6 mm long; plants often taller 2 _ сш; plants of Nigeria or south tropical Africa, including Botswana and 17a. Inflorescence a lax panicle with the main terminal bene eae 8- owered; upper tepal suberect to curving forward over the stamens 3. L. rivularis 17b. Inflorescence a lax to dense panicle with the main terminal branches 1-5-flowered; upper tepal erect to patent, spreading outward, thus not m diam., ud to terete in section and without an evident midrib; perianth tu tube 3-5 mm long and tepals 5-14 mm long; main inflorescence hos with 1 or 2 flowers. 19a. Perian th tube exceeding the bracts by 2-3 mm and only slightly shorter than the tepals; tepals 5-6 mm long ........... . teretifolia 19b. O tube barely exceeding the bracts and less than half s long as the tepals; tepals 12-14 mm long ........................... Te L чш 18b. Leaves plane, narrow to broad, 0.5-8(-11) mm wide and alwa E a es midrib; perianth tube 7-14 mm long and visis PRA at least some inflorescence branches with more E A 8) fl 20a. Plants cay taller than 12 cm; bracts herbaceous : an thesis; branching usually contorted . 5. L. ҮТ 20b. Plants (15-)20-45 cm high; bracts generally membranous and dry above at anthesis; branching not contorted ............. 4. L. erythrantha Volume 77, Number 3 1990 Goldblatt 445 Systematics of Lapeirousia in Tropical Africa For the southern African countries of Botswana, Namibia, and South Africa, cited specimens are arranged according to the grid reference system based on geographical degree coordinates of lati- tude and longitude currently used in floristic treat- ments for the subcontinent (Edwards & Leistner, 1971). All type material cited was seen unless otherwise indicated. Descriptions are based on both fresh and the available dried material, but measurements are based on living plants whenever possible. A shrink- age factor of 20% or more may be expected for flower parts, and 10% for leaves, depending on the care and method of preservation. Flower color fades progressively in dry specimens, eventually changing completely, becoming darker or lighter and sometimes ultimately disappearing. Color notes on collected specimens are desirable and are fre- quently mentioned by collectors. SUBGENUS PANICULATA GOLDBL. & MANNING SECTION PANICULATA 1. Lapeirousia coerulea Schinz, Verh. Bot. Verein. Brandenburg 31: 212-213. 1890. Baker, Handbk. Irideae 168. 1892; Fl. Trop. Africa 7: 351. 1898. Sólch, Prod. Fl. Sud- westafrika 155: 8-9. 1969 (including L. avas- montana). TYPE: Namibia, (Upingtonia) Om- bale, súdost Ondonga, Mar. 1886, Schinz 13 (lectotype, К, here designated; isolectotype, COI); Hereroland, Luderitz 28 (syntypes, B Z); Otjitambi im Kaoko, Belck 50 (syntype, not seen). Figure 4. » Ixia dinteri Schinz, Mém. Herb. Boissier 20: 14. 1900. TYPE: Namibia: Karibib, Spitzkop p. marshy ground, Dinter 22 (32 on the type) in 1898 (holotype, Z). Plants (12-)15-30 cm high, branched repeat- edly. Corm 12-16 mm diam. at the base, light brown, the tunics of pale compacted fibers decaying to become coarsely to finely fibrous and теси ай. Cataphylls 2, pale to light brown. Leaves 2-4, only the lowermost inserted at the base, this leaf longest and reaching to the middle of the inflores- cence or beyond, + linear, 2-4(-5) mm wide, the upper leaves decreasing in size above. Stem com- pressed, 2—3-angled, often intricately branched. Inflorescence a lax panicle with ascending branch- es, the larger terminal branches bearing (2-)3-5 sessile flowers; bracts 3-4 mm long, herbaceous below, membranous apically and becoming com- pletely dry in fruit, the inner bracts about as long as the outer. Flower actinomorphic, + stellate, blue to light purple with a white hastate marking outlined in dark blue to violet (less often reddish) in the lower half of each tepal; perianth tube 1— 1.5 mm long, cylindric below, widening above, the cylindric part shorter than 1 mm; tepals spreading below, curving upward distally, subequal, lanceo- late, 7-9.5 mm long, 3-5 mm wide. Filaments inserted below the mouth of the tube, united basally for less than 0.5 mm by an obscure coronalike ring, erect, ca. 4 mm long, nearly contiguous around the style; anthers diverging, 3-3.5 mm long, curv- ing inward after anthesis; pollen light blue-purple (white to yellow when dry). Ovary ca. 2 mm long, style erect, 6-7 mm long, dividing near the anther apices, branches ca. 1.2 mm long, barely notched apically. Capsules obovoid-globose, weakly 3-lobed, ca. 4 mm long; seeds dark brown, globose to weakly angled by pressure, 1.3-2 mm diam., tapering near the attached funicle. Chromosome number 2n — 8. Flowering time. (January)February to April. Distribution and habitat. of damp, low-lying places, Lapeirousia coerulea is widespread throughout the summer-rainfall part of Namibia (Fig. 4) where it occurs in seeps, sea- sonal vleis, damp poorly drained grassland, and Typically a species shallow soil in rock outcrops where water accu- mulates in the wet season. There are a few records from northwestern Botswana, where it is apparently ч m e. Several collections are accompanied by notes indicating that Lapeirousia coerulea is eaten by natives, particularly the Kung Bushmen for whom the corms are a staple, eaten after roasting in hot ash (Story 6121), or sometimes raw (Rodin, 1985). Lapeirousia coerulea and the closely allied L. avasmontana Diagnosis and relationships. appear to be taxonomically isolated among the tropical African members of the genus in their short-tubed and completely actinomorphic flowers. However, their paniculate inflorescences indicate a relationship with the tropical African sect. Panic- ulatae. The actinomorphic, blue flowers with their vestigial perianth tube and unusually short, undi- vided style branches make the two species easy to recognize. The differences between L. coerulea, a species of wet low-lying places, and L. avasmon- tana, which grows in well-drained stony banks and hills (discussed in detail under L. avasmontana), involve vegetative and floral features, including corm tunics, branching pattern of the inflores- cence, and flower size and patterning. These mor- phological differences, combined with the disparate 446 Annals of the Missouri Botanical Garden E RE 4. Morphology and distribution of Lapeirousia coerulea (A) (and hatched area) and L. avasmontana : x 2 IGU (B) (and closed circles). Habits (Drawn by J. C. Manning.) and corms habitat preferences and karyotypes (discussed be- low), compel their separation even though Sölch (1969) united them A diploid number of 2n = 8 has been recorded even populations of Lapeirousia coerulea (Goldblatt, 1990b). The karyotype consists of two larger and two smaller chromosome pairs. This low and apparently derived number also occurs in the Ethiopian L. setifolia, both of which have karyotypes comparable to but some- abyssinica and in L. x 0.5; single flowers and fruit full size; details of stamens and style what different from that of L. coerulea. Lapei- rousia avasmontana has 2n = 16, but it is clearly not a direct polyploid derived from ancestors such as L. coerulea. Its karyotype comprises one long and seven short chromosome pairs, and it has about the same amount of chromosome material per cell as L. coerulea. Base number in L. coerulea is consistent with its placement in sect. Paniculatae. Despite its actinomorphic flowers, L. coerulea is probably most closely related to the L. erythrantha Volume 77, Number 3 1990 Goldblatt 447 Systematics of Lapeirousia in Tropical Africa group of species, all of which have longer-tubed, zygomorphic flowers. Chromosome numbers in the section are 2n = 12, 10, 8, and 6 (Goldblatt, 1990b), and the karyotypes in the group are weak- ly, if at all, bimodal. The simple flower structure of L. coerulea makes it appear to be close to the basal stock of the tropical African species. History. The widespread Lapeirousia coe- rulea was first collected in 1879 by the Swedish traveler and explorer Ture Een on his second major expedition to Namibia. This early record received no attention, and collections made after 1885 in- dependently by Hans Schinz, August Lúderitz, and Waldemar Belck formed the basis for the proto- logue published by Schinz in 1890. In 1900 Schinz also described Ixia dinteri, a species undoubtedly onspecific with L. coerulea, based on specimens collected by Kurt Dinter in 1898. Additional specimens examined. | BOTSWANA. NGA- MILAND: 19.21 (Aha Hills) Xangwe, Mar. 1961 (CB), Gibson 186 (MO, WIND); Dobe region, north of Aha Hills near Namibian border, dry pan, 25 Apr. 1980, Smith 3496 (MO, PRE, SRGH); Dobe, 19 Sep. 1964, Lee 13 (SRGH); hard clay in dry pan, Quangwa River catchment, 19°35” 21%”, 23 Арг. 1981, Smith 3674 (BR, PRE). GHANZI: 21.22 (Kobe) pan on farm 102, Kuki, 21 Feb. 1970 (AC), Brown & Brown 8723 (С, PRE, SRGH); 53 km NE of Ghanzi on the road to Maun, 22°05” 21°26”, seasonally flooded limestone outcrops, Dec 1977, Skarpe 213 (К, MO, PRE, UCBG). NAMIBIA. OVAMBOLAND: 17.15 (Ondangua) 3 km S of Oshikango, З Apr. 1973 (BD), Rodin 9203 (K, M, MO, PRE, WIND). ETOSHA: 18.15 (Okahakana) Etosha Pan, large vlei 5 mi. W of Okondeka, 27 Mar. 1963 (DD), Giess et al. 6056 n WIND); NW Ecke des Etosha National Park, trock- ne randzone von vlei, 18 Mar. 1974, Mermüller & Giess 30378 (М. PRE). GROOTFONTEIN: 19.17 (Tsumeb) 25 km from Tsumeb toward Ondangua, 14 Feb. 1983 (BA), Lavranos & Pehlemann 21101 (WIND); farm Toevlug, lime marl, 5 Feb. 1971, Giess 11292 (K, MO, P SRGH, WAG, WIND); farm Goab, spring meadows, 29 Jan. 1978 (CA), Giess 14955 (M, MO, PRE, WAG, WIND); 19.18 (Grootfontein) Gr о TA red loam flats (CA), Jan. 1935, Schoe me Е); 13 m N of age 4 Jan. 1 Sohoenyelder 434 (PRE). farm Kumkaus к їп у о 7 Маг. 1974, Мегх muller & Giess boo (M). OKAVANGO: 19.20 (Tsumkwe pans at Tsumkwe, 14 Jan. 1971 (DA), Giess al. 11076 , S, PRE, WIND) Tsumkwe, 14 Jan. 1958, Story 6121 (M, PRE, SRGH). outjo: 22.15 о. farm Goreis, red loam sand, 17 Feb. 1971 (BB), Giess 11248A (K, M, PRE, bbs OTJIWARONGO: 20. UU din Otjiwarongo (BC), s.d., Barnard 198 (SAM); 3 of Otjiwarongo, 1985 (CB ) diae MD. zm E i usen omuramba at Om 1 Mar. 1940 (CA), Volk 2845 (WIND). GOBABIS: T pL aun farm ге 10 Feb. 1982 (DC), Rauh 57762 (WIND). 17 (Windhoek) Neudam Experimental Я 1960 (AD), van Vuuren 1027 (К, М РКЕ, SRGH, WIND); Ongombo, № of Neudamm, 26 Dec. 1963, Giess 243 (M); farm Otjikundua 67 mi. WSW of Steinhausen, open vlei, 19 Feb. 1955 (BA), de Winter E — 9 У 2409 (К, М, МВС, jc WIND); farm Aris, S of Wind hoek, omuramba, 1 . 1953 (CA), Walter & Walter 1550 (B, BR, Ы OMARURU: 21.15 (Karibib) Ohere- Oos, granite flats, 14 Feb. 1958 (BA), Merxmú Cies 1583 (M, PRE, WIND); 30 km NW Omaru AG, WIND); 22.15 (Trekkopje) Okongawa, Cranitbankberg, 4 Feb. 1934 (BB), Dinter 6957 (В, BOL, , М, PRE, S, WIND, Z). OKAHANDJA: 21.16 (Okahandja к Omatako View. 15 Feb. 1974 (BA), Woortm M, PRE, WIND); farm Omongongua, shallow pene in omuramba, 27 Mar. 1960 (DB), to Karibib, 21 Deo em (D MO, PRE, WIND). г 23. 16 (Nauchas) farm Gollschau, 27 Jan. 1972 (BC), Giess & Húbsch 11600 KEETMANSHOOP: 26.18 (Keetmanshoop) Spitzkoppe, 7 Apr. end bear 5 1533 (WIND). WITHOUT PRECISE BIA: Damaraland, 1879, Een s.n. (BM); шуы, 26 Feb. hari, 1886, Ross s.n. ex gelegen, 22 Feb. 1961, Seydel 2609 (BR, COI, M, MO, WAG). 2. Lapeirousia avasmontana Dinter, Feddes Rep. 29: 256. 1931. TYPE: Namibia: Lich- tenstein, Апаз Mountains, 20 Feb. 1923, Din- ter 4454 (holotype, B; isotypes, GH, K (photo), 5, Z (3)). Figure 4 Plants 20-30 cm high, branched repeatedly. Corm 15-20 mm at the widest diam., tunics dark brown to blackish, the inner layers + woody, the outer decaying irregularly, often into vertical strands. Cataphylls 2, membranous, usually brownish, the inner one reaching shortly above the ground. Leaves 3-4, linear, 2-3 mm wide, only the midrib prominent, the lowermost leaves inserted near ground level and longest, ii slightly lon- ger than the inflorescence, ot eaves inserte above the ground and nU smaller above, those subtending branches becoming bractlike. Stem + divaricately branched above, rounded to lightly triangular, sometimes obscurely winged or ridged at one or more of the angles. /nflorescence a round- ed corymbose panicle, the ultimate branches with l(or 2) flowers; bracts 5-6 mm long, herbaceous below, membranous in the upper half and bent outward by the tepals, becoming completely dry in fruit, the inner bract about as long or longer than the outer. Flower actinomorphic, blue to light pur- ple with a white heart-shaped mark feathered pur- ple on the edges in the lower midline of each tepal; perianth tube ca. | mm long, cylindric in the lower 448 Annals of the Missouri Botanical Garden half, widening above; tepals spreading outward and curving upward apically, subequal, lanceolate to elliptic, 11-15.5 mm long, (4-)5- 7 mm wide. Fil- aments inserted at top of the slender part of the tube, erect, 4-5 mm long, in the lower half nearly contiguous around the style, diverging above; an- thers diverging, 4-5 mm long, curving inward after anthesis; pollen light blue-purple to whitish. Ovary ca. 2 mm long; style erect, 6-7 mm long, dividing near the anther apices, branches ca. 1.2 mm long, barely notched apically. Capsules + globose, weakly 3-lobed, showing the outline of the seeds, 6-8 mm long, ca. 7 mm diam.; seeds globose, dark brown, 1.5-2 mm diam., rounded to weakly angled by pressure, tapering toward the attached funicle. Chromosome number 2n = 16 February to April. Flowering time. Distribution and habitat. Lapeirousia avas- montana is endemic to central interior Namibia, where it is locally common on the hills and moun- tains around Windhoek and in the north toward Okahandja (Fig. 4). Apparently preferring well- drained sites, it grows on open, stony, sloping ground. Depending on the rainfall, flowering may last between three and eight weeks. New branches continue to be produced from the cauline leaf axils as long as the ground remains moist, and a late rainfall can stimulate production of a second flush of flowering from new branches on the same flow- ering stem. The bright blue flowers are visited by a variety of insects including bees, wasps, and butterflies. No measurable nectar is produced, and presumably the only reward to insect visitors is pollen. The very short perianth tube and blue, stellate flower suggest that the species, like the morphologically similar L. coerulea, depends on short-tongued pollen-foraging bees for pollen trans- Diagnosis and relationships. The large, pale blue (blue-lilac) flowers with a white center, com- pletely actinomorphic perianth and stamens, and large umbrellalike, paniculate inflorescence distin- guish Lapeirousia avasmontana from other trop- ical and subtropical African species of the genus. It is most easily confused with the more widespread Namibian species L. coerulea, which has similar but smaller actinomorphic flowers. Lapeirousia avasmontana has dark brown, more or less woody corm tunics that break into vertical strips as the decay and a divaricately branched inflorescence, the terminal branches of which bear a single (rarely two) flowers. Both features contrast strongly with L. coerulea, which has light brown tunics that become fibrous and reticulate as they decay, and a less strongly branched inflorescence of ascending branches, some of which terminate in spikes of 3- 5 flowers. The flowers of the two species also differ, those of L. avasmontana having somewhat larger, pale blue tepals 11-15.5 mm long, with heart- shaped white markings, compared with those of L. coerulea having tepals 8-9 mm long, with hastate white markings. These differences were noted in the protologue by Dinter (1931), who was aware of the possible confusion between L. avasmontana and L. Chromosome cytology provides additional infor- coerulea. mation about the status of Lapeirousia avasmon- tana. Its diploid chromosome number is 2n = = and the strongly bimodal karyotype consists long and 14 short chromosomes (Goldblatt, e Lapeirousia coerulea has, in contrast, 2n = 8 and a karyotype of four long and four shorter chro- mosomes. Despite the apparent numerical polyploid relationship, their karyotypes differ so much that it is clear that a more complex situation is involved. Karyology supports the separation of the two species and suggests that they are not particularly closely related. History. Lapeirousia avasmontana was dis- covered in 1923 by Kurt Dinter who described it in 1931. In the protologue Lapeirousia avasmon- tana was compared closely with the similar L. coerulea which, as Dinter pointed out, grows in a different habitat and differs in the several vege- tative and floral morphological features mentioned above. Although L. avasmontana was included in L. coerulea in the Prodomus Flora Südwestafrika (Sölch, 1969), I have no hesitation in recognizing it as a separate species. Additional specimens examined. NAMIBIA. WIND- HOEK: 22.16 (Otjimbingwe) farm Onduno, Hochflaeche, 27 Feb. 1966 (BD), Meyer 115 en WIND); farm Terra 73 (CD), Giess 13498 Mar. 1982 (DB), Milter & Kolberg "20: 39(PRE, WIND); 16 Apr. 1939 (Н € fr), Gassner 134 (М); 22.17 (Wind- hoek) Elisenheim, Erosberge, below Wachter, smooth shale slopes, 28 Feb. 1974 (AC), Merxmúller & Giess 30017 (K, M, S, PRE, SRGH, WAG, WIND); near Brakwater on the Windhoek-Okahandja road, 14 Mar. 1959, de Winter & Giess 7136 (M, PRE, WIND); Auasberge near Windhoek, Moltkeblick, 24 Mar. 1969 (CA), Meyer sub Giess 10725 (WIND); 5 km W of Windhoek on Daan Viljoen road, stony quartz hill slopes, 14 Mar. 1988, Goldblatt & Manning 8798 (E, K, M, MO, NBC, PRE, sesta . Lapeirousia rivularis Wanntorp, Svensk. Bot. Tidskr. 65: 53-56. 1971. Roessler & Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa 449 Merxmiller, Mitt. Bot. Staatssamml. Mun- chen 15: 394-395. 1979. TYPE: Namibia: Karibib, farm Ameib, ca. 25 km NE of Usa- kos, granite kopje S of track, ca. 5 km E of farmhouse, 15 Apr. 68, Wanntorp « Wanntorp 907 (holotype, S; isotype, M). Fig- ure Plants 30-45 cm high, paniculately branched. Corm 12-22 mm wide at the base, tunics dark brown to blackish, coarsely fibrous or cancellate. Cataphylls 2, membranous, pale to dark brown, the inner one reaching shortly above the ground. Leaves 3-5, the lower 2 + basal, bifacial and channeled for at least half their length, reaching to about the base of the inflorescence, linear and unifacial above, 2-3 mm wide, firm-textured, mid- rib lightly raised, the upper leaves subtending the branches of the inflorescence, bifacial for all or most of their length, progressively shorter above. Stem weakly compressed and 3-4-angled, usually laxly branched. Inflorescence a rounded to elon- gate panicle, the ultimate branches bearing (1-)3- 5 flowers, laxly arranged; bracts 4-6(-9) mm long, green below, membranous and reddish in the upper half, apically dry and brown-tipped in bud, later becoming dry for 2 of their length, inner bract about as long as the outer and entire or apically forked. Flower zygomorphic, pale blue-mauve with white to cream nectar guides on the lower 3 tepals outlined in deep violet, sometimes with a median red streak, the reverse of the tube bluish; perianth tube ascending, narrowly funnel-shaped, ca. 1.3 mm wide at the base, 3 mm at the mouth, 7-9 mm long, weakly curved in the upper third; tepals lanceolate, subequal in size, (7-)10-13 mm long, 4-5 mm wide, the margins somewhat undulate, the upper tepal suberect to arching over the sta- mens, others directed forward, the lower 3 nearly horizontal, joined for ca. 1 mm longer than the upper, sometimes each with an obscure to con- spicuous bump or callus in the median lower half. Filaments unilateral and arcuate, 8.5 mm long, inserted 3-4 mm below the mouth of the tube; anthers parallel and contiguous, 3.5-4 mm long, blue-mauve; pollen pale blue-mauve, fading to yel- low. Ovary globose, ca. 2 mm long, style arched behind the stamens, dividing near the anther api- ces, 18-20 mm long, branches ca. 3 mm long, each shortly to deeply divided (occasionally entire) for up to half their length, ultimately recurving. Capsule 4-5 mm long; seeds + globose, 16-21 mm diam., brown. Chromosome number 2n = 12. Flowering time. (Mid January) February to early April. FIGURE 5. Morphology of Lapeirousia rivularis. Habit and corm x 0.5; flower full size. (Drawn by M. L. Branch.) Distribution and habitat. laris appears to be largely a species of locally wet sites in semiarid southwestern tropical Africa (Fig. Lapeirousia rivu- 6). The type locality is in western Namibia at the southern end of the Erongo Mountains, and it ex- tends from here to northern Namibia and southern Angola. There are also several records from south- ern and central Zambia where plants are more robust, perhaps a reflection of the wetter climate. The usual habitat is along temporary streams, seeps, or rock flushes, and in Zambia I have seen it in rocky grassland at poorly drained but not notably wet sites. The species probably also occurs in Zim- 450 Annals of the Missouri Botanical Garden io: FIGURE 6. Distribution of Lapeirousia rivularis. babwe, but I have seen no specimens that | can confidently refer here although some specimens (e.g., Eyles 1952; Drewe 68; Mitchell 1319) may belong to this species. The zygomor- phic, pale blue flower with a perianth tube 7-9 mm long, an upper tepal erect to somewhat hooded over the arcuate stamens and style, and the upper Diagnosis and relationships. lateral tepals directed forward are the principal features that distinguish Lapeirousia rivularis from other tropical African species of Lapeirousia. It is probably closely related to the L. erythrantha complex and is most often confused with the blue- flowered form of L. erythrantha, from which it is difficult to distinguish when dry. The flowers of L. erythrantha and L. rivularis are similarly colored and proportioned, with a perianth tube 7-9 mm long (- 15 mm in L. erythrantha) and tepals about as long as the tube. When seen alive, the difference in flower form is striking. Lapeirousia erythrantha and its allies have the upper tepal erect to recurved and the upper laterals reflexed. Thus when the flowers are fully open the upper tepals lie in more or less the same horizontal plane as the liplike lower three. Sometimes close examination of herbarium material makes it possible to see this difference, but often the flowers are too distorted. In general L. rivularis has a more lax, open panicle typically with 3—5 flowers on the major terminal branches, whereas in L. erythrantha the most common type of inflorescence is comparatively dense with flowers crowded terminally on the branches. t is uncertain whether Lapeirousia wel- witschii, described by Baker in 1878 and based on specimens collected by Welwitsch in central Angola, is conspecific with L. rivularis. The gen- eral aspect of the type specimens is similar to dwarfed forms of L. rivularis, especially its type, but the flowers are so shrunken and distorted that I hesitate to decide whether L. welwitshii is con- specific with L. rivularis or is a stunted form of L. erythrantha. The name is excluded until plants can be re-collected at the type locality, Pungo Andongo. The collection from southern Angola, Baum 958, is somewhat unusual. Most of the many plants of thus the inflorescence is а simple spike, others have one the collection have unbranched stems, branch, and only a few have more than two branch- es (one specimen at Z has five branches). That the reduced branching is the result of abnormal grow- ing conditions is as likely as the alternative that the population represents a genetic variant. History. Lapeirousia rivularis was described by the Swedish botanist Hans-Erik Wanntorp in 1971, although he was not the earliest to record the species. It appears to have been first collected in Namibia by the South African zoologist K. H. Barnard, who found it on the northern border of Namibia along the Cunene River in 1921. The type locality of L. the farm Ameib at the southern end of the Erongo Mountains, is the rivularis, most southern station for L. rivularis, and plants from here can be recognized by the toothlike callus present in the midline of the lower three tepals. This tooth has little taxonomic significance. It ap- pears occasionally in several species of Lapeirou- sia, and I have noted it in a few individuals in populations of the western Cape L. anceps and L. divaricata, and in some Namibian populations of gracilis. А tepal tooth is a common feature in Tritonia (Ixioideae—Ixieae), in L. bainesii and L. which it defines some sections of the genus (de Vos, 1982), but even in this genus the expression of the character may sometimes vary; a callus or tooth may develop on only one tepal or on none. Їп Lapeirousia the presence of a tooth alone cannot be treated as evidence for recognition of species or even infraspecific taxa. Additional specimens examined. ANGOLA. CUNE between Kiseve and Humbe, 1,100 m, (16°40” 14957” ) 1 Apr. 1900, Baa 958 (BM, BR, COI, E, G, K, M, S, Z). NAMIBIA. OVAMBOLAND: 17. 15 (Ondangua) Engela Mis- sion, shallow vlei in black sticky clay, 16 Feb. 1959 (BD), de Winter & Giess 7061(B, K, M, PRE, SRGH, WIND). OTJIWARONGO: 20.16 (Otjiwarongo) Otjiwarongo, not on summit but high up, 16 Mar. 1980 (BC), Craven 1129 (WIND). KaRiBIB: 21.14 (Uis) Brandberg, Г com- sr on top of Sonuseb saddle at water hole, 5 (BA), Craven 2277 (WIND); 21.15 а fm naeh below Jatow cave, somewhat marshy granite soil, 19 Mar. 1963 (DC), Giess 13132 (B, M, PRE, WIND); 22 Mar. 1965, Giess 8452 (K, M, MO, WAG, WIND); Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa 17 Mar. 1963, Giess et al. 5795 (B, M, PRE, WIND); D); 5 Apr. 1974, Merxmúller & Giess 30699 (B, M, K, PRE, SRGH, WAG, WIND); Ameib, Philips Caves, edge of stream, 19 Mar. 1963, Hardy & de Winter 1420 (В, К, М, PRE, WAG). ZAMBIA. CENTRAL: Lusaka District, campus of the Univ. of Zambia, 19 Jan. 1973, Strid 2884 (С, К, MO); Lusaka, off Church Road in grounds of Evelyn Hone College, 11 Jan. 1986, Gold- blatt 7537 (E, K, MO, NBG, PRE, S, WAG, WIND); Lusaka, Concord Mr 3 mi. S of Chisamba Station, d, n. 1973, Benson 214 (К); Mamb- wa, 27. Dec. 1665, van Rensburg 2644 (K, SRGH); Kabwe District, ETC 10 km NW of Kabwe, dambo 1973, Kornás 3023 (K). SOUTHERN: Kalene Hill, Mwinilunga, seasonally damp ground on sand- stone outcrops, 15 Dec. 1963, Robinson 605 1 (M, WAG). ES : Machili, moist 1 : 1921, Barnard 201 (SAM); u and the Kunene banks, Barnard 142 (SAM); Mafa, Barnard 197 (SAM 4. Lapeirousia erythrantha (Klotzsch ex en Baker, J. Linn. Soc. Bot. 16: 155. 1878; Handbk. ies 168. 1892; Fl. Trop. Africa 7: 351-352. 1898. Geerinck et al., Bull. Soc. Roy. Bot. Belgique 105: 335-344. 1972 (but including L. teretifolia and L. welwitschii). Ovieda erythrantha Klotzsch ex Klatt in Pe- ters, Reise Nach Mossambique, Volume 6 Bot. 2: 516, t. 58. 1864. TYPE: Mozambique: Bo- ror, к Peters s.n. hoc В). Fig- ure Lapeirousia sandersonii sensu Baker in Fl. Trop. Africa 7: 352. 1898. Lapeirou sia erythra е ntha var. briartú (de Wild. & Dur- Mussimu, Haut Lualaba, Briart s.n. (holotype, BR). Lapeirousia erythrantha var. ти (№. Е. Вг.) arais ех Сее "e he 2 ull. Roy. Bot. Belgique 105. 336. >. Lapeirousia rhodesiana N. E 7 Бигу апа Umtali, Cecil 154 (holotype, К). Lapeirousia graminea Vaupel, Bot. Jahrb. Syst. 48: 533. 2. TYPE: Mozambique: 25 Mile Station (Dondo), in forest, Schlechter 12238 (holotype, B). Lapeirousia spicigera Vaupel, Bot. Jahrb. Syst. 48: 547- 548. 1912. TYPE: Angola, Huilla: Antunes 256 (ho- lotype, B. Lapeirousia plagiostoma Vaupel, Bot. Jahrb. Syst. 48: M . o A 7 y Na 4 / FIGURE 7. Habit х 0.5; flower full size. (Drawn Morphology of Lapeirousia erythrantha. by J. C. Manning.) 547. 1912. TYPE: Mozambique, Station Howesa, rocky slopes, Tiesler 46 (holotype, B). Lapeirousia montana Hutchinson, Kew Bull. 1921: 403. 21, nom. illeg., non L. montana Klatt (from S. Africa, Goldblate 1972: 52). Hutchinson & Dalziel, rop. Africa, Ist edition, 2: 376. 1935. Foster, Contrib. Gray 1936, nom. nov montana Hutchinson. TYPE: Nigeria: top af Zaranda Moun- tain, 5,800 ft., Lely 189 (lectotype, K, here des- ignated); Меш» Valley, 2 mi. E of Government Sta- tion, Naraguta, Lely 271 (syntype, K? not seen). Plants (15-)20-45 cm high, paniculately branched. Corm 8-16 mm wide at the base, tunics blackish, of densely compacted fibers, the outer layers coarsely fibrous, sometimes becoming + re- ticulate and fine, then forming a matted layer. — 452 Annals of the Missouri Botanical Garden n | | t 1 „|, | pm ELE NEN. ES RUNE * | + | j 1 FIGURE 8. Distribution of Lapeirousia erythrantha. The question mark indicates the doubtful locality of the type collection of L. spicigera from southwestern Angola. Triangles represent red-flowered populations and dots those with blue flowers. Cataphylls 2, the inner one reaching shortly above ground level, usually dark brown. Leaves 3-4, the lower 2-3 longest and %- as long as the stem, the upper decreasing in size and becoming pro- gressively bractlike above; the basal leaves 2-3, linear to lanceolate, sometimes somewhat falcate, (2-)4-8(-1 1) mm wide; only the midrib prominent when live. Stem compressed and 2-angled below, branched repeatedly, 3-angled above. /nflores- cence a several- to many-branched panicle, often assuming a corymbose appearance, the ultimate branches bearing (2-)3-6(-8) flowers, these often crowded terminally; bracts 3-6 mm long, green below, dry and membranous above in bud, becom- ing entirely dry and transparent, brownish in the upper third, the outer bract obtuse, usually slightly longer than the inner one, the inner acute or api- cally bifurcate. Flower zygomorphic, either blue- violet having the lower tepals each with a hastate white mark outlined dark blue to purple, or the flower crimson red and then usually without mark- ings but occasionally the lower tepals each with a white median streak (rarely white, either uniformly or with dark markings); perianth tube (6-)7-11 (714) mm long, slender below, expanded somewhat and slightly curved near the throat; tepals subequal in size, lanceolate-spathulate, acute to obtuse, (6—)7-11 mm long, 2-3(-4) mm wide, lying nearly in one plane, the uppermost tepal held apart from the others and reclined, the lower 3 narrower at base thus + clawed, set closely together and joined for ca. 1 mm longer than the upper. Filaments unilateral and arcuate, 4-6 mm long, inserted ca. 1.5 mm below the mouth of the tube; anthers parallel and contiguous, (2.5-)3-4 mm long, pale bluish to white, purple or red; pollen whitish. Ovary globose, 1-2 mm long, style arched behind the stamens, dividing at mid to upper anther level, ca. 15 mm long, branches 1-1.5 mm long, recurving and forked for 4—% their length. Capsule 3-lobed, 3-4 mm long, outline of the seeds distorting the walls; seeds globose, 2-5 per locule, nearly globose, 1.6-1.9 mm diam., dark brown, weakly to strongly reticulate. Chromosome number 2n = 14 (? = 12 + 2B), 12+0-3B. Flowering time. (Mid December) late Jan- uary to March, depending on when the summer rains begin. Distribution and habitat. Аз circumscribed here, Lapeirousia erythrantha has a wide distri- bution across south tropical Africa (Fig. 8). It ex- tends from eastern Angola and southern Zaire through Zambia and northeastern Botswana to Zimbabwe, Malawi, western and southern Tanza- nia, and the coast of central Mozambique. It is a variable species and has differentiated into a num- ber of local races or forms, which are dealt with below. The most common species of Lapeirousia in central Africa, L. erythrantha, can frequently be found in rocky outcrops throughout its range, or less often in open grassland or woodland. Its frequency in thin soils in rocky sites suggests that the corms are protected in such habitats, while in deeper ground they are accessible to rodents, por- cupines, and perhaps other animals. There seems no a priori reason for the apparent preference for rocky habitats, and plants grow well in cultivation in almost any type and depth of soil. The small flowers, conspicuous in the dense inflorescences, are visited and presumably pollinated by large num- ers of various bees and wasps that forage for nectar and pollen in warm weather. Annotations on two specimens indicate that La- peirousia erythrantha is edible (Simpathu 90, Victoria Falls; Buchanan 426, Shire Highlands, eaten in time of great famine). However, there appears to be no record of the species being sig- nificant in the diet of any human population. One of the sev- eral species of Lapeirousia with a strongly rami- Diagnosis and relationships. fied, paniculate inflorescence, L. erythrantha, can be distinguished by its relatively small flowers, typ- ically (but not always) soft-textured leaves with a conspicuous midrib, and perianth tube seldom ex- ceeding 12 mm. The flowers are somewhat crowded at the apices of the branches of the inflorescence, and the terminal branches are spikes of 3-8 sessile Volume 77, Number 3 1990 Goldblatt 453 Systematics of Lapeirousia in Tropical Africa flowers. The orientation of the tepals is also dis- tinctive; the tepals lie in more or less the same plane (Fig. 7), with the three lower tepals connate for about 1 mm and held closely together while the upper tepal is recurved and held apart from the others. Similar flowers also characterize L. san- dersonii and L. masukuensis, both of which have a perianth tube 15-25 mm long. Lapeirousia san- ics and an inflo- rescence in which the ultimate branches have only dersonii has different corm tun one or two flowers. The Ethiopian Lapeirousia abyssinica is closely allied to L. erythrantha but has a less ramified inflorescence, and the plants are seldom taller than 15 cm high. The flowers are virtually indistinguish- able from those of L. erythrantha. The difference in chromosome number, 2n = 12 in L. erythrantha vs. 2n = 8 in L. abyssinica, was a significant reason for my decision to continue to recognize the two species as distinct Also closely related is Lapeirousia setifolia, a dwarf species of rocky, high-altitude sites in Ma- lawi, Tanzania, and Zimbabwe. Again the flowers of the two species are nearly identical, but the short stature of L. setifolia combined with its rather intricately branched inflorescence and chromosome number of 2n = 8 suggest its separate status Lapeirousia rivularis is almost indistinguishable from L. erythrantha when pressed, but they have different flowers. In L. rivularis the upper tepal is arched forward over the stamens and style branch- es. This important difference between the two species is generally obscured in herbarium speci- mens. Lapeirousia rivularis grows in seasonally wet sites from central Zambia across southern An- gola to northern Namibia, and it thus overlaps in distribution with L. erythrantha. I have raised to species rank Lapeirousia tereti- folia, a plant of western Zambia and western Sha- ba, treated by Geerinck et al. (1972) as a variety of L. erythrantha. No doubt closely related to L. erythrantha, it has particularly small flowers with a pale bluish to white perianth and a highly ramified inflorescence with only one, or rarely two, flowers on the terminal branches. The leaves are terete, a condition that does not occur in any of the several forms of L. erythrantha. Variation. Comprising a number of more or less distinct races or forms, Lapeirousia erythran- tha presents a difficult problem in circumscribing the species and its close allies. I treat the species fairly broadly and do not attempt to establish an infraspecific taxonomy, such as that of Geerinck et al. (1972) who recognized four varieties in Zaire. The difficulty is the existence of numerous inter- mediates and parallel forms in different parts of the range. However, it is useful to discuss the main variants here, each of which has been described at some time as a distinct species. The most distinct of the variants is the typical form, which has small to moderate-sized flowers with a deep red perianth (occasional white-flowered plants are albino sports). It occurs in relatively dry es at low to nn elevations coasta Beira inland to southern Malawi and southeastern Zimbabwe. Plants grow in open ground or rocky sites in Mopane or dry Brachystegia woodland. The unusual intense red perianth (dark maroon on drying) suggests that this form merits taxonomic recognition, but after examining the ample нот now at hand, and living populations of red- blue-flowered forms, I have concluded that чам are simply color forms. Populations of plants with flowers more or less the same size but with purple or blue perianths occur in southern and central Malawi and are iden- tical in all respects but color with plants from populations with either uniformly red or blue flow- ers. In the Masvingo area of Zimbabwe there are red-flowered populations with large flowers that correspond in size to those of blue-flowered plants that occur in central Zimbabwe and are often re- ferred to L. In the main part of its range across Zambia and southern Zaire Lapeirousia erythrantha is rela- tively uniform, although flower size seems to de- crease moving from east to west and south to north. The small-flowered plants correspond to L. briartii, a species described from Zaire. Plants more or less matching this form but with short stature, unusually mm wide), rhodesiana. narrow leaves (generally about 1.5- and small flowers occur in northwestern Zambia and adjacent western Tanzania. This combination of characters suggests affinity with L. teretifolia. However, the 2-4-flowered terminal inflorescence branches of the northeastern Zambian plants and their flat leaf blades are unlike the predominantly 1 -flowered inflorescence branches and terete leaves of L. teretifolia. 1 believe that the general similarity here is superficial and does not indicate close re- lationship. The northeastern Zambian plants pre- sumably represent a regional form of L. erythran- tha. The name L. erythrantha var. welwitschii sensu Geerinck et al. (1972) applies to narrow- leafed specimens of L. erythrantha otherwise cor- responding to the type of L. briartii. The type of L. welwitschii is from central Angola, and I have not been able to identify it satisfactorily owing to its poor preservation (see excluded species). 454 Annals of the Missouri Botanical Garden A particularly unusual form of Lapeirousia erythrantha occurs locally in western Shaba at Etoile (Kalukuluku) Mine (Symoens 10141; Ndjele 1087; Lisowski 115, 117). It has rigid leaves, which appear in the dried material to lack a distinct midrib, and has rather open ramified panicles. The flowers appear to be similar to those in other pop- ulations of L. erythrantha from Zaire. This may be a race adapted to highly mineralized soils. In eastern Zimbabwe, populations appear to con- sist largely of plants with a very open inflorescence and a reduced number of flowers (1—3) on each inflorescence branch. These plants correspond with L. graminea, the type of which is from Tete Prov- ince, Mozambique. In western Zimbabwe and adjacent parts of Bo- tswana, plants have rather large flowers with peri- anth tubes usually 10-14 mm long. These popu- lations are the least well understood of the variants of Lapeirousia erythrantha and are sometimes referred to L. rhodesiana (or L. erythrantha var. rhodesiana). The type of L. rhodesiana is from Headlands in eastern Zimbabwe and does not cor- respond exactly to the western Zimbabwe and Bo- tswana plants, which may on further study be found to be a separate species. A few populations in central and northern Ma- lawi stand out in having flowers with a particularly long perianth tube and have sometimes been re- ferred to L. and the number of flowers on the inflorescence branches correspond exactly to L. erythrantha, as does the chromosome number, 2n = 12, and karyo- type in one population sampled. Moreover, indi- viduals within these populations vary for tube length, which ranges from 12 to 20 mm (e.g., Goldblatt 7535) sandersonit. However, their corms History. Lapeirousia erythrantha was first collected by the German physician and explorer Wilhelm Peters in the years 1842-1848 when he traveled in Mozambique under the a e of King Friedrich Wilhelm IV. Later the Scottish explorer Sir John Kirk, who кауыз David Livingstone on his Zambezi expedition in 1858, also recorded the species in the lower Shire valley in Mozambique. Kirk collected it again in the fol- lowing years on subsequent expeditions to the Af- rican interior. The species was described by F. W. Klatt (1864), who saw only the Peters collection, and referred to Ovieda, a synonym of Lapeirousia used from 1815 to 1876. In the Flora of Tropical Africa, Baker (1898) recognized L. erythrantha for red-flowered plants, and he referred blue-flowered plants to the Trans- vaal species, L. sandersonii. A collection from Zaire of the common blue-flowered form of L. erythrantha was described as L. briartii by de Wildeman in 1900. Similar plants from Zimbabwe were assigned to L. rhodesiana by N. E. Brown (in 1906), while those from Mozambique were treated by Vaupel (1912) as L. graminea and L. plagiostoma. A specimen collected in southwestern Angola was referred to L. spicigera by Vaupel. Lapeirousia spicigera is unusual in having up to 10 flowers crowded on the terminal branches of the few-branched inflorescence. No other speci- mens of L. erythrantha have been recorded from this part of Angola, and there is consequently some doubt about the provenance of the collection. To this growing number of named blue-flowered forms of Lapeirousia erythrantha John Hutch- inson in 1921 added L. montana, based on spec- imens from Nigeria. A homonym for a southern African species (Goldblatt, 1972), this name was replaced by L. nigeriensis by R. C. Foster (1936). The Nigerian populations were referred to L. rho- desiana by Hepper (1968). They correspond most closely to western Zimbabwe plants of L. erythran- tha in flower size, but the plants are comparatively short in stature. The treatment of some variants of Lapeirousia erythrantha as formal taxonomic varieties by Geerinck et al. (1972) was an attempt to deal with the variation in the species in Zaire. W. Marais (pers. comm.) proposed treating L. sandersonii (including L. masukuensis) and the southern Af- rican and Angolan L. bainesii (including L. ota- viensis) as varieties of L. erythrantha, but all four of these are in my opinion separate species, as is L. teretifolia, which Geerinck et al. treated as a variety of L. erythrantha. dí JA nd blue- Additional specimen flowered forms, CORDE ing to the types of L. ery- mt and L. briartii, respectively, are listed sepa- rate Red lowered Forms. MALAWI. SOUTHERN REGION: Blan- tyre, savanna woodland, Michiru, 4,000 ft., 26 Dec. 1966, a ge 9 (K, MAL, SRGH); Blantyre, Chi- chiri сарп = > rm bod es) p 3d et xxi Hd O iz Ф e = e + wm patamanga 969, Eccles 223 (к, MAL, SRGH); Li- rangwe, са. о. mi. N of Blantyre, open ground іп wood- land, 16 Jan. 1967, Hilliard & Burtt 4500 (E, К, MAL); ); Manganja Hills, Dec. 1861, Kirk s.n. t к гу, open woodland, 27 Dec с ар man 8, О). MOZAMBIQUE. MOZAMBIQUE: Nampula, 7 Jan. 1937, Torre 1278 (COD); Nada monte Nas- Volume 77, Number 3 1990 Goldblatt 455 Systematics of Lapeirousia in Tropical Africa sapo, andados 23 km de Nampula para Meconandados, 13 Jan. 1964, Torre en Paiva 9911 к: ТЕТЕ: Tete, Chioco, at km 4 e road to Mocubur Feb. 1968, Torre & aa Н 7670 (LISC). ZAMBEZIA: Serra Tumbine, E of Mlanje town hall, 17 Jan. 1971, Hilliard & Burtt 6295 (E, LMU); 97 km NE of Mopeia Velha on the road to Quelimane, woodland gh a vlei, Pope & Muller 547 (LISC, LMA, MO, ya amagoa Estate, Mocuba District, Namagoa, 200 ft., Dec.-Jan. 1943, Faulkner 156 (BR, COI, K, P, PRE, S, SRGH); een the mouth and Morrumbala, La: Do ed GH); Manica, Мог- m 814 к 25 ке Miller 4893 (К); Halt on Triangle road, wet vlei at foot of granite hill, 15 Jan. 1963, Leach 11592 (К, LISC, MO, PRE); C District, kopje near Madzivire Drive, grassland, 30 1962, Moll 484 (K, SRGH); Fort Velas Duvali Ranch, 21 Jan. 1948, Fisher 1387 (PRE), 1390 (SRGH); 5 mi. from Fort Victoria on the Sabie road, wet pan, Jan. 1969, Goldsmith 10/69 (K, ae ce PRE, d H). Blue-flowered Forms. ANG oxico: W of River Kaperu, grassland, 10 Jan. 1938, Milne- Redhead 40 030 (BR, K, LISC, PRE). HUILA: исса 256 (В) [can this locality Бе correct?]. BOTSWANA i . 1 243 (MO, SRGH); border near Plumtree, sandy flat side, 7 Mar. 1961, Richards 14552 (K, мата CENTRAL: Bakalaka area, between Francistown- Maun road and Marapong, 26 Jan. 1967, McClintock K73 (К); Between Francistown and Nata River, on Maun road, 21 K, LISC, PRE), 22.27 (ВВ). Selebi, Jan. 1978, Kerfoot 8007 (PRE). MALAWI. CENTRAL REGION: Lakeview, Dedza Plateau, roadside, 5,500 ft., 22 Dec. 1976, Pawek 12010(K, MAL); 15 km N of Ncheu, around granite dome, 9 Sep. 1986, Goldblatt 7534 (MO); 2 mi. E of the State House, Lilongwe, rock outcrop, 11 Jan. 1984, May 1 (MAL); Dedza, Brachystegia woodland on Kanjoli Hill, 13 Feb. 1967, Salubeni 563 (K, LISC, PRE, SRGH); Lilongwe, Dzalanyama Forest Reserve, val- ley NW of Kazuzu hill, 24 Feb. 1982, Brummitt 16078 (K); road to Dedza Forestry School, around quarry Jan. 1986, Goldblatt 7535 (MAL, MO, PRE); Chongoni Forest Reserve, 24 Feb. 1986, La Croix 2698 (MO). SOUTHERN REGION: Zomba, Old Naisi Road, 11 Jan. 1978, Masiye 26 (M, MAL, MO, SRGH, Z); Zomba, Old Naisi Road, 1 mi. E of herbarium, rock outcrop, 23 Dec. 1980, Chapman 5500 (BR, MAL, MO), 5 Jan. 1986, Goldblatt 7514 (MO); Blantyre District, grassland north of Lundu, 7 Jan. 1986, Goldblatt 7525 (MO); Zomba, rock outcrop opposite university across Mponda stream, 6 Jan. 1986, Goldblatt 7521 (MO); Matope, Blantyre, 6 Jan. 1956, Jackson 1777 (BR, K, MAL); Shire Highlands, Oct. 1879, 1880, Buchanan 50 (E); 1891, 426 “eaten in time of great famine.” NORTHERN REGION: Rumphi District, Luwachi dispensary ridge, 24 Dec. 1972, Pawek 6122 (MAL); Karonga District, Sangilo Point, 705 m, 2 Jan. T 1973, Pawek 6304 (MAL, SRGH), 24 Feb. 1978, Pawek 13822 (BR, MO, WAG); Rumphi District, Chiweta, 474 m, near lake shoe 30 Dec. 1986, La Croix 4255 (MO). MOZAMBIQUE. ТЕТЕ: Angónia, 2 Dec. 1980, Macuácua 1363 (MO, PRE, WAG); Chioco, km 49 estrada para Mocubura, 15 Feb. 1968, Torre & Correia 17677 (LISC); n . 1973, Torre O (LISC). Niasa: Mandimba, 9 Dec. 1941, Hornby 3518 (PRE). NicERIA. BAUCHI: Vom, Bauchi Pla- teau, shallow soil on rocks, Dec. 1 a I 61986, ^ Lowe 1330 (K, WAG); Jos, July 1974, Shar: land 429 (К); Jos District, Naraguta, 10 May 1965, Olorunfemi s.n. (FHI 55801 in K). TANZANIA. MTWARE: Lindi, Mar. 1952, Semsei 701 (B, K, PRE); Tendaguru, 100 km NW of „к 17 Feb. 1935, Schlieben 6010 e cx x M, Z). RUKWA: Ufipa, Ilembe, 2,100 yo 8781 (K). ZAIRE. SHABA: Elisa- bill, 28 Mar. 1912, Bequaert 294 (BR); mont Na- ntamba, forét didi, 13 Feb. 1987, Billiet & Jadin 4137 (BR); entre village Kamina et Kyalwe, 3 Jan. 1972, Bulaimu 311 (BR); Welgelegen, 1912, Corbisier & Flo- rent 632 (BR); Kafubu, forét, 21 Mar. 1970, De Georgi 7 (BR, S); mont Mukuen, 8 Jan. 1957, Detilleux 361 (BR); Elisabethville, 22 Jan. 1926, Hirschberg 60 (К, PRE); Feb. 1912, Homblé 118 (BR); vallée de Kapiri, Feb. 1913, Homblé 1192 (BR); prés de Kipushi, 10 Mar. 1970, Lissowski 111 (BR); roadside 12 km from Lubum- bashi to Likasi, 1,200 m, 18 Feb. 1970, Lisowski 112 (BR); open forest on hill near Lukuni, 21 km NW of Lubumbashi, 1,300 m, 25 Jan. 1970, Lisowski 114 (B, К); Lupembe valley, ca. 28°28” 12°, 27 Jan. 1905, Kassner 2391 (BR, E, К, Р, 7); Chabara, colline cupri- fere, 1,430 m, 10 Jan. 1981, Malaisse 11446 (BR); Mwashya, 935 m, 30 Jan. 1981, Malaisse 11558 (BR); route Gombela à Poste Luishi, 30 km NE Gombela, 8 Feb. 1982, Malaisse & Robbrecht 1851 (BR); Fungu- b. 1983, Malaisse & Robbrecht 2181 (BR); ESE Kolwezi, colline cuprifére, 1,300 m, 17 Feb. 1982, Malaisse & Robbrecht 2407 (BR); Haut- Eanes 12 km S of Elisabethville, 17 Nov. 1928, Quarré 1017 (BR, K); Ha E valley of the Lubumbashi near Elisahethville, Mar. 1933, Quarré 3103 (BR, K, P, PRE, n. 1938, o (BR, К); Lubumbashi, Feb. nga, sol 946, Quarré 8178 (BR); Keyberg, 8 km SW а Doni нен rocky wooded hill, Jan. 1947, Schmitz и (ВВ); Jan. 1953, Schmitz 4303 (ВВ); rocky hill 10 km S ar paio nds: Feb. 1954, Schmitz W de Elisabethville, foret — ios de c 21 Feb. 1971. Liawehi 115 (BR, K) 14 Jan. 1971, Lisowski 117 (B, BR, K). ZAMBIA. COPPERBELT: Luan- NDO); "d gravelly y pan, 18 Jan. 1955, Fan- 2s 1800 (K, N 2211 (K, NDOJ); ан chipya dambo, 9 Mutimushi 2895 (NDO the Lumwana River, near ana Mission, shallow soil 456 Annals of the Missouri Botanical Garden over rock, 11°49” 25%7”, 19 Jan. 1975, Brummitt, Pol- hill & Chisumpa 13867 (K, NDO, SRGH, WAG), Mu- fulira, 4,000 ft., rocky shallow soil at riverside, 8 Feb., Cruse 187 (BR, К); S of Mufulira near Kafue bridge, cliffs and rock outcrops, 17 Jan. 1986, Goldblatt 7575 (MO); Parklands, Kitwe, bush at the end of Lincoln Ave., 2 Mar. 1961, Linley 85 (К, LISC, MO, SRGH); Kitwe, hill есте 6 Маг. 1964 (fr), Mutimushi 668 (К, NDO, SRGH); Kitwe, laterite in Miombo, 16 Feb. 1967, кект 1810 (BR, К, NDO); Chati Forest Station, ocky hill in Pees mpala Nature Reserve, 14 Jan. 1986, Goldblatt 7 7 (MO, NDO). NORTHERN: Abercorn Dis- trict, Tanya, n Jan. 1952, A 460 (K); flat wet site t dier owelo, 5 Mar. 1952, Rich- ards dea er UT ain of Set. Chilongwelo, 4,800 ft., 15 T hards 4495 (ВВ); Mningi pans, йшнде among rocks, 22 Feb. 1959, Richards 10957 (BR, K, MO); арни (Mbala), Itembwe Gap, 19 Jan., ES hards 18829 (K); Mbala District, St. Paul road, near edro, 5,000 ft., 22 Feb. 1968, Sanane 52 (K); 4 mi. NORTHWESTERN: Mwiniunga District, just N of Mwinilun- ga, shallow soil overlying laterite, 26 Jan. 1938, Milne- Redhead 4361 (BR, K). CENTRAL: Sanje Hill, Faden et al., 74/83 (MO, US); күрү ыг headwaters, 100-129 km E of Lusaka, 14 Feb. 1965, Robinson 6384 (B, M). EASTERN: Fort Jameson Ghee: Sumbi Hills, 950 m, 3 Jan. 1959, Robson 1025 (BR, K, LISC, PRE, SRGH); Fort Jameson District, Zingali Hill, 14 Feb. 1961, Grout 256 (BR, K); pred: Valley, siltstone areas near Ka- jp (NDO, SRGH). SOUTHERN: zh , 40 km N of Choma, 9 Feb., a 454 (K ADO); Mazabuka District, S of Lusaka, open ' woodland, l Jan. 1958, Noak 316 (К, SRGH); Mazabuka District, 4 mi. from Chirundi Bridge, mopane woodland, 6 Feb. 1958, Drummond 5493 (BR LISC, PRE, SRGH); Batoka Gorge, 3 Feb. 1963, Mitchell 17151 (B, K, SRGH). ZIMBABWE. MASHONALAND WEST: Lomagundi, Silverside Mine, 20 Jan. 1962, Jacobsen 1614 (PRE). MASHONALAND EAST: Domboshawa Mission, grassland, 27 Dec. 1971, Norrgrann 80 (MO, SRGH); Dombashawa, granite slopes, 16 Feb. 1958, Leach s.n. (BR, GRA, K, P, SRGH 83649); 3 Apr. 1977, Grosvenor & Renz 1302 (К, MO, PRE, SRGH); 3 Apr. 1986 Bayliss 10367 (MO); Bamps et al. 902 (BR, SRGH Salisbury District, Ruwa, farm Tanglewood, Dec. 1958, Miller 5583A (BR, SRGH). MASHONALAND CENTRAL: Dar- win District, Kandeya Native Reserve, 3,200 ft., 17 Jan. 1960, Phipps 2293 (BR, MO, PRE, SRGH); Sipolilo, below с М of Sipolilo, 30 Jan. 1948, Whellan 296 (K, SRGH). MANIC к Melsetter, Mt. Selinda, farm ua Dec. 1939, Obermeyer 2283 (M, PRE); Inyanga, Nyamaropa ee Trust, Biegel 1741 (MO, SRGH); Makoni, between Umtali and Spr near Inyanga road, 29 Feb. 1930, Friess et al. Pi S); Mtare District, Zimunya's Reserve, он country, 25 Jan. 1959, vei 7045 (BR, K, LISC, PRE, SRGH); Kelly's Park, 18 mi. NW of Mtare, 27 Nov. 1948, Chase 983 (BR, LISC. Е pou p Dis- ict, Gi 957, Phipps 34 (К, M ~ we. uw andhlovu Pasture Station, Jan. MO, LISC, SRGH); Shangani, Mar. 1918, Eyles 953 (K, P, SAM, SRGH); Wankie y Victoria Falls village, 17 Jan. 1974, Gonde 47/73 (K, S, SRGH); Victoria , 10 Feb. 1912, Rogers pe (PRE) Wankie National Park, Mopane savanna, 25 Feb. 1967, Rush- worth 229 (BR, K, SRGH); Bulawayo, May 1915, Rogers Jan. 1976, Cross 352 (K, MO, PRE, SRGH). MATABE- LELAND SOUTH: Matobo, 6 Dec. 1947, West 2442 (MO, SRGH); Matopos, Nov. 1922, Eyles 3752 (SAM); Tuli Experimental Station, Gwanda District, 14 Jan. 1965, Norris-Rogers 586 (К, SRGH); southern outskirts of Bu- lawayo, Jan. 1972 Goldblatt 606 (BOL). MIDLANDS: Gwelo District, Mlezu school, 8 Feb. 1 = > PRE). O PRECISE LOCA beleland, Feb. 1886, Elliot s.n. (K). South metas Gold. fields, 1870, mc s.n. (K) 5. Lapeirousia setifolia Harms, Bot. Jahrb. Syst. 30: 278. 1902. Lapeirousia erythran- tha var. setifolia (Harms) Geerinck et al., Bull. Soc. Roy. Bot. Belgique 105: 344. 1972. TYPE: Tanzania. Eastern Livingstone Moun- tains, Ubena, Tsausingewe District, 2,100 m, Mar. 1899, Goetze 812 (holotype, B; iso types, BR, E). Plants small, 5-10(-15) cm high. Corm 10-12 mm diam. at the base, tunics with the inner layers firm and unbroken, becoming decayed and + fi brous with age, dark brown. Cataphylls usually 2, pale and membranous below, brownish apically. Leaves 3-7, the lower 2-4 clustered basally, as- cending, or the lower leaves spreading, the low- ermost longest and exceeding the inflorescence, those above decreasing in size progressively, linear above and 0.5-1 mm wide, usually channeled to above the midline and noticeably broadening to- ward the base. Stem irregularly flexuose and some- what twisted, several- to many-branched, 3-4-an- gled. Inflorescence a congested pseudopanicle, the ultimate branches bearing 2-4 flowers, the inter- nodes half as long as the bracts; bracts herbaceous, often purple-flushed, 4—5 mm long, the inner as long or slightly shorter than the outer, the apices dry and apiculate, recurved. Flower zygomorphic, blue to violet, the lower tepals each with a white and dark blue marking in the lower midline; peri- anth tube 8-10 mm long, slender, expanded and curved in the upper 1.5 mm; tepals narrowly lan- ceolate, 7-9 mm long, ca. 2 mm wide, spreading at right angles to the tube. Filaments unilateral, exserted 2-2.5 mm from the tube; anthers parallel and contiguous, ca. 1.5-2 mm long, yellow (at least when dry). Ovary ca. 1 mm long, style di- viding near mid anther level, the branches ca. 1 mm long, divided for about half their length. Cap- sule globose-trilobed, ca. 2 mm long; seeds globose- Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa oblong, sometimes flattened or lightly angled on the raphal side, 1.3-1.8 x 1-1 Chromosome number 2n = 8. .3 mm, red-brown. Flowering time. February to March. Distribution and habitat. folia has a scattered distribution across south trop- ). Its northern limit is in the highlands of western Tanzania at the northeast end of Lake Malawi, and across the lake on the Nyika Plateau of Malawi. Populations also occur in the Inyanga Highlands of eastern Zimbabwe and in the lower and drier Matopos Hills of southwestern Zim- babwe. The habitat is one typical of Lapeirousia, Lapeirousia seti- ical Africa (Fig. exposed rock outcrops in shallow soil. Dwarf plants from the Chimanimani Highlands of eastern Zim- babwe possibly also belong here, but they have larger flowers and more likely are depauperate L. erythrantha. Diagnosis and relationships. Lapeirousia setifolia can usually be distinguished with ease from L. erythrantha and other members of this complex by its low stature, seldom exceeding 15 cm, leaves narrower than 2 mm, and somewhat contorted inflorescence branches. The flowers are fairly typical of the complex, having a violet peri- anth and relatively short perianth tube, but they are generally smaller than those of most forms of L. erythrantha, and they have narrow tepals ca. 2 mm wide and a particularly narrow perianth tube. Slightly more robust plants from rocky sites in the Matopos area of western Zimbabwe are in- cluded here with some hesitation. It is possible that they are dwarfed specimens of the form of Lap- eirousia erythrantha that occurs widely across south central Africa. However, L. erythrantha in western Zambia has large flowers with unusually broad tepals unlike L. setifolia. Plants from the higher Chimanimani Mountains of Zimbabwe per- haps belong in L. setifolia, but the available spec- imens are more like L. erythrantha although low in stature and with a few straight branches. Their branching pattern and relatively large flowers with broad tepals are quite typical of L. erythrantha. A single gathering from Zaire, Lisowski 9954b from the Marungu Plateau, that was assigned to Lapeirousia setifolia (as var. setifolia) by Gee- rinck et al. (1972) has larger flowers and bracts than L. setifolia and has a poorly ramified inflo- rescence of weakly twisted branches. It does not seem appropriate to place Lisowski 9954b in L. setifolia. It is perhaps a stunted form of L. ery- thrantha, which is well represented in Zaire, pos- sibly growing under unusual edaphic conditions, RE 9. Distribution of Lapeirousia setifolia (dots), I L. teretifolia (open circles), and L. angolensis (triangle). h ndicates uncertainty about the exact placement in Angola of the single collection each of L. teretifolia and L. angolensis. not uncommon in this area of highly mineralized soils. Additional specimens examined. MALAWI. NORTH- Nyika Plateau, Chelinda Bridge, 7,500 ft., Pawek 3421 (B, K, MAL, MO); Pa Mn 12431 (BR, MO, WAG); N Richards 14395 (K); Chosi ue Nike Plateau, 17 Feb. 1976, Phillips 1227 (K, MAL, MO, WAG); rocks above Chelinda Bridge, 2 Mar e Tones & E 269 (MAL); Nyika, just after Chosi turnoff, , 14 Feb. 1987, га Croix 4321 (MO); Nyika Plateau, 5. pe а сатр п the road to Mt. Chosi, among г | . 1977, CG aa. & Renz 1118 (MO). CANA, IRINGA: East- s, Ubena, Tsausingewe — 1899, as 812(B, BR, E). ZIMBABWE MASHONALAND EAST: Inyanga, shallow лісой soil on granite at Matemma, 11 Jan. 1967, Plowes 2841 (BR, P). MATABELELAND SOUTH: Matobo District, farm Велла Kobila, shallow soil over rock, Dec. 1954, Miller 2568 K, PRE); Mar. 1960, Miller 7261 (PRE, SRGH); Besna Kobila, grassland, Jan. 1954, Miller 2063 (K, LISC, SRGH); Matopos, Feb. 1903, Eyles 1176 (SRGH); Ma- topos, Amatyundula, 5,000 ft., 22 Dec. 1920, Borle 43 (PRE, SRGH); Matobo, farm Quasinga, 4,700 ft., Dec. 1953, Miller 2004 (B, BR, LISC, PRE, S). ~ 6. Lapeirousia teretifolia (Geerinck et al.) Goldblatt, comb. et stat. nov. Lapeirousia ery- thrantha var. vici Cage Geerinck, Lisowski, Malaisse & Symoens, Bull. Soc. Roy. Bot. Belgique 105: 342. 1972. TYPE: Zaire. Shaba: Plateau de la Manika, env. 2 km de Ka- tema, 20 Jan. 1969, Lisowski, Malaisse & Symoens 182 (lectotype, BR, here designat- ed—no holotype was indicated from among 458 Annals of th Missouri а Сагаеп the several duplicates of the type collection; isotypes, K (also EBV, POZ not seen)). Plants 20-40 cm high. Corm 9-13 mm diam., tunics dark brown to blackish, woody, the outer layers breaking into vertical parallel segments. Cataphylls 1 or 2, the upper (or only) one reaching shortly above the ground, dark brown, apparently dry by anthesis. Leaves 3-4, the lowermost in- serted near the ground and longest, the upper leaves decreasing in size above, the longest about half as long to as long as the stem, + terete to elliptic in section, without a discrete midrib, rigid, 1-1.5 mm diam. Stem + terete below, lightly 4-angled above. Inflorescence a rounded to colum- nar panicle, the main axis usually dominant, the terminal branches 1-2(-3)-flowered; bracts (2.5-) 3-4 mm long, + membranous, transparent below, rust-brown above or brownish entirely, the inner bract slightly longer than the outer. Flowers weakly zygomorphic, whitish to pale lilac, the lower tepals each with a darker blue to violet median streak in the lower midline; perianth tube 4-5 mm long, slender, slightly expanded above; tepals subequal, spreading + at right angles to the tube, 5-6 mm long, the uppermost held apart from the others, the lower 3 held closely together. Filaments uni- lateral, erect, 3.5-4 mm long, exserted 2-2.5 mm from the tube; anthers parallel and contiguous, 2— 3 mm long; pollen yellow. Ovary globose, 1-1.5 mm long, style arching behind the stamens, divid- ing at mid anther level, branches ca. 1 mm long, usually divided for about half their length, some- times for less. Capsule and seeds unknown. Chro- mosome number unknown. Flowering time. | February to April. Distribution and habitat. Lapeirousia teret- ifolia is a fairly local endemic of interior central Africa. It occurs in southern Zaire on the Manika Plateau where it extends from the Parc National de l'Upemba in the north to Musokantanda in the southwest; in Zambia it is restricted to the Mwini- lunga District in the northwest; and there is one record from Cacumbe, a locality I have not been able to place, in northeastern Angola. Collection information indicates that L. teretifolia grows in seasonally moist or waterlogged ground. Diagnosis and relationships. Lapeirousia teretifolia is a member of the tropical African L. erythrantha complex, with which its general form and flower correspond, yet it merits species rec- ognition. It has a more or less terete leaf, an open paniculate inflorescence with the main terminal branches 1-2- or rarely 3-flowered, a low stature, and somewhat smaller flowers than are normally found in L. erythrantha. Flower color also differs; the perianth is whitish to pale blue or lilac, whereas L. erythrantha has blue or red flowers. Lapeirou- sia teretifolia occurs partly within the range of L. erythrantha and, as indicated by collection data, in rather wetter habitats. First collected in 1938 by E. W. Milne-Redhead, Lapeirousia teretifolia has re- History. mained difficult to place. It was treated as var. teretifolia of the widespread L. erythrantha by Geerinck et al. (1972), but it seems altogether more distinctive than the several forms of this widespread species and has accordingly been raised to full species rank. Additional specimens examined. ANGOLA. LUNDA: Cacumbe, near River Cacu du 6 Dec. 1946, adn 13942 (K). ZAIRE. SHABA: env. de Katema, 12 Jan. 1971, Lisowski et al. 13243 (BR): кча и de Manika, oris de Katema, 12 Jan. 1971, Lisowski et al. 13302 (B, BR, R, K); Parc National de Г Upemba, 13 Feb. 3341 (BR, K, WAG). ZAMBIA. NORTHWESTERN: Mwinilunga Pres Kalenda plain below the W side of Matonchi Hill, 1,300 m, 18 Feb., Hooper & Townsend 1937, Milne-Redhead 3301 (BR, K Ridge W of Matonchi farm, laterite, among Vellozia, 22 Jan. 1938, Milne-Redhead 4280 (BR, К, LISC, PRE); Mwinilunga District, Luakera Falls, sandy ia 25 Jan. 1938, Milne-Redhead 4339 (BR, K, LISC, PRE); Ma- tonchi farm, peaty soil, 19 Nov. 1962, Richards 17294 (К); 15 km W of Kalene Hill, 14 Jan. 1963, Robinson 6031 (BR, K, M, SRGH); Ikelenge, Mwinilunga District, 16 Apr. 1963, Robinson 6597 (K, SRGH), 2.1960, Pinhey ? (SRGH) 7. Lapeirousia angolensis Goldblatt, sp. nov. TYPE: Angola. Moxico: a few mi. W of River Kaperu (Kapelu), boggy bu 10 Jan. 1938, Milne-Redhead 4037 (holotype, K; isotypes, BM, BR, LISC, P, PRE). Plantae 24-30 cm altae, tunicis cormi atrobrunneis, foliis teretibus vel ellipticis, inflorescentis paniculatis pau- 2 mm longis, tepalis 13-14 mm йй, filamentis са. б mm longis, exsertis 5 mm tubo Plants 24-30 cm high. Corm 13-16 mm diam., tunics dark brown to blackish, composed of hard layers of densely compacted fibers, the outer layers breaking into parallel vertical sections. Cataphylls 2, membranous and dark brown, the inner one reaching shortly above the ground. Leaves 2-3, the lowermost inserted near the ground and longest, about half as long to as long as the inflorescence, Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa 459 the upper leaves decreasing in size progressively, + terete to elliptic in section, without a discrete midrib, rigid, 1-1.5 mm diam. in the middle. Stem weakly compressed, 2-angled below, 3-angled above the branches; branching divaricate. Inflorescence a rounded panicle with relatively few flowers, the main axis usually dominant, the terminal branches 1-2-flowered; bracts (4-)5-6 mm long, + mem- branous, green below, rust-brown above, becoming brownish entirely, the inner slightly longer than the outer. Flowers zygomorphic, pale violet, the lower tepals each with a pale yellow median strea edged in purple on the lower midline; perianth tube 3.3 mm long, slender, slightly expanded above; tepals subequal (their orientation uncertain), + differentiated into a claw below, 12-14 mm long, the limb lanceolate, 1.5-2 mm wide, the margins undulate, the upper tepal apparently held apart from the others, the lower 3 tepals united below for ca. 2 mm and held closely together. Filaments unilateral, erect, ca. 6 mm long, exserted 5 mm from the tube, reddish; anthers parallel and con- tiguous, 3.5 mm long, violet; pollen pale. Ovary globose, ca. 1.5 mm long, style arching behind the stamens, dividing toward the apex of the anthers, branches ca. 1.4 mm long, undivided, recurving. Capsule and seeds unknown. Chromosome num- ber unknown. Flowering time. February to April. Distribution and habitat. golensis is known only from three gatherings made in western Zambia and eastern Angola (Fig. 9). It occurs in boggy grassland, probably in seasonally inundated areas. This type of habitat is most often associated with L. rivularis and may indicate a phylogenetic relationship between the two. Lapeirousia an- Diagnosis and relationships. The most un- usual feature of Lapeirousia angolensis is its flow- er with long narrow tepals that greatly exceed the fairly short perianth tube. The tepals appear to be subequal, but the lower three are connate for ca. mm more than the upper three, and hence are shorter. I have not seen live material, and the flower description is a reconstruction that may not B» entirely accurate. Clearly the flower is sul different from that of any other species of the L. erythrantha complex. The tube is only 3.3 mm long and does not or barely exceeds the bracts, and the unusually long tepals are thus at least three times as long. The union of the lower three tepals for ca. 2 mm is an additional feature that separates L. angolensis from the other species in the com plex. The narrow terete leaf is reminiscent of L. teretifolia, but the lax panicle and larger bracts and flowers are quite different from this small- flowered species. The style branches are undivided, a feature that is not uncommon in the tropical African species of Lapeirousia, and this helps little in placing the species. History. Lapeirousia angolensis is poorly known, having been collected just three times, first by a Mlle. Kiener prior to 1896, at Haut-Zambése, a locality too vague to be placed. Later collections are from eastern Angola and western Zambia, in areas that are seasonally inundated. Until now the species has not been associated with any other, although it bears a fair resemblance to L. ery- thrantha. Specimens examined. ANGOLA. MOXICO: a few mi. W of River Kaperu n boggy grassland, 10 Jan. "eie Milne- Следат 37 (ВМ, ВК, К, LISC, PRE). МВІА. е ER ngu flood plain, damp grassland, 29 Jan. 1966, be. 6830 (К). WITHOUT PRECISE LocaLiTY: ?ZAMBIA: Haut-Zambése, Kiener s.n. before 1896 (P). 8. Lapeirousia abyssinica (R. Br. ex A. Rich- ard) Baker, J. Linn. Soc. Bot. 16: 155. 1878; Fl. Trop. Africa 7: 351. 1898. Andrews, Flow. Pl. Sudan 3: 293. 1956. Cufodontis, Enum. Pl. Aethiopiae Sperm. 2: 1592. 1972. Geisso- rhiza abyssinica R. Br. ex A. Richard, Tent. Fl. Abyssinica 2: 308. 1850. TYPE: Ethiopia: Maigoigoi ad Dobre Sina, Quartin Dillon & Petit s.n. (lectotype, P, here designated; iso- lectotype, BR); Selleuda prope Adoua, Quar- tin Dillon s.n. (syntype, P); Maigoigoi, Schim- per s.n. (syntype, P). [Geissorhiza abyssinica R. Br. in Salt, Voyage to Abyssinia, Appendix 1. 1814, nom. nud.] Figure 10. Montbretia abyssinica Hochst. ex A. Richard, Tent. Fl. Abyssinica 2: 308. 1850. TYPE: acum collibus prope Adoua, flor. et fruct. Octobre, Schimper 329 (lectotype, P, here одане sri BM, K, M, P). Montbretia gallabatensis Schweinf. ms (Schweinfurth 1, B K, P). Plants generally small, 9-15 cm high but oc- casionally to cm, sparsely branched. Corm 8- mm diam., tunics brown to gray, densely fi- brous, the outer layers becoming finely fibrous and reticulate. Cataphylls 2, the inner one pale and membranous, reaching shortly above the ground, the outer shorter and dark brown. Leaves 3, the lower 2 at least usually inserted near ground level, the lowermost longest and about as long as to slightly exceeding the inflorescence, lanceolate, 3— 5 mm wide in the midline. Stem compressed and 460 Annals of the Missouri Botanical Garden — о 100/400 800 800 1000 кы 10 to FIGURE 10. 2-3-angled, sometimes narrowly winged above. In- florescence a spike or few-branched pseudopanicle, the main axis 5-7-flowered, the secondary axes with fewer flowers; bracts herbaceous, often flushed red to purple, becoming dry in late flower, 6-8 (— 10) mm long, the outer bract nearly always ex- ceeding the inner. Flower zygomorphic, violet, the lower 3 tepals each with a white median streak edged with a darker band of purple in the lower midline; perianth tube + straight, narrowly funnel- shaped, ca. 9 mm long; tepals unequal, lanceolate, the lower 3 horizontal to descending, held close together and forming a lip, ca. 9 mm long, to 2 mm wide, the upper 3 larger, the uppermost + erect, the upper laterals reflexed, ca. 9 mm long, to 3 mm wide. Filaments ca. 6 mm long, unilateral, exserted 3 mm from the tube; anthers parallel and contiguous, ca. 3 mm long, pale gray; pollen whit- ish. Ovary ovate-obovate, ca. 2 mm long; style unilateral, arching behind the stamens, dividing between the middle and apex of the anthers, branches ca. 2 mm long, barely notched apically. Capsules globose-trigonous, 3-4 mm long, showing the outline of the seeds; seeds red-brown, globose, 2 mm diam. Chromosome number 2n = Flowering time. Late August to early October. ¿a > A ХЕ. 1 ay + L> \ % ч T7 n Pe i кү агае o j f H Wit c > | | „ 4 + LÀ e + t + \ Ol | | à | N yt f Ñ EN 4 ALA € L5. v КАЛ am s. M t — и ы N \ AAA 20 WM f 4 Га aq o 10 20 30 “0 so Morphology and distribution of Lapeirousia abyssinica. Habit х 0.5. (Drawn by J. C. Manning.) Distribution and habitat. Lapeirousia abys- sinica is fairly widespread in the northern half of Ethiopia, where it occurs in shallow soils, usually in rocky sites from northern Shoa Province in the south to Eritrea in the vicinity of Keren in the north. The record indicates that the species is most common in Tigray, the origin of numerous collec- tions. It is recorded from a variety of substrates but is apparently most often found associated with limestone rocks. A collection from “Gallabat near Matamma” (Schweinfurth 1) is probably from eastern Sudan close to the Ethiopian border, and is so cited by Andrews (1956). Diagnosis and relationships. Evidently closely related to Lapeirousia erythrantha, and clearly a member of this tropical African species complex, L. abyssinica is a plant of low stature with a relatively few-branched to unbranched in- florescence that is essentially a spike. The main axis has 5-7 flowers and the branches typically fewer. The flowers are virtually identical to those of L. erythrantha in their blue-violet color, white and dark blue nectar guides on the lower tepals, and erect to reflexed upper tepal. The chromosome number is 2n — 8 in the one population counted Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa (Goldblatt, 1990b), whereas L. erythrantha has a basic diploid number of 2n = History. First collected by Henry Salt in Ti- gray, northern Ethiopia in 1810-1811, Lapei- rousia abyssinica was listed without description by Robert Brown as Geissorhiza abyssinica in the Appendix to Salt’s Travels (Salt, 1814). Schimper used the name for some of his collections of the species, and Achille Richard validated it in 1850. In the same work Richard described Montbretia T и Hochstetter’s manuscript e that appeared on the labels for Schimper 329, iba undoubtedly conspecific, Geissorhi- za abyssinica and Montbretia abyssinica are based on different type specimens. Geissorhiza abyssin- ica was transferred to Lapeirousia by J. G. Baker in 1876. Additional Vig e examined. ETHIOPIA. ERI- TREA: near Asm ‚600 ft., thin soil, flat open land, 29 Aug. 1954, Colville 19 (К); of Asmara, sandy waste ground around granite boulders, 26 Aug. 1959, Mooney 8071 (BR, K, S); Mt. Presso, Scimenzana, 2,500 m, 28 Aug. 1902, Pappi 814 (BR, nzana, Monti Presso, Senafé, ft., 11 957, Hallier 931 (К). GONDER: Gondar, Rochet de Hericourt s.n. (P). SHOA: Royaume de Choa 1847, Rochet de Hericourt s.n. (P); Muger valley, ca 100 km N of Addis pr £m аы 1,800 m, 9°30” 38°25", Edwards et al. 97 (MO). TIGRAY: 40 len S of Kwi m, gent a. slopes on i esto 13°11” wiha, 2, m 39°32”, 12 Aug. 1973, Gilbert & Getachaw 2625 (K); S of Qui 2,150 m, 5 Sep. 1970, de Wilde 6 WAG); 8 km SW of Inda Selassie on ы road, 1, 830 m, shallow soil over basalt, seasonally waterlogged, 29 Aug. 1973, Gilbert & Getachaw 2919 (С, К); near the pass of Atbara, sand, Salt 37 (BM); valee d'Adoua, Aug. 1839, Quartin Dillon & Petit s.n. (P); prope Adouam, 1839, Schimper 645 (BM, С, Р); Shire, 1853, Schimper s.n. (Р); 37 km S of бш. road to Maichew, dry limestone hills, low shrubland, 5 Sep. 1970, Amshoff 6984 (MO). SUDAN: Gallabat, near Matamma, 24 June 1861, *Montbretia gallabatensis Schweinf.', Schwein- furth 1 (BM, С, К, P) (possibly in Ethiopia). WITHOUT РкЕСІЅЕ LocaLiTY: ETHIOPIA: Abyssinia, Pearce s.n. in 1820 (BM); 1861, bes n. (К); vallée Mareb, Rochet de Hericourt s.n. (BM); Aaosa, 6,000 ft., 8 Sep. Wo (fr), Schimper 563 (P (2); Woina, 6,000-7,000 ft., Aug. 1852, Schimper 563 (BR, K, P). 9. Lapeirousia masukuensis Vaupel & Schlechter, Bot. Jahrb. Syst. 48: 545-546. 1912. TYPE: Mozambique. Inhambane: Ma- suku forest, 10 Feb. 1898, Schlechter 12109 (holotype, B; isotypes, BR, COI, G, K, P, PRE, SAM) Plants 40-60 cm high. Corm campanulate, ca. 15 mm diam. at the base, light brown, tunics co- riaceous with a reticulate surface, the outer layers decaying to become coarsely fibrous. Cataphylls usually 2, membranous, light brown, the upper one longer and reaching shortly above ground level. Leaves 4-6, the lowermost inserted close to the ground and largest, the remaining becoming pro- gressively smaller above, the longest usually slight- ly exceeding the inflorescence, narrowly linear- lanceolate, the midribs prominent, 3-6 mm wide near the midline. Stem compressed and 2-angled to winged below, 3-angled and lightly winged above the branches. /nflorescence a branched spike or pseudopanicle, the main axis + straight and dom- inant, the major ultimate branches forming spikes of 5-9 flowers, the flowers often crowded with the bracts overlapping and (1-)1.5-2 internodes long; bracts herbaceous at anthesis, becoming dry and scarious toward the end of flowering, then either pale throughout or brownish above and pale below with streaks of brown on the nerves, 5-7 mm long, the inner bract about as long as the outer or slightly shorter. Flowers zygomorphic, either blue to violet or greenish cream, the lower tepals each with a purple to red and white hastate median streak in the lower half; perianth tube cylindric, (15-)20- 25 mm long, slightly expanded in the upper 4 mm; tepals subequal, 8-10 mm long, lanceolate, ca. 3 mm wide, spreading at right angles to the tube and lying in + the same plane, the upper tepal held apart from the others. Filaments unilateral, + erect, ca. 5 mm long, exserted 2.5-3 mm from the tube; anthers parallel and contiguous, 3.5 mm long, probably pale yellow when live. Ovary ca. 1.5 mm long, style unilateral and arching behind the stamens, dividing near the upper 4% of the anthers, the branches ca. 2 mm long, divided for about 4 their length. Capsules depressed-globose, + 3-lobed and showing the outline of the seeds, seeds globose, ca. 2 mm diam. Chromosome number unknown. ca. 5 mm diam.; Flowering time. February to April. Distribution and habitat. sukuensis is restricted to southeastern tropical Af- rica, where it extends from coastal central and Lapeirousia ma- southern Mozambique in the provinces of Inham- bane and Maputo (Sul do Save) westward into the lowlands of the eastern Transvaal of South Africa and southeastern Zimbabwe (Fig. 11). There is an apparent gap in the distribution between the Kru- ger National Park on the Transvaal border and the Mozambique coast, but this may be due to inad- equate sampling in interior Mozambique. A single 462 Annals of the Missouri Botanical Garden FIGURE 11. (triangles). Habit and inflorescence X 0.5; leaf detail x 1.5. (Draw collection from northern Malawi is provisionally referred to this species although its identity remains uncertain. The collection data make it clear that L. masukuensis prefers relatively wet habitats. Most specimens are recorded as growing in vleis, the edge of swamps, or in seasonally waterlogged flats. Diagnosis and relationships. The compara- tively tall stature, usually over 45 cm high, alter- nate branching pattern, ultimate inflorescence branches bearing 5—9 apically crowded flowers, and the long straight perianth tube 20-25 mm long are the main distinguishing features of Lapei- rousia masukuensis. Plants from the Mozambique coast are described as having greenish flowers with Morphology and distribution of Lapeirousia sandersonii (dots) and distribution of L. masukuensis n by Б) Ј. С. Маппіп red markings, an unusual color in the genus. How- ever, Zimbabwe, and Malawi L. masukuensis has flowers with a dark blue-violet perianth with white markings that are typical of related species, such as L. erythrantha and L. sandersonii. The long perianth tube and general appearance of the flowers, including their color, suggest a relationship wit Transvaal and Botswana species L. sandersonii, and it is usually under this name that specimens of L. masukuensis have been placed in herbaria until now. Lapeirousia sandersonii is a shorter plant, rarely exceeding 35 cm, with a dichotomous and often intricately branched habit, and the ul- timate branches of the inflorescence have one, two, in the eastern Transvaal, the western Volume 77, Number 3 1990 Goldblatt 463 Systematics of Lapeirousia in Tropical Africa or rarely three flowers. The corms of the two species also differ, those of L. masukuensis being relatively small and having the outer tunic layers fibrous and reticulate, whereas the larger corms of L. sander- sonii have dark brown tunics that decay irregularly into smooth vertical strips. The corms o ma- sukuensis resemble closely those of L. суша with which it is perhaps most closely allied. Plants from the Transvaal and Zimbabwe are more variable than those from Mozambique and particularly so in regard to the length of the peri- anth tube and the number of flowers on the terminal branches of the inflorescence. Generally plants from the Transvaal and Zimbabwe have five or six flow- ers per branch compared with 6-9 in Mozambique, and the perianth tube is 15-22 mm long. A notable example is Mauve 4326, 10 m N of Abel Erasmus Pass, which has some flowers with a perianth tube just 15 mm long, while others have a tube up to 22 mm. It is difficult to explain such gross variation in a feature such as perianth tube length, but this degree of variability is noted in a few collections of other tropical African Lapeirousia, e.g., L bainesii, L. sandersonii, and L. erythrantha. History. Discovered in 1898 by the widely traveled German botanist and prolific collector Ru- dolf Schlechter, Lapeirousia masukuensis was de- scribed by Vaupel and Schlechter in 1912. It ap- pears to be comparatively rare, particularly in Zimbabwe and Mozambique. An early collection made by the French missionary Henri Junod at Shilouvane in the Transvaal bears the manuscript name L. junodii N. E. Br., indicating that Brown also considered the species distinct from L. san- dersonii. Additional mo examined. MALAWI. NORTH- ERN PROVINCE: 20 mi. NW of Rumphi, 1,400 m, 11 Mar. 1978, hard, sig pne soil, Pawek 14048 (K, MAL). ОА. INHAMBANE: Quissico, 28 Feb. 1955, Exell, Mendonca & му ТОЗ үз, as H); between Mor- г ene and Massinga, 26 xell, Mendonca & Wild 652 (LISC. SRCH): екеш Velho. June a Gomes & Sousa 2133 (COI, К). MAPUTO: Manhiga, v n Incomati, 26 Mar. 1979, de Koning 7353 (K, LISC, A); ier ве ев the меде СУ e and gh Pati, om the pontoon, 24 Mar. 1954, Barbosa & Баште 5450 (LISC, LMA): Vila Luisa, 1974, papa pre Balsinhas 2699 y Marracuene ao 2 de Vila Luiza para eb. 1969, Correia e Marques 591 (WAG). UTH AFRICA. TRANSVAAL: 1 (Pafuri), Kruger Na- tional Park, Kla cereo 1 Jan. 1953 (CA), van der Schijf 1858 (PRE) 23.30 (Tzaneen) Hans Merensky ature Reserve, mopane veld in damp earth along sloot, 11 Feb. 1971 (DA), Oates 371 (PRE); Merensky Nature о. waterlogged clay loam in vlei, 2,000 ft., 15 Mar. 1977, Zambatis 737 (PRE); 24.30 (Pilgrims Rest), E Jan. 1919 (AB), Junod 4139 (С, M, PRE); Shilouvane Plaine, s.d., ‘L. junodii N. E. Br.’ Junod 736 G, K); 10 m N of Abel Erasmus Pass, grass in bushveld, 16 Dec. 1964 (DA), Mauve 4326 (K, PRE) 24.31 (Acornhoek), farm Grootdraai, stony flats, 1,500 ft., 15 Nov. 1973 (AA), Zambatis 543 (МО); Klaserie, farm Sark, seasonal са іп sandy clay, 28 Jan. 1982 (AC), Zambatis 1345 (PRE); Kruger National Park, 25 Feb. 1953 (AD), van a A йө (PRE); Pumbe, Satara, sandy soil, Mar. 1967 (BB), van ү 4783 (PRE); Man- ame Mea clay, 400 ft., 9 Mar. 1977, Bre- denkamp a (PRE). ZIMBABWE. m CTORIA: Bikita Dis- ugh-Fort Victoria + oad, 16 May 1962, nster Mission, 17 Mar. 1958, Leach 8219 (SRGH); Kyle National Park, clay slope above granite outcrop, 9 Feb. 1972, Gibbs Russel 1465 (SRGH); Kyle Dam, shade in gully by granite outcrop, 8 Jan. 1972, Gibbs Russel 1452 (SRGH); Be- lingwe, near the Mnene road, among rocks, Norlindh & Weimarck 5202 (BR, PRE, S, SRGH) — 10. Lapeirousia sandersonii Baker, Handbk. Irideae 169. 1892; Fl. Cap. 6: 95. 1896; Fl. Trop. Africa 7: 352. 1898, excl. specimens cited. van Druten, Fl. Pl. Africa 31: 1226. 1956. Letty, Wild Flowers of the Transvaal 77, t. 37. 1962. TYPE: South Africa: Trans- vaal, Sanderson s.n. (lectotype, K, here des- ignated); Rhenosterpoort, /Velson 402 (syn- types, K, ; Transvaal, without precise locality, Todd 20, 21 (syntype, K). As no type was indicated in the protologue, a lectotype has been selected from among the four spec- imens from South Africa cited in Flora Ca- pensis (1896). None of the several more from tropical Africa cited by Baker (1898) are this species. The Sanderson collection from the Transvaal chosen as the lectotype is m so annotated in the Kew Herbarium (by N Brown ?). Figure 11 Lapeirousia bainesii var. d Baker, J. Linn. Soc. Bot. 16: 156. 1856, n no specimen cited, but “Nelson 402, see e: is so annotated). Plants 18-35 cm high. Corms 2.5-3 cm diam., tunics dark brown, coriaceous internally, decaying somewhat irregularly into vertical segments, sel- dom becoming fibrous and never reticulate. Cata- phylls usually 2, dark brown, the inner one reach- ing to shortly above ground level, the outer much shorter. Leaves 2-4, + linear, 2-3 mm wide, firm to rigid, the midrib and lateral veins fairly prom- inent and closely set, the lowermost longest and usually exceeding the inflorescence, the upper leaves progressively shorter. Stem compressed, 2-angle to winged below, triangular above the branches and often lightly winged on the angles. /nflorescence sometimes intricately so, the branches + divaricate but unequal and a main axis much branched, 464 Annals of the Missouri Botanical Garden usually evident, the ultimate branches 1-2(-3)- flowered; bracts (4.5-)5-8(-10) mm long, her- baceous with brown tips in bud, often flushed pur- ple, becoming scarious throughout in later flower and brown almost entirely. Flowers zygomorphic, blue to violet, the lower 3 tepals each with a deep red to purple and white spear-shaped nectar guide 15-18(-20) mm long, slender, but slightly wider in the upper in the lower midline; perianth tube 2-3 mm; tepals subequal, 10-11 mm long, lan- ceolate, 3-3.5 mm wide, the upper held apart from the others and reflexed, the lower 3 joined for about 1 mm and forming a lip, when fully open all held in + the same nearly horizontal plane. Filaments unilateral, + erect, ca. 5 mm long, exserted 3 mm from the tube; anthers parallel and contiguous, 3— 4 mm long, purple; pollen pale yellow. Ovary ovoid, ca. 2 mm long, style dividing near the apex of the anthers, the branches simple or divided for a short distance, ca. 1.5 mm long. Capsules depressed globose, ca. 5 mm diam., 4-5 mm long; seeds nearly globose, ca. 2 mm diam. Chromosome num- ber 2n — Flowering time. December to April. Distribution and habitat. Lapeirousia san- dersonii is native to the interior of eastern Botswa- na and the adjacent part of South Africa, where it occurs in the relatively dry northern and western Transvaal and the northern Cape (Fig. 11). It ap- pears to be most common in the Pretoria and Rustenburg areas of the Transvaal, and, according to the collection record, relatively rare in the north- ern Transvaal and Botswana. The easternmost rec- ords from the cool, high, well watered Dullstroom, Belfast, and Middleburg areas of the eastern Trans- vaal are surprising in view of the rest of the range in semiarid country, but the eastern populations differ in no significant way from those occurring further west. The habitat is always rocky, and usually well drained, such as hill slopes, ridges and summits. Corms are seldom collected, which prob- ably reflects the difficulty in extracting them from rocky ground. Diagnosis and relationships. Clearly a mem- ber of the Lapeirousia erythrantha complex, L. sandersonii has the repeatedly branched, pseu- dopaniculate inflorescence and blue-violet flowers that characterize most tropical African species of Lapeirousia. It can be distinguished from the re- lated L. erythrantha by its longer perianth tube 15-20 mm long, and by its usually highly ramified and divaricately branched inflorescence with 1-2 (—3) flowers on the major terminal branches. La- peirousia sandersonii is most easily confused with occasional longer-tubed forms of L. erythrantha that occur in eastern Zimbabwe and Malawi, but these plants have shorter bracts 4-6 mm long, and less-branched inflorescences with 3-5 flowers per main terminal branch. These long-tubed forms of L. erythrantha are particularly variable and con- sist of plants with tubes ranging from 12 to 15 mm. А population from near Dedza in central Ma- lawi that is particularly variable for perianth tube length has a chromosome number of 1 exactly corresponding to that in surrounding pop- ulations of L. erythrantha and unlike the karyo- t of subsp. sandersonii, which has 2n = 10 (Goldblatt. 1990b). so easily confused with L. sandersonii is L. masukuensis of the eastern Transvaal, southeast- ern Zimbabwe, and Mozambique. This species has flowers with a particularly long perianth tube, typ- ically 20-25 mm long, soft-textured leaves, coarse- ly fibrous to reticulate outer tunic layers, and 5- 9 flowers per major terminal inflorescence branch. The two have often been confused, and the later name, L. masukuensis, has not been used in her- baria. It seems distinct from L. only in several morphological characters but also in the lowland distribution, from 2,000 ft. to near sea level, and preference for wet sites such as vleis, sandersonii, not seeps, or seasonally waterlogged flats. A few collections of Lapeirousia from the Wa- terberg in the western Transvaal broadly resemble L. sandersonii in habit, flowers have a perianth tube 10—12 mm long (e.g., Werdermann & Oberdieck 1640). The short peri- anth tube in £. sandersonii from this area is puz- leaf, and corm, but the zling. These plants are indicated by an asterisk in the exsiccatae. History. Collected first by John Sanderson, probably in 1852 when he traveled to Rustenberg and the Magaliesberg on his only journey to the Transvaal, Lapeirousia sandersonii first appeared in the literature as “L. bainesii var. breviflora,” Baker (1876). Baker sandersonii in 1892, it was so treated in Flora Capensis and Flora of Tropical Africa. W. Marais (pers. comm.) sug- gested that it be assigned infraspecific status in L. erythrantha although he preferred varietal rank. a nomen nudum of J. G. subsequently described L. Additional des irn examined (unusual short-tubed ms ated by an asterisk [*]). BOTSWANA. 2 24.25 (Gaborone) near Molepolole, shale, 15 Apr. 1930 (BC), van Son s.n. (PRE 28664); W of Ga- b 1976, Mott 928 (SRGH, UCBC). NGWAKETSE: 25. 25 (Mafeking) 6 mi W of Kanye, Feb. 1971 (AB), van Rensburg B4226 (PRE). SouTH Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa 465 AFRICA. CAPE: 26.24 (Vryburg) Moshesh near Mosita, ironstone koppie, 10 Apr. 1945 (BB), Brueckner 470 (К, PRE); а 5 Feb. 1948 (DC), Rodin 3501 (К, , P, PRE, S); 27.23 (Kuruman) Bathlaros, Apr. 1921 (AC), Silk 231(K); 7 mi. SW of Kuruman, Wisura 2003 NBG). TRANSVAAL: 22.29 (Waterpoort) Zoutpansberg, 1 mi. from Dandy farm, road to Sand River, S slopes in rocky outcrop, 3 Apr. 1957 (DC), Meeuse 10212 (K, M, PRE, SRGH); farm Thornhill 743, red sand in riverine woodland, 18 Mar. 1985 (DD), Raal 434 (PRE); 22.30 (Messina) Louis Trichardt, farm Rietbok 226, among rocks in montane scrub forest, 1,476 m, 25 Mar. 1982 (CC- CD), Venter 7812 (PRE); 23.27 (Ellisras) йылы d 67 mi. N of Vaalwater on Beauty road, eb. 1983 (DB), Reid 656 (PRE); бое bermeen Groothoek and Vaalwater, 1,000 m, 6 Jan. 1959 (?), Werdermann & Oberdieck 1640 (B, BR, GH, i MO, PRE, WAG); 24.28 (Nylstroom) *Vlakfontein, 3 mi. N of P. O. Palala, 29 Jan. 1960 (BC), Codd 9975 (PRE); Sterkrivier Nature Reserve, rocky slope in mixed woodland, below slabs of bedroc . 1973 (BD), Jacobsen 2794 (PRE); koppie 19 km from Warmbaths on Kwaggasnek road, W slope, 13 Feb. 1981 (CC), Herman 227 (PRE); 24.30 = PRE, SRGH); 25. 26 (Zeerust) Zeerust, 28 Jan. (CA), Thode A1503 (GH, K); Zeerust, Feb. 1912, Jenkins 1M1 1660 (K, PRE); 15 m E of Swartru ко. bashýeld on ridges (DB), 14 Feb. 1956, Acocks 1874 mi. E of Pus ed on the Rustenburg diei 25 Feb. 1947, Sidey 83 (S); Koster, among rocks in 1929 (рї, Gilmore 1915 (G); 25.27 (Rustenberg) Rus- tenberg, 4,000 ft., stony hillside near town, July 1904 (fl & fr) (CA), Pegler 1108 (GRA, K, PRE, SAM); 22 m W of Rustenberg, 14 Feb. 1956, Leistner 539 (K, PRE); Rustenberg, Tierkloof, moeras by wildreservaa- thek, 13 Mar. 1976, Venter 654 (K, MO, PRE, SRGH, WAG) Rustenberg Nature Reserve, rocky hillsides in grassland, 25 Feb. 1970, Jacobsen 745 (PRE); 16 km from Palala on road to Bamboeskloof, 9 Mar. din (CC), Poo 735 (PRE); Breedtsnek, grass on open top, 7 Apr. 1939 (fr) (CD), Connell 42 (PRE); Brits, Silikaatsnek, top of poort in shallow sand over quartzite (DB), Codd 736 (PRE); 14 Feb. 1946, Story 0 (PRE); Krugersdorp, Jack Scott Nature Reserve, 2 Feb. 1961 (DC), Wells 2305 (К); 25.28 (Pretoria) Horns- nek Pass, Magaliesberg, W side, S slope in grassland, 31 Jan. 1983 (CA), Perry 2017 (NBG); Hornsnek, 12 km W of dien 12 Jan. 1956, Schlieben 7751 (B, BR, G, HBG, К, M); Hornsnek, light shade on summit among rocks, 10 Feb. 951, Prosser 1604 (K, NBG, PRE); N slopes of со 25 Mar. 1945, Mogg s.n. (К, M, PRE 27329, S, SRGH); The Willows, Pretoria, 9 Mar. 1906, Burtt Davy 5333 (GRA, K), 1703 (K); Cullinan, farm Vaalwaterkrans, 25.2 km from Pretoria on Roodeplaat Dam road, 28 Feb. 1980 (CB), €: Herman 159 (MO, PRE); Pienaarspoort, 17 mi. Pretoria, 20 Feb. 1959, Brent 138 (K, PRE); altos Vakansieoord, E of Pretoria, sandstone rock crevices an slopes, 8 Mar. 1977 (CD), van Jaarsveld 1870 (NBG); Donkerhoek, 20 mi. E of Pretoria, 19 Mar. 1959, Codd 9908 (K, PRE); 19 Mar. 1959, Letty 428 (K, PRE); Donkerhoek, rocky koppie, 11 Mar. 1943, Verdoorn 1908 (PRE); Premier Mine, Jan. 1919 (DA), Rogers 22415 (К,Р, SAM); June 1921, Rogers 24146 (В, BR, O, P, S, SAM); Renosterkop, Bronk- horstspruit District, 7 Feb. 1932 (DB), Young 2116 (K, PRE); farm Valsspruit, 19 km N of Bronkhorstspruit, 10 Feb. 1984, Crosby 64 (PRE); 25.29 (Witbank) Loskop Dam, Nooitgedacht, shallow des on su 9 1967 (AD), Theron 1214 (P i 30 Jan. 1929 (CD), [эн vias 2716 (K, belo, Renosterpoort, mountain summit, M > son 402 (K, PRE); 25.30 (Lydenburg) hills above Dullst- room, 23 Feb. 1937 (AC), van der Merwe 1255 (В, К, PRE); Dullstroom, among dolerite rocks on farm Valleis- pruit, 6,500 ft., 30 Jan. 1933, Galpin 13369 (K, P, PRE); farm Onverwacht 99, E aspect on shallow stony ground, s.d., Engelbrecht s.n. (PRE 664971); Belfast, Feb. 1909 (CA), Doidge 4800 (К); 26.27 (Potchefst- room) Dassiesrand, Potchefstroom, 23 Mar. 1940 (CA), van der Westerhuizen 1114 (PRE). WITHOUT PRECISE deme Pals AFRICA: Transvaal, Oct. 1869, Bu- chanan 21 (or Todd sub Buchanan) Jan. 1924, Rogers oe (G, K) (as Natal, Weenen, Culvers 6, 000 ). ft., which is almost certainly incorrect 11. Lapeirousia gracilis Vaupel, Bot. Jahrb. Syst. 48. 548. 1912. Sólch, Prod. Fl. Súd- westafrika 155: 9. 1969. TYPE: Namibia: Great Namaqualand, Doorns, dolomite, 1,450 m, r. 1907, Range 292 (holotype, B; K (pho- to), M, fragment). Figure 12 Plants (12-)15—30 cm high, often fairly slender, but occasionally robust, few- to several-branched. Corm 12-18 mm at the widest diameter, narrowly campanulate, tunics light brown, coriaceous, out- ermost layers becoming coarsely fibrous, the fibers extending upward as short spines. Leaves 2-3 (-4), the lowermost inserted just below the ground, the lower or second leaf usually the largest, the upper decreasing in size, becoming bractlike, linear to narrowly lanceolate, 3-5(-8) mm at the widest, the longest about as long or somewhat longer than the inflorescence, somewhat thickened around the midvein, a second vein on either side of the midvein also sometimes evident. Stem erect below, branch- ing + divaricately and the branches ascending but flexed to become + erect at the base of the flowers, 2-angled and winged below the first branch, 3-angled and winged above, the wings often promi- nent. Inflorescence + paniculate or the branching sparse, the ultimate branches short spikes of (1-) 2-6 flowers; bracts (4-)6-7 mm long, herbaceous below, becoming + membranous above, sometimes flushed purple, sometimes membranous and com- pletely dry before anthesis, then usually transpar- ent or with fine brown veins, subequal or the inner or outer larger. Flowers zygomorphic, white to pale blue or mauve, the lower 3 tepals yellow at the base and outlined distally by a dark violet margin, and each marked with a purple median streak or spot, occasionally one or more of the lower tepals with a small median toothlike callus in the midline, lightly fragrant; perianth tube cylindric, slightly 466 Annals of the Missouri Botanical Garden FIGURE 12. details of stamens and style x 2. (Drawn by J. С. curved outward at the apex, (21-)25-38 mm long; tepals nearly equal in size, lanceolate, ca. 10 mm -5 mm wide, the margins lightly undulate, the lower 3 closer together, apparently forming a lip, the uppermost held apart and nearly erect to reflexed, rarely the lower 3 tepals each with a short median toothlike callus. Filaments unilateral, exserted for ca. 2 mm; anthers parallel and con- tiguous, ca. 5 mm long, pale lilac; pollen cream. Ovary obovoid, ca. 2 mm long, style arching be- hind the filaments, dividing near the apex of the anthers, branches 1.5-2 mm long, forked for ca. 0.5 mm, ultimately recurving. Capsules + globose, obtusely trigonous, 4-5 mm long, 5-6 mm diam.; seeds globose to weakly angular, ca. 2 mm diam., dark brown. Chromosome number 2n — 12. Flowering time. Late January to April. Distribution and habitat. cilis is endemic to Namibia, where it occurs in a relatively broad band along the west central part of the country (Fig. 12). It occurs in rocky sites or sometimes on sandy flats; especially in more Lapetrousia gra- Morphology and distribution of T E ua gracilis. Habit and corm x 0.5; single flower full size; nning.) arid areas it is often associated with springs or other places where additional water supplements the sparse rainfall. The range of L. gracilis extends from the Fish River Canyon in the south to the Hoanib River in the Kaokoveld in the north. Flow- ering normally takes place at the end of summer, mostly in February and March but sometimes as late as May. An October-blooming collection, Cra- ven 1341, from the Brandberg is difficult to ex- plain; perhaps it indicates an unusual rainfall pat- tern in а particular year. The spring-blooming specimens do not appear to differ from those col- lected flowering in the summer. Diagnosis and relationships. The slender perianth tube 25-38 mm long, white to pale blue flower color, and comparatively slender habit make most specimens of Lapeirousia gracilis easy to recognize. It most closely resembles L. bainesii and L. long-tubed flowers. Lapeirousia bainesii has white to pale pink flowers with dark red to brown mark- ings and a tube 35-47 mm long. It also has a otaviensis, both of which have similarly Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa 467 more robust habit and longer floral bracts 7-12 mm long; the highly ramified inflorescence has only 1(-2) flowers per ultimate branch. Occasional ro- bust specimens of L. gracilis must be examined carefully to avoid confusing them with depauperate L. bainesii. Confusion with Lapeirousia otaviensis is more likely since this species, like L. gracilis, has an inflorescence with the main terminal branches bear- ing 2—5 flowers. The flowers and bracts of L. ota- viensis are larger and the perianth tube slightly longer than in L. gracilis. The latter typically has short and transparent floral bracts 4-7 mm long, a useful guide when flower size is intermediate and perianth color is unknown. Sólch (1969) distin- guished L. gracilis by its shorter perianth tube and more slender habit but not by flower color, which he described as white in both this and L. bainesii. This is generally incorrect for L. gracilis; although the flowers fade after drying and appear white, I have seen few collections describing the living flower as white and several that indicate shades of blue, evidently the more common peri- anth color in £. gracilis. Phylogenetic relationships are difficult to pos- tulate with confidence in Lapeirousia, but it seems plausible that L. gracilis, L. bainesii, and L. ota- viensis are immediately allied, sharing the derived features of pale-colored flowers having an extended perianth tube and pale, reticulate corm tunics (Ta- ble 3). The chromosome number of L. gracilis, 2n — 12, and bimodal karyotype (Goldblatt, 1990b) differ from that in L. otaviensis and L. bainesii, which have 2n — 10 or 6 and nonbimodal karyo- types. Discovered in 1907 by the German geologist Paul Range near Doorns in southern Na- mibia, Lapeirousia gracilis was described in 1912 by Vaupel. It was for many years poorly known but botanical exploration in the 1960s and 1970s has revealed a coherent distribution pattern. Additianat specimens examined. NAMIBIA. KAOKO- pr. Moss & Jacobsen K311 (РКЕ, WIND). OMARURU: 21.14 (Uis) Brandberg, 17 nid 1981 b Craven 1341 (WIND); Bran M Orabeswand, 2,000 m, 6 Apr. 1964, 6 (M, S). KARIBIB: 21.15 (Karibib) farm n Spit na coarse aru sand, 22.15 (Trekkopje) in marble rock around spring, 19 farmi Tsabichab, cracks May 1973 (BA), Giess d (K, M , 5, WIND) 22.16 (Otjimbingwe mibrand, Karibib, Otjosandu, 1 Mar. 1963 (AA), Seyde 13419 (B, C C, GH, K, MO, S, SRGH, WAG, WIND). REHOBOTH: 23. 16 (Nauchas) farm Weisenfels, 11 Mar. 1953 (AD), Walter & Walter 1675 (B, WIND); 23.17 (Rehoboth) Buellsport, flats, Mar. 1949 (?), Strey 2700 (PRE). MALTAHOHE: 25.1 hard clay vi ч Pease May 1913 3061 (SAM); , Goldblatt Man i e (K, MO, UE "PRE, S, WIND). KEETMANSHOOP: вае itzkoppe, 7 Арг. 1984 (AD), Craven 1666 (WIND). ae 27.17 (Chamaites) Fish anyon, river camp, 30 Mar. 1953 (DA), Walter & ‘Walter 2263 (B, WIND); Fish River Canyon, middle plateau, 23 Feb. 1963, Leipert 422 1 (WIND); Fish River anyon Reserve, edge of the can : NAMIBIA: Quartel/Re- hoboth, 10 Apr. 1911 (fr), Dinter 2152 (SAM). 12. Lapeirousia otaviensis R. Foster, Contr. Gray Herb. 127: 45-46. 1939. (Lapeirousia bainesii sensu Solch, Prod. Fl. Südwestafrika 155: 8. 1969.) TYPE: Namibia: Auros farm near Otavi, 11 Feb. 1925, Dinter 5577 (ho- lotype, а isotypes, В, С, PRE, SAM, Z (3)). Figure Plants 30-55 cm high, repeatedly branched. Corm 2-2.5 cm diam., tunics light brown, coria- ceous, becoming reticulate and coarsely fibrous with age, sometimes ultimately finely fibrous. Cata- phylls 2, membranous, the inner reaching to short- ly above the ground. Leaves 4-7, + linear (to narrowly lanceolate), gray-green, only the midrib prominent, the lower 2—4 basal and longest, 5-9 mm wide, exceeding the inflorescence by 5-10 cm; upper leaves cauline and decreasing in size above, those subtending the branches becoming bractlike. Stem weakly compressed below, trian- gular in the middle part and rectangular above, the angles lightly winged, the main axis straight, branches diverging at 45—80°. Inflorescence a + corymbose pseudopanicle, the main terminal branches with (1-)3-5 flowers in a short spike; bracts herbaceous, flushed red above, becoming dry apically, usually quite dry when the flowers have wilted, (7-)8-11 mm long, the inner often longer than the outer. lanceolate and acute, often Flowers zygomorphic, white to cream, sometimes flushed pale lilac, rarely purple, the lower 3 tepals each marked with violet streak in the midline and with darker double purple to red lines near the base, rarely the upper tepals with a pale purple median line, unscented and opening in mid morn- ing; perianth tube 40—45 mm long, straight, very 468 Annals of the Missouri Botanical Garden FIGURE 13. of stamens and style x 2.5. (Drawn y J. C. Manning.) gradually flared from the base, about 1.2 mm diam. below, 2.2 mm diam. at the mouth; tepals sub- equal, lanceolate, widest in the middle, the margins straight, 15-19 mm long, 4-5 mm wide, spreading almost at right angles to the tube, the uppermost slightly larger than the others, weakly to strongly acute. Filaments unilateral, exserted for 4-5 mm from the tube, white; anthers parallel and contig- uous, 5-6 mm long, gray-purple; pollen white. Ovary + ovoid, ca. 2 mm long; style nearly straight, lying behind the filaments, the branches usually dividing at (or to 3 mm beyond) the anther apices, branches ca. 2 mm long, forked for 44-12 their length, diverging but barely or not at all recurved. Capsules globose trigonous, ca. 7 mm long; seeds not known. Chromosome number 2n = 10 Flowering time. February to April. а апа e of Lapeirousia otaviensis. Habit x 0.5; single flower full size; details Distribution and habitat. Of relatively re- stricted distribution, Lapeirousia otaviensis oc- curs in a wide arc from the Erongo Mountains in western Namibia through the hills south of Etosha Pan to the Otavi Hills in the northeast and locally also in southern Angola (Fig. 13). It is apparently confined to rocky outcrops, usually growing in granite, but in the Otavi Hills it occurs on loca shale outcrops, not in the dolomite that predomi- nates in this area. Diagnosis and relationships. The large flow- er with a long perianth tube 4-5 cm long, white to lilac perianth with lanceolate tepals, and the ultimate branches of the inflorescence having up to five flowers (and rarely fewer than two) distin- guish Lapeirousia otaviensis. It has often been confused in herbaria with the apparently related Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa 469 L. bainesii and was even regarded as identical to this species by W. Marais (pers. comm.; Geerinck et al., 1972), under the name L. erythrantha var. bainesii. Lapeirousia otaviensis and L. bainesii both merit specific status. They can be separated by a series of characters, and they do not intergrade with each other or with any of the forms of L. erythrantha and its close allies. bainesii has flowers with a slightly shorter perianth tube 30-40 mm long, this often purplish in color, and has more or less clawed, white tepals with Lapeirousia undulate margins. The inflorescence of L. bainesii branches divaricately, and the ultimate branches have one or rarely two flowers. Lapeirousia ota- viensis is restricted to rocky sites of granite or shale, whereas L. bainesii occurs in deep hard sand in level savanna or bushveld. Ап interesting color variant of Lapeirousia ota- viensis occurs on the plateau and higher parts of the Erongo Mountains (Craven & Craven 115; Lavranos 22693). This has purple flowers in con- trast to the white or pale lilac elsewhere. Apart from the perianth color, there seem to be no sig- nificant differences between this and the more wide- spread color form. A collection from southern Angola made in 1909, Pearson 2738, is assigned here, but the poorly preserved plants are either depauperate or possibly belong to the related L. gracilis. The perianth tube is 35-48 mm long, within the range of variation encountered in L. otaviensis. The brown-tipped bracts are 7-9 mm long, which corresponds with L. otaviensis but is also in the upper range for L. gracilis. The latter is centered in southern to west central Namibia, whereas L. otaviensis occurs across northern Namibia so that the latter deter- mination is phytogeographically the more likely. History. Although first collected in 1879 in “Damaraland” by Ture Een, an early botanical explorer in Namibia, Lapeirousia otaviensis was described in 1939 by the American R. C. Foster from a gathering made in 1925 by Dinter at Auros farm near Otavi. This was the first collection of L. otaviensis to be distributed widely to herbaria. It seems to have been recognized as a distinct species by accident, as Foster compared it to the very different, short-tubed L. avasmontana in the pro- tologue. The reason for this is obscure and may have been because he had misidentified material of the latter. Apparently Foster was unaware at this time of the similarities that L. otaviensis shared and with whic became confused later. In the Prodromus Flora Sudwestafrika (Sólch, 1969) L. otaviensis and L. with L. ИНН Г. otaviensis bainesii were treated as conspecific and distin- guished in the key by having several-flowered ul- timate inflorescence branches. Sólch reserved the name L. vaupeliana Dinter for what is here called L. bainesii. The type material of L. vaupeliana has all the characters of L. bainesii, including one- flowered ultimate inflorescence branches. Additional specimens examined. ANGOLA. HUILA: damp sandy places near Monino, Humpata Pass, 3 Apr. 1909, Pearson 2738 (K). CUNENE: top of Ruacana Falls, ank, crevices in moist springs, 30 Apr. 1967, Rycroft 2443 (NBG, WIND). NAMIBIA. KAOKOVELD: 17.12 (Posto Velho), granite slope at Ombepera, rock crevices, corms aes 10 Apr. 1957 (BD), de Winter & Leistner 5481 (B, M, PRE, WIND); Orumana, stony flats below ee ridge, 20 Mar. 1974 с. егин & Giess 30407 (M, PRE, SRGH, WAG, W unene River banks, Mar. 1925, Barnard 1381 e KARIBIB: 21.15 (Karibib), 1 60, 27 Mar. 1976 (DC), Craven & Craven os (WIND); Erongo Plateau, among granite boulders, 985, Lavranos 22693 (E, K, M, MO, P, S); 22.15 rekkar e) Okongava, granite slope on Kalkberg, 4 Feb. 1934 (BB), Dinter 6962 (B, BM, G, HBG, K, M, PRE, ND, Z). OMARURU: 20. 15 (Ohjihorongo) 90 km кш Omaruru on the road to Fransfontein, Table Моип- Un nda К, S). ourjo: 19. 14 (Kamanjab) 8 km N manjab-Nord, granite domes, 18 Mar. 1974 (DB), Merx- müller & Giess 30392 (К, M, PRE, WAG, WIND); Kamanjab, dry granite hills in rock crevices, 2 Mar. 1957, de Winter & Leistner 5132 (B, K, M, PRE, WIND); Kamanyab, Mar. 1925, Thorne s.n. (SAM 31741); 20.16 о. Paresis Mts. (AD), Barnard 201 (SAM). GROOTFONTEIN: 19.17 (Tsumeb) farm Auros, slopes behind о E m E. e ~ WIND); Auros Farm, shale hills near the farm NOR 21 Mar. 1988, Goldblatt & Manning 8837 (E, K, M, MO, NBG, PRE, S, WAG, WIND). лен PRECISE LOCALITY: NAMIBIA: Damaraland, 1879, n. (BM); Kunene River banks, Mar. 1925, Bid 1381 (SAM). 13. Lapeirousia bainesii Baker, J. Bot. 14: 8. 1 Südwestafrika 155: 8. 1969 (but applied to L. otaviensis). TYPE: Botswana: Kobe Pan (in- ter Koobie et N Shaw valley), Baines s.n. (lectotype, annotated as “ТҮРЕ” by М. E. Brown, and confirmed here as lectotypified, K); South Africa. Transvaal: Todd s.n. (syn- type, K—as Todd 19). Figure 14. я vaupeliana Dinter, Feddes Rep. 18: 436. 1922. Sólch, Prod. Fl. Südwestafrika 155: 8. 1969. ibi 5 O u. Okanjatu, Dinter 3374 (syntype, SAM) Plants 30-60 cm high, usually repeatedly branched. Corm 13-20 cm diam., tunics middle 470 Annals of the Missouri Botanical Garden FIGURE 14. Mor details of stamens and style x 2. (Drawn by J. C. Ma to dark brown, coriaceous to cartilaginous inter- nally with the veins sharply outlined, decaying to become + fibrous, the fibers wiry and coarse. Cata- phylls 2, pale and membranous, the inner reaching 2-4 cm above the ground. Leaves narrowly lan- ceolate to linear, glaucous, 5-7 mm wide, usually slightly longer than the inflorescence, midrib prom- inent and a lateral vein on either side of the midvein evident. Stem compressed and 2-winged below, 3— 4-angled and winged above, the wings sometimes slightly crisped to serrulate. /nflorescence a pseu- dopanicle, the branching divaricate, ultimate branches with 1(-2) flowers, the axes angularly trigonous; bracts 7-10(-12) mm long, herbaceous below, dry before anthesis, becoming membranous, dry and light brown especially above and the apices often darker brown, the inner slightly larger than the outer. Flowers zygomorphic, white to cream, 11 | 1) | | ; i + ES - a | Р A | ч A TELA H Tt SN LAM Nd ^s VI ы. M ү” | Ме S> | p i— | | < s Kk t L | | М] \ ey ha к= A 2 ШР n EE | [ e. EN M | | | | I—31——— 1 — | | e -4 \ | | | | pea — ALA J | YA [| Lo A GN MN E | | | | | y т | 2 ж Ф e p in and distribution of Lapeirousia bainesii. Habit and corm x 0.5; single flower full size; nning.) sometimes flushed pale pink, the tube pale purple, the lower 3 tepals each usually marked with a red to brown streak in the lower half and a dark mark at the base, the top of the throat red on the lower side, sweetly scented and opening in the mid to late afternoon; perianth tube cylindric, slightly expanded in the upper 5 mm, 25-34(-40) mm long, usually inclined; tepals subspathulate, widest in the upper third, + clawed (rarely with a toothlike callus on each of the 3 lower tepals), the margins undulate, nearly equal in size or the uppermost often slightly larger, 9-12(-15) mm long, 3-4 mm wide, the lower 3 closer together, forming a lip, the uppermost often slightly larger than the others and held apart, + upright to reflexed at right angles to the tube. Filaments unilateral, exserted for 3.5- 5 mm, white; anthers parallel and contiguous, 4.5- 6 mm long, light purple; pollen cream. Ovary ca. Volume 77, Number 3 1990 Goldblatt 471 Systematics of Lapeirousia in Tropical Africa 2 mm long, ovoid; style arching over the stamens, dividing between the middle and apex of the anthers or sometimes exceeding them, branches spreading, virtually undivided or notched apically, ca. 2 mm long. Capsules obovoid to globose, 5-6 mm long; seeds globose, ca. 2 mm diam. Chromosome num- ber 2n = 10, 6. Flowering time. January to April, rarely late November or December Distribution and habitat. A relatively com- mon species of the dry interior plains of southern Africa, Lapeirousia bainesii extends from near Windhoek in central Namibia to the Cunene River in the north, into southern Angola, and across the Kalahari to eastern Botswana and the northwestern Transvaal (Fig. 14). It favors hard flat sandy ground where the corms may be lodged up to 30 cm below the surface. I have never seen it in rocky terrain, where it apparently is replaced by the related and morphologically similar L. otaviensis. The flowers open in the late afternoon and last through the night and into the following day, then they wilt rapidly in the dry, hot daylight conditions. The freshly opened flowers have a strong sweet and pleasant fragrance and produce abundant nectar. Lapeirousia bainesii is probably pollinated by hawkmoths. Like several other species of Lapei- rousia in Namibia (see Ethnobotany), the corms of L. bainesii are reportedly eaten raw or roasted ) by the native population (Rodin, 1985 Diagnosis and relationships. A long perianth tube, whitish perianth with pink to red markings near the base of the lower tepals, and a divaricately branched inflorescence with the branches bearing one or rarely two flowers immediately distinguish Lapeirousia bainesii. The perianth tube is usually (« 20)25-35(-40) mm long. The corm tunics are composed of light brown, densely compacted fibers, the outer layers of which become increasingly re- ticulate with age. These corm tunics, combined with the pale, long-tubed flower and a basic chro- mosome number of x = 5, suggest a close rela- tionship with L. otaviensis and L. schimperi, and to a lesser extent with L. gracilis, which has a different karyotype (Goldblatt, 1990b) but also n — 5. The similarity in the size and general shape of the flowers of L. otaviensis, L. gracilis, and L. bainesii has led to much confusion in the past, but field studies in Namibia have indicated convincingly that these are three separate species with different ranges and habitat preferences as well as slightly different flowers and inflorescence structures (Figs. Specimens from central Botswana near Ma- cheng, Hansen 3357, appear to represent a very odd form of L. bainesii. The flowers have a short tube 14-20 mm long (vs. 25-35 mm long in most other collections of the species). In other respects the plants accord with L. bainesii, including the fairly distinctive perianth coloring. The collection may represent a hybrid population, then most likely with L. sandersonii, the only other species of La- peirousia recorded from this part of Botswana. One of the few collections that I have seen from Angola, Barbosa & Moreno 10181 from Chibia, is unusual in having a wine red perianth tube. The reverse of the outer tepals is a similar dark color, although the condition in the living plants may have been somewhat different. The significance of this variation cannot be properly assessed until more material from Angola can be examined. History. Apparently first collected in 1863 by the nineteenth-century English landscape artist Thomas Baines and named in his honor, Lapei- rousia bainesii was described by J. G. Baker in 1876. The species was based largely on Baines's collection but also on a second gathering from the Transvaal made by a certain Mr. Todd, about whom I have not been able to obtain information. Baines's plants were collected in western Botswana near Kobe Pan, an area still poorly known botan- ically. d ¡had Plants from west central Namibia in 1922 as the separate Lapeirousia vaupeliana by Dinter, based on three of his own collections. The distinction that Dinter made between this and the avowedly related L. bainesii was the red-violet flower with darker veins in his species compared with the white flowers of L. brown markings. These differences do not seem significant and certainly do not merit recognition of L. vaupeliana as a separate species. Lapeirousia otaviensis was regarded as a syn- onym of L. bainesti by Solch (1969) in his treat- ment for the Prodromus Flora von Sudwestafrika. However, he upheld £. vaupeliana, distinguishing it in the diagnostic key by the ultimate branches of the inflorescences each bearing a single flower in contrast to L. bainesii, in which the ultimate bainesii with dark branches are spikes (i.e., with two or more sessile flowers). In fact the type specimen of L. bainesii has single flowers on the ultimate inflorescence branches and is conspecific with the later L. vau- peliana, whereas L. otaviensis, with its 3-5-flow- ered inflorescence branches and different flowers, is distinct from L. bainesii. Additional specimens examined. ANGOLA. HUILA: 16 472 Annals of the Missouri Botanical Garden km from Quihita to Vila de Jodo de Almeida (Chibia), black soil under Acacia, Barbosa & Moreno 10181 (COI, LISC). CUNENE: Namakunde, Barnard 139 (SAM); Rau- tanen 702 (Z). BOTSWANA. NGAMILAND: үз: (Aha Hills) Aha Hills, 110 km W of Nokaneng (CB), Wild & Drum- mond 6919 (BM); Dobe, 26 km N of Aha Hills, SWA border (CA), Wild & Drummond 7202 (К, PRE); 20.21 (Koanaka Hills) near SWA border fence, sandy grassland, (2), Smith 3330 (MO); 20.23 (Kwebe Hills) Ngamiland, Lugard 179(GRA, K). CENTRAL: 20.25 (Mompse) Odiakwe, savanna, W of Francistown just N of Mkarikari Pan (AB), Wild & Drummond 6826 (К); 21.25 (Loth- lekane) Ngamiland, near Bachakuru, white and chocolate (DD), Lugard 242 (K); 23.26 (Mahalapye) Mahalapye Exp. Morale, shallow gritty sandy loam (BB), Yalala 356 (K, LISC, PRE); Mahalapye, Camerik 215 (PRE); 23.27 (Ellisras) 10 mi. W of Macheng towards Mahalapye, wood- land, 23°10” 27°20” (AB), Hansen 3357 (С, К, PRE, WAG). KGATLENG: 24.26 (Mochudi) Mochudi (AC), Har- bor sub Rogers 6569 (G, K, PRE); Harbor s.n. (PRE 14060). NAMIBIA. ETOSHA: 18.14 ee Etosha Na- tional Park, S of Okawao, red chalky sand (DD), Giess & Loutit 14195 (WIND); e : (Kamanjab) Etosha Na- tional Park, Kaross (B), k & Le d 808 (PRE, WIND). OKAVANGO: 17. or u), 14 i. E of Runtu on Sambiu road (DD), de Winter & n 4558 (К, ster 2788 (PRE); 18.19 (Karakuwise) Cigarette, NE of Сз чирли че ics Magutre ris (NBG), 2381 ed PRE); 19.20 (Tsumkwe) 300 m S of main Tsumkw igi MA road, border of iu Dept. farm (CB), Hines 363 (WIND); 6 km E of Tsumkwe on the road to Botswana, white sand (DA), Giess et al. 11033A (WIND). OVAMBOLAND: 17.15 (Ondangua) Odanga, no date (DD), Barnard 195 (SAM), 199 (SAM); 100 km E Oshikango, corm eaten raw or roasted (BD), Rodin 9295 (K, M, MO, PRE, WIND). Kaprivi: 18.21 (Andara) Bagani Camp, Kaprivi side of river (BA), de Winter & Wiss (PRE). GROOTFONTEIN: 19.16 (Gobaub) farm Norabis 387. W of Otavi, thornveld on red sandy loam (DD), он & Manning 8826 (MO, PRE, WIND), 7 (Tsumeb) Otavi (CB), Dinter 5755 (B, G, PRE, Z); e Otjirukaku (DB), Seydel 2068 (B, BR, C, ЄН, К, M, M > AG, WIND); 19.18 (Grootfontein) farm Olie weniof. sehr háufig auf der Palmflache (CB), MOD & Giess 30153 (K, M, WIND, PRE, S, SRGH, WAG); 20.16 (Otjiwarongo) farm Wittenberg 90, red sand, Gold- blatt & Manning 8832 (MO). oKAHANDJA: 21.16 (Oka- handja) Omatako View (BA), Woortman 152 (M, WIND); 20.17 (Waterberg) Quickborn, sand (AA), Bradfield 196 (K, PRE). ourjo: 20.16 (Otjiwarongo) Outjo (AA), Rau- tanen. 389 (Z). WINDHOEK: 21.17 (Otjosondu) 48 km along Kapps Farm road from Steinhausen, compact sand (DC), и & Manning 8808 (Е, К, М, MO, NBG, WAG ND). GOBABIS: 21.18 (Steinhausen) deep a sand, fum Mex, pink to white, (AB), /mmelman 526 (K, PRE); farm Mex, 45 km N of Witvlei, red sand, Germishuizen 2662 (PRE, WIND); 21.19 (Epikuro) Epi- kuro Reserve 30 mi. NE of Epata, Eiseb Omuramba Giess 9746 (WIND); 23.18 (Leonardville) sandfeld, Ga- moros (cult.) (AB), Dinter 2788a (SAM). KEETMANSHOOP: 25.18 (Tses) Kalaharirand, Tutara, farm Okamatangara (DB), Seydel 2549 (BR, Y: E MO, SRGH, т сш ÁFRICA. Т e 9 (Waterpoort) Dong Re- serve, 10 bs SE Reserve, А Coila (AD), € odd P Dyer 3780 (E, К, PRE); Dongola, Zoutpansbere (BC), Pole Evans un (K, MO, PRE, SRGH); Soutpansberg, 1.5 > > NW of Wyliespoort, 2,800 ft., Codd 8367 (PRE); 23. 27 (Ellisras) Potgietersrust District t, 2 m NE Tom- burke, red gritty flats (BB), Codd 6614 (K, PRE, SRGH); mi. E of Ellisras, cultivated land (DB), Louw 3506 (AAU). WrrHouT Precise LocaLiTY: BOTSWANA: Bak- wena Territory, Sirorume River S of Tropic of Capricorn, (K Holub s.n 4. Lapeirousia schimperi (Aschers. & Klatt) Milne-Redhead, Kew Bull. 307. 1934. Que- zel, Fl. Veg. Pl. Darfur & Jeb Gourgeil (Dos- siers Rech. Coop. Prog. 45: 134. 1969. Cu- fodontis, Enum. Pl. Ethiopiae Sperm. 2: 1592. 1972. Wickens, For. Bull. new ser. 14: 39. 1969; Fl. Jebel Marra 158. 1976. Tritonia schimperi Aschers. & Klatt, Linnaea 34: 697. 1866. Schweinfurth, Beitrag Fl. Aeth. 1867. [ Acidanthera unicolor Hochst. ex Baker, J. Linn. Soc. Bot. 16: 160. 1878; Handbk. Irid- eae 188-189. 1892; Fl. Trop. Africa 7: 359. 1898, nom. superfl. pro Tritonia schimperi Aschers. € Klatt (1866) (Schimper, Plantae Abyssinicae 2304).] TYPE: Ethiopia. Tigray: woods and thickets near Goelleb on the river Tacazze, 4,000 ft., Schimper 2304. (lecto- type, B, here designated; isolectotypes, K, P); Yemen: Schimper s.n. (syntype, B?, not seen and perhaps lost). Figure 15. Lapeirousia angolensis (Baker) R. Foster, Contr. Gray Herb. olensis Baker.] TYPE: Angola. Without precise locality, cult. Kew, Monteiro s.n. (holotype, K) Lapeirousia fragrans Welw. ex т Trans. Linn. Soc. London (Bot.) ser. 2, 1: 272-273. 1878. TYPE: Angola. Huila: ad n + Lopollo, 5,200 ft., stony 59, Welwitsch 1552 (ho- ^ ү: COI, G, P). Lapeirousia cyanescens Welw. ex Baker, Trans. Linn о ndon (Bo 2, 1: 272. 1878. Solch, Pr Südwestafrika 155: 9. 1969. Cufodontis, Enum. Pl. Ethiopiae Sperm. 2: 1592. 1972. TYPE: Angola. Huila: hills near Humpata, Apr. 1860, Wel- witsch 1553 (lectotype, K, here designated; еня types, B, BM, COI, С). rb. Boissier 4 (Ap- Amboland, W Lapeirousia edulis Schinz, bis He x З PE: Namibia: А of н 5 йш. 1893, Rautanen 106 T type, Z, here designated; isolectotypes, K, P(2,Z Lapeirousia porphyrosiphon Baker, Fl. Trop. Africa 7: 353. 1898. TYPE: Botswana. Ngamiland: Kalahari Desert near Mamunwe, 26 Feb. 1897, Lugard 338 holotype, K). usw erythreae Chiovenda, Ann. Bot. Roma 9: 139. 1911. TYPE: Ethiopia. Eritrea: Acchelé Cuni s ege near Loggo Sarda, Deggahen, 2,600 ‚ 15 Sep. 1902, Pappi 1414 (lectotype, E pres ынена о GH, МО); Bogos near Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa 473 ft | § AAN | AU ) UN | AH] d Pa a Жы * LT | = | po j 53 me | | mo | | і RH == EM ___4— ( | \ \ | | | | ү ү E ao Эи a. —M— о 2001400 $00 800 1000 км ——— Я 20 to Y FIGURE 15. with a ? Purus verification and rem and style x 2.5. (Drawn by J. C. Manning. ) Cheren, 24 Aug. 1903, Pappi 7088 (? a not seen); Dembelas towards the Mai Albo, 25 Se 1903, Pappi 6072 (? FI, not seen). ч montaboniana hide Ann. Bot. Rom . 1911. TYPE: Ethiopia. Eritrea: Bogos near Lo 24 Aug. 1906, Pappi 7087 (? FI, not see Lapeirousia uliginosa Dinter, Die vegetabilische Veld- ost Deutsch-Súdwest-Afrikas 13-14. 1912. TYPE: none cited. Lapeirousia dinteri Vaupel, Bot. Jahrb. Syst. 48: 544 45. 1912. ТҮРЕ: Namibia: Damaraland, доа Morphology апа е of Lapeirousia schimperi. Stations in Malawi п doubtful. Habit and corm x 0.5; and Yemen indicated single flower full size; details of stamens Kreyfontein, 28 Dec. 1908, Dinter 810 (holotype, B; isotypes, B, SAM (2)). Plants (20-)30-80 cm high, usually several- branched. Corm 18-22 mm diam. at the base, tunics composed of compacted fibers, light to dark г the outer layers becoming loosely fibrous and reticulate. Leaves linear, 3 or more, the lower 2 largest and usually slightly longer than the in- florescence, decreasing in size above and becoming bractlike, narrowly lanceolate, 5-10(-15) mm wide, Annals of the Missouri Botanical Garden the midrib lightly raised. Stem rounded below, to nearly square and 4-angled to 4-winged above. Inflorescence a lax pseudopanicle, the ultimate branches with 1-3 sessile flowers; bracts + equal, (10-)20-35(-45) mm long, herbaceous becoming membranous above to almost completely dry and papery, then light to dark brown, apices dark brown. Flowers zygomorphic, white to cream, rarely pale violet, when whitish sometimes fading or drying lilac especially on the tube, opening in the evening and then sometimes scented; perianth tube cylin- dric, slender, 10-14(-15) em long; tepals lanceo- late, extended + at right angles to the tube, 1 22 mm long, 6-7 mm at the widest. Filaments unilateral, erect, exserted 5-7 mm from the tube; anthers parallel and usually contiguous, 6-7 mm long, cream; pollen cream. Ovary ca. 4 mm long, rapidly elongating after fertilization; style filiform, arching behind the filaments, dividing at or up to 3 mm beyond the anther apices, branches ca. 2 mm long, forked for ca. Y4 their length. Capsules obovoid-oblong, 8—12 mm long, partly enclosed in the bracts; seeds globose to slightly angular, some- times tapering to a persistent funicle, 2— diam. Chromosome number 2n = Flowering time. Mid to late summer, Decem- ber to March, in central and south tropical Africa; north of the equator mostly in September and Oc- tober, in Kenya and Ethiopia also April to June. Distribution and habitat. Lapeirousia schimperi has a remarkably wide distribution across Africa (Fig. 15). It extends from southwestern Ап- gola, north-central Namibia, and northern Botswa- na through southern Zambia to Zimbabwe in an Further north its distri- bution is scattered, and it occurs in northern Tan- almost continuous belt. zania, eastern Kenya, southern Ethiopia, disjunctly in northern Ethiopia, and the Jebel Marra and Jebel Gurgeil highlands of northwest Sudan. It is almost certainly absent from Zaire, Uganda, central and western Kenya, and southern Tanzania, but the species probably occurs in western Somalia (al- though I have seen no records from there). A plant from Yemen is cited in the protologue, but | have seen no collections from this country. I have also had seeds of plants said to have been collected on the Nyika Plateau in Malawi. The Malawi and Yemen records require confirmation. Despite the questions about the distribution of L. schimperi, there is no doubt that it is absent from large parts of tropical Africa, and its unusual discontinuous range is noteworthy. Lapeirousia schimperi grows in moist situations in otherwise largely arid country. It typically occurs in washes, the so-called omurambas of Namibia, dambo margins, stream sides, seasonal marshes, and in damp grassland. The flowers open in late afternoon and last through the night, but gradually wilt the next day. Some collections note a sweet fragrance when the flowers are open, but live plants from Namibia that I have examined were scentless, suggesting variation in this character. The edible corms are eaten either raw or roasted by the Ovam- bo of northern Namibia (Rodin, 1985) and the Kung Bushmen (Marshall, 1976). Diagnosis and relationships. The large white flower with a perianth tube 10-15 cm long and the laxly branched inflorescence distinguish Lapei- rousia schimperi readily from most other species in the genus. The paniculate inflorescence, plane leaves with a well-defined midrib, and corm tunics of compacted fibers place it in sect. Paniculata. It has no obviously close relatives in this alliance, composed largely of plants with small flowers and typically multibranched pseudopanicles with crowded flowers. Its basic chromosome number, л = 5, and karyotype (Goldblatt, 1990b) and usually light brown, fairly fibrous corm tunics are shared with L. otaviensis and L. bainesii, suggesting that they may be the closest allies of schimperi. Those two species, centered in northern Namibia and adjacent Botswana, also have pale flowers with relatively long perianth tubes, but their inflores- cences are more compact, and they are not likely to be confused with L. Unusually long-tubed white flowers characterize schimperi. two other tropical African species of Lapeirousia, both of which have corrugate leaves and woody corm tunics, which L. odoratissima and L. schinzii, place them in sect. Lapeirousia; the floral simi- larity must be due to convergence. Presumably all are pollinated by hawk moths. Lapeirousia schimperi exhibits ey В variation for a species with such а Some specimens from Angola and Namibia in the southwest of its range have unusually short bracts in the 15-20 mm range (e.g., Giess & Loutit 14123), but at least some plants from Namibia have bracts up to 35 mm long, thus comparable to those in plants from tropical and northeastern Africa in which the bracts range from 35 to 45 m. Some collections from Ethiopia and Sudan (e.g., Schimper 2304, 431) also have pale or fairly dark brown, particularly short bracts 10-15 mm long. Their flowers have a perianth tube 7-8 cm long. Such plants may be depauperate owing to a par- ticularly dry growing season, or they may represent Volume 77, Number 3 1990 oldblatt 475 Systematics of Lapeirousia in Tropical Africa an ecotype or race that constantly has these fea- tures. History. Although a well-defined species, Lapeirousia schimperi has a long and confused history. This is largely due to its wide distribution rather than to any conviction that its synonyms were species distinct from those from distant parts of its range. The type locality is in northern Ethio- pia, where collections were made in the 1840s by Quartin Dillon and Petit and later by Schimper in 1852-1864. The species was named in Schimper’s honor by Ascherson & Klatt (1866), who referred it to the otherwise southern African genus Tritonia. Three separate collections from southwestern An- gola made at this period by Joachim Monteiro and Friedrich Welwitsch were described by J. G. Baker as Anomatheca angolensis (1876), L. cyanescens (1878), and L. fragrans (1878), respectively, with- out reference to the Ethiopian species. Later col- lections from Namibia were named independently by Hans Schinz as L. edulis in 1896 and Friedrich Vaupel as L. dinteri in 1912. Plants from Botswa- na, collected by E. J. Lugard, were assigned to L. porphyrosiphon (Baker, 1898), while in 1911 A. Chiovenda described L. erythreae and L. monta- boniana based on contemporary collections from Eritrea. Only in 1934 was Tritonia schimperi transferred to Lapeirousia, but for many years the name L. porphyrosiphon was used for the species in south tropical Africa. Additional specimens examined. ANGOLA. CUANZO SUL: road from Nova Redondo to Lobito, 60 m, 17 Apr 1969, Teixeira et al. 11412 (LISC). CUNENE: Rocades, centro do Estudos do Cunene, da Silva 2991 (BR, К, onnefoux & Villain 58 (Р). HUAMBO: Cuima, Eleude Mission, Dec. 1940, Faulkner 4390 (К, PRE); road from Nova Lisboa to Luanda, Serra do Cus- sava, 26 Dec. 1970, Moreno 311 (LISC, M); vicinity of Nova Lisboa near Cruzeiro, banks of Cuando River, 26 . 1971, da Silva 3522 (LISC, PRE); Chianga, ca. 1,700 m, 5 Mar. 1962, Teixeira & Sra 6542 (COI, т HUILA: hills near Humpata, Feb. 1899, Antunes 24 (P), Mar. 1902, DeKindt 3213 (449) (LISC, P); m 14°50” 13°22”, 2,230 m, 7 Mar. 1973, Bamps et al. 4057 (BR, K, LISC); Guilengiós-Chinsoral. 5 Dec. 1962, de Menezes 384 (К, LISC, PRE, SRGH); Chitanda munding, 1,100 m, 28 May 1905, Baum 949 (BM, COI, E, G, K, M, S, Z); Caconda, Mar. 1880, Anchieta 15 (LISU); Mar. 1882, Anchieta 34 (LISU); 882, Anchieta 131 (LISU); Ganguelas, Vila Artur de i Cutato, Mendes 3329 (LISC); Vila Artur de TUS ie de Ponte), granja da administracao, 1,470 m 1 960, Mendes 1919 (LISC); near Huila, Me мее Жн. Sá da Bandeira, Torre 8629 (LISC). NAMIBE: hills on road between Namibe (Mossamedes) and St. colau, Kers 3646 (PRE). m NA. NGAMILAND: 17. 25 ivi of C i above ne, May 1972, Sheppe 167 (БЕСН); 18. 21 (Andara) SW end of Tsodilo Hills, open savanna, 11 Mar. 1985 (DD), Long 12330 (E); 19.23 (Maun) Khwai/Maxwe road, Moremi Wildlife Reserve, 16 Mar. 1977 (BC), Smith 1936 (BR, К, SRGH). ETHIOPIA. SIDAMO: near the Sagan River, ca. 500 m, 37°45” 4°47”, Acacia woodland in grass, 27 May 1974, Sandford sub Ash 2498 (K). TIGRAY: near Goelleb on River Tacazze, 4,000 ft., 28 Aug. 1854, “А. unicolor’ Schimper 2304 (BM, К, Р). GONDER: Simen, Dsha Dsha, 22 Aug. 1853, Schimper 431 (P). KENYA: Somali border, open grasslamd, 23 Aug. 1961, Gillespie 250 (BR, K, 1952, Gillett 13426 (K). NAMIBIA. OVAMBOLAND: 17.15 (Ondangua) 100 km E of Oshikango, 19 Apr. 1973 (BD), Rodin 92954A (M, MO, PRE, WIND). OKAVANGO: 17.18 (Kuring Kuru) Katui-Tui, maize land in sandy loam, 15 May 1965 (fr) (AD), Вагпага 191 (W Sani s s.n. (WIND 3938); 17.19 (Runtu) swampy marshes below Runtu, 31 Jan. 1956 (AD), de Winter & Marais 4469 (К, М, PRE, WIND, Z); 18.19 (Karakuwisa) grass flats, Karakuwisa, 4 Mar. 1958 (DC), Merxmuller & Giess 1795 (BM, K, M, PRE, WIND); 18.21 (Andara) Kaprivi side of the river at Andara Mission, crevices on rocky outcrops, 23 Feb. 1956 (AB), de Winter & Marais 4820 (K, PRE, WIND); Okavango River, 19 km N of Shakawe on the Botswana border, 16 Mar. 1965 (BA), Wild & Drummond 7093 (K, LISC, M, PRE, SRGH). KAPRIVI: 17.24 (Katima Mulilo) Katima Mulilo, vlei, 30 Jan. 1975 (AD), Vahrmeijer & du Preez 2496 (MO, PRE). ErosHa: 18.16 (Namutoni) Etosha National Park, Bigales Huh, 15 Feb. 1974 (C), Le Roux 644 (PRE, WIND); 19.15 (Okakuejo) Etosha Pan, black peat soil E of Okakuejo-Ombika road, 5 Mar. 1976 (BD), Giess & Loutit 14123 (K, M, MO, PRE, WIND); gray-black peat flats N of Ombika, Le Roux eer A WIND); 19.16 (Gobaub) Etosha Game Par r Homob water hole, 11 Feb. 1966 (AA), Tinley 1285(M. PRE, WIND). GROOTFONTEIN: 19.16 (Gobaub) farm Neidaus North 78, heavy black clay with limestone, seasonally waterlogged, 19 Mar. 1988 (DC), Goldblatt & Manning 8831 (MO, WIND); 19.17 (Tsumeb) farm Malta, 5 Feb. 1971 (AB), Giess 11226 (M, PRE, WIND); farm Kumkauas, large Mors among tough grasses, 30 Jan. 1971 (CA), Giess p WAG, WIND); 9 Mar. 1974, Merxmüller 0175 (K, M, PRE, S, SRGH, WAG, WIND). SUDAN. DARFUR: Jebel Marra, E of Zalinjea, 3,500 ft., poorly I 13 Aug. 1964, Wickens 2101 (K); Suni uora Tanje, 7,800 ft., moist oar 21 Sep. 1964, Wick- ens 2695 (K); Nyertete, 3 Wickens 2126 (К); ;ur Lambang, lava soils, 6,300 P ‚ 17 Sep., Wickens m Taurotonga to Kilokitting, ,000 m, basalt, 15 Sep., Jackson 4073 Euh ARUSHA: Varas dapi vig Hills, 70 mi sha, 3,500 ft., Mar. 1967, Beesley 265 (BR, K); X District, Tarangire National Park, 1, m, 14 Fe 1970, penty 254 . DODOMA: Great North Road, Kalo, 15 mi. N of Ko ndoa, 5,050 ft., black clay in vlei, 11 Jan. 1962, Polhill & Pau lo 1132 (BR, K, LISC, P, PRE). Mwanza: Mwanza, s.d., Davis 180 (К). SHINYANGA: near Shinyanga, Jan. 1933, Bax 399 (BR, K) Huru- huru- Mantini road, Shinyanga, 3,800 ft., 16 Jan. 1932, Burtt 3511 (BM, BR, K); Nindo, Jan. 1972, Stefanescu 125 (K). ZAMBIA. SOUTHERN: Livingstone, 20 Jan. 1929, 476 Annals of the Missouri Botanical Garden Grant 4507 (MO, РКЕ); Victoria Falls road, Livingstone, 20 Jan. 1919, Young 17315 ); Kafue basin, Monze near Lochinvar R xed Acacia woodland, 31 Jan. basin, ice Ca: margins, 9 Ja Fanshawe 6098 (BR, K, NDO, SRGH). ZIMBABWE. MASHONALAN CENTRAL: N of Sipolilo, 1948, Whellan 31 3 (K, LISC). MASHONALAND sub Moss 17315 (K). ict, Mense Pan, 11 Di strict, mi. ESE of Косу oi mad 5340 d BR, SRGH); Jan. 1906, 256 (K, B Jan. 1910, Rogers ine > (K, SRGH, WAG, 4); Wankie National du Sina- tella Dam ca. 1 mi. from the ca 24 Feb. 1967, Кы! 174 (MO, SRGH); Wankie, ee Camp, Kalahari sand, 27 Feb. 1967, Rushworth 269 (BR, K, LISC, PRE, SRGH); Wankie, Shapi road, in vlei, 15 Feb. 1956, Wild 4747 (К, PRE, SRGH); Insiza District, Shan- gani, farm Bon Accord, pu in granite whalebacks and uA black vlei soil, 26 Jan. 1976, G RGH); Nyamandhlovu, БНА Station, nie 1661 (K, LISC). MATABELELAND SOUTH: Matobo District, 3 Feb. 1948, West 2678 (MO, SRGH); 7.2 km S of Bulawayo Post Office, S side of Johannesburg road, pus woodland, 22 Jan. 1976, Cross 340 (K, MO, PRE, SRGH). MIDLANDS: Que Que to Gwelo, main road, Davey 1 (K, PRE, SRGH). WirHouT PRECISE LOCALITY: ?ANGOLA: 1878, Capello 20 (LISU); Welwitsch 4108 (P). BorswANA: Ngamiland, Jan. 1931, Curson 4 75 (PRE) Erniopia: Tigre & Begemdir, 21 Aug. 1862, Schimpe 909 (BM); Habab, 5,000 ft., Sep. 1872, Hildebrandt 374 (BM); Abba Heruke, 6 Aug. 1852, Schimper s.n. or 431 (P); Masser, 4 Sep. 1853 (fr), Schimper s.n. or 431 (P). Namibia: Kapichu, Mar. 1923, Barnard 137 (SAM). SUBGENUS LAPEIROUSIA SECTION SOPHRONIA (LicHT. EX prem Е SCHULTES Согрві.. & MANNI 15. Lapeirousia littoralis Baker, Trans. Linn. Soc. London, Bot. ser. 2, 1: 273. 1878. TYPE: Angola. Namibe: sandy coastal hills ad Praia da Amelia prope Villa de Mossamedes, July 1859, Welwitsch 1546 (holotype, BM discussion of the type specimen). Synonyms are listed under the two subspecies. Plants (5-)10-35 cm high, simple or often with 1 -few(or many) branches, these either long or clus- tered near the base. Corm campanulate, 10-14 mm wide at the base, tunics woody, brown, the outer layers breaking irregularly, rarely becoming fibrous, the basal margin crennate or rarely bluntly denticulate. Cataphylls usually 2, the outer one short and the inner reaching almost to ground level, membranous. Leaves few to several, linear, 1.5- 4 mm wide, lightly corrugate, the lowermost in- serted at or just below the ground and longest, often as long as the inflorescence or rarely some- what longer, ascending to aia or trailing, upper leaves progressively shor Stem simple ог branched, the branches ine crowded below or laxly arranged, + rounded to weakly angled below the nodes. Inflorescence comprising |~several lax to fairly congested branches of 5-12 flowers; bracts herbaceous, weakly keeled, the outer 10-20(-25) mm long, the inner smaller, becoming membra- nous, apically forked. Flowers zygomorphic, white to cream, greenish yellow, or light purplish brown, with a strong sweet fragrance; perianth tube + dimorphic, slender below, curving outward and ex- panded above, (20-)30—45(- 70) mm long, the up- per part ca. 6 mm long; tepals subequal, narrowly m long, 1.3- 3) mm wide, acute to attenuate, spreading lanceolate to linear-filiform, 13-30 mm equally at right angles to the tube, vertical. Fila- ments unilateral and arcuate, exserted 2-3 mm; anthers parallel, 4-5 mm long, cream; pollen cream. Ovary globose, ca. 3 mm long; style arching behind the stamens, branching between the base and middle of the anthers, the branches divided for about % their length, ascending below, recurved above and usually tangled in the anthers. Capsules globose, 6-8(-11) mm long; seeds globose, ca. 2 mm diam. Chromosome number 2n — Flowering time. Usually late December to March in tropical Africa (subsp. caudata and subsp. littoralis), September and October in southern Na- mibia and South Africa (subsp. littoralis) but also at other times. Distribution and habitat. toralis occurs in a wide swath across south tropical Lapeirousia lit- Africa (southwestern Angola, northern Namibia, Zambia, Zimbabwe, Botswana, southern Mozam- bique) with southward extensions into the arid parts of southern and western Namibia and the northern Cape and Namaqualand in South Africa (Fig. 16). It blooms in the wet season, usually not long after the first soaking rains. Thus the Namaqualand and southern Namibian populations usually flower in the spring (August to October) following the winter rainfall that prevails along the southwestern Afri- can coast. Elsewhere populations generally flower in early to late summer, not long after the beginning of the summer rains of tropical and eastern south- ern Africa. In southern Mozambique populations appear to bloom in almost any month, reflecting the nonseasonal rainfall pattern of the southeast coast. Lapeirousia littoralis favors sandy, well- drained soils, sometimes occurring on sand dunes. The probable pollinators are moths, given the pale- Volume 77, Number 3 1990 Goldblatt 477 Systematics of Lapeirousia in Tropical Africa colored, very fragrant, long-tubed flowers. How- ever, the style branches are tangled with the an- thers, and self-pollination seems probable in the absence of insect-mediated pollen transfer. Diagnosis and relationships. Characterized by pale, uniformly colored flowers with a long, dimorphic perianth tube that curves outward and is expanded above, Lapeirousia littoralis can usu- ally be recognized by its flowers alone. Its vege- tative and floral morphology are remarkably vari- Zimbabwe, and Mozambique have relatively long, sometimes lax spikes with short bracts 10-18 (-23 mm long and are usually 20-30 cm tall. The flowers in all collections from these relatively well- watered areas have long, narrow tepals 25-30 mm long and about 1.3 mm wide. As the flowers fade and dry, the tepals become distinctively filiform. Plants from arid southwestern Angola, western and southern Namibia, and from the northwestern Cape and Namaqualand in South Africa tend to be shorter, often having the branches crowded at the base. In these plants the spikes are shorter, usually 10-15 cm high, and sometimes congested, and the bracts are usually 15-20(-25) mm long. The flow- ers in these populations have tepals about 20 mm long and 2-2.5 mm wide. When dry they do not always assume the filiform shape of the northern populations. These two series of populations are usually easy to separate and might be regarded as different species except for the morphologically intermediate populations in central Namibia and southern Bo- tswana that are often not easily assigned to either major group. The intermediates suggest that sep- aration at subspecific rank is the most appropriate treatment for the southern and tropical populations of the species. The shorter form, first described by J. G. Baker in 1878 as L. littoralis, later as L. burchellii (1892), and then by Dinter as L. ra- mosissima, was first regarded as a subspecies of L. caudata by Goldblatt and Marais (Goldblatt, 1972), essentially for the reasons outlined above. Lapeirousia streyi from the Namib Desert south of the Kuiseb is regarded as belonging here, but it is close to being intermediate between the shorter subsp. littoralis and the taller subsp. caudata. Plants from southern Mozambique, mostly from immediately around Maputo (Lourengo Marques), seem best treated as belonging to subsp. caudata. They stand out in having particularly long-tubed flowers and in often being very robust. The tubes are sometimes up to 7 cm long, compared with the 5-4 cm usual in central African plants. Floral FIGURE 16. subsp. littoralis represented by triangles, subsp. caudata by dots. Distribution of Lapeirousia littoralis: variation is notable also in plants from western ambia, and a collection from Mongu (Robinson 6749) has flowers with perianth tubes 8-13 mm long and tepals ca. 18 mm long. Such isolated variation, particularly in tube length, also occurs in other species of Lapeirousia and appears to have no taxonomic significance. Lapeirousia littoralis has a diploid number of 2n = 16 and a dimorphic karyotype of one very long and seven short chromosome pairs. This is characteristic of subg. Lapeirousia, a largely west- ern Cape and Namaqualand alliance comprising 17 species (Goldblatt, woody corm tunics, and spicate or tufted inflores- cences. There seems no doubt that L. littoralis belongs in this section despite its predominantly tropical distribution. The tufted tropical species, L. odoratissima, which has a larger flower with a 1972) with corrugate leaves, perianth tube 10-14 cm long, seems most likely the closest relative of L. littoralis, an assumption based largely on the similarity of the flowers of the two species. Other species of sect. Lapeirousia that seem morphologically close to L. littoralis include the South African L. arenicola and L. anceps, both of which have long-tubed flowers with narrow tepals. In these two species the perianth has ier: markings on the lower tepals unlike L. littorali History. The fruiting type collection of Lapei- rousia littoralis was made by Friedrich Welwitsch in July 1859, and was described by J. G. Baker in 1878, based solely on this collection. The t locality is near Mossamedes on gravelly hills around Praia da Amelia in southwestern Angola, now the province of Namibe. The type material is in poor condition and consists of depauperate plants with 478 Annals of the Missouri Botanical Garden branches produced from near the ground, bracts ca. 12 mm long, capsules ca. 9 mm long, and linear, lightly corrugate leaves. Alone the specimen cannot be matched with confidence, and until now L. littoralis has not been identified with any known species of the genus. However, another collection from southwestern Angola, Torre 8824, previously assigned to L. caudata, comprises plants that are clearly the same as those collected by Welwitsch, and i have flowers that correspond to the south- n African and Namibian L. caudata subsp. bur- chellii (Goldblatt, 1972). The latter taxon, based on Lapeirousia bur- chellii, the English botanist and explorer William Burchell reached the northern Cape. Burchell's collections were described formally by J. G. Baker in 1892 and were not considered to be a subspecies of L. was first recorded in October when caudata, the e used until now for L. littoralis, until later (Goldblatt, sissima, collected near Grúndorn in southern Na- mibia by Dinter and described by him, matches closely the type collections of L. burchellii and L. streyi from west- 1972). Lapeirousia ramo- littoralis. A third species, L. central Namibia, is close to being intermediate be- tween L. littoralis and plants from tropical Africa that have somewhat longer and narrower tepals and taller stems. These characteristics correspond to L. caudata, which was based on plants collected at Olukonda in northern Namibia and described by Hans Schinz in 1890. Plants matching L. caudata are now known to occur widely across south-central Africa from northern interior Namibia, across Zam- bia to Zimbabwe, with an isolated series of popu- lations in southern Mozambique. Currently regard- ed аз L. caudata subsp. caudata (Goldblatt, 1972), it now becomes L. littoralis subsp. caudata. The history of this subspecies is discussed below. KEY TO THE SUBSPECIES OF LAPEIROUSIA LITTORALIS la. Perianth tube 28-35 mm long; tepals 13-15 mm long y 2-3 mm wide; bracts -2 (725) mm long „sooi 15b. subsp. littoralis lb. Perianth le К 25-)30-45(-70) mm jen tepals 8-30 mm long E .3-2 mm wide; bracts mm long ............... 1 10-18(-23) m 9a. subsp. caudata 15a. Subsp. caudata (Schinz) Goldbl., comb. et stat. nov. Lapeirousia caudata че hinz, Verh. ot. Vereins. Brandenburg 31: i Baker, Handbk. Irideae 172. 1892. Sélch, Prod. Fl. Súdwestafrika 155: 10. 1969. Gold- blatt, Contrib. Bolus Herb. 4: 30-33. 1972. TYPE: Namibia: Amboland, Olukonda, Rau- tanen 2 (lectotype, Z, designated by Solch on the sheet; isolectotype, Z); Olukonda, Schinz 2, 15, or s.n. (probable syntypes, G, K, Z). кые delagoensis Baker, Handbk. Irideae 171- 172. 1892; Fl. Cap. 6: 94. 1896. TYPE: Mozam- bique: ied a Bay, Lourengo Marques, sandy places, H. Bolus 7618 (lectotype, K, here designated; is- olectotypes, BOL, G, SAM). "e lacinulata acis Bot. Jahrb. Syst. 546-547. 1912. : Zambia: Kantanina Hills Kassner 2170 (Кеш В; isotypes, BM, BR, E, HBG, К,Р, 7). Plants 20-30 cm high, usually with а few long branches produced from near the base. /nflores- сепсе a spike of 8-12 flowers; bracts 10—18(-23) mm long. Flowers with a perianth tube (25-)30- 45(-70) mm long; tepals (18-)25-30 mm long and 1.3-2 mm wide, usually + filiform, especially when dry. Distribution. Subspecies caudata has a wide range across south-central tropical Africa (Fig. 16). It extends from northern Namibia across western and southern Zambia to Zimbabwe. There is also a series of isolated populations in southern Mozam- bique. History and typification. Subspecies cau- data was first gathered in 1885 at the Finnish Mission Station at Olukonda in northern Namibia by the Swiss botanist Hans Schinz. Schinz later described it in 1890 based on his and the Finnish missionary Martti Rautanen's ample but confus- ingly labeled collections at the Zurich Herbarium. sheet collected by Rautanen in 1887 was des- ignated the lectotype by Sólch in 1959 and this choice appears to be suitable. orresponding closely with Lapeirousia cau- data, although from northern Zambia, acin- ulata was collected by T. Kassner in 1906 and described by Vaupel in 1912. He distinguished L. lacinulata on the basis of its slender habit, low stature, and shorter perianth tube only 2.5 cm long (actually 2.5-2.8 cm), in contrast to L. caudata in which the perianth tube is 3-4(-7) cm long. 'The distinction seems minor and is not sufficient for the separation of the species. А few other col- lections of subsp. caudata have a similar short perianth tube but not the slender habit. The isolated Mozambican populations near Maputo were dis- covered by the missionary Henri Junod in 1890 and subsequently by the Cape botanist Harry Bolus in 1886. Bolus's collection was referred by J. G Baker (1892) to the new species L. delagoensis. Additional specimens examined. BOTSWANA. NGA- MILAND: .23 (Siambiso) sandy floodplain of Kwand River, 18°5” 23°20” (AB), Smith 2222(K. PRE, SRCH). Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa 479 KGALAGADI: 23.20 (Ukwi) 120 km WNW of Hukuntsi on track to Ncojane, 23°41” 20°47”, open sandy savanna (DC), 13 Mar. 1979, Skarpe 333 (K, PRE, SRGH, UCBG); 23.22 (Kang) Kang, 320 km W of Gaborone, 3,500 ft. (DD), 20 Oct. 1975, Mott 772 (SRGH, UCBG). MOZAMBIQUE. MAPUTO: Delagoa PRE, SAM); Rogers 22475 (K, PRE); 1890, Junod 24 (G, Z); 1896, Junod 486 (G, P, Z); Costa do Sol, sand dunes, 2 Mar. 1960, Balsinhas 126 (В, BR, К, LMA, P, PRE); 22 July 1965, Caldeira & Marques 595 (LMU, WAG); Sul do Save, Marracuene, between Lourenço Mar- ques and Costa do Sol, Jan., Pedro & Bachir 3846 (LMA); Exell et al. s.n. (LISC); between Vila Luiza and Manhiga, 17 Nov., Myre & Carvalho 1043 (LMA), Polana Beach, Delagoa Bay, Thoday 169 (SAM). Na- MIBIA. OVAMBOLAND: 17.15 (Ondangua) Ondangua, open sandy flats, 5 Feb. 1959 (DD), de Winter & Giess 6877 (М, WIND); 1924, Liljeblad 190 (7); 17.16 (Enana) Oniipa, eroded field, 7 Mar. 1967 (CC), Soini 429 (WIND); Olukonda, 30 Jan. 1887, Rautanen 2 (Z), Feb. 1894 (CC), Rautanen 170 (B, G, Z); Schinz 2 and s.n. (B, K, Z); 30 Dec. 1885, Schinz 15 (Z), 18.15 (Okahakana) (DD), Kruger 1 (РКЕ); 19.20 of Tsumkwe on road to Botswana, 10 Jan. 1971 (DA), Giess et al. 11196 (WIND); Tsumkwe, hard black sand, Oberflachenkalk, 13 Jan. 1971, Giess et al. 11050 (WIND); Simkue, 157 mi. E of Grootfontein, Acacia- Combretum parkland, 17 Jan. 1958 (DB); ites 6162 (M, PRE, SRGH, WIND). KAPRIVI: 17.24 (Katim WIND). REHOBOTH: 23.17 (Rehoboth) farm Arovley, red sand, 14 Feb. 1965 (AA), Giess 8401B (M, WIND). MARIENTAL: 25.17 (Gibeon) sandbodem, Sudkalahari bis Gibeon, May 1912 (BA-BB), Range 1455 (SAM). SouTH AFRICA. CAPE: 24.20 (Unions End) Kalahari Park, Khaa- pan, Jan. к (BD), van der Walt (PRE); 28.21 (Up- ington) ca. km N of Upington at the S edge of the Kalahari, 1 Ney 1980 (AB), Snijman 235 oe 28. 22 (С1еп Lyon) sandveld W of Padkloof, on dune running Langebergen, 15 Mar. 1937 (DA), Asoc Shiwa Ngandu, 17 Jan. 1937, Ricardo 147 (BM). LUAPULA: Lake Mweru, sandbank above = water mark, 13 Nov. 1957, Fanshawe 3941 (BR, CENTRAL: Serenje District, в July 1968. William- son 1337 (K, MO, PRE, SRGH). SOUTHERN: Mazabuka, Ridgeway road, stony red зо en 2 Dec. 1931, Trapnell 538 (BR, K, PRE); Machili, 10 Dec. 1960, Fanshawe 5959 (K, NDO, SRGH); 27 Dec. 1960 (fr), Fanshawe 6026 (K, NDO, SRGH); 3 mi. from Namwala on Ngama road, 16 Dec. 1962, van Rensburg 1101 (BR, K, LISC, t., van eura 2039 ( of Senaga, moist sandy soil at edge of plains, 1 Aug. 1952, Codd 7329 (BM, BR, K, MO, PRE, S, ; Kalabo, river bank, 2 Aug. 1962, Robinson 5437 (К, Р, SRGH); Kalabo District, between Sandaula pontoon and Kalabo, edge of grassy floodplain, 12 Nov. 1 ookson 6389 (E, К, LISC, MO, SRGH); Mongu, Barotseland. 22 Dec. 1965, Robinson 6749 (K, M, SRGH, WAG); Sesheke District, no date, H). WESTERN: Вагон Gairdner 69 (К). ZIMBABWE. MATABELELAND NORTH: angani River, NE of Bulawayo, 7 Jan. 1898, Rand 229 (BM). MASHONALAND EAST: Salisbury, Nov. 1919, Eyles 1885 (K, PRE, M H); Hunyani District, Old Charter road, 27 Dec. 1926, Eyles 5 (5, SRGH); Marandallas, Looe, near stream, 22 1948, Wild 2716 (BR, K, S, SRGH). MIDLANDS: Gw иза whitewaters dam, Loveridge 596 (К, SRGH); Charter District, 27 Dec. 1 Charter 4622 (K). WITHO OUT PRECISE o л № omuramba, 14 Мау 1939, Volk 2056 (М) 15b. Subsp. littoralis Lapeirousia caudata subsp. burchellii sgr. Marais & Goldbl., Ann. Bolus Herb. 4: 30-33. 2. Lapei- rousia burchellii Baker, Handbk. Sade dE 1892; Fl. Cap. 6: 93-94, 1896. TYPE: South Africa. Cape: 26.23 (Morokweng) Chooi Desert, Oct. 1812 (AC), Burchell 2350 (lectotype, K, designated by Gold- blatt, 1972; Wiss e бы — ramosissima Dinter, Feddes Rep. 29: 255. 1931. - Namibia. Lüderitz: dunes near Grün- dorn, Dinter 5043 ише В; isotypes, BOL, BR, Lapeirousia streyi eec „ Mitt. Bot. Staatssamml. München [1 (3): 88. 1951. TYPE: Namibia. Lúderitz: dones S of the Kuiseb, Strey 2587 (holotype, M; isotype, PRE (cited in error as Sb by Goldblatt, 1972)). Plants 10-15(-20) cm high, usually with 1- several short branches produced from near the base. Inflorescence usually a crowded cluster of short spikes, each 4-8-flowered; bracts 15-20 (-25) mm long. Flowers with a perianth tube 28- 35 mm long; tepals 13-15 mm long and 2-3 mm wide, not usually filiform even on drying. Distribution. an arc from a somewhat isolated Subspecies littoralis occurs in extensian in southwestern Angola through western sud south- ern Namibia and the northern Cape Province of South Africa into southern Botswana (Fig. 16). In Namibia it extends from near the Brandberg in the central west to the Lüderitz District in the south. In South Africa subsp. littoralis occurs in interior Namaqualand east of Springbok and in the north- ern Cape. NAMIBE: 2. Additional specimens examined. ANGOLA. near Mossamedes close to Vila Arriaga, 7 Feb. Torre 8824 (LISC). BOTSWANA. NGWAKETSE: 25.25 (Ma- fekeng) q ranch, 15 Sep. 1978 (CA), y e 3455 p Mug ie OMARURU: 21. foit, Brandberg, , sand i B. W. Mar. Tem yr em 942 (WIND): M Tumasberg at campsite (BA), Giess 13550 (M, WIND). 480 Annals of the Missouri Botanical Garden | | | FIGURE 17. and style х 2. (Drawn by J. С. Mann MALTAHOHE: 25.16 (Helmeringhausen) farm Duwisib, 17 May 1956 (BC), Volk 12782 (M, MO). LÚDERITZ: 26.16 (Aus) E 15 mi. W of Aus (СВ) Giess & van Vuuren 828 (WIND); 13 mi. W of Aus, coarse sand, Giess & van ae n 824(M, WIND). ean AFRICA. CAPE: 25.20 (Mata-Mata), Kalahari Gemsbok Park, between the Nos- sob and Auob dunes, 25 Apr. 1960 (BC), мона 807 ©); Kalahari Gemsbok Park, 6 mi. SE of Kaffer Pan, loose red sand on dune top, 22 Apr. 1960 (CC), Leistner 1874 (M, MO, PRE); 26.23 (Morokweng) Chooi Desert, Giraffe £ Station, Oct. 1812 (AD), Burchell 2341 (К); 27.18 (Vioolsdrif) a ropa 15 km S of Vioolsdrif, Koubank LI July 6 (DA), Giess WIND); 28.18 о Оно 27 July 1950 (CD), бер 6263 (NBG); 2 mi. © of Goodhouse, 27 July 1950, Lewis 2272 (SAM); 28 8.21 (Upington) p ington, sandveld, s.d. (AC), Mostert 1416 (PRE); 28.2 61, Leistner 2866 (B, G, PRE); 29.17 (Spring- bok) pu. mi. N of Okiep on Goodhouse road (BD), Lewis 5517 (NBG); 29.18 (Gamoep) Little Bushmanland, Keu zabies (AB), Schlechter s.n. or 103 (B, BOL, COI, E, G. , PRE); Aggenys, 10 mi. W of the farmhouse, 4 Sep. 1971 (BD), Wisura 2222 (NBG); pa 7 Sep. 1950 (CA), Barker 6755 (BOL, МВС); 25 mi. Springbok, van Breda 1370 (PRE); 29.19 DRM 20 mi. from Pofadder on the road to Springbok, Strauss 123 NBG). 16. Lapeirousia odoratissima Baker, Trans. Linn. Soc. London (Bot.) ser. 2, 1: 273. 1878; Morphology and oe of Lapeirousia odoratissima. Habit and corm x 0.5; detail of stamens by Б.) Handbk. Irideae 173. 1892; Fl. Trop. Africa 7: 354. 1897. Sólch, Prod. Fl. Súdwestafrika 155: 10. 1969. Geerinck et al., Roy. Bot. Belgique 105: 344-346. 1972. TYPE: Angola: sandy woods near Lopollo, Wel- witsch 1551 (lectotype so annotated, BM; isolectotypes, В, BM, С, С, К). Figure 17. cxi d Vaupel, Bot. Jahrb. Syst. 48: 548. PE: Namibia: Omaheke, 1,300 m, in brown sand, n Mar i nated by S ólch on sheet, В); Om Dinter 642 (syntype, SAM, as ыл Lapeirousia congesta Rendle, J. Linn. Soc. Bot. 30: 435. 189 aker, Fl. Trop. Africa 7: 354. 1 TYPE: Bad i "between Zanzibar and Uyui" [the latter ar Tabora], Taylor s.n. in 1886 (holotype, BM). Lapeirousia juttae Dinter, Die RECTOR Veldkost Deu sch-Súdwest-Afrikas 13-14. 1912. TYPE: not Plants 10-18(-25) cm high, with a condensed aerial axis sometimes rosettelike, rarely the stem partly aerial and with expanded internodes. Corm campanulate, 15-20 mm wide at the base, tunics woody, brown, the outer breaking irregularly, rare- ly becoming fibrous by decay. Cataphylls usually 2, the outer one short, the inner reaching almost to ground level, membranous. Leaves few, often Volume 77, Number 3 1990 Goldblatt Systematics of Lapeirousia in Tropical Africa only 2 (but hardly distinguishable from the bracts except by position), linear, corrugate, the lower- most inserted below the ground, exceeding the bracts and up to twice as long, to 30 cm long, 3-5 mm wide. Stem comprising 1 long basal internode to 8 cm long, usually reaching to just below ground level, upper internodes contracted, rarely 5-10 mm long, the aerial stem 2-5(- or with several branches, each subtended by leaves or by leafy Brace Inflorescence comprising 1 or d spikes, + umbellate in appearance, cm long; simple "n 3- -6 per branch; bracts herbaceous, 6-15 cm long, lanceolate, the outer lightly corrugate, the inner about 4 shorter than the outer, + mem branous. Flowers actinomorphic, hypocrateriform, white to ivory, usually strongly scented especially in the evenings; perianth tube cylindric, 10-14 cm long; tepals narrowly lanceolate, attenuate, 35-40 mm long, extended horizontally or some- times slightly oe widest in the lower third and there 4-5 mm wide. Filaments ca. 4 mm long, exserted for 1.5-2 mm, symmetrically disposed; anthers 6-8 mm long, linear, white; pollen yellow. Ovary ca. 3.5 mm long, style usually ultimately reaching to about the apex of the anthers, some- times shorter, branches 3-4 mm long, forked for ca. half their length, diverging and recurved. Cap- sules obovoid-oblong, 15-18 mm long, concealed in the leaf and bract bases; seeds + globose to somewhat angled by pressure, 2-2.3 mm at the widest diameter. Chromosome number 2n — 16, 18. Flowering time. December to March, occa- sionally in April; flowers opening near sunset and white, ivory by morning and gradually wilting dur- ing the day. Distribution and habitat. Although distrib- uted widely across south tropical Africa, Lapei- rousia odoratissima is apparently common only in Namibia where it occurs most often in sandy flats in the central and upper half of the country. Collections from a number of scattered sites indi- cate a fairly wide distribution across south tropical Africa except along the east coast and near interior. It extends from southwestern Angola, Zambia, and adjacent Shaba Province of Zaire to central and northern Malawi, Zimbabwe, and in locally dry sites in western and central Tanzania. Lapeirousia odoratissima probably also occurs in Botswana although there are no records from that country. Largely a plant of semiarid habitats, L. odoratis- sima also occurs in Brachystegia woodland and in exposed places in montane sites, such as the Nyika Plateau in Malawi and the Inyanga High- lands in Zimbabwe, as well as open grassland and Acacia savanna. Little is known about its biology but the large, long-tubed, and usually intensely fragrant flowers are presumably pollinated by hawkmoths. My ob- servations on plants in Namibia indicate that flow- ers open toward sunset, at which time they are creamy white. By morning the perianth has turned ivory to buff, and the tepals droop slightly below the horizontal; they wilt by the end of the day. The flowers are most intensely fragrant when they first open but maintain their scent through the following day. Ample nectar is produced in the perianth tube. Up to 25 ul was measured in one flower in which the nectar sugar concentration ranged from 16% to 25%. Observations made by Jean Pawek in Malawi confirm my phenological studies: she recorded that buds open between 5:30 P.M. and 6 P.M. but have wilted at least by 10 A.M. (earlier than in Namibia). Flowering phenology is seldom recorded and the presence of fragrance only rarely. The corms of Lapeirousia odoratissima are reported to be edible and to comprise part of the diet of the Khu Bushmen of Namibia (Marshall, 1976). Diagnosis and relationships. The usual tuft- ed or rosettelike growth habit combined with the large white flowers with an exceedingly long peri- anth tube 10—14 cm long set Lapeirousia odora- tissima well apart in the genus. The other tufted to rosette-forming species of sect. Sophronia are all southern African and have smaller flowers, col- ored pale blue to violet. The chromosome number, 2n — 16 or 18, is the same as in the largely southern African subg. Lapeirousia and different from most of the tropical African species with pseu- dopaniculate inflorescences. Subgenus Lapeirou- sia includes the south tropical African L. littoralis, which has somewhat similar white flowers with a shorter but still substantial perianth tube usually 25-45 mm long. Lapeirousia littoralis has an aerial stem, and is not likely to be confused with L. odoratissima. Lapeirousia odoratissima is probably most closely related to L. littoralis rather than to the rosettiform southern African species of sect. Sophronia. Flowers very similar to those of Lapeirousia odoratissima are found in L. schimperi (sect. Pa- niculata), but clearly this species is only distantly related to L. odaralasina, aad i has the plane char leaves and lax acteristic of sect. рані ога. The long-tubed flow- ers of these two species are presumably pollinated 482 Annals of the Missouri Botanical Garden by the same agent, most likely a hawkmoth, and are a notable example of convergence in two dis- tantly related species of the same genus. History. Lapeirousia odoratissima was dis- covered in southern Angola in 1859 by Friedrich Welwitsch, and his collection was described by G. Baker in 1878. Another early collection was made by the Rev. W. E. Taylor in central Tanzania in 1886 and was described as L. congesta by A. В. Rendle, who distinguished it from L. odoratis- sima on the basis of a more congested habit. Rendle described the plant as having a stem forming a dense sessile head above the first leaf. This is not strictly true, although the internodes of the stem and inflorescence axis are shorter than in the type of L. internodes found in the species. From the ample odoratissima, which has among the longest material available it now seems clear that the vari- ation between the extremes represented by the two collections is continuous, and there is no reason to consider them separate species. ater collections from then German South West Africa made by Kurt Dinter and Franz Seiner were described as Lapeirousia stenoloba by Vaupel ( 2). He considered these Namibian populations to constitute a species related to L. odoratissima, differing by a robust habit and particularly narrow and attenuate tepal apices. Many Namibian spec- imens of L. odoratissima differ from those in trop- ical Africa by their branched stems and numerous flowers, but intrapopulational variation is consid- erable in this species, and there are often few- branched or even unbranched plants in populations of more robust ones with several branches, thus the difference does not seem significant taxonom- ically. Additional specimens examined. ANGOLA. HUILA: 58 km on road from Sá de Bandeira to Vila Paiva Couceiro, 9 Jan. 1973, Couto 287 (K, LISE, SRGH); Huila Plateau, dry sandy terrain, Nov.-Dec. 1895, Berthelot 342 (Р); Tchivinguire, 22 Jan. 1962, Barbosa & Moreno (СОТ); Huila, sandy prairies, 1,740 m, Jan. 189 5 Dekindt 733 (LISC, P); Huila, 1883, Newton 231 (COI, Z); between Huila and Palanca, 26 Jan. 1956, M 1 423 (LISC); Ganguelas, 12 km from Vila Artur toward Galangue, 1,500 m, 5 Jan. 1960, qu her 1970 (LISC); Chicungo, ca. 1,700 m, 12 Feb. 1973, Texeira & Andrade 8446 (LISC). HUAMBO: DA Plateau, near River Catumbela . 1906, buellas, Vila da Ponte, Dec. 1932, Gossweiler 4022 (COI, ; Mt. Moco, Anhara de Meca (12.15 CD), 19 Dec. 1973, Huntley et al. 113 (PRE). BIÉ: Andulo, Cruza- mento N Lubia a 7 km from Nharea, 1,400 m, 24 Nov. 1965, Texeira et al. 9518 (LISC). CUNENE: Cuvelai ca. 20 km from Cuvelai to Chamutete, 10 Feb. 1973, Me- nezes, Barosso & Sousa 4444 (LISC, SRGH); Cuvelai ca. 16 km from Cuvelai to Bambi, 12 Feb. 1973, Me- 4506 (LISC, SRGH). CUANDO-CUBANGO: Menongue, Caiundo, Capico, near Mis- . 1960, Mendes 22. 34 (LISC). . NORTHERN PROVINCE: i Mzuzu, Kasitu River, 3,800 Pawek 8611 (MO); Mzimba District, 7 mi. towar ukuru River, ca. 1,260 m, 31 Jan. 1976, Pawek 10798 (K, MAL, ag тл Plateau, grassland at radio transmitter, 4,10 9 Feb. 1976, Phillips 1251 (МО); Mzimba нчы 5 ең W of 549 toward Vuvumwe bridge, 20 Jan. 1978, Pawek 13642 (BR, MO, SRGH, WAG); Mzimba District, edge of dirt road in sand, behind Ekwendeni, 26 Feb. Ble Ta 13931 (K, MAL, MO); Mzimba District, c m N of Mphe- rembe, lo Croix 4304 (MO). CENTRAL PROVINCE: Chitipa District, Kaseye Mission 10 mi. E of Chitipa, 1,270 m, closed in daytime, 26 Dec. 1977, Pawek 13376 (K, MAL); Lilongwe- “Dzalanyamas road near Ketete bridge, dry sandy dambo, 6 Feb. 1957, Robson 1483 (BM, K, MAL. PRE, SRGH); Kasungu Game Reserve, white sand, 20 Jan. 1970, Hall. Mn 577 (PRE); Dowa District, Xerophyta, 7 Mar ; NAMIBIA. KAOKOVELD: 18. 13 (Ohopoho) Okakura, 28 Feb. 1913 (DA), Dinter 3322 (SAM). KAOKOVELD: 19.14 (Ka- manjab) N of Otjovasandu in red sand, 11 Mar. 1976 (BA), Giess & us tit 14191 (WIND). ourjo: 19.14 (Kamanjab) Etosha National Park, Kaross, 9 Apr. 1974 (fr) (B-), Volk & Le Roux 807 (WIND); between Kaross and Kamanyab (BC), Thorne s.n. (SAM 313743); 19.15 (Okaukuejo) 10 km W of farm Uitzig, ca. 60 km ENE of Otjiwarongo, grassveld on red sandy loam, 9 Feb. 1983 (CB), Lavranos & Pehlemann 21059 (MO, WIND). OVAMBOLAND: 17.15 (Ondangua) Ondonga (DD), Barnard 196 (SAM); 18.15 (Okahakana) Onolongo-Onambeke, Apr. 1923 (?BB), Barnard 136 (SAM). GROOTFONTEIN: 19.16 (Gobaub) farm Norabis 387, thornveld, 18 Mar 1988 (fr) (DD), Goldblatt & Manning 8824 (MO); 19.17 (Tsumeb) Auros (Otavi), 10 Tob 1925 (DA), Dinter 5599 »G 19.18 (Grootfontein) 30 mi. , Story 6460 (f Bg N of Ga ; (PRE); Oliewenhof farm, sandy flat (CB), Merxmuller & Giess 30147 "(M, WIND). отл- WARONGO: 20.17 d и ongo District, farm Okosongomingo, 4 Mar. 19 CA), pc & Giess 30023 (M, PRE, WAG, WIND) KAHANDJA: 21.16 Okahandja) Omatako View, red sa ^4 Mar. 1974 (BA), Woortman 256 (WIND); 21.17 (Otjosondu) Okahandja District, farm Hochveld, brown sandy loam, 30 Apr. 1963 fr) (BD), Giess et al. 6672 (WIND); Okahandja, 22 Feb. 1928 (DD), Bradfield 385 (PRE). WINDHOEK: 21.17 онш 35 km from Steinhausen to Windhoek on Kapps farm road, 15 Mar. 1988 (DD), Goldblatt « Malos 8807 d 22.17 (Windhoek) farm Boden- hausen, near the river, 7 Mar. 1959 (BC), Seydel 1771 WIND). coBaBis: 19.20 (Tsumkwe) Grootfontein Dis- trict, ca. 3 mi. S of Nama Pan (DC), Story 6275 (M, PRE, WIND) 20.20 (Kaukauveld) v eee 56 mi. a toward Kano r. 1967 — ~ N of Eiseb Omuram Ap (fr) (AD), Giess 9815 (WIND); 21. үз retin 15 km from Steinhausen to Windhoek on Kapps Farm road, . 1988 (fr) (CC), slang es e Manning 8803 (MO). TANZANIA. IRINGA: 12 mi. SE of Iringa, 5,500 ft., well-drained red soil, 8 Feb. 1962, Polhill £ Paulo 1390 (B, BR, K, LISC, P, PRE, SRGH). RUKWA: Sumbawanga, Ufipa, among roadside grasses, 6,300 ft., 29 Jan. 1950, Volume 77, Number 3 1990 Goldblatt 483 Systematics of Lapeirousia in Tropical Africa Bullock 2357 (B, BR, K, MO, S); Ufipa Plateau, Michel- more 1065 (K). ZAIRE. SHABA: Mamea — ear 1938, Paterson s.n. (K); plateau de Manika, W de geen x Jan. 1983, Schaijes 1779 (BR b ue a km W de Kat , 21 Jan. 1969, Lisowski et al. 179 (BR). Тиш COPPERBELT: | Kos asempa Тин, Chati Forest Re- serve, W of Kasempa road, sandy edge of path to dambo, 12 Jan. 1961, En 55 (MO, SRGH); Chati dambo, 6 1962, Odgers 675 (NDO). sare о. prd r Mupomadzi River, 2,000 ft., 5 Jan 4415 (K, SRGH). N farm, dam, grassland, 13 Jan. 1965, Richards 19554 (К); Abercorn-Lunzua Falls road, red sand at top o escarpment, 1, , 26 Jan. 1962, Richards 15962 t, 1,500 m (K); Mbala District, road to Inono k 1955, Richards 4142 (BR, К). NORTH- WESTERN: Mwinilunga District, S of Samuteba on Solwezi Mwinilunga road, sandy dambo, 19 Jan. 1975, Brummitt et al. 13875 (K, NDO, SRGH); Mwinilunga Subdistrict, plain, dem d 136 (K). SOUTHERN: Machili, wood- рас A ud 5994 (K, NDO, SRGH); к: tallow в oil in miombo, 20 Jan. 1956, 35 (NDO). Victoria Falls, 3,000 ft., Rogers 5393 Inyanga District, Pteridium grassland W of Punch Rock, in fire break, 1,925 m, 24 Dec. 1972, Biegel 4122 (К, LISC, MO, PRE, SRGH); Inyanga National Park, near Maroro bridge, 23 Jan. 1975, Burrows 705 (NBG, SRGH); Inyanga ЕА sandy level river bank, Chase 581 (ВМ, К, SRGH); Inyanga, 6,000 ft., Bayliss 10641 (MO); Iya, grass- land along the road, 1,700 m, 13 Jan. 1931, Fries et al. 4244 (BM, BR, PRE, S); Rusape, > Jan. Hopkins s.n. (SRGH 7042); Manica District, Odnani Riv- er valley, 1915, Teague 334 (K). MASHONALAND NORTH: Lomagundi, Audley farm, Darwendale, 24 Jan. 1969, Biegel 2843 (PRE, SRGH); Bindura, 8 Jan. 1932, Brain 8063 (SRGH). MASHONALAND EAST: Macheke, 5,000 ft., sweet scent, Eyles 1989 (K, PRE, SRGH); near Salisbury, i istrict, Lake a SRGH); Nuza Plateau near Panga 1408 (BM, К). мА disponi a RTH: Victoria Falls, Jan же le 5393 bon Н); на /Bubi District, Gwa or Reserve, E lahari sand in Baikia woodland, Goldsmith 87/56 (K, PRE. LISC. SRCH). MIDLANDS: Gokwe, Brachystegia woodland on sand, 2 Feb. 1964, Bingham 1201 (K, cu эн е PRE- СІЅЕ LOCALITY: NAMIBIA: 217.14 (AC ne River banks, н 1923, Barnard 135 (SAM); n 15 (BB) Okakuja, 1912, Dinter 2574 (BM, K, SAM); Apr. 1912 (fr), inter 2649 (SAM); Mar. 1913, Dinter 2788 (SAM). EXCLUDED SPECIES Bot. Jahrb. Lapeirousia graebneriana Harms, 6 (holotype, B; isotype, BR) laxa (Thunb.) Goldbl. Fairly ed of the species, with a slender narrow perianth tube, but the bracts are 10—15 mm long, the lower flower with the largest bracts. Lapeirousia holostachya Baker, Kew Bull. 390. 894; Fl. Trop. Africa 7: 354. 1897. TYPE: Tanzania. Fwambo, Carson 14/1893 (holo- type, K, not seen). = Radinosiphon lep- tostachya (Baker) N. E. Br. (Carter, Fl. Pl. Africa 35: pl. 1384. 1962). Lapeirousia welwitschii Baker, Handbk. Irideae 168-169. 1892. ngola: Malange, Pungo Andongo, 9°40” 15°35”, Jan. 1857, Welwitsch 1531 (BM, annoted as lectotype in unknown hand; isolectotypes, C, COI, G, K, P). Th served, especially the flowers, and as discussed under L. rivularis, may be either L. rivularis or L. erythrantha. The type locality must be visited to resolve this question. е type specimens are poorly pre- Lapeirousia erythrantha var. welwitschii (Baker) Marais ex Geerinck, Lisowski, Malaisse & Symoens, Bull. Soc. Roy. Bot. Belgique 105: 340-341. 1972. LITERATURE CITED ANDREWS, F. W. udan 1956. The Flowering Plants of the , Volume 3. Sudan Government, Arbroath, 1876. New Gladioleae. J. Bot. London 3-339. . 1878a. Systema Iridearum. J. Linn. Soc. Bot. 16: 61-180. 1878b. Report on the Liliaceae, Iridaceae, Hypoxidaceae, and Haemodoraceae of Welwitsch’s Angolan Herbarium. Trans. Linn. Soc. ue. Bot Ser. E 1: 245-273 892. Handbook of the Irideae. George Bell & Sos, London 896. Ir ideae. In: W. T. Thiselton-Dyer (ed- # же Capensis 6: 88-89. Lovell Reeve, Lon- d 1898. Irideae. In: W. T. Thiselton-Dyer (ed- itor), Flora of Tropical Africa 7: 337-376. Lovell Reeve, London. CUFODONTIS, С. 1974. Iridaceae in Enumeratio Plan- tarum Ethiopiae Spermatophyta 2. Jardin Botanique National de Belgique, Meise. DE Vos, M. P. , The genus кш in South Africa. J. S. African Bot. Suppl. Volum . 1982. The African genus Tritonia. Ker Gawler (Iridaceae): Part 1. J. S. African Bot. 48: 105-163. DINTER, К. 1912. Die vegetabilische Veldkost Deutsch- Südwest-Afrikas. Selbstverlag, Okahandja. Epwarps, D. 8 O. A. LEISTNER. 1971. A degree square ern África. Mitt. Bot. Staatssam], München 10: 501- 5 кш R. С. . Notes on nomenclature in Irida- ae. Contr. Gray Herb. 114: 37-50. Annals of the Missouri Botanical Garden Fox, F. W. & M. E. М. Younc. 1982. Food From the Veld. Delta, ies urg. GEERINCK, D., S. wSKI, F. MALAISSE & J. J. SYMOENS. Aia Le ru Lapeirousia Pourr. (Iridaceae) au e. Bull. Soc. Roy. Bot. Belgique 105: 333-351. GorpBLATT, P. 1971. Cytological and morphological studies in the southern African lridaceae. J. S. Af- rican Bot. 37: 317-460. 1972. A revision of the genera Lapeirousia Pourr tand Anomatheca Ker in the winter rainfall Mesi of South Africa. Contrib. Bolus Herb. 4 . 1977. Systematics of Moraea (Iridaceae) in Hie Africa. Ann. Missouri Bot. Gard. 64: 243- 29: ———. 1982. Revision of n а African genus Freesia. J. S. African Bot. 47: 39-91. 4. A revision of а (Iridaceae) in the winter к area of southern Africa. J. 5. dale Lus 50: 141. Notes on the systematics of Hesper- aña Тн Ixioideae) in tropical Africa. Апп. bagy Bot. Gard. 73: 134-139. 89. The southern African genus Watsonia. Ann. кол Bot. Gard. 17. 1990a. Phylogeny and classification of Iri- nicas: Ann. Missouri Bot. Gard. 77(4): (in press). 1990b. Cytology and chromosome variation in Lapeirousia (Iridaceae). Ann. Missouri Bot. Gard. T4: -15 —— —— & J.C. MaNNING. 1990. Leaf and corm struc- ture in Lapeirousia (Iridaceae- Ixioideae) in relation to phylogeny and infrageneric classification. Ann. Missouri Bot. Gard. 1-150. & W. бор, 1979. nov., a segregat Lapeirousia (Irida of deae). Ann. И Bot. Gard. 66: 84 5.8 M е уа реп. na id НЕРРЕВ, К. М. 1968. Flora of West Tropical Africa 3(1). London. 1805. v E Ordo. Kónig & Sims Ann. 2 : hose Pp. 515-517 in W. C. H. Peters (editor), Reise Nach Mossambique. Bota- nik. 6(2). Berlin. 1866. Revisio Iridearum (addenda, emendata et corrigenda). Linnaea 34: 690-73 Lewis, С. J. Some aspects of ie morphology, phylogeny and taxonomy of the African Iri- daceae. African Mus. 40: 15-1 13. MarsHaLL, L. 1976. The 'Kung of Nyae Nyae. Harvard Univ. Press, Cambridge, Massachusetts. RENDLE, A. B. . Catalogue of the African plants collected by Dr. jocum Welwitsch in 1853-61 British Museum, Lon RicHARD, А. 1850. i205 Florae Abyssinicae. 2. Bertrand, Paris Корм, R. 1985. ‘The ethnobotany of the Kwanyama Bot. Missouri Bot. Gard. 9 же of African Studies. Travels and Narratives 16. Frank Cass, London. [Reprinted, 1967 SOLCH, А. 1969. Iridaceae. In: Н. Маеги Dr see rodromus einer Flora von Südwestafrika 155: 6- 10. J. Cramer, Lehre. . Iridaceae Africanae novae. Bot. Jahrb. 49 1971. Lapeirousia rivularis, a new species of Iridaceae from South West Africa. Svensk. Bot. Tidskr. 65: 53-56. REYER-BRANDWIJK. 1962. denal ünd Eu Plants of Southern and Eastern Africa, 2nd edition. Livingstone Ltd., Edinburgh. LÉON CROIZAT’S PLANT COLLECTIONS FROM THE HEADWATERS OF THE RIO ORINOCO! Bruce K. Holst? and Carol A. Todzia? ABSTRACT 1951 Léon Croizat participated in the Franco-Venezuelan expedition to the headwaters of the Rio Orinoco and the collecting localities. Léon Croizat-Chaley (1894-1982) is best known for his writings on biogeography (Croizat, 1952, 1958, 1961), systematics (mainly the Euphorbi- aceae), and morphology. Born and raised in Italy, he moved to Venezuela in March 1947 after study- ing for a few years at the Arnold Arboretum of Harvard University. In Venezuela he held various positions including Professor of Botany and Ecol- ogy at the University of the Andes in Mérida, and later Technical Director of the Xerophytic Botan- ical Garden in the state of Falcón. He was a prolific writer, amassing a bibliography of over 15,000 pages including his monumental Panbiogeogra- phy (1958), Principia Botanica (1961), and Space, Time, Form: The Biological Synthesis (1964). Less well known are his plant collecting efforts. In 1951 at the age of 57, Croizat partic- ipated as botanist in the Franco-Venezuelan ex- pedition to the headwaters of the Rio Orinoco. He was decorated with the “Order of the Liberator” by the Venezuelan government for his contributions to the expedition (Steyermark, 1983). ! We thank Prof. Joseph Ewan and the late Dr. Julian Steyermark for their discussions of the project. Carol Blaney, Linda Albert de Esc e ar, Paul Berry, Gustavo Romero, nts on the manuscript. Doris Lee Tischler provided the illustration. Rogers provided va comme We Brian Boom, Otto Huber, Germán Carnevali, and George e also thank i following people for MW re W. R. Anderson (MICH), Malpighiaceae; L. Andersson (GB), Musaceae; M. M. A ae Menispermaceae, Mimosaceae; H. Bed Arbo (CTES), M wis D. F. Aus liaceae; W. D'Arcy (MO), Solanaceae; C. Dodso Harley (K), Lamiaceae; G. Harling (GB), C but oue А. Hender : A), Polygonaceae; су Huft (МО), синони (їп рагї); J. Kallunki (NY), Bache 1 (VDB), Xyridaceae; A. Kra , Marantaceae; R. Kra in (FAU), Convolvulaceae; R. Barneby (NY), ell (GH), А С.С. (Ternstroemia); E. Christenson (SEL), posi ped (in part); T. Cro n (MO), puces A part); A. Gentry (MO), атат К. Caesalpiniaceae, Berg (BG), Moraceae; B. Boom (NY), Theaceae at (MO), Araceae; C. Cristóbal (CTES), Stercu- n (NY), Arecaceae; W. H. A. Hekking (U), (HBG), Clusiaceae (Caraipa a), Dilleniaceae; A. J. M. L Liesner (MO), miscellaneous; J. pasa AS Ericaceae; P. J. M. Maa daceae, Zingiberaceae; L. Marcano Berti (MER), Vochysiaceae; W. Meijer (KY), Tiliaceae; A. ege ga (U), Hip- pocrateaceae; J. Mickel (NY), үрне (Elaphoglossum); J. Miller (MO), Boraginaceae; 5 NY), Lecythi- daceae; M. Nee MOL ir e; B. one aard EI A T bi. + ington (K), Meliaceae, Sapotaceae; nce (K), "el y ip рны n Royen, ds L. Skog mith (ОС), ай ылан aie: Elaphoglossum - Lycopodiaceae; C. Stace (LTR), Саке В. Stein NM Conservancy), Campanulaceae; P. Taylor (K), Lentibulariaceae; W. Thomas (NY), Simaroubaceae; S. Tillett (MYF), Passifloraceae; K. Vincent (NY), Hei edis iaceae; D. Wasshausen (US), Acanthaceae; С. Webster (DAV), Euphorbiaceae (in part); А. Weitzman (US), DES (Freziera); J. J. Wurdack (US), Melastomataceae, Polygalaceae; H. van der Werff (MO), Lauraceae; E. Zardin (MO), Onagraceae ? Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. з Department of Botany, University of Texas, Austin, Texas 78712, U.S.A. ANN. Missouni Вот. Ganp. 77: 485-516. 1990. 486 Annals of the Missouri Botanical Garden The expedition was conceived by Joseph Grelier, a French geographer and hydrographer, who de- sired to find the source of the Orinoco. In addition to reaching the headwaters of the Rio Orinoco and establishing its coordinates, the objectives of the expedition included contacting the various Indian tribes and making archaeological, anthropological, and ethnographic studies, studying in detail the flora and fauna of the region, investigating the geology of the area, and obtaining photographs to increase the knowledge of the area and serve as a base for future investigations (Contramaestre, 1954). Grelier recruited two other Frenchmen, Pierre Couret, a pharmacologist and botanist, and Ray- mond Pélegri, a radio technician, plus Frantz La- forest, a Canadian ethnologist and archaeologist. When the French group arrived in Caracas, the Minister of Defense appointed Major Franz Ris- quez-lribarren to assist in the preparations and to recruit a Venezuelan team. Quickly the expedition grew to include Captain Félix Cardona Puig, his son Félix Cardona Johnson, geologists Marc de Civrieux and Carlos Carmona, archaeologist José- Maria Cruxent, entomologist and parasitologist Pa- blo Anduze, Luis Carbonell, René Lichy, and Léon Croizat. In total the expedition involved 54 mem- bers, including 14 scientific personnel and 40 assis- tants and members of the military. Such a large and diverse cast of characters with conflicting in- terests and personalities resulted in five books, each airing a different viewpoint on the expedition (An- duze, 1958; Contramaestre, 1954; Grelier, 1954, 1957 (English translation); Lichy, 1978; Risquez- Iribarren, 1962). The five books are similar in that they all provide an account of the expedition and include maps and numerous photographs. Lichy (1978) provided a day-by-day log of the expedition apparently derived from his journal. As military commander of the expedition, Risquez-Iribarren presented a more technical account including much cartographic information. Anduze (1958) com- mented on each of the expedition members and provided a narrative of the journey and a long chapter of observations on the Yanomami, an in- digenous group in the upper Orinoco region. He included photographs of the personnel on the ex- pedition and of the Yanomami, as well as appen- dices on the mammals, birds, pani grasses, and palms collected on the trip. A more loquacious narrative is Grelier's (1954, 1957, 1971), who also provided some historical background as well as photographs of the river and its surroundings. Contramaestre (1954) gave a very brief account with photographs of everyone on the expedition or associated with it. It is mainly from these five books that the following account is drawn On April 9, 1951, after four months of prep- aration in Caracas, expedition members flew from Caracas to San Fernando de Atabapo, closest air- strip to La Esmeralda, to construct a base camp and landing strip. Base Camp Number 1, as La Esmeralda was called, served the dual function of the first study area and the first operational base. In June, Croizat left Caracas along with other mem- bers of the expedition by small plane headed for La Esmeralda, but because of bad weather were routed to Puerto Ayacucho. They then traveled by river to La Esmeralda. On this journey of seven days, Croizat collected his first 60 numbers of the expedition. On the expedition Croizat dressed in khaki garb, a tropical pith helmet, and was well-armed with two knives, a revolver, and a rifle (Lichy, 1978, p. 112). Anduze (1958, p. 48) observed that of all the expedition members Croizat was the most ded- icated to documenting the trip and kept very clear and meticulous notes. Unfortunately, his copious notes on the expedition do not survive. Situated on the north bank of the Orinoco, La Esmeralda is surrounded by a mixture of savanna, forest, and sandstone hills. Croizat and Couret searched this area bringing back plants to their camp, which was constantly cluttered with flowers, fruits, and piles of newspaper and blotting paper. They dried their specimens on a makeshift dryer using a continuously burning fire. Croizat and Cou- ret thus were dubbed the *'vestal virgins” by other expedition members (Grelier, 1957, pp. 63-64). In the early 1950s the Orinoco above Esmeralda was still fairly unknown, and no experienced guides were available. In mid-July after an advance party scouted the area, the expedition moved on to the Guaharibo Rapids where they were to set up the second base and research center. The Guaharibo Rapids are the first major barrier on the Upper ЕС and there the river is approximately 150 wide. Croizat spent 11 days in this leal over 200 numbers. Although there must have been many harrowing experiences on the expedition, few specifically mention Croizat. How- area and ever, one evening Croizat and Couret went in the curiara, the dugout canoe made by the native In- dians, above the rapids in search of plants, and as they pulled away from shore, the current sucked them into midstream and hurled them through the rapids. On August 11, Croizat and four other scientists set off from the Guaharibo Rapids for their next stop, Base Camp 3, Salto Salas, named in honor Volume 77, Number 3 1990 Holst 8 Todzia Croizat's Collections from the Franco-Venezuelan Expedition 487 65° 64° Caracas Ciudad Boliva U La Esmeralda 3° 7 39 RIQ ОСАМО X 4 B t t з х ®10 I О» E No n hh 2 С "i B uw tM о Or z наа Guana Islo ‹ де © d à > t ›Чаг M H < СА > Isla del Esfuerzo y( [/ 7 2 Pto Esperanza Ta oo ALL aet 1 `+ а Man сай, ге — b om у^ "i 2 M 1 L 1 1 ] :| О 20 30 40 50km 65° н 64° FIGURE 1. Мар showing the upper Rio Orinoco including Croizat's collecting localities. of a friend of Major Risquez. Croizat stayed in the area of Salto Salas for 11 days and collected 84 numbers. Base Camp 4 was established on an island below a broken stretch of water that they named Effort Rapids or Isla del Esfuerzo. It was here that the “Guaharibo Gripe,” the fever that had been plaguing the expedition for several weeks, reached epidemic proportions. The constant humidity, fa- tigue, and necessity of standing in rushing water for long periods of time began to wear on the crew. The camp had as many as 10 sick men at a time. Major Risquez felt it best that the sick be evacuated at this time, and the first group returned to La Esmeralda where a plane could pick them up. Croi- zat continued on with the expedition to the next camp, Potsherds, or Los Tiestos, where the ex- pedition stayed for only three days. It was decided that those who were well would form an exploratory group taking 25 men and six boats, continue up- river for seven days, and then send the boats and men back to the main party. On September 11, the party set out, and after an entire day’s effort, no more than a mile of river had been traversed. And on that day, to Croizat's despair, the boat carrying the botanical specimens was submerged in water. The party moved upriver slowly, spending nights camped at El Motín, Raudal Bobadilla, El Queso, Raudal Montserrat, localities where Croizat collected specimens. Croizat continued on with the advanced exploratory part to the confluence with the Rio Ugueto. At the mouth of the Rio Ugueto, the expedition pitched camp and decided to rest so that eac scientist could continue his own research. Here Croizat collected 65 numbers. The rains were be- ginning to let up and the river started to fall. The expedition once again set off, this time loaded with fresh provisions that had been parachuted in. The river now was only 20 m wide and consisted mostly of rocky shoals. It was decided not to take the boats any further upstream, although it was esti- mated that the source of the Orinoco was 40 miles 488 Annals of th Missouri ota Garden further. In these rapids above Ugueto, Croizat lost his field notebooks and journal (letter from Croizat to J. Steyermark, 14 December 1964). Croizat had been sick for some time with toothaches and gastroenteritis (Anduze, 1958, p. 48). He had be- come so sick that he needed assistance to dress and to get into his hammock (Grelier, 1954, p. 166). At Ugueto, Croizat decided he would go no further and returned to Esmeralda with a party of three others, a journey of 22 days. Although Croizat’s notes and journal were lost, some insights gleaned from his trip were published in his Panbiogeography. “In the Upper Orinoco the riverine forest generally alternates between two main types, one denser and richer in high vege- tation in stretches where deep, heavy clays dom- inate; the other rather more open and lower by canopy holding belts or rocky seams and thinner soil (these belts mark as a rule a rapid)” (Croizat, 1958, 1: 416). Some members of the expedition gave exaggerated reports of the botanical richness of the area: “We know, however, that the botanical collection contains a great number of new species and many others which, though known, had not been previously met with in these latitudes” (Gre- lier, 1957, p. 182), and “Upon leaving the Rau- dales Atures and Maipures which are the limit of the zone, we find out that the flora and fauna, mostly endemics, are products of the physical con- dition of the earth" (Anduze, 1958, p. 165). Croi- zat, however, noted that few of his plants were endemic and that the flora of the upper Rio Orinoco was essentially Amazonian (Croizat, 1958, 1: 755). After the expedition, a full set (or nearly so) of the plants were assembled at VEN and shipped to the U.S. in 1952 (Couret, 1966). These apparently arrived first at F, but were not accepted for fear that they y removed from Venezuela (Steyermark, Sn. comm.). Later, they were ac- cepted by Maguire at NY (Steyermark, pers. comm.) where the bulk of them sat in the backlog until 1984. Duplicate collections of certain families (no- tably the Pteridophytes) exist at several other her- baria, F, MO, P, and VEN, but it is not know w or when they were distributed. Couret (1966) indicated that 1,200 species were collected, though he was probably referring to the total number of collections and about 5,000 samples, so it is pos- sible that if not already destroyed, additional du- plicates may eventually turn up. As Croizat indicated, the specimens are gener- ally riverine plants and thus widely distributed. However, two new species, Persea croizatii van der Werff (Lauraceae) and Ouratea croizatii Ma- guire & Steyerm. (Ochnaceae), have been de- scribed, each from a single Croizat collection ap- parently collected on higher ground away from the river. Several prominent plant families of the re- gion, notably Cyperaceae and Poaceae are absent from the list (Appendix I). It is possible that they were separated from Croizat's collections for study upon return to Caracas. Anduze (1958, appendices 4 and 5) presented lists for these two families in Territorio Federal Amazonas but did not cite spec- imens. Not all of the botanical collections of the ex- pedition were made by Croizat. Couret, as was previously mentioned, assisted Croizat at times and also independently gathered cryptogams, which he later sent to P for study (Couret, 1966). After Croizat turned back at Ugueto, Pablo Anduze and Luis Carbonell continued collectings plants (An- duze, 1958, p. 133; Huber & Wurdack, 1984) None of their names, however, appear on the plant labels. Cruxent apparently made a separately num- bered collection, also from the area above Ugueto (B. Boom, pers. comm.). As with the Croizat col- lections, Cruxent's are nearly all deposited at NY. LITERATURE CITED ANDUZE, P. J. 1958. Shailili-Ko: Descubrimiento de las fuentes del Orinoco. The Author as. CONTRAMAESTRE TORRES, A. 1954. La expedición Fran- zolanas. Mem: Soc. Ci. Nat. La Salle 26(7 3): CRoizAT, L. 1952. Manual of Phytogeography or a Account of Plant- шры throughout the World. W. Junk, The Ha не а ЕЕ or an Introductory Synthesis of Zoogeography, Phytogeography, and Geology; with Notes on Evolution, Systematics, Eco ogy, Anthropology, etc. The Author, Caracas. 19 Principia Botanica or Beginnings of Botany. The Author, Caracas. Space, Time, Form: the Biological Syn- thesis. The ‘Author, Caracas. GRELIER, J. 1954. Aux Sources de l'Orénoque. La Table Ronde, Paris. 1 To the Source of the Orinoco. Herbert Jenkins, London. [Translated by Н. A. С. Schmuck- E 1971. Aux Sources de l'Orénoque, 2nd edi- tion. La T жыен. Paris HUBER, O. & J. WURDACK. 1984. History of Botanical Exploration in “Territorio Federal Amazonas, Vene- zuela. Smithsonian Contr. Bot. 56. Licuy, R. Ya Kú; Expedición Franco-Venezolana del Alto Orinoco: Monte Avila Editores, Caracas. RÍSQUEZ-IRIBARREN, F. 1 эже Donde Nace el Orinoco. Ediciones Grecco, Carac STEYERMARK, J. A. 1983. Leo Croizat-Chaly [sic]. Tax- on 32: 530-531. Croizat's Collections from the Franco-Venezuelan Expedition Volume 77, Number 3 Holst 8 Todzia 489 1990 APPENDIX I. Croizat collections from the headwaters of the Rio Orinoco. 1. NUMERICAL LIST The collections are arranged numerically by Croizat’s collection numbers and are preceded by collecting locality noted or two different collections given the same number. A question mark after the plant name indicates uncertainty by the determiner. Many of the locality names used during the expedition, and in many of the books resulting from it, are not accepted by Cartografia Nacional of Venezuela. These include the following: Salto Codazzi, Isla de las Hormigas, Pto. Esperanza, Isla del Esfuerzo, Los Tiestos, El Motin, Raudal Bobadilla, El Queso, Raudal Montserrat, and La Mantequilla. They appear on the map (Fig. 1) only as aids in locating the collecting areas. PUERTO AYACUCHO, 27 JUNE 1951; 05%40'N, 67*38'W; 80 m elev. 1 Rudgea maypurensis Standley Rubia 5 Manihot brachyloba Muell. ¡e ? Euphorbiaceae 6 Phyllanthus minutulus Muell. Ar Euphorbiaceae 7 ari esvauxil var. triumeiralis H. Irwin & Barneby Cateslpinia aceae 9 oe orinocensis (HBK) Mei Haemodoraceae 10 Burse Burseraceae 11 Irlbachia a (Aublet) P. Maas subsp. alata Gentianaceae BETWEEN RÍO ATABAPO (04°03'N, 67*40'W), and LA ESMERALDA (03?10'30"N, 65°32'44”W); 1-6 JULY 1951 12 Ficus nymphiifolia Р. Miller Moraceae 13 Dalechampia affinis Muell. Arg. Euphorbiaceae 15 Solanum sp. Solanaceae 16 Distictella ia e E (Vahl) Sandw. Bignoniaceae 17 Talisia firma Rad Sapindaceae 18 Heterostemon сын ЖЕМ Benth. т 19 Dalbergia riedelii (Radlk.) Sandw. Fabace 20 Dalbergia riedelii (Radlk.) Sandw. Fabaceae 22 Hibiscus furcellatus Desr Malvaceae 23 Myrcia deflexa (Poiret) DC Myrtaceae 24 Roucheria calophylla Planc Linaceae 25 Vismia guianensis (Aublet) Choisy subsp. persicoides Ewan Clusiaceae 26 Dichorisandra is (Aublet) Standley Commelinaceae 27 Ocotea cymbarum HBK Lauraceae 28 Posoqueria Ша би Aublet Rubiaceae 29 Myrtaceae 30 Croton trinitatis Millsp. Euphorbiaceae 31 Маргоипеа guianensis p Euphorbiaceae 32 C жайбы arborea (L.) S. К. Blake Violaceae 33 Galeandra devoniana Ex ex Lindley Orchidaceae 34 Amazonia sp. e 35 Gurania d Cogn. Cucurbitaceae 36 Turnera acuta Willd. Turneraceae 37 Кыа енши (Hoffsgg.) А. С. Smith d iue ы 40 Psychotria poeppigiana Muell. Arg. subsp. barcellana (Muell. Arg.) Stey- Rubiace erm. 4l Croton cuneatus Klotzsch Eu ume 42 a poeppigiana Muell. Arg. subsp. barcellana (Muell. Arg.) Stey- Rubiacea 43 Pe е рапигепѕіѕ Spr. & Sandw. Loganiaceae 44 Talisia firma Radlk. Sapindaceae 46 Mikania micrantha Kunth Asteraceae no locality, no date 48 Passiflora securiclata Masters Passifloraceae 49 Croton trinitatis Millsp Euphorbiaceae 50 Passiflora vitifolia HBK Passifloraceae 51 Phyllanthus orbiculatus Rich Euphorbiaceae 52 Phyllanthus sp. Euphorbiaceae Sta. Bárbara del Orin 53 Miconia pin (Schldl. & Cham.) Naud. Melastomataceae 490 Annals of the Missouri Botanical Garden 54 Chamaecrista desvauxii var. mollissima (Benth.) Н. Irwin «€ Barneby Caesalpiniaceae 56 Inga nobilis Willd. imosaceae 57 Bixa urucurana Willd. Bixaceae Sta. Bárbara del Orinoco 58 Miconia rufescens a. DC. Melastomataceae 59 Utricularia subulat Lentibulariaceae 60 Dipteryx кена la Amshoff Caesalpiniaceae LA ESMERALDA, 9-20 JULY 1951; 03?10'30"N, 65?32'44"W; 120 m elev. 61 Ocotea sp. Lauraceae 63 Byrsonima crassifolia (L.) HBK Malpighiaceae 64 Alchornea discolor Poeppig Euphorbiaceae 65 Heteropterys atabapensis W. R. Ander Malpighiaceae 66 Ladenbergia lambertiana (A. Braun ex ra Martius) Klotzsch Rubiaceae no locality, no date 69 Cassytha filiformis L. Lauraceae 70 Irlbachia eta (Kunth) Cobb 8 P. Maas Gentianaceae 71 Habenaria leprieuri eichb. Orchidaceae 72 Remijia hispida Spruce ex x Schuman Rubiaceae 74 Calathea inocephala (Kuntze) Kennedy & Nicolson Marantaceae 76 a ot subtilis HBK Polygalaceae 81 Piriqueta sp. Turneraceae 82 Paye а egensis I Arg. Rubiaceae 83 Palicourea calophylla A. DC. Rubiaceae 84 Sipanea hispida Ben c ex Wernham var. iris Rubiaceae 85 P latifolia (Aublet) Schumann var. latifolia Rubiaceae 86 Sipanea pratensis Aublet var. dichotoma (HBK) Steyerm. Rubiaceae 87 Psychotria bracteocardia (A. DC.) Muell. Arg. Rubiaceae 90 Clusiaceae 91 Crotalaria sagittalis L. Fabaceae 92 Сига tenuifolia (Aublet) Knobl. subsp. tenuifolia Gentianaceae 93 Polygala hygrophila HBK Polygalaceae 94 Polygala hygrophila HBK Polygalaceae 95 Polygala adenophora DC. var. robusta Chodat Polygalaceae 96 Meriania urceolata Triana Melastomataceae 97 Passiflora nitida HBK Passifloraceae 99 ococa macrophysca Spruce ex Triana Melastomataceae 100 Pterogastrus divaricata (Bonpl.) Naud. Melastomatacea 106 Vismia laxiflora Reichardt Clusiaceae 108 Buchnera palustris (Aublet) Spreng. Scrophulariaceae 110 Phyllanthus hyssopifolioides Kunth Euphorbiaceae 112 ое bicolor С. Martius Burmanniaceae 113 Mac a thyrsiflora DC. Melastomataceae 116 Li Verbenaceae 117 ad longicornis Lindley Orchidaceae 118 Dichaea brachypoda Reichb.f. Orchidaceae 119 Epidendrum nocturnum Jacq. Orchidaceae 120 Passiflora foetida L. var. foetida Passifloraceae 72121 Microgramma tecta ( el г Polypodiaceae 2121 Metaxya rostrata (НВК) Р Metaxyaceae 122 Lycopodiella caroliniana a Pichi-Serm. var. meridionalis (Underw. & Lycopodiaceae Lloyd) В. Ollg. & Wind 123 Lindsaea stricta (Sw.) Dryander Dennstaedtiaceae 124 Lycopodiella camporum В. Yllg. € Wind. Lycopodiaceae 125 Monotagma laxum (Poeppig & Endl.) Schum. Marantacea 126 Byttneria genistella Triana - Planchon Sterculiaceae 127 Abolboda macrostachya Spruce ex Mal ar. macrostachya Xyridaceae 130 Clidemia hirta (L.) Don var. tiliifolia (DC. ) “Mache ide Melastomataceae 131 Bellucia grossularioides (L.) T Melastomataceae 132 rcgraviaceae 133 Phyllanthus aff. lindbergii Muell. Arg. Euphorbiaceae 134 Acosmium nitens (J. Vogel) Yakovlev Fabaceae 135 Tococa nitens (Benth.) Triana Melastomataceae 136 Clidemia rubra (Aublet) C. Martius Melastomataceae 144 Selaginella parkeri (Hook. & Grev. i Sprir Selaginellaceae Raudal de los Guaharibos, Foot of Mt. Rimbaud, 30 July 1951 Volume 77, Number 3 Holst 4 Todzia 491 1990 Croizat's Collections from the Franco-Venezuelan Expedition 145 Hirtella racemosa Lam. var. hexandra (Willd. ex Roemer & Schultes) Chrysobalanaceae rance 147 de willdenowiana Stued. subsp. willdenowiana Dilleniaceae 148 miria balsamifera (Aublet) St. Hil. iriac 149 Utricularia amethystina St. Hil. Lentibulariaceae 150 Comolia е Benth. s.l. Melastomataceae 153 Clusiace 154 Ormosia coccinea Jackson Fabaceae 155 Clusiaceae 156 огои linearis НВК асе 157 Vochysia Vochysiaceae 157-A da rubiginosa (Stafl.) Marcano-Berti Vochysiaceae Raudal de los Guaharibos, toward the top of Mt. Rimbaud, 30 July 1951 158 Jacquemontia tamnifolia (L.) Grise Convolvulaceae 159 Calycobolus glaber (HBK) Hous Convolvulaceae 161 Sauvagesia ramosa (Gleason) Sastre Ochnac 162 Catasetum discolor Lind Orchidaceae 2163 Brassavola martiana Lindley Orchidaceae ?163 Quassia amara L. Simarouba 164 Lindsaea cf. portoricensis Desv. Dennstaedtiaceae 165 Trichomanes pilosum xs Hymenophyllaceae 166 Heteropterys nervosa A. J alpighiaceae ?167 Actinostachys pennula C ) Hook. Schizaeaceae ?167 ursera 168 Melochia villosa (Mill.) Fawcett & Rendle var. villosa Sterculiaceae 169 M ema xyridoides Gleason apateaceae 170 Vismia japurensis Reichardt Clusiaceae 171 Clusiaceae 172 Smilax sp. Liliaceae 174 Fla т 175 Doliocarpus dentatus (Aublet) Standley subsp. esmeraldae (Steyerm.) Dilleniac Kubitzki 176 Solanum pensile Sendtner Solanaceae 177 Xylopia aromatica (Lam.) C. Martius nnonaceae 178 Buchnera rosea Kun Scrophulariaceae 180 d setigera (C. Martius) H. A. Wendl. 183 Catasetum pileatum Reichb.f. Orchidaceae 184 Iechnosiphon cannoideus L. Anderss. Marantaceae 186 Genipa spruceana Steyerm. Rubiace 187 a sp. Droseracea 188 Pitcairnia patentiflora Lyman B. Smith var. patentiflora Bromeliaceae 189 Piriqueta cistoides (L.) Griseb. erac 19] Chamaecrista desvauxii var. a шыш ) Н. Irwin & Barneby Caesalpiniaceae 192 Marsdenia rubrofusca Ben г Asclepiadaceae 199 Catasetum mein i P Orchidacea 200 Tibouchina spru Melastomataceae 201 ps iconia горе eh, 8 Cham.) Naud. Melastomataceae 203 Melastomataceae 204 к Mose (Miller) Fa awcett & Rendle var. villosa Sterculiaceae 205 Clusiace 206 Epidendru Orchidaceae 207 к, villosa е Fawcett & Rendle Sterculiaceae 208 Anthurium bonplandii Bunting subsp. bonplandii Araceae 209 Poychotria эллан Muell. Arg. subsp. barcellana (Muell. Arg.) Rubiaceae Ste 210 Burseraceae 211 Irlbachia 9 (Aublet) Р. Мааз subsp. angustifolia (Kunth) Pers. & Gentianaceae . Maa 212 лн affinis Кое чы сөнө 213 е аісосса Chan & Schldl. subsp. dicocca var. guianensis Rubiac Steyer 214 Guatteria. cf. ne” Kunth Annonaceae of era 215 Phyllanthus а Kun Euphorbiaceae 218 Rhynchanthera ra a DC. Melastomataceae 221 Cupania scrobiculata Ric Sapindacea 224 cotea s Lauraceae 225 Psychotria bahiensis A. DC. var. cornigera (Benth.) Steyerm. Rubiaceae 492 Annals of the Missouri Botanical Garden 226 Siparuna guianensis Aublet Monimiaceae 228 Pagamea coriacea Spruce ex Benth. Rubiaceae 229 Drosera sp. Droseraceae Cerro de la Ban 230 Cochlospermum orinocense (Kunth) Steudel Cochlospermaceae no locality, 24 July 1951 231 Duckeella ром Сагау Orchidaceae Cerro de la Bandera 232 Acacallis cyanea Lindley Orchidaceae Cerro de la, Bandera 233 Curculigo scorzonerifolia (Lam.) Baker Amaryllidaceae 235 Clidemia capitellata (Bonpl.) Don var. dependens (Don) Macbride Melastomataceae Cerro de la Bandera 236 Bredemeyera d (Benth.) A. W. Bennett Polygalaceae Cerro de la er 237 Guarea б м (Rich.) A. Juss. subsp. pubescens Meliaceae 237-A Meliaceae Raudal de los Guaharibos, 21 July 238 ш schomburgkii ш У и Arg. Rubiaceae cidentale 239 C aa rista orenocensis (Benth. ) H. Irwin & Barneby Caesalpiniaceae Slopes of Cerro Bandera RAUDAL DE LOS GUAHARIBOS, 21 JULY-1 AUG. 1951; 02?18'N, 64°39'W; 150 m elev. Note: The locality names listed below for some collections, Mt. Rimbaud and Mt. Maunoir, though not officially recognized, correspond to due in = Sierra Guaharibo, a mountain range stretching along the Orinoco River to the north of Raudal de los Guahari Hiraea faginea (Sw. ) Niedenzu См & Y % ч E © S ~ е ® Y ~ € = La Esmeralda, 21 July 1951 Apinagia staheliana (Went) van Royen Hylocereus s Heteropterys йе ensis (HBK) A. Juss. Lindackeria a SaS ) Gilg Between La Esmeralda and Raudal de Los Guaharibos Selaginella asperula die Catasetum barbatum Lindley Ficus matiziana Dugand Miconia cf. brevipes Benth. Xiphidium coeruleum Aublet Passiflora vespertilio L. Salto Salas, 23 Aug. 1951 Passiflora longiracemosa Ducke Selaginella parkeri (Hook. € Grev.) Spring Anthurium gracile (Rudge) Schott Cattleya violacea (HBK) Rolfe Dichorisandra hexandra (Aublet) Kuntze Between La Esmeralda and Raudal de los Guaharibos, 23 July 1951 Operculina sericantha (Miq). Ooststr. Bauhinia glabra Jacq. on Clidemia hi dg (L.) Don var. tiliifolia (DC.) Macbride Annona sp. Mata yba P opaca Radlk. Pachira aquatica Aublet? Cattleya sp. Asplenium claussenii Hieron. ? Pteris propig Agardh Mixed с Monstera a Schott + Monstera dubia (HBK) Engl. Voyria flavescens Grise Tournefortia ара НВК Malpighiaceae Flacourtiaceae Podostemaceae Cactaceae Malpighiaceae Flacourtiaceae Selaginellaceae Orchidaceae oraceae Melastomataceae Haemodoraceae Passifloraceae Passifloraceae Selaginellaceae raceae Orchida Gain ee Convolvulaceae Clusiaceae rat eae urseraceae Adiantaceae Araceae Gentianaceae Boraginaceae Volume 77, Number 3 Holst 4 Todzia 493 1990 Croizat's Collections from the Franco-Venezuelan Expedition 82 Couepia guianensis subsp. guianensis (Miq.) Prance пв 282-bis Inga sp. Mimosaceae 283 Crotalaria micans кн laa 284 Siparuna guianensis Monimiaceae 285 Microgramma persicarifalia ( (Schrader) C. Presl Polypodiaceae 286 Hyptis atrorubens Lamiaceae 289 Araceae 290 Dioscorea sp. Dioscoreaceae 291 Costus guanaiensis уез уаг. а (Schumann) Р. Мааз Zingiberaceae 292 Costus scaber Ruiz Lopez эй is avo эе инш 293 Mansoa kerere (Aubl.) А. С Bignoniaceae 295 Adiantum жо аан wi ex сев гу ee еае 296 Paullinia novemalat Sapindaceae 297 Ludwigia latifolia ( (Benth. ) H. Hara Onagraceae 298 icus sp. Moraceae 300 Cydista aequinoctialis (L.) Miers Bignoniaceae 301 Mikania guac Asteraceae no locality, no date 03 riplaris americana L. Polygonaceae 303-A Polygonum acuminatum HBK Polygonaceae 04 ntana sp. Verbenaceae 304-A а ѕр. Verbenaceae 306 Securidaca cf. warmingiana Chodat Polygalaceae 307 Geophila Баш (L.) Т. М. Johnston Rubiaceae 308 Mixed collecti im d ubia Es E {еч var. dubia + Scoparia dulcis L. нисе 309 Phaseolus campes Fabac 310 Polygala sprucean и ИЯ а Esmeralda, 24 July res 311 Tragia Kis ilis L тыа ЫШ 312 Coussa sp. Mora 313 Hyptis ша Benth. is o locality, no date 2315 Thelypteris hispidula (Decne. ^ C. e A ue ?315 Cy an eltis semico тең a (Sw.) J Aspleniac sn ality 317 кол Y pusifolia subsp. rhombifolia (G. Mey) A. Gentry Bignoniaceae 318 Monstera dubia (HBK) Engl. aceae 320 Clavija lancifolia Desf. Theophrastaceae 321 Picramnia latifolia Tul. Simaroubac 322 Clibadium sp. Asteraceae 323 Apeiba cf. albiflora Ducke Tiliaceae 2324 Hamelia patens oe var. glabra Oersted Rubia 2324 Utricularia triloba Ben Lentibulariaceae no locality, no date 325 Guarea о (Rich.) A. Juss. subsp. pubescens Meliaceae 327 Miconia ciliata (Rich.) DC. Melastomataceae 330 Croton palanostigma Klotzsch Do a 331 Philodendron smaragdinum Bunting Arace 333 Montrichardia linifera (Arruda) Schott raceae 334 Selaginella umbrosa Lem. ex Hieron. Selaginellaceae 335 Tynanthus ci (Bureau) Sandw. Bignoniaceae 337 Scoparia dulci Scrophulariaceae 338 Inga dein (lin DC. Mimosaceae 338-bis Inga edulis C. Mar Mimosaceae 339 Pithecellobium (Zvgia) divaricatum Benth. imosaceae 340 Wulffia stenoglossa (Cass.) DC. Asteraceae 341 Eupatorium (Chromolaena) odoratum L. Asteraceae no locality, no date 342 Cochlospermum orinocense (Kunth) Steudel Cochlospermaceae 343 Coussapoa asperifolia Trécul subsp. magnifolia (Trécul) Akkerm. & Moraceae C. C. Ber 345 Senna silvestris (Vell. Conc.) H. Irwin & Barneby Caesalpiniaceae 346 Erythroxylum divaricatum Peyr Erythroxylaceae 347 ectaria incisa Cav. Aspleniaceae 348 Cissus erosa Rich Vitaceae 348-A Cissus sicyoides L. Vitaceae 494 Annals of the Missouri Botanical Garden 349 Sauvagesia erecta L. Ochnaceae 350 Byttneria divaricata Benth. Sterculiaceae 352 Vanilla pompona Schiede Orchidaceae 353 Renealmia alpinia (Rottb.) Р. Maas ingiberaceae 355 Guettarda divaricata (Humb. & Bonpl. ex Roemer & Schultes) Standley Rubiaceae 356 Bunchosia decussiflora W. R. A Malpighiaceae 357 Clonodia complicata (HBK) v R. Anders Malpighiaceae 358 Selaginella parkeri (Hook Grev.) Spring Selaginellaceae 358.bis Microgramma PEU AU (Desv.) Sota Polypodiaceae Tiestos, no 360 Aciotis indecora (Bonpl.) Triana var. macrophylla Cogn. Melastomataceae 361 Aristolochia sp Aristolochiaceae 363 364 Cordia nodosa Lam Boraginaceae 365 Campylocentrum poeppigii (Reichb.f.) Rolfe Orchidaceae 366 eobroma cacao L. Sterculiaceae 367 Doliocarpus major Gmel. subsp. major Dilleniaceae 368 Clusiaceae 369 — sp. Arecaceae о locality, no d 371 Cordia d aa Чаш. ) 1. М. Johnston Boraginaceae 372 Burseraceae 373 Triplophyllum — id a Holttum Tectariaceae 375 Combretum dere ats bretaceae 376 Leandra solenifera Melastomataceae 377 Arrabidaea pubescens (L. ) A. Gentry Bignoniac 378 Tetrapterys muc ronata Cav. Malpighiaceae 379 Adenocalymma inundatum var. surinamensis Bureau & Schumann Bignoniaceae 380 Dalechampia magnoliifolia Muell. Arg Euphorbiaceae 2381 Philodendron smaragdinum Bunting Araceae 7381 myriaden Mimosa 383 Selaginella flagellata Spring Selaginellaceae 384 Turnera odorata Urban Turneraceae 385 Psychotria ауа Rubiaceae 386 me ie ernstü Eichler var. ernstii Menispermaceae no loc ate 387 Diodia отра (Willd.) Bremek. Rubiaceae 389 Calathea propinqua (Poeppig & Endl.) Коеп. Marantaceae 391 Rollinia exsucca (Duna Annonaceae 391-A Cymbopetalum brasiliense (Vell Conc.) Benth. Annonacea 392 Solanum stramonifolium Solanacea 392-A Cestrum latifolium var. aes (HBK) O. Schulz Solanaceae 393 Burseraceae 396 — чалан smaragdinum Bunting Araceae 397 Sterc Sterculiaceae 397-A Bac prm Benth. Sterculiaceae no locality, dat 398 Inga edulis C. Martius Mimosaceae 398.bis Inga leiocalycin Mimosaceae 99 Costus guanaiensis Rosy var. macrostrobilus (K. Sch.) P. Maas Zingiberaceae 399.A Menispermaceae 400 Philodendron smaragdinum Bunting Araceae 401 Ludwigia leptocarpa (Nutt.) H. Hara банган 405 Triplophyllum funestrum (Kunze) чн Tectariaceae 406 Strychnos matogrossensis S. Moo Loganiaceae 407 Hibiscus bifurcatus Ca Malvaceae 408 Vanilla palmarum Lindley Orchidaceae 408-A Vanilla palmarum Orchidaceae 409 Dichorisandra hexandra (Aublet) Kuntze Commelinaceae 410 Stromanthe ай. jacquinii (Roemer & Schultes) Kennedy € Nicolson Marantaceae 411 Renealmia aromatica (Aublet) Griseb. Zingiberaceae 412 Inga stipularis DC. Mimosaceae no locality, no date 413 Acalypha villosa Jacq. Euphorbiaceae 414 Iresine н Willd. Amaranthaceae 415 Oxalis Oxalidaceae 416 Geophila. repens (L.) I. M. Johnston Rubiaceae Volume 77, Number 3 1990 Holst 8 Todzia Croizat's Collections from the 495 Franco-Venezuelan Expedition Solanum stramonifolium Vahl etaxya rostrata (HBK) Presl Caladium bicolor (Aiton) Vent. Salto Salas, 18 Aug. 1 Renealmia e Жон, ) P. Maas Piper aequale Agarista duckei (Huber) Judd Upper ranges of Mt. Rimbaud Hamelia patens Jacq. var. glabra Oersted Slopes of Mt. Rimbaud Epiphyllum phyllanthus (L.) H Slopes toward the top of Mt. bam sp. Slopes of Mt. Rimbaud Isertia parviflora LE var. parviflora orest toward imbau еке. crassifolia ( (Focke) Morton Forest toward slopes of Mt. Rimbaud Inga nobilis Willd. Dicranopteris flexuosa (Schrader) oe Underw. About 3 hours navigation upstre Clidemia du ал Benth. Forest toward Mt. Mau Palas af, crocea (Sw) Roemer & Schultes Mt. Rimbaud Selaginella parkeri (Hook. & Grev.) Spring oot of Mt. Rimbau Ternstroemia punctata (Aublet) Sw. Toward summit of Mt. Rimbaud Polypodium bombycinum Maxon Toward Mt. Maunoir Xiphidium coeruleum Aublet Monstera obliqua Miq. Isla de las Hormigas, 30 Aug. 1951 i ele io, A ee 9 ) Јас̧. То Ri owa Tillandsia NC n Mez Cyrilla racemiflora er ranges of Mt. Rimbaud Huperzia linifolia (L. ) Trev. St. Léon. var. jenmanii (Underw. & Lloyd) B. Ollg. & Wind. Duroia eriopila L.f. var. eriopila Microgramma iM (Schrader) C. Presl Vellozia tubiflora (A. Rich.) H Top of Mt. Maunoir Navia sp. Slopes of Mt. Maunoir Solanaceae Metaxyaceae Araceae Araceae Zingiberaceae e Ericace Rubiaceae Family Indet. Cactaceae Gesneriaceae Rubiaceae Gesneriaceae Mimosaceae Gleicheniaceae Ericaceae Melastomataceae Melastomataceae Rubiaceae Lauraceae Selaginellaceae Theaceae Polypodiaceae Haemodoraceae Araceae Verbenaceae Bromeliaceae Cyrillaceae Melastomataceae Cyrillaceae Lycopodiaceae Rubiaceae Mimosaceae Cyrillaceae Araceae Moraceae Ericaceae Polypodiaceae Velloziaceae Bromeliaceae 496 Annals of the Missouri Botanical Garden SALTO SALAS, 10-21 AUG. 1951; 02?16'N, 64?19'W, ca. 150 m. elev. Spathiphyllum m (Dryander) Schott Mikania psilosta . DC. Erythrochiton brasiliensis Nees & C. Martiu Stromanthe aff. jacquinii ay & Schultes) Kennedy & Nicolson Polypodium bombycinum Maxo Paullinia s rymonia ore (Jacq.) C. Martius Coccoloba latifolia Lam. q: Selaginella umbrosa Lem. ex Hieron. Pouteria ? plicata Penn. Rudgea stipulacea (A. DC.) е Iresine diffusa Willd. Adiantum и 1. йн e Trichomanes pin es um Hedwig Ble шын AMEN Rich. ani latifolium Lam. Trichomanes pinnatum Hedwi Erythro Dum vernicosum O. Schulz Asplenium serratum L. Pityrogramma calomelanos (L.) Link Dichorisandra hexandra (Aublet) Standley Astrocaryum gynacanthum C. Martius o locality, no date Paphinia cristata Lindley Piper demeraranum (Miq.) C. DC. fn Lii villosa Jacq. cotea Tresine diffusa Catasetum ag as (Lindley) Lindley, vel aff. Euter no кай no date eds monticola Barb. Rodr. o locality, no date Сес) eS кш С. Martius Ipomoe L. round adco Sw. var. sylvestris Mikania чаш no locality, п Mikania leiostachya Benth. no loc dat коз nitidella (Muell. Arg.) Standley Guazuma ulmifolia Lam. Clidemia iaga (Aublet) C. Martius no locality, no date Apinagia richardi {апа (Tul.) van pie El Motin (lower ranga 13 Sep. 195 Cecropia latiloba Miq. Stromanthe aff. jacquinii (Roemer & E Kennedy & Nicolson Aphelandra deppeana Schldl. & Cham Micropholis egensis (A. DC.) Pierre Aniseia martinicensis (Jacq.) Choisy raceae Asteraceae Melastomataceae raceae Нушеторнуйегае Blechna Adianta a ИРА Arecaceae Orchidaceae Arecaceae Arecaceae raceae Convolvulaceae Flacourtiaceae Asteraceae Asteraceae Rubiaceae Sterculiaceae Melastomataceae Podostemaceae Melastomataceae Asteraceae Convolvulaceae Volume 77, Number 3 Holst & Todzia 1990 Croizat's Collections from the 497 Franco-Venezuelan Expedition 519 Bauhinia glabra Jacq. 520 Nectandra cuspidata Nees 521 Psychotria calviflora Steyerm 523 Thoracocarpus bissectus (Vell Conc. 2 Harling 524 Campyloneurum phyllitidis (L.) C. Presl 525 a glomerata (Nees) Mez 2527 -bis Pithecellobium divaricatum Benth. 2527 -bis Inga nobilis Willd. 528 Bactris corossilla Karsten no locality, no 529 Monstera obliqua Miq 530 Sarcoglottis sp.? 531 Maxillaria camaridii Rei 533 Stanhopea grandiflora (Lud. ; Lindley 534 Catasetum barbatum (Lindley) Lindley 536 Quiina parvifolia. Pulle 538 Aechmea setigera C. Martius ex Schultes f. 539 Aechmea angustifolia Poeppig & Endl. 540-A Piper sp. no оо, no date 541 Cous p- 542 miu nudum (L.) Pal. 543 Begonia humilis Aiton SALTO CODAZZI, 30 AUG. 1951; 02*15'N, 64?20'W; ca. 150 m elev. 543-A Begonia humilis Aiton ISLA DE LAS HORMIGAS, 30 AUG.; 02°15'N, 64?18'30"N 544 Ficus paraensis (Miq.) Miq. PTO. ESPERANZA, 31 AUG.; 02?14'58"N, 64?16'44"W; 177 m elev. 546 Rinorea pubiflora (Benth.) Sprague & Sandw. ISLA DEL ESFUERZO, 1-5 SEP. 1951; 02°15'N, 64°17'W 547 Possible xd collectio Guat от С. Martius + Pa a dolichocarpa Sprague & Sandw. 548 Thoracocarpus bissectus en Cone.) Harling sla de ещ 7 Sep 1 550 Pii sicyoides L. 951 aipa sp. 552 arica abysmophila Maguire & Steyerm. 553 Passiflora vespertilio L. 554 Sabicea о Benth. subsp. duidensis Steyerm. 556 Cissampelos p 557 ae fasciculata O. Berg 558 Turnera odorat 559 Physalis pea йн s L. 560 Apteria aphylla (Nutt.) Barnhart ex Small 561 Monotagma plurispicatum (Koern 562 Schefflera morototoni (Aublet) Maguire, Steyerm. & Frodin 563 Epiphyllum phyllanthus (L.) Haw LOS TIESTOS, 7-10 SEP. 1951; 02*13'N, 64°10'W 564 Aechmea rubiginosa Ме? 565 Croton boultonianus Cro 566 Erythrochiton пме инн “Nees . Martius 568 з ый ен megalophylla т ; Sota 569 Hylaea 570 Abuta erandifolia (C. Sras, Sandw. 2571 Metaxya rostrata ( 2571 Microgramma tecta (Kaulf. a Alice 572 Apeiba schomburgkii Szyszyl. 573 Philodendron scandens C. Koch Caesalpiniaceae uraceae Piperaceae oraceae Psilotaceae Begoniaceae Begoniaceae Moraceae Violaceae Annonaceae Cyclanthaceae Araliaceae Cactaceae Bromeliaceae Euphorbiaceae Rutaceae Polypodiaceae Polypodiaceae Tiliaceae Araceae 498 Annals of the Missouri Botanical Garden 574 пое par is (Miers) Barneby & Krukoff Menispermaceae 575 Monst adanso Укен. var. laniata (Schott) Madison cea 576 Adiantum але ines m Willd. Adiantaceae 578 Scaphyglottis sickii Р rchidaceae 580 eromi w.) Loudon Piperaceae 580-A Piper de gl Sw. f. hispidum Piperaceae no loc o date 581 рыма Из (НВК) Eichler Loranthaceae EL MOTIN, 12-14 SEP. 1951; 02°13'N, 64°10'W 582 Ficus guianensis Desv. Moraceae 583 Eugenia sp. Myrtaceae 584 Socratea exorrhiza (C. Martius) Wendl. Arecaceae no locality, no da 585 Oncidium nanum Lindley Orchidaceae 586 Gustavia poeppigiana O. Ber Lecythidaceae La Esmeralda, 16 July 1 586-A Eschweilera pedicellata (Rich.) S. Mori Lecythidaceae 587 Calycolpus goetheanus ) r Myrtaceae 588 Psychotria aff. platyp Rubiaceae 589 andia venezuelensis Steyerm Rubiace 590 Matelea stenopetala . Asclepiadaceae 591 Apinagia longifolia (Tul.) van Royen Podostemaceae 592 Esenbeckia pilocarpoides Kunth subsp. pilocarpoides Rutaceae 592-A Esenbeckia pilocarpoides dise subsp. pilocarpoides Rutace 593 Passiflora longiracemosa Passifloraceae ?594 sychotria racemosa (Aublet) Raeusch. Rubiaceae ? Morinda tenuiflora (Benth.) Мешкей var. leiophylla (Steyerm.) Steyerm. ^ Rubiace 595 Rubiaceae? 596 Dieffenbachia seguine (L.) Schott raceae 597 Angostura trifoliata (Willd.) T. Elias Rutaceae 598 Acaly, . тасг Euphorbiaceae 599 Eschweilera pedicellata (Rich.) S 2d Lecythidaceae 600 Eschweilera pedicellata о ) S. Mor Lecythidaceae 600-A Eschweilera coriac C. Martius. ex O. Berg Lecythidaceae no ligi no date 601 Oxandra s Annonaceae 603 Vismia cayennensis (Jacq.) Pers. Clusiaceae 605 Calliandra stipulacea Benth. dris 605-bis Calliandra stipulacea Benth. Mimos 06 Hae a ee 607 Campyloneurum phyllitidis (L.) С.Р de eas 608 Pouteria caimito (Ruiz Lopez & Pavón) Rad, Sapota 610 Vanilla pompona Schiede Orchidecen 611 cf. Stromanthe jacquinii чаш & Schultes) Kennedy & Nicolson Marantaceae 612 Piper francovilleanum 2 Piperaceae 613 Calathea densa (Koch) R ara 614 Mecardonia arrear ie Small Scrophulariaceae Salto Salas, 15 Sep. RAUDAL BOBADILLA, 15-16 SEP. 1951; 02?13'N, 64°07'W; ca. 200 m. elev. 617 Dioclea virgata (Rich.) Amshoff var. virgata 618 Caladium bicolor (Alton) Vent. 619 Mabea sp. 620 Miconia cf. bubalina (Don) Naud. 621 Passiflora coccinea Au d Ugueto, 15 Sep. 622 Brosimum guianense ЕМ Н. E. Huber 623 Rinorea sp. 624 Ouratea croizatii Maguire & Steyerm 625 Jacaranda obtusifolia subsp. rhombifolia ~ Mey) A. Gentry 626 Oldenlandia lancifolia (Schumann 627 Brassia bidens Lindley 628 Mabea occidentalis Benth. 630 Ocotea sp. le race Бироа со as Passifloraceae Violaceae Ochnaceae Bignoniaceae Rubiaceae Orchidaceae Euphorbiaceae Lauraceae Volume 77, Number 3 Holst 8 Todzia 1990 Croizat's Collections from the Franco-Venezuelan Expedition EL QUESO, 17 SEP. 1951 631 Dichorisandra hexandra e) Standley 633 Guarea мы а (L.) Sleu 634 Hylocere 635 ое reptans (Cav.) А. К. Smith 637 Pithecellobium latifolium Bent 638 Anthurium sinnuatum Clan. 639 Chamissoa altissima (Jacq.) HBK RAUDAL MONTSERRAT, 17 SEP. 1951; 02%08'N, 63°53'W 642 Polypodium в Splitg.? 643 Tectaria incisa 644 Triplophyllum Ge (Kunze) Holttum RAUDAL BOBADILLA, 17-18 SEP. 1951; 02*13'N, 64%07'W 645 Teliostachya cataractae Nee 646 Aphelandra deppeana аны. & Cham. 647 lada 5 audal Monts 648 Campylocentrum ae (Reichb.f.) Rolfe RAUDAL MONTSERRAT, 19-20 SEP. 1951; 02?08'N, 63*53'W 649 Guettarda divaricata (Humb. € Bonpl. ex Roemer & Schultes) Standley 650 Psychotria bracteocardia DC.) Muell. Arg 651 Asplenium claussenii Hieron.? 652 ee cal (L.) Link 653 Trichilia s 654 Sauvagesia erecta L. 655 oe ка Kunth & Bouché 656 Paypayro 657 Epidendrum alias group) 657-A Epidendrum secundum Jacq. 659 Hamelia patens Jacq. var. glabra Oersted 660 Mendoncia sp. 661 Stizophyllum Pr i Sandw. Los Tiestos, 9 Se 662 Psychotria deflexa A. pc. iden campyloneura wey Arg.) Steyerm. 663 мат lancea (L.) Bedd. x L. divaricata Klotz 664 Duguetia sp. 666 ^i di denm Ruiz Lopez & Pavón 667 Pleurothallis sp. 670 Mendoncia sprucei Lindau 671 Palicourea calophylla A. DC 672 Pouteria surumuensis Baehni 673 Nidema ottonis Reich 74 Pteris punge illd 677 4-bis etaxya rostrata uL ш 75 Adiantum petiolatum 675-bis Lomariopsis japurensis (С. “Mart tius) J. Sm 76 Bulbophyllum pachyrhachis (A. Rich.) Griseb. 678-A ¡per arboreum et var. arboreum no locality, no 679 Heteropsis tenuispadix Bunting 680 Piper bartlingianum ы, .) C. DC. 681 Besleria laxiflora 682 ac, a [* ) Dodson & Dressler 682-A Psygmorchis pusilla (L.) Dodson & Dressler 682-B Psygmorchis pusilla (L.) Dodson & Dressler La Mantequilla, 24 Sep. 1951 682-C Pleurothallis sp. La Mantequilla, 683 Scaphyglottis к Rolo) B. R. Adams, vel aff. 685 Odontocarya sp. Commelinaceae Meliaceae Cactaceae Polypodiaceae Mimosaceae Araceae Amaranthaceae Polypodiaceae Aspleniaceae Tectariaceae Acanthaceae Acanthaceae Acanthaceae Orchidaceae Rubiaceae Ru Bignoniaceae Rubiac Dennstaedtiaceae ace Piperaceae Araceae Orchidaceae Orchidaceae Orchidaceae Menispermaceae 500 Annals о Missouri с Garden 687 Oxandra 688 Свин fasciculata O. Berg 689 691 Vittaria costata Kun Pica de la Mañicquills, 20 Sep. 1951 692 edo em (Benth.) Sw. 694 Adiant osum e 695 Blechnum ee 696 Corynostylis на 7 ) S. F. Blake 697 Spathiphyllum canniifolium (Dryander) Schott LA MANTEQUILLA, 21-23 SEP. 1951; 02°08'N, 63°52'W 698 Batemannia colleyi Lindley 699 Vittaria costata Kunze 700 Selaginella umbrosa Lem. ex Hieron. 701 Tectaria plantaginea (Jacq.) Maxon var. macrocarpa аш С. Могїоп 702 Aciotis indecora (Bonpl.) Triana var. macrophylla Cog Raudal Montserrat, 21 Sep. 1951 703 Doryopteris == ee Raudal Mon at, 21 Sep. 704 Peristeria puse Коз & ain ., vel aff. 705 Trichomanes ekmanii W. Boer 706 Diplazium grandifolium (Sw.) Sw 707 Campyloneurum angustifolium (Sw.) Еее 708 Trichomanes collariatum Bosc 710 Campyloneurum repens (Aublet) С. Presl 711 Asplenium serratum L. 712 Asplenium auritum ay. s.l. ?713 Asplenium cirrhatum Rich. ex Willd. ?713 Perama galioides (HBK) e La Esmeralda, 14 July 714 Diodia ocimifolia (Willd. ) one 715 Masdevallia norae Luer 716 Pleopeltis percussa (Cav.) Hook. & Gre 717 Polypodium polypodioides (L.) Watt. var. burchellii (Baker) Weath. 1 Mollinedia sp. 718-A Mollinedia sp. Salto 2 25 Sep. 1951 718-B Mollinedia no locality, no date 119 Hecistopteris pumila (Sprengel) J. Sm. 720 Campyloneurum phyllitidis pe A C. Presl 721 Antrophyum guayanense Hier 722 Renealmia monosperma Miq. 723 Asplenium haplophyllum Domin 724 Dichaea trinitensis Gleason 725 Campylocentrum micranthum (Lindley) Rolfe 726 Diplazium cristatum (Desr.) Alston 727 Trema dia ura (Beurling) Standley 728 Tectaria trinitensis Maxon 729 Pleopeltis percussa (Cav.) Hook. & Grev. 730 Trichomanes ekmanii W. Boer 731 Aspasia variegata Lindle 732 Microgramma persicariifolia arog C. Presl. 7 Niphidium crassifolium (L.) L 733-bis rammitis mollissima (Fée) P Triplophyllum ie ~ Holttum 735 Asplenium auritum Sw SALTO ROJAS, 25 SEP. 1951; 02?08'N, 63°50'W; ca. 300 m elev. 736 Maranta humilis Aublet 2737 Thelypteris x rolandii (C. Chr.) R. Tryon 2737 Drymonia coccinea (Aublet) Wiehler 738 Ardisia guianensis (Aublet) Mez 740 Virola elongata (Benth.) Warb. 741 Sciaphila purpurea Benth. Annonaceae Myrtaceae Myristicaceae Adiantaceae Burseraceae Araceae Orchidaceae eae Me on Adiantaceae Orchidaceae Hym enophyllaceae olypodiaceae Monimiaceae Monimiaceae Adiantaceae Hymenophyllaceae Orchidace AA Polypodiaceae Aspleniaceae Marantaceae Triuridaceae Volume 77, Number 3 Holst 8 Todzia 501 1990 Croizat's Collections from the Franco-Venezuelan Expedition 742 Loreya mespilioides Miq Melastomataceae 743 Souroubea dasystachya Gilg & Werderman Marcgraviaceae 744 с Orchidaceae o locality, n 745 Miconia ee (Rich.) Don oe 746 Pleurothallis sp. Pda 747 Psychotria acuminata Benth. Rubiac 748 Bauhinia ungulata innu M 750 Dieffenbachia duidae (Steyerm. 4 Bunting racea 751 Nectandra globosa (Aublet) Mez Laurace 753 Sciaphila purpurea Benth. Triuridacea 754 Cheiloclinium serratum (Cambess.) A. lu Hippocrateaceae 755 Ludwigia foliobracteolata (Munz) H. Onagraceae 755-A Ludwigia чи (Munz) Н. m Onagraceae no locality, n 755-B Ludwigia a (Nutt.) H. Hara Onagraceae no locality, no date 756 Codonanthe calcarata (Miq.) Hanst. Gesneriaceae 757 Cyathea pungens (Willd.) Domin Cyatheace 757-bis Thelypteris tristis (Kunze) R. Tryon Thelypteridaceae 58 Urera baccifera (L.) Gaudich. Urticacea 758-A Urera baccifera (L.) Gaudich. Urticaceae no locality, no date UGUETO, 29 SEP. 1951; 02*%08'22"N, 63%48'47"W; 300 m elev. 762 Clarisia sp. Moraceae 763 Stelis s Orchidaceae Desmodium wydlerianum Urban Fabaceae 764-bis Metaxya rostrata (HBK) Presl Metaxyaceae Raudal Montserrat, 19 . 1951 765 Diplazium cristatum (Desr.) Alston Aspleniaceae 767 Thoracocarpus bissectus (Vell. Conc.) Harling Cyclanthaceae mixed with inflorescence of Asplundia sp. 768 Brosimum lactescens (S. Moore) С. С. Ber Moraceae 769 Bixa orellana Bixacea 770 Dryopteris guianense (Klotzsch) L. D. Gómez кое 771 Adiantum humile Kunz diantac 771-bis Trichomanes collariatum мену vel aff. Hym cc И ?772 Psychotria е Steye ubiaceae ocality, no da 2772 Miconia dispar Benth Melastomataceae 773 Miconia lateriflora Cogn Melastomataceae 774 Byttneria ed hie Jacq subsp. catalpifolia Sterculiacea 775 Gurania trialata Cog Cucurbitaceae 776 Clusiaceae 777 Pleurothallis Orchidaceae 778 a ү а (Fee) Т. Moore Aspleniaceae 780 Psychotria racemosa (Aublet) Raeusch Rubiace. 782 Cyclodium guianense (Klotzsch) L. D. Gómez Aspleniaceae 783 Microgramma tecta (Kaulf.) Alsto Polypodiaceae 784 Nephrolepis biserrata (Sw.) jos 2 Davilliaceae 784-bis Asplenium claussenii Hieron. aff. Aspleniaceae 785 rugmansia — ын (Wili: ] [nm & Presl Solanac 787 Peperomia aa w.) Piperaceae 788 Adiantum term uum e ex x Mig. Adiantaceae 789 Paullinia эзше Radlk. Sapindaceae 790 од fintelmanii Wagener ех Schldl. Tropaeolaceae 791 Psychotria horizontalis Sw. var. glaucescens (HBK) Ste Rubiaceae 792 Morinda tenuiflora (Benth.) sy ines var. leiophylla (Steyerm.) Steyerm. Rubiaceae 793 Bolbitis semipinnatifida (Fée) A Aspleniaceae 794 Asplenium delitescens (Maxon) "p. Gómez Aspleniaceae 795 Polypodium decuman illd Polypodiaceae 796 Mendoncia cardon Acanthace 797 Hymenophyllum hirsutum (L.) Sw. Hymenophyllaceae 797-A Hymenophyllum polyanthos (Sw.) Sw. Hymenophyllaceae 797-B Hymenophyllum polyanthos (Sw.) Sw. Hymenophyllaceae 502 Annals of th Missouri Botanical Garden 798 Saccoloma elegans Kaulf. Dennstaedtiaceae 799 Conceveiba guianensis Aublet Euphorbiaceae 800 Bertiera guianensis ceo subsp. guianensis var. guianensis Rubiaceae 801 a grandifolia DC. Meliaceae 803 Dendropanax arboreum (L.) Decne. & Planchon Araliaceae 804 Chamissoa altissima (Jacq K Amaranthaceae 805 Adenocalymma impressum (Rusby Bignoniaceae 806 Dicranoglossum desvauxii (Klotzsch) Proctor Polypodiaceae 808 Adiantum terminatum Kunze ex Miq. Adiantaceae 809 Bunchosia decussiflora W. R. Anders. Malpighiaceae 809-A Clusiaceae no locality, no date 810 Apeiba aspera Aublet Tiliaceae 811 Катана diversifrons ere Mett. ex = Hymenophyllaceae 811 -bis Selaginella parkeri (Hook. & Grev.) S Selaginellaceae 812 Psyc d poeppigiana Muell. Arg. S barcellana (Muell. Arg.) Rubiaceae Steye 813 NES A aea РУР Совп. Cucurbitaceae 814 Manettia reclinata Mutis Rubiaceae 814-A Psychotria ulviformis Steyerm. Rubiaceae 815 Sagotia racemosa Baillon Euphorbiaceae 816 Rinorea pubiflora (Benth.) vv & Sandw. Violaceae 817 Lindackeria paludosa (Benth.) Gilg oo 818 Rinorea lindeniana (Tul.) Kuntze Vi 819 Micropholis egensis (A. DC.) Pierre кола 821 Didymochlaena truncatul Aspleniaceae 822 Dicranoglossum desvauxii (Klotzsch) Proctor Polypodiaceae 823 Hylaeanthe sp. Marantaceae 824 Trichomanes pinnatum Hedwi Hymenophyllaceae 825 Ryania speciosa Vahl. var. bie olor (A. DC.) Monach. Flacourtiaceae 826 lacio excelsum (Be us е Caesalpiniaceae 827 Theobroma bicolor Humb. & "Bon Sterculiaceae RAUDAL DEL QUESO, 17 OCT. 1951 828 Inga sertulifera DC. Mimosaceae UGUETO, 23 OCT. 1951 829 Polypodium caceresii Sodiro Polypodiaceae The following ари use the Croizat numbering series оп the labels, but mention по localities or dates. It is believed that these are the collections of Pablo Anduze and Luis Carbonell from the Rio Orinoco, above Ugueto. gueto was the ша point that Croizat traveled on the expedition. The collections seem to have been arranged by family, but many were improperly identified and so do not follow the numerical sequence exactly. 830 Ruellia rubra Aublet Acanthaceae 831 Mendoncia cardonae Leonard Acanthaceae 832 о altissima (Jacq.) НВК Amaranthaceae 833 Anno xs Annonaceae 834 Dugue Annonaceae 835 с, cardoniana R. E. Fries Annonaceae 836 Mixed collect Annonaceae + А (A + Clusiaceae Clusiaceae 837 Guatteria Аа К. E. Fries Annonaceae 7838 Annona sp. Annonaceae 7838 Clusiaceae Guatteria recurvisepala R. E. Fries Annonaceae 7840 Ferdinandusa goudotiana Schumann var. eciliata Steyerm. Rubiaceae 2840 Cymbo Ber. brasiliense (Vell. Conc.) Benth. Annonaceae 841 Tabernaemontana heterophylla Vahl Apocynaceae 842 Monstera obliqua Miq. Araceae 842-A Monstera dubia (HBK) Engl. Araceae 843 Monstera lechleriana Schott Araceae 843-А Monstera obliqua Araceae 8 Piper cernuum Vell. Conc.? Piperaceae 844-A Monstera dubia (HBK) Engl 845 Heteropsis tenuispadix Bunting Araceae Volume 77, Number 3 Holst 8 Todzia 503 1990 Croizat's Collections from the Franco-Venezuelan Expedition 846 Dieffenbachia жү (Jacq.) Schott Araceae 847 Monstera obliq Araceae 848 Monstera dubia (HBR) Engl. Araceae 849 Philodendron scandens C. Koch & Sello Araceae 850 Heteropsis melinonii (Engl.) A. Jonker & Jonker Araceae 851 Araceae 852 Philodendron maguirei Buntin Araceae 853 Anthurium Mio e a Araceae 854 Monstera obliqua M Araceae 855 Araceae 856 Tabebuia pilosa A. Gentry Bignoniaceae 856-A Crescentia amazonica Duc Bignoniaceae 856-B Paragonia d suia (Rich. ) шын Bignoniaceae 856-C Cydista id is (L.) Mie Bignoniaceae 857 Bixa orellan Bixaceae 858 Aechmea Мур ЕВ André Bromeliaceae 859 Anacardiaceae 860 Burseraceae 863 Mixed collectio Hylaeanthe 2 + leaves of cf. Marantaceae Calathea splendens (Lem.) Regel 864 Melothria pendula Cucurbitaceae 866 oanea guianensis (Aublet) Benth. iral ы 867 Satyria и к ) Benth. & Hook.f. Ericac 867-A ibaudia nutans шм 867-B Agarista duckei шо, Judd Ericaceae 867-C Befaria sprucei Meissne Ericace 68 Conceveiba guianensis Aublet Euphorbiaceae 868-G Conceveiba guianensis Auble Euphorbiaceae 868-H Hieronima laxiflora (Tul.) Muell. Arg Euphorbiaceae 868-1 Hieronima oblonga (Tul.) Muell. Arg Euphorbiaceae 869 Maprounea guianensis Aublet Euphorbiaceae 869-F Alchornea sp. Euphorbiaceae 870 Hieronima laxiflora (Tul.) Muell. Arg. Euphorbiaceae 870-E Acalypha villosa Jacq Euphorbiaceae 871 Aparisthmium lia (A. Juss.) Baillon Euphorbiaceae 871-D Mabea occidentalis Benth. Euphorbiaceae 872 Alchornea s Euphorbiaceae 872-C Phyllanthus orbiculatus Rich Euphorbiaceae 873 Aparisthmium cordatum (A. Juss. ) Baillon Euphorbiaceae 873-B Mabea sp. Euphorbiaceae 874 Acalypha macrostachya Jacq. Euphorbiaceae 874-A Acalypha macrostachy Euphorbiaceae 875 splenium cirrhatum Rich. ex Aspleniaceae 876 Pleopeltis percussa (Cav.) Hook. & Grev. Polypodiaceae 877 splenium m Sw Aspleniaceae 878 Polypodium peaje Sw Polypodiaceae 879 Polypodium triseriale Polypodiaceae 880 Triplophy llu aos (Kunze) Holttum Tectariaceae 881 anoglossum desvauxii (Klotzsch) Proctor Polypodiaceae 882 in citrifolium (L.) Splitg. Adiantaceae 883 Selaginella diffusa (С. Presl) е Selaginellaceae 884 Cochlidium tepuiense (A. C. Smith) L. E. Bishop Polypodiaceae 885 Flacourtiaceae 886 Casearia arborea (Rich.) Urban Flacourtiaceae 887 Flacourtiaceae 888 Ryania speciosa Vahl var. bicolor (A. DC.) Monach. Flacourtiaceae 889 Besleria sp. Gesneriaceae 900 Caraipa tereticaulis Tul. Clusiaceae Of doubtful occurrence in Venezuela, det. K. Kubitzki 900-A Clusiaceae 901 Caraipa punctulata Ducke Clusiaceae 902 Clusiaceae 903 Clusiaceae 904 Clusia 904-A ta guianensis Aublet subsp. guianensis Chrysobalanaceae 905 Clus Clusiaceae 504 Annals {өңе а Garden 906 Clusiaceae 907 Clusiaceae ?908-B Caraipa punctulata Duck Clusiaceae ?908-B Irlbachia ppe Sae Р. Maas subsp. alata Gentianaceae 908-C Epistephium s Orchidaceae 2909 Clusiaceae 2909 Hyptis mutabilis (Rich.) Briq. Lamiaceae 2910 Clusiaceae 2910 Ocotea aciphylla (Nees) Mez uraceae 911 Ocotea aciphylla group, flowers diseased, probably O. aciphylla (Nees) Mez Lauraceae 912 Ocotea cernua (Nees) Mez auraceae 913 cf. и cuspidata Nees Lauracea 914 Ocotea Lauraceae 915 Nectandra reticulata (Ruiz Lopez & Pavon) Mez Laurac 916 Ocotea guianensis Auble Lauraceae 917 Rhodostemonodaphne kunthiana (Nees) Rohwer Lauracea 918 Persea croizatii van der Werff Lauraceae 919 Laurac 919-B Endlicheria gracilis Kosterm. Lauraceae 919-C Nectandra cus t е Lauracea 919-D Aiouea guianensis Aublet Lauraceae 919-E Nect a cuspidata Ne Lauraceae 921 Senna silvestris (Vell. Conc.) H. Irwin & Barneby Caesalpiniaceae 923 Senna multijuga (Rich.) H. Irwin & Barneby Caesalpiniaceae 924 Phaseolus appendiculatus Benth. Fabaceae 2930 s e DNE Mimosaceae ?930 Muc sp. Fabaceae 931 P ns basillaris (L. xi H. Irwin & Barneby Caesalpiniaceae 933 Mucuna urens (L. Fabaceae 934 Inga pezizifera Buon. Mimosaceae 938 Smila xs Liliaceae 939 Centropogon cornutus (L.) Dru Campanulaceae 943 Mascagnia macrodisca (Triana. & » i Niedenzu Malpighiaceae 944 Bunchosia decussiflor . R. And Malpighiaceae 945 Stigmaphyllon hypoleucum Miq Malpighiaceae 946 Bunchosia decussiflora W. R А Malpighiaceae 947 Monotagma laxum (Poeppig & Endl.) Schum. Marantacea 948-A Heliconia episcopalis Vell. Conc. Musaceae 948-B Renealmia alpinia (Rottb.) P. Maas Zingiberaceae 948-C Marantaceae 948-D Renealmia aromatica (Aublet) Griseb. Zingiberaceae 948-E Renealmia aromatica (Aublet) Griseb. Zingiberaceae 948-F Costus arabicus L. Zingibera 949 Miconia cf. minutiflora (Bonpl.) DC. Melastomataceae 949-А Bellucia grossularioides (L.) Triana Melastomataceae 950 Miconia s Melastomataceae 950-A Miconia amplexans (Crueg.) Cogn. Melastomataceae 951 Miconia spennerostachya Nau Melastomataceae Surely from Peru or Bolivia, not Venezuela, det. Wurdack 952 Miconia centrodesma Naud. Melastomataceae 953 Bellucia grossularioides (L.) Triana Melastomataceae 954 Miconia dispar Benth. Melastomataceae 955 Aciotis purpurascens (Aublet) Triana Melastomataceae 956 Ficus maxima P er Moraceae 957 Cecropia membranacea Trécul Moraceae 958 icus sp. Moraceae 959 Coussapoa о Cuatrec. Moraceae 960 Ficus eximia Moraceae 961 Coussapoa Moraceae 962 ourouma melinonii Benoist Moraceae 963 ‘icus nymphaeifolia Р. Miller Moraceae 964 Brosimum guianense (Aublet) H. E. Huber Morace 968-A Iryanthera нар (Benth.) Myristicaceae 968-B Virola sebifera Au Myristicaceae 969 Osteophloeum нр ИЕ (DC.) Warb. Myristicaceae 970 Osteophloeum rp ders up ) Warb. Myristicaceae 971 Cybianthus viridiflor us А. mith Myrsinaceae Volume 77, Number 3 Holst 8 Todzia 505 1990 Croizat's Collections from the Franco-Venezuelan Expedition 973 Calycolpus goetheanus (DC.) O. Berg Myrtaceae 973-A Calyptranthes fasciculata O. Berg Myrtaceae 973-B 'yrcia bracteata (Rich.) DC. Myrtaceae 973-С Myrcia dichasialis McVa Myrtaceae La Esmeralda, 10 July 1951 975 Caularthron bicornutum (Hook.) Raf. Orchidaceae 975-A Mixed collection: Orchidaceae caphyglottis sp. + Epidendrum strobiliferum Reichb.f. 76 Notylia sp. Orchidaceae 976-A pata ee Reichb.f. Orchidaceae 977 Oncidium nanum Lindle Orchidaceae 977-A as тыш (L.) Lindley Orchidaceae 978 Orlean 5р. Orchidaceae 978-A Brassia iib m (L.) Lindley Orchidaceae 979 Brassia sp. Orchidaceae 979-A Dichaea brachypoda Reichb.f. Orchidaceae 980 Maxillaria amazonica Schltr. Orchidaceae 981 Sigmatostalix amazonica Schltr. Orchidaceae 982 Oncidium s Orchidaceae 983 Sobralia fragrans Lindley Orchidaceae 984 Oncidium cebolleta e. ) Sw. Orchidaceae 985 Vanilla pm Schiede Orchidaceae 986 Orleanesia Orchidaceae 987 Seaphyglotis prolifera Cogn., vel a Orchidaceae 988 Clo a warscewiczii (Reichb.f.) Dodson Orchidaceae 989 Dichaea pendula (Aublet) Cogn., vel aff. Orchidaceae 990 ollea hemixantha Reichb. f. Orchidaceae 99] Epidendrum secundum Jacq., vel aff Orchidaceae 992 Epidendrum schomburgkii Lindley Orchidaceae 993 Que microscopica Li Orchidaceae 994. Iriartella setigera (C. Martius) Wendl Arecaceae 995 Iriartella setigera (C. d Wendl. Arecaceae 995-A yospathe elegans C. Mart Arecaceae 995-B p- Arecaceae 995.C Bactris corossilla Karste Arecac 96 Passiflora cf. maguirei Killip Passifloraceae 1000-A iper cernuum Vell. Conc.? Piperaceae 1001 Piper sp. Piperaceae 1001-A Piper cf. augustum Rudge Piperaceae 006 Piper s Piperaceae 1016 аш вр. Polygalaceae 1017 Rhamnaceae 1019 ошно barbiflor га А Rubiaceae 1020 Palicourea guianensis Aublet subsp. occidentalis Steyerm. Rubiaceae 1021 Palicourea guianensis Aublet subsp. occidentalis Steyerm. Rubiaceae 1022 Psychotria tepuiensis (Steyerm.) Steyerm. Rubiaceae 1023 elect nitidella (Muell. Arg.) Standley Rubiaceae 1024 Psychotria berteriana A. DC. subsp. Ma (Rusby) Steyerm. Rubiaceae 1025 Pa buen nitidella (Muell. Arg.) Standley Rubiaceae 1026 udgea woronowii ai ley Rubiaceae 1027 Psychotria barbiflora A. DC. Rubiaceae 1028 Psychotria mico Ruíz dew ex Standley biaceae 1029 Psychotria poeppigiana Muell. . subsp. E (Muell. Arg.) Rubiaceae Steyerm 1030 Ferdinandusa goudotiana Schumann var. eciliata Steyerm. Rubiaceae 1031 Schradera polycephala A. DC. Rubiaceae 1032 Psychotria alba Ruíz Lopez & Pavón, new to Venezuela Rubiace 1033 Paullinia ingaefolia Rich. Sapindaceae 1035 Paullinia ingaefolia Rich. Sapindaceae 1036 otaceae 1037 Chrysophyllum sanguinolentum (Pierre) Baehni subsp. balata (Ducke) Penn. Sapotaceae 1038 seracea 1040 Anacardiaceae 1040-B Burseraceae 1040-C Toulicia pulvinata Radlk. Sapindaceae 1040-D Cupania scrobiculata Rich. Sapindaceae 1040-E Paullinia leiocarpa Griseb. Sapindaceae 506 Annals of the Missouri Botanical Garden 1042 Virola sebifera Aublet Myristicaceae 1043 Myrsinaceae 1044 Besleria sp. Gesneriaceae 1044-A Besleria laxiflora Benth Gesneriaceae 1044-B Codonanthe usi (Fo м Могїоп Gesneriaceae 1045 Aechmea penduliflora A Bromeliaceae 1045-A Aechmea cf. chantinii ун A Baker Bromeliaceae 1046 Caraipa tereticaulis Tul. Clusiaceae oubtful occurrence in ее det. Kubitzki 1047 Amphirrhox latifolia C. Mart Violaceae 1048 Clusiaceae 1049 Pouteria? filipes Eyma Sapotaceae 1051 Byttneria catalpifo olia Jacq. a catalpifolia Sterculiaceae 1052 Tontelea ovalifolia (Miers) A. С. Smith Hippocrateaceae 1052-A Pouteria? filipes Eyma Sapotace 1052-B Anaxagorea dolichocarpa Sprague & Sandw. Annonaceae 1052-С Endlicheria sp. Lauraceae 1053 Solanum rugosum Dunal Solanaceae 1054 Solanaceae 1055 Solanum subinerme Jacq. Solanaceae 1059 Freziera calophylla Triana & Planchon acea 1060 Schlegelia spruceana Schumann Bignoniaceae 1061 Schlegelia spruceana шй Bignoniaceae 1062 Rinorea pubiflora (Benth.) Sprague & Sandw. Violacea 1063 Violace 1064 Burseraceae 1065-A Rinor Violac 1065-B Кїлогеа pom (C. iP Kuntze Violaceae 1066 Schradera polycepha Rubiaceae 1067 Palicourea nitidella (Muell Arg. ) Standley (probably) Rubiaceae 1068 Palicourea nitidella (Muell. Arg.) St Rubiaceae 1069 Tabernaemontana heterophylla Va Apocynaceae 1070 Tabernaemontana heterophylla Vahl Apocynaceae 2. FAMILY LisT This list is arranged into two major groups: ‘Pterido- genus, and species, followed by Croizat’s collection num- ber(s). PTERIDOPHYTES ADIANTACEAE Adiantum cajennense Willd. ex 295 Klotzsch Adiantum dolosum Kunze 694 Adiantum humile Kunze 771 ue latifolium Lam. 486 Adiantum pet е pipa 675 vine pulverulen 482 Adiantum рентна "Willd. 5176 i al terminatum Kunze ex 188, 808 Poem m (L.) Splitg. 882 Antrophyum guayanense Hieron. 721 Puce sagittifolia (Raddi) J. 703 Smith Hecistopteris pumila (Sprengel) J. 719 Smith Por calomelanos (L.) 490, 652 Link Pteris propinqua Agardh Zig Pteris pungens Willd. 674 Vittaria costata Kunze 691, 699 ASPLENIACEAE Asplenium auritum Sw., s.l. Asplenium auritum Sw. Asplenium cirrhatum Rich. ex illd. Asplenium claussenii Hieron.? Asplenium delitescens (Maxon) L. ómez Asplenium haplophyllum Domin Asplenium serratum BLECHNACEAE Blechnum occidentale L. Blechnum serrulatum Rich. CYATHEACEAE Cyathea pungens (Willd.) Domin DAVALLIACEAE Nephrolepis biserrata (Sw.) Schott DENNSTAEDTIACEAE Lindsaea lancea (L.) Bedd. x L. divaricata Klotzsch? Lindsaea cf. portoricensis Desv. JN stricta (Sw.) Dryander Saccoloma elegans Kaulf. 712, 735 877 713, 875 275, 651, 784- bis 794 723 489, 711 695 485 757 784 Volume 77, Number 3 Holst & Todzia 507 Croizat's Collections from the Franco-Venezuelan Expedition DRYOPTERIDACEAE POLYPODIACEAE Bolbitis semipinnatifida (Fée) 793 tS шаш angustifolium 707 Alston ée Cyclodium guianense (Klotzsch) 782 bind phyllitidis (L.) 524, 607, 720 . D. Gómez C. Presl Cyclopeltis semicordata (Sw.) J. 315 Campyloneurum repens (Aublet) 710 mi C. Presl Didymochlaena truncatula (Sw.) 821 Dicranoglossum desvauxii 806, 822, 881 J. Smi (Klotzsch) Proctor Diplazium cristatum (Desr.) 726, 765 Microgramma megalophylla 358-bis, 568 Alston (Desv.) Sota Diplazium grandifolium (Sw.) Sw. 706 Microgramma persicariifolia 285, 456, 732 A Een (Klotzsch) 710 (Schrader) C. Presl Microgramma reptans (Cav.) A. 635 я flaccidum (Fee) 778 R. Smith T. Moore Microgramma tecta (Kaulf.) 121,571, 783 Lomariopsis japurensis (Mart.) J. 675-Ыѕ Alston Smith Niphidium crassifolium (L.) 733 Tectaria incisa Cav. 347, 643 Lellinger Tectaria plantaginea (Jacq 701 ds ея percussa (Cav.) Hook. 716, 729, 876 axon var. macrocarpa 1 (Fée) & . Morton Poen bombycinum Maxon 443, 468 Tectaria trinitensis Maxon 728 olypodium caceresii Sodiro 829 Triplophyllum и (Kunze) 375, 405, 644, Polypodium decumanum Willd 795 Holttum 734, 880 Де гона hygrometricum 642 Split GARICHENIACEAR Polypodium polypodioides (L.) 717 Dicranopteris flexuosa (Schrader) 434 - burchellii (Baker) L. Underw. Wea са Polypodium triseriale Sw. 878, 879 GRAMMITIDACEAE Cochlidium tepuiense (A. C 884 icc Smith) L. E. Bishop Psilotum nudum (L.) P. Beauv. 542 Grammitis mollissima (Fée) 733-bis Proctor SCHIZAEACEAE HYMENOPHYLLACEAE Actinostachys pennula (Sw.) 167 оок hirsutum (L.) 797 Book. il polyanthos (Sw.) 797-А, 797-B AS Sw Selaginella asperula Spring 248 Thichomásiós collariatum Bosch, 771-bis Selaginella diffusa (C. Presl) 883 vel aff. pring Tr Paes collariatum Bosch 708 Selaginella flagellata Spring 383 Trichomanes кечын ыш (Вогу) 811 Selaginella parkeri (Hook. € 144, 258, 358, Mett. ex Sadebac Grev. i 441, 811-bis Trichomanes prep W. Boer 705, 730 Selaginella umbrosa Lem. ex 334, 475, 700 Trichomanes pilosum Raddi 16 i Trichomanes pinnatum Hedwig 484, 487, 824 THELYPTERIDACEAE LYCOPODIACEAE Thelypteris hispidula (Decne.) C. 315 Huperzia linifolia ( (L.) Trev. St. 450 ee éon var. jenmanii (Underw. & Thelypteris x rolandii (C. Chr.) 737 Lloyd) B. Øllg. & Wind. ‚ Tryo Lycopodiella camporum B. Ollg. 124 Thelypteris tristis (Kunze) R. 757-bis & Wind. Tryon Lycopodiella caroliniana (L.) 122 meridionalis (Underw. & Lloyd) B. Øllg. & Ашина Wind. FAMILY INDET. 426 МЕТАХҮАСЕАЕ АСАМТНАСЕАЕ Metaxya rostrata (HBK) Presl 121, 418, 571, Aphelandra 647 513, 646 674-bis, 764- bis sp. ar deppeana Schldl. & Cham 508 Annals of the d Botanical Garden Mendoncia sp. ndoncia cardonae Leonard Mendoncia sprucei Lindau Ruellia rubra Aublet Teliostachya cataractae Nees AMARANTHACEAE Chamissoa altissima (Jacq.) HBK Iresine diffusa Willd. AMARYLLIDACEAE Curculigo scorzonerifolia (Lam.) Baker ANACARDIACEAE Genus indet. ANNONACEAE Mixed esed Anaxagorea sp. + Clusiac ЕУ “dolic hocarpa Spr & Sandw. Annona 5р. C ‘ymbopetalum brasiliense (Vell. zonc. nth. Duguetia sp. Guatteria maypurensis Kunth Guatteria cardoniana R. E. Fries Guatteria schomburgkiana C. Martius Охапага 5р. Rollinia exsucca (Dunal) А. DC. Xylopia aromatica (Lam.) С Martius APOCYNACEAE Tabernaemontana heterophylla ARACEAE Genus indet. Anthurium bonplandii Bunting subsp. bonplandii Anthurium gracile (Rudge) Schott Caladium Dieffenbachia duidae (Steyerm.) unting Dieffenbachia seguine (Jacq.) Scho Heteropsis melinonii (Engl.) A Jonker & Jonke Heteropsis tenuispadix Bunting Monstera adansonii Schott у laniata ia t) Madison e Schott + nstera dubia (HBK) Engl. Monstera dubia (HBK) Engl. Monstera lechleriana Schott 639, 804, 832 414, 480, 497 233 859, 1040 836 547, 1052-B 833, 838 391-A, 840 664, 834 2 835, 837, 839 547 01, 687 391, 465 177 es D 289, 419.A, 517, 851, 855 208 259, 853 596, 846 850 679, 845 575 278 318, 842-A, 844-A, 848 843 Monstera obliqua Miq. шо ш linifera (Arruda) T м maguirei Bunting Philodendron scandens С. Koch Sello Philodendron smaragdinum unting Spathiphyllum canniifolium (Dryander) Schott ARALIACEAE Dendropanax arboreum (L.) ecne. & Planchon Schefflera morototoni (Aublet) aguire, Steyerm. & Frodin ARECACEAE Astroc ve gynacanthum С. artiu Bactris nes Karsten Bactris monticola Barb. Rodr. Hy йы elegans С. Martius Iriartella setigera (C. Martius) Wendl. Socratea exorrhiza (C. Martius) Wendl. ARISTOLOCHIACEAE Aristolochia sp. ASCLEPIADACEAE Marsdenia rubrofusca Benth. ex Fourn. Matelea stenopetala Sandw. ASTERACEAE Clibadium sp ата rabia) atum i Mikania o Kunth Mikania idis qn Benth. Wulffia ia ee om ) DC. BEGONIACEAE Begonia humilis Aiton BIGNONIACEAE Adenocalymma impressum (Rusby) Sandw. Adenocalymma inundatum var. surinamensis Bureau & Schumann Arrabidaea pubescens (L.) A Gentr entry Crescentia amazonica Ducke 445, 483, 529, 842, 843-A, 847, 854 333 852 573, 849 331, 381, 396, 400 459, 697 562 492 528, 995-C 499-A 499 369, 995-B 995-A 180, 994, 995 584 322 341 301 505 46 460, 510 504 340 543, 543-A 856-A Volume 77, Number 3 Holst 8 Todzia 509 1990 Croizat's Collections from the Franco-Venezuelan Expedition Cydista aequinoctialis (L.) Miers 300, 856-C Bauhinia ungulata 748 Distictella magnoliifolia (Vahl) 16 hamaecrista desvauxii va 191 Sandw brevipes (Benth.) H. Irwin La Jacaranda obtusifolia sub 317, 625 arneby rhombifolia (G. Mey) А. Cas Chamaecrista desvauxii va 54 Mansoa kerere (Aubl.) A. Gentry 293 mollissima (Benth.) H. Irwin & Paragonia pyramidata (Rich.) 856-B Barneby Bureau Chamaecrista n var. 7 Schlegelia spruceana Schumann 1060, 1061 triumvaralis H. Irw Stizophyllum riparium (HBK) 661 Barneby Sandw. Chamaecrista orenocensis 239 Tabebuia pilosa A. Gentry 856 (Benth.) Н. Irwin & Barneby Tynanthus polyanthus (Bureau) 335 Dipteryx punctata (Blake) 60 Sandw. Amshoff Heterostemon mimosoides Benth. 18 BIXACEAE ie = excelsum (Benth.) 826 Glea Bixa orellana L. 769, 857 Bixa urucurana Willd. 57 M en p illaris (L.f.) H. Irwin & 931 BOMBACACEAE Senna multijuga (Rich.) H. Irwin 923 Pachira aquatica Aublet? 271-A Senna po (Vell. Conc.) H. 345, 921 rwin & Barneb BORAGINACEAE Cordia nodosa Lam 364 SUME US Cordia polycephala (Lam.) I. M. 371 Centropogon cornutus (L.) Druce 939 Johnsto ironia cuspidata HBK 281 CHRYSOBALANACEAE TARA Couepia сш subsp 282 glandulosa (Miq.) Pra Aechmea angustifolia Poeppig & 539 Couepia guianensis nen sen 904-A Endl. ulanensis ed cf. chantinii (Carriére) 1045-А Hirtella racemosa Lam 145 hexandra (Willd. ex ha 8 pene penduliflora André 858, 1045 Schultes) Prance Aechmea rubiginosa Mez 564 Aechmea setigera C. Martius ex 538 CLUSIACEAE oo Genus indet. 90, 153, 155, avia sp. 458 171. 205 ы ан Lyman В. 188 274. 368. r. patentiflora 776, 809.A Tillandsia pu итин Mez 446 838, 900 A. 902, 903, BURMANNIACEAE 904, 906, Apteria aphylla (Nutt.) Barnhart 560 907, 909, x Sma 910, 1048 Burmannia bicolor C. Martius 112 Mixed collection: Clusiaceae + 836 Anaxagorea s BURSERACEAE Caraipa sp. : Caraipa punctulata Ducke 901, 908-B Genus indet. er ane ~ d tereticaulis Tul. 900, 1046 Clusia sp. 905 360 1038, Vismia cayennensis (Jacq.) 603 1040-B, 1064 Persoo Bursera sp. 10 : Ch 25 Hemicrepidospermum rhoifolium 692 "nd guianensis d UNE Benth.) Sw. subsp. persicoide (Ben Vismia japurensis Reichardt 170 Vismia laxiflora Reichardt 106 CACTACEAE Epiphyllum phyllanthus (L.) 427, 563 COCHLOSPERMACEAE Haw Cochlospermum orinocense 230, 342 Hylocereus sp. 244, 634 (Kunth) Steudel CAESALPINIACEAE COMBRETACEAE Bauhinia glabra Jacq. 263, 519 Combretum rotundifolium Rich. 375 510 Annals of the Missouri Botanical Garden COMMELINACEAE Dichorisandra hexandra (Aublet) Standley CONVOLVULACEAE Aniseia martinicensis (Jacq.) oisy C alye aiu glaber (HBK) House Ipomoea alba Jacquemontia tamnifolia ( (L.) Griseb. Operculina sericantha (Miq.) Ooststr. CUCURBITACEAE Gurania spruceana Cogn. Gurania trialata Cogn. Melothria pendula L Posadaea sphaerocarpa Cogn. CYCLANTHACEAE Thoracocarpus bissectus (Vell. Conc.) Harling — mixed with Thoracocarpus bissectus (Vell. onc.) Harling CYRILLACEAE Cyrilla racemiflora L. DILLENIACEAE Dolioc и Саак (Aublet) Standley е (Steyer mi "Kubitz Doliocarpus major Gmel. subsp. major Tetracera willdenowiana Steudel subsp. willdenowiana DIOSCOREACEAE Dioscorea sp. Dioscorea mune Maguire & Steyerr DROSERACEAE Drosera sp. ELAEOCARPACEAE Sloanea guianensis (Aublet) Benth. ERICACEAE Agarista iind pa eed Judd Befaria s i Meissn de panurensis (Benth. ) . & Hook.f. Thibaudia nutans Mansf. ERYTHROXYLACEAE Erythroxylum divaricatum Peyr. O. Erythroxylum vernicosum Schulz 26, 261, 409, 491, 631 767 523, 548 452, 447, 449 175 367 147 290 552 187, 229 422, 867-B 867-C 867 435, 455, 867-A 346 488 EUPHORBIACEAE Acalypha cf. macrostachya ы Acalypha macrostachya Jac ee villosa Jacq. Alchorne sp. Mas discolor Poeppig Aparisthmium cordatum (A. Juss.) Baillon C Onc eveiba a guianensis Aublet Dalechampia affinis Muell. Arg. Dalechampia magnolüfolia uell. Arg. Hieronima laxiflora (Tul.) Muell. rg. Hieronima oblonga (Tul.) Muell. Ar 8- Mabea sp. bea occidentalis Benth. Manihot brachyloba Muell. Arg.? Maprounea ee Aublet Phyllanthus Phyllanthus pa TE unt Phyllanthus aff. lindbergii Muell. rg. ji uil minutulus Muell. Phyllanthus orbiculatus em Sagotia racemosa Baillo Tragia volubilis L FABACEAE Acosmium nitens (J. Vogel) Yakovlev Crotalaria micans эга Crotalaria sagittalis & Dalbergia riedelit ГЛ ) Sandw. Desmodium wydlerianum Urban Dioclea ы (Rich.) Amshoff var. virgat oo mitis Jacq. Mucun Mucuna urens ds ) DC Ormosia coccin ackson Phaseolus appendio ulatus Benth. Phaseolus campestris C. Martius Phaseolus pcs HBK FLACOURTIACEAE Genus indet. Casearia arborea (Rich.) Urban Casearia sylvestris Sw Casearia sylvestris Sw. var. sylvestris Lindackeria paludosa (Benth.) zilg Ryania speciosa var. bicolor (A. Monach. GENTIANACEAE Curtia PE ned Knobl. subsp. tenuif 874, 874-A 413, 495, 870-E 869-F, 872 64 871, 873 199, 868, 868-С 565 868-H, 870 868-1 619, 873-B 628, 871-D 5 31, 869 110, 215 133 6 51, 872.0 815 311 174, 885, 887 886 242 503 246, 817 825, 888 Volume 77, Number 3 1990 Holst 4 Todzia Croizat's Collections from the Franco-Venezuelan Expedition 511 Irlbachia alata (Aublet) Maas ubsp. alata Irlbac hia alata (Aublet) Maas subsp. angustifolia (Kunth) Pers. & Maas Irlbachia pratensis (Kunth) Cobb & Maas Voyria flavescens Griseb. GESNERIACEAE Besleria Besleria p Benth. Codonanthe calcarata (Miq.) anst. Codonanthe crassifolia (Focke) Morton Drymonia coccinea (Aublet) Wiehler Drymonia serrulata (Jacq.) C. Martius Nautilocalyx sp. HAEMODORACEAE Genus indet Sc hiekia orinocensis (HBK) Xiphidium coeruleum Aublet HIPPOCRATEACEAE Cheiloclinium serratum (Cambess.) A.C. Smith Peritassa шен (Hoffsgg.) A.C. Sm Tontelea ovalifolia (Miers) A. C. Smith HUMIRIACEAE Humiria balsamifera (Aublet) A. St. Hil. LAMIACEAE Hyptis atrorubens Poit. Hyptis dilatata Benth. Hyptis mutabilis (Rich.) Briq. LAURACEAE Genus indet. Aiouea guianensis umet Cassytha filiformis L. Endlicheria s Endlicheria gracilis Kosterm. cf. Nectandra cuspidata Nees Nectandra cuspidata Nees Nectandra globosa (Aublet) Mez, s.l. Nectandra reticulata (Ruiz Lopez & Pavón) Mez Ocotea sp. Ocotea aciphylla group, flowers diseased, probably O. aciphylla (Nees) Mez Ocotea aciphylla (Nees) Mez 11, 908-B 279 889, 1044 681, 1044-A 756 432, 1044-B 606 9 255, 444 754 440, 919 919-D 69 1052-C 919-B 520, 919-C, 61, 224, 496, 630, 914 911 910 Ocotea cernua (Nees) Mez Werff Rhodostemonodaphne kunthiana (Nees) Rohwer LECYTHIDACEAE Eschweilera coriacea "is ) С. Martius ex O. Ber Esc Lom аа (Rich.) 5 Мог Сома poeppigiana О. Berg LENTIBULARIACEAE Utricularia amethystina A. St. Hil. Jtricularia subulata L. Utricularia triloba Benj. LILIACEAE Smilax sp. LINACEAE Roucheria calophylla Planch. LOGANIACEAE Strychnos matogrossensis S. Moore Strychnos panurensis Spr. & Sandw. LORANTHACEAE Phthirusa pyrifolia (HBK) Eichler MALPIGHIACEAE аа decussiflora W. К. An er Byrsonima a crassifolia (L.) HBK Clonodia complicata (HBK) W . Anders. Heteropterys atabapensis W. R. ers. Heteropterys nervosa А. Jus Heteropterys orinocensis (HBK) uss. Hiraea faginea (Sw.) Niedenzu Mascagnia macrodisca (Triana & Planchon) Niedenzu Stigmaphyllon hypoleucum Miq. Tetrapterys mucronata Cav MALVACEAE Hibiscus bifurcatus Cav. Hibiscus furcellatus Desr. MARANTACEAE Genus indet. Mixed ee Hylaeanthe m + leaves of cf. 600-A 586-A, 599, 600 586 172, 938 356, 809, 944, 946 512 Annals of the Missouri Botanical Garden Calathea splendens (Ї етп.) ege Calathea densa (Koch) Regel pa acd cens (Kuntze) icolson Galathea propinqua (Poeppig & Endl.) Koe Hylaeanthe talon cannoideus L. Anderss. Maranta humilis Aublet ео нЕ (Poeppig & Endl.) S Moa кен atum Koern.) Schumanr n mo aff. jacqui r& сыш Kennedy & ‘Nicolson MARCGRAVIACEAE Gent indet mute dasystachya Gilg & Werder MELASTOMATACEAE Aciotis indecora (Bonpl.) Triana var. macrophylla Cogn Aciotis purpurascens (Aublet) Triana Bellucia grossularioides (L.) Triana Clidemia capitata Benth. — a (Bonpl.) Don . dependens ці, Macbride Cli der emia denta Clidemia rubra (Aublet) C. Martius Clidemia umbonata DC. Comolia kesi» Benth. s.l. ж. cf. brevipes Ben Miconia cf. bubalina (Don) Naud. Miconia dispar B Miconia жөк Cogn: haei Naud. Miconia cf. шш aa (Bonpl.) DC. Miconia rufescens (Aublet) DC. Miconia scorpioides (Schldl. & Cham.) Naud. Miconia spennerostac hya Naud. surely from Peru or Boliva, not Venezuela, det. J. Wurdack 613 74 389 969, 823 184 736 125, 947 561 410, 467, 512, 611 132 743 131, 949-A, 953 438 235 264 130, 266 136, 509 Miconia tomentosa (Rich.) Don Pterogastrus divaricata (Bonpl.) Naud. Rhynchanthera grandiflora (Aublet) DC. Tibouchina spruceana Cog Tococa macrophysca ie ex Triana Tococa nitens (Benth.) Triana MELIACEAE Genus indet. с. grandifolia Guarea guidonia (L.) Sleumer Guarea pubescens (Rich.) A. Juss. subs escens Trichilia sp. Trichilia ? pallida Sw. MENISPERMACEAE Genus indet ш ТҮ (С. Martius) Sandw Cissampelos pareira L. Disc iphania ernstii Eichler var. ernstu Odontocarya sp. Orthomene schomburgkii (Miers) Barneby & Krukoff MIMOSACEAE Acacia polyphylla DC. Calliandra stipulacea Benth. nga sp Inga bourgonii ол DC. Inga edulis — Inga 1 а кла Benth. Inga nobilis Willd. Inga pezizifera Benth. al е DC. In пва fs Min сани adena Be Pithecellobium dancin Zygia 2 latifolia (L.) Fawcett & Rendle MONIMIACEAE Mollinedia sp. Siparuna guianensis Aublet MORACEAE Brosimum guianense (Aublet) Huber Brosimum lactescens (S. Moore) C. C. Ber Cecropia latiloba Miq. Cecropia m membranacea Tré Cec ih sciadophylla C. oss Clarisia C ни жае еда Trécul 745 100 200 930 605, 605-bis 282-bis 338 338-bis, 398 bi 398-bis 56, 433, 451- bis, 527-bis 934 828 412 381 527-bis 637 718, 718-A, 718-B 226, 284 Volume 77, Number 3 1990 Holst & Todzia Croizat's Collections from the Franco-Venezuelan Expedition 513 subsp. ard a о АККег Coussa ip à Ap viridifolia Cuatrec. icus 8 Ficus eximia Schott F А Ficus nymphiifolia Р. Miller Ficus paraensis (Miq.) Miq Ficus vs. trigonata L. Pourouma melinonii Benoist MUSACEAE Heliconia episcopalis Vell. Conc. MYRISTICACEAE Genus indet. н hostmanii (Benth.) сорив platyspermum (DC.) Warb. Virola elongata e ) Warb. Virola sebifera Auble MYRSINACEAE Genus indet Ar disia guianensis ru Mez Cybianthus viridiflorus A. C. Smith MYRTACEAE Genus indet po Boetheanus (DC.) O. UE suites fasciculata O. Eugen MP cad ia troc teata (Rich.) DC. Myrcia a (Poiret) DC. Myrcia оа McVaugh OCHNACEAE Ouratea croizatii Maguire & Steyerm Dea ci erecta L. Sauvagesia ramosa (Gleason) ONAGRACEAE Ludwigia аиа (Munz) H. Har н “latifolia ( (Benth.) Н. Har Ludwigia leptocarpa (Nutt.) H. ORCHIDACEAE Mixed collection: Scaphyglottis Epidendrum тант Reichb.f. 312, 541, 961 959 298, 958 960 464, 582 252 948-A 689 968-A 969, 970 740 968-B, 1042 29 587, 973 557, 688, 973-A 583 973-B 23 973-C 624 349, 654 161 755, 755-A 297 401, 755-B 975-A Acacallis cyanea Lindley Brassavola martiana Lindley Bras Brassia pm Lindley Brassia caudata (L.) Lindley Bulbophyllum pachyrhachis (A. Rich.) Griseb. Ca icr micranthum (Lindley) R snc буле poeppigii (Reichb.f.) Rolfe Catasetum barbatum (Lindley) indley Catasetum discolor Lindley е кн Reichb.f. Cattley aa ео (НВК) Rolfe Caularthron bicornutum (Hook.) Raf. C i sad warscewiczii (Reichb.f.) odso Dichaea or hypoda Reichb.f. Dichaea pendula (Aublet) Cogn., vel a Dichaea trinitensis Gleason Duckeella diei a Garay Epidendrum procis a mud group) Epidendrum nocturnum Jacq bonorum schomburgkii indle ¿pidendrum secundum Jacq., vel aff. Epidendrum secundum Jacq. Epidendrum strobiliferum Reichb.f. Epistephium sp. Galeandra devoniana Schomb. ex Lindle Habenaria leprieuri Reichb.f. Masdevallia norae Luer Maxillaria amazonica Schltr. Maxillaria camaridii Reichb.f. AD ui ottonis Reichb.f. Notylia Onc ris m sp. есиги d (Jacq.) Sw m Lindley ten Pa phinia t cristata Lindley Peristeria guttata Knowls & vel aff Pleurothallis sp. Psygmorchis pusilla (L.) Dodson & Dressl sler a microscopica Lindley Sarean иб sp. Scaphyglottis boliviense (Rolfe) . Adams, ve Стя a Cogn., vel aff. 6 977-A, 978-A 676 725 365, 648 249, 498, 534 162, 199 1 Annals of the Missouri Botanical Garden Scaphyglottis sickii Pabst Sigmatostalix amazonica Schltr. Sobralia fragrans Lindley Stanhopea grandiflora (Ludd.) ind Vanilla palmarum Lindley Vanilla pompona Schiede OXALIDACEAE Oxalis sp. PASSIFLORACEAE Passiflora coccinea Auble Passiflora foetida L. var. [o Passiflora longiracemosa Ducke Passiflora Passiflora nitida HBK Passiflora securiclata Masters Passiflora vespertilio L. Passiflora vitifolia HBK PHYTOLACCACEAE Phytolacca rivinioides Kunth & Bouché PIPERACEAE Peperomia serpens (Sw.) Loudon 2; > Piper aequale Vahl iper arboreum Au blet Piper UM Aublet var. arboreum Piper ef. augustum Rudg Piper асан (Miq. ) C. DC. Piper cernuum Vell. Conc.? Piper demeraranum (Miq.) C. DC Piper francovilleanum C. DC Piper hispidum Sw. f. hispidum PODOSTEMACEAE Apinagia longifolia (Tul.) van Royen Apinagia richardiana (Tul.) van oyen Apinagia staheliana (Went) van Roy POLYGALACEAE Bredemeyera lucida (Benth.) A. ‚ Benne Polygala adenophora DC. subvar. robusta Chodat Polygala hygrophila HBK жое spruceana А. W. Polygala subtilis HBK Securidaca sp. Securidaca cf. warmingiana Chodat 408, 408-A 352, 610, 985 655 580, 787 540-A, 1001, 1006 421 473 678-A 1001-A 680 844, 1000-A 494 612 580-A POLYGONACEAE Coccoloba з Гат. Mixed collec Pol ygonum acuminatum HBK + Scoparia dulcis L. Triplaris americana L. QUIINACEAE Quiina parvifolia Pulle RAPATEACEAE Cephalostemon affinis Koern. Monotrema xyridoides Gleason RHAMNACEAE Gouania sp. RUBIACEAE Genus indet Bertiera ен Aublet subsp. r. guianensis жо latifolia (Aublet) chumann var. latifolia Diodia к (Willd.) reme Duroia eriopila L.f. var. eriopila Ferdinandusa goudotiana Schumann var. eciliata Steyerm Genipa spruceana Steyerm. e d ин (L.) Т. М. Johns Gonz ada dicocca Cham. & . subsp. dicocca var. mb. Bon ha x Roemer & бу п Hameka. Lm Jacq. var. glabra Oerste Isertia DN Vahl. var. parvi iaa lambertiana (A. Br. x C. Martius) Klotzsch Manettia reclinata Mutis Morinda tenuiflora (Benth.) teyerm. var. leiophylla (Steyerm.) Steyerm. е yu (Schum Pagamea coriacea а ех Bent Palicourea о А. DC. Palicourea ай. с а (Sw.) Roemer & Schultes Palicourea aa Aublet subsp. occidentalis Steyerm Райс оигеа Mp (Muell. Arg. ) Standley Perama galioides (HBK) Poiret Piye hotria alba Ruiz em & Pavón 303-A 1017 387, 714 451 840, 1030 186 307, 416 213 355, 649 324, 423, 659 430 66 814 594, 792 626 228 83, 671 439 1020, 1021 506, 1023, 1025, 1067, 1068 713 28 747 1032 Volume 77, Number 3 1990 Holst 4 Todzia Croizat's Collections from the Franco-Venezuelan Expedition 515 Psychotria bahiensis A. DC. var. cornigera (Benth.) Steyerm. subsp. luxuriana (Rusby) Steyerm. Psychotria bracteocardia (A. DC.) Muell. Ar Psychotria calviflora Steyerm. Psychotria deflexa A. DC. subsp. campyloneura (Muell. Arg.) Steyerm. Ne dies egensis Muell. ш glaucesce Psychotria ion: Ruiz Lopez ex Standley B dens aff. platypoda . des il Psychotria poeppigian subsp. bar s ME Arg (сш анн Bou: racemosa (Aublet) c aeusch. i eds tepuiensis (Steyerm.) Psychotria ulviformis Steyerm. enezuelensis Steyerm. Remijia hispida Spruce ex Schumann Retiniphyllum schomburgkii (Benth.) Muell. DE subsp. occidentale Stey Rudgea ma шы es а 68 Rudgea US (A. DC.) Steyerm Rudgea woronowii Standley Sabicea velutina Benth. iubes duidensis Steyerm Schradera pod а рс. Sipanea hispida d Wernham var. his Sipanea pratensis june var. dichotoma (HBK) Steyerm. RUTACEAE Angostura trifoliata (Willd.) T. Elias Erythrochiton brasiliensis Nees & С. Martius Esenbeckia pilocarpoides Kunth subsp. pilocarpoides SAPINDACEAE Doa scrobiculata Rich. Mata oe opaca Radlk. Paullin Paullinia ae Radlk. Talisia firma Toulicia риса Radlk. SAPOTACEAE Genus indet. 225 1019, 1027 1024 87, 650 478, 521 662 82 385 791 1028 40, 42, 209, 812, 1029 594, 780 1022 772, 814-A 589 238 1 477 1026 554 1031, 1066 84 86 597 466, 566 592, 592-А 221, 1040-D 271 469 789 1033, 1035 1040-E 1036 Chrysophyllum sanguinolentum (Pierre) ec ubsp. balata (Ducke) Micropholis nn (A. DC.) Penn. Pouteria caimito (Ruiz Lopez & Рауоп) Radlk. Pouteria? filipes Eyma Pouteria? plicata Penn Pouteria surumuensis Baehni SCROPHULARIACEAE Buchnera palustris (Aublet) Spreng. Bou rosea Kunth Mixed collection: eed dubia (L.) puros var. + Scoparia du Mec ardonia procumbens (Miller) Sm paña dulcis L. SIMAROUBACEAE Picramnia и Tul. Quassia amar SOLANACEAE Genus indet. Brugmansia suaveolens (Willd.) Bercht. & Presl Cestrum latifolium Cestrum кн Ruíz Lopez Pav Physalis к L. Solanum sp. Solanum pensile Sendtner Solanum subinerme Jacq. STERCULIACEAE Byttneria catalpifolia Jacq. subsp. catalpifolia Byttneria divaricata Benth. Byttneria genistella Triana & Planchon Guazuma ulm Melochia villosa (Miller) Fawcett & Rendle Melochia On Qm Fawcett & Rendle il Sterculia sp Theobroma. bioota Humb. & onpl. Theobroma cacao L. THEACEAE Freziera calophylla Triana & Planchon Ternstroemia punctata (Aublet) Sw. THEOPHRASTACEAE Clavija lancifolia Desf. 1037 516, 819 608 1049, 1052-A 476 672 108 774, 1051 350, 397-A 126 507 207 168, 204 397 827 366 1059 516 Annals of the Missouri Botanical Garden TILIACEAE VIOLACEAE Apeiba cf. albiflora Ducke 323 Genus indet. 1063 Apeiba aspera Aublet 810 Amphirrhox latifolia C. Martius 1047 Apeiba schomburgkii Szyszyl. 572 С LAB arborea (L.) Blake 32, 696 ap 650 TRIURIDACEAE Rinore 623, 1065-A Sciaphila purpurea Benth. 741, 753 Rinorea fale ашы Манн) BOO: В . Rinorea lindeniana 2 ) Kuntze 818 ИИМ 2 дий (Ben h.) 546, 816, 1062 Tropaeolum fintelmanii Wagener 790 e & San ex Schldl. VITACEAE TURNERACEAE Cissus erosa Rich. 348 Piriqueta sp. 8l Cissus sicyoides L. 348-A, 550 Piriqueta cistoides (L.) Griseb. 189 Turnera acuta Willd. 36 VOCHYSIACEAE Turnera odorata Urban 905,995 Ruizterania rubiginosa (Stafl.) 157-A пи Marcano-Berti DS Vochysia sp 157 Celtis iguanea (Jacq.) Sarg. 472 Trema integerrima (Beurling) 727 X YRIDACEAE Standley Abolboda macrostachya Spruce 127 m" ex Malme var. macrostachya URTICACEAE Urera baccifera (L.) Gaudich. 758, 758-A ZINGIBERACEAE — Costus arabici 948-F И Y Costus шша; Rusby var. 291, 399 Vellozia tubiflora (A. Rich.) НВК 457 macrostrobilus (Schumann) P. aas VERBENACEAE Costus scaber Ruíz Lopez & 292 or ee tr Pavon Aegiphila integrifolia (Jacq.) 445 Renealmia alpinia (Rottb.) Р. 353, 420, 948-B Jac P Amaz uis - 34 Maas — "P Renealmia aromatica (Aublet) 411, 948-D, oe m 304 Griseb h р 116 : he . ^ 948-E Cori ею ep. 304-A enealmia monosperma Miq. 722 A PHYLOGENETIC REEVALUATION OF THE OLD WORLD SPECIES OF FUCHSIA (ONAGRACEAE)! Jorge V. Crisci? and Paul E. Berry’ ABSTRACT The four Old World species of Fuchsia — F. cyrtandroides, F. excorticata, E екан, and F. procumbens — ogenetic reevaluation of the g have a consistency index of 0.75 when of the other species in the respective cladogra defined by constricted floral t on flavonoids, which grouped F. procumbens in a clade with F. perscanden nce of flavones in all species. маг nds in the section were reanalyzed, and ЗР а Ман сһагасїегѕ vei employ group, using the rest of the g used, 7 were phylogenetically informative, resulting in two equally most cn cladogram all noninformative characters are exclu of the Tahitian F. cyrtandroides and the New Zealand F. procumbens, each of ms. In both t ubes. These results differ from a prior laced F. cyrtandroides ае as the sister species of the rest of the section ап nus as the outgroup. е ms 23 е long. Bot ded. The two trees differ in the position whic trees, F. excorticata and F. persca cladistic easi of the section based primarily d Fuchsia is a genus of about 105 species found primarily in mountainous regions of the Neotropics. Four species comprising the distinctive sect. Skin- nera are confined to the Old World, however, with three species in New Zealand and one on the island of Tahiti. This section is of considerable biogeo- graphical interest for its marked disjunction from the American species and the presence of closely related taxa on distant islands in the South Pacific. As part of a study of foliar flavonoid compounds Skinnera, Williams € Garnock-Jones (1986) presented a cladistic analysis of the group, in sect. based on their flavonoid results and several other characters. A single an tree resulted from their analysis, in. whic e itian Ё. cyrtan- droides is the sister species of a a clade comprising all the New Zealand taxa, and the arborescent F. excorticata is the sister species of a clade formed у F. procumbens and F. perscandens, a creeping and a scandent species, respectively. Independently, Averett et al. (1986) conducted a comprehensive survey of foliar flavonoids in the entire genus, including a greater number of samples in sect. Skinnera than in the earlier study by Williams & Garnock-Jones. Their study resulted in the detection of several compounds not previ- ly found in the same taxa by Williams & Gar- nock-Jones (1986). This and new information de- rived from a revision of the section by Godley, Berry, and Smith (in prep.) prompted us to reassess the phylogenetic relationships within sect. Skin- nera, using a reevaluation of flavonoid characters as well as a larger number of nonflavonoid char- acters. MATERIALS AND METHODS We recognize four species as terminal taxa in sect. Skinnera (Table 1). ree occur in New Zealand: the widespread tree fuchsia, F. excorti- cata, the lianoid F. perscandens, and the rare creeper, F. procumbens, which is restricted to the northern third of the North Island. Following Allan (1927, 1928), we treat F. Xcolensoi as a series of interspecific hybrids between F. excorticata and F. perscandens and do not include it in our anal- ' Work on this paper was supported by grants from the National Science F oundation (DEB- 8518906) and the John D. and Catherine T. MacAr thur Foundation to Peter H. Raven, whom we thank for We thank Peter C. Hoch for his help to both authors during work carried out at the Missouri Botanical Garden. Suggestions on нин versions = the manuscripts were a? made by L. Brako, S. E. Freire, P. C. Hoch, A. A. Lanteri, P. G. Martin, P. H. Raven, and P. M. Richardso 2], aboratorio m teles y “Biología Evolutiva, Museo de La Plata, Paseo del Bosque s/n, 1900 La Plata, Argantina requeste 5 Missouti Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A.; d. to whom reprints should be ANN. Missour! Вот. GARD. 77: 517-522. 1990. 518 Annals of the Missouri Botanical Garden TABLE l. Таха and geographical distributions in Fuchsia sect. Skinnera. Taxa Geographical distribution F. cyrtandroides J. W. Moore F. excorticata (]. В. & С. Forster) Linnaeus f. F. perscandens Cockayne & Allan F. procumbens R. Cunningham Tahiti (Society Islands) Throughout New Zealand, Chatham Islands, Stewart Island New Zealand: throughout South Island, in southern half of North Island New Zealand: restricted to northern third of North Island ysis. The fourth species, F. cyrtandroides, grows as a small tree and is endemic to a few high peaks on Tahiti, in the Society Islands. 1981; Maddison et al., 1984), collectively using the other sections of the genus as the outgroup since there is no single section that is a clear sister group. If sect. Jimenezia i is used as the outgroup, based on the p li y results of Sytsma & Smith's (1988) survey of chloroplast DNA restriction site mutations in Fuchsia, there are no differences in polarity assignment from using the other sections collectively. The sources of the flavonoid data were Williams & Garnock-Jones (1986) and Averett et al. (1986). Where the two studies differed in the flavonoid profile of a particular species, the presence of a compound was employed in our analysis rather than its absence, since lack of detection can be due to presence in low concentrations, inadequate amounts of sample material, environmentally in- duced flavonoid changes, developmental differ- ences, or natural intraspecific variation (Harborne, 1975; Bohm, 1987). Data on the sexual systems of the species were taken from Godley (1955, 1963, 1979). Field observations and analysis of preserved specimens were used to define the re- maining floral and vegetative characters. The data matrix used in this analysis is contained weighting procedure (Farris, 1989), which calcu- lates weights from the best fits to the most parsi- monious trees and applies them in the weighting procedure until there are no changes in succes- sively produced trees. CHARACTER DEFINITION AND CODIFICATION 1. Flavones: the different biochemical pathway leading to the synthesis of flavones makes the pres- ence of this class of compounds an important dis- tinguishing character from taxa that have only flavonols (Gottlieb, 1975). The presence of flavones is generally an advanced trait (Harborne, 1975), but the secondary loss of such compounds can also constitute a third, highly advanced stage in fla- vonoid evolution (Gornall & Bohm, 1978; Averett & Raven, 1984). In view of this situation, we use the overall presence or absence of the flavone class in Fuchsia as a character, rather than treating each flavone compound individually. Presence of flavones is considered apomorphic in sect. Skin- nera, since they are absent in sect. Jimenezia and in nearly all the other species of the genus. Absence = 0, presence = 2. Pollen: blue pollen is very rare in the an- giosperms and is found in Fuchsia only in sect. Skinnera. Anthers of F. excorticata are rich in three different anthocyanins (Crowden et al., 1977) and probably form a blue metallo-flavone-antho- cyanin complex in living flowers (N. H. Fischer, pers. comm.). Cream-colored pollen = 0, blue pol- len = 1. 3. Ovule number: all other sections of Fuchsia have fewer than 200 ovules per ovary, with ovules arranged in two rows per locule. This is also the case for F. procumbens, but the other three species in sect. Skinnera have more than twice the max- imum number of ovules found in the other sections and have more than two rows of ovules per locule. < 200 ovules/ovary = 0, > 200 ovules/ovary =] 4. Flavone sulphates: these compounds are un- usual among angiosperms (Harborne & Williams, 1982) and occur in Fuchsia only in sect. Skinnera. Absence = О, presence = 1. 5. Flavonol diglycosides: flavonol monogly- cosides were detected in all 80 taxa of the genus studied by Averett et al. (1986), whereas flavonol diglycosides occur in only a few taxa, including F. excorticata in sect. Skinnera. Absence = 0, pres- ence = ]. 6. Eriodictyol 7-glucuronide: this uncommon flavanone has only been found within the genus in Volume 77, Number 3 Crisci 8 Berry 519 1990 Phylogenetic Reevaluation of Old World Fuchsia TABLE 2. Data matrix for the taxa of Fuchsia sect. Skinnera; the other sections of Fuchsia are used as the outgroup. Character 7 is additive; characters 14 and 15 are nonadditive Character Taxa 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Outgrou 0 0 0 0 0 0 0 0 0 0 0 0 0 о 0 о 0 F. eyrtandroides 1 1 1 0 0 0 0 0 0 0 0 0 1 2 0 1 0 F. procumbens 1 1 0 1 0 0 2 1 1 1 1 0 0 0 2 0 1 F. perscandens 1 1 1 1 0 0O 1 0 0 0 0 1 0 1 1 1 1 F. excorticata 1 1 1 1 1 1 1 0 0 0 0 1 0 2 0 1 1 F. excorticata (Williams et al., 1983). Absence = О, presence = 1. 7. Sexual system: gynodioecy is derived from hermaphroditism in F. procumbens and F. excor- ticata by the presence of a dominant gene for male sterility. The addition of a linked gene dominant for female fertility in F. procumbens then leads to subdioecy. The other sections of Fuchsia are her- maphroditic or else have male sterility determined by a recessive gene (Breedlove et al., 1982). Her- maphroditic = 0, gynodioecious = 1, subdioecious = 2; additively (= sequentially) coded. ‘lower position: hanging flowers character- ize all but one other section of Fuchsia, with erect flowers in F. procumbens clearly derived. Hanging = 0, erect = 1. 9. Petals: petals are present in all but one other section of the genus. In sect. Skinnera, the petals are either much reduced or completely lacking in F. procumbens. Present = О, lacking = 10. Stamens: all species in the genus except F. procumbens have the stamens in two whorls of conspicuously different lengths. Distinct whorls = О, stamens subequal = 11. Sepals: sepals are spreading at anthesis in all but a few species of the other sections of the genus. In sect. Skinnera, F. procumbens is the only species with completely reflexed sepals dou- bling back и the floral tube. Spreading = 0, reflexed = 12. бө tube: constriction is the prevalent condition in the out- group. In sect. the tubes are either strongly constricted above the nectary and then abruptly widened above, or else cylindrical. Un- constricted = 0, strongly constricted = 13. Leaf texture: membranous or subcoria- ceous leaves occur in the rest of the genus, with F. cyrtandroides unique in its considerably thicker, crassate leaves. Membranous = О, thick-crassate floral tubes with little or no Skinnera, 14. Leaf underside: the presence of a silvery- white leaf underside is caused by the absence of chlorophyll in the spongy parenchyma of F. ex- corticata (Suckling, 1914) and is more specialized than the normal green leaf underside in the genus. Fuchsia perscandens has chlorophyll in the spongy parenchyma but also has a whitish underside. Green = 0, whitish = 1, silvery-white = 2; nonadditive (= unordere abit: үн outgroup is primarily shrubby and is here taken to include the small to medium- sized trees found in two species of sect. Skinnera. The unique creeping habit of F. procumbens and the unusual lianoid habit of F. perscandens are both specialized for the genus. Shrubby or tree = O, lianoid = 1, procumbent creeper = 2; nonad- ditive. 16. Flower color change: species of the section pass through an abrupt floral color change from green to red (Delph « Lively, 1985; Berry, unpublished). 0 = no distinct color phases, 1 = two distinct color phases 17. Leaf phyllotaxis: leaves are initially op- posite in all species of the section at the seedling stage, but in three species they become alternate soon thereafter. Opposite = 0, alternate = 1. flowers in three RESULTS Two equally most parsimonious hypotheses were generated by our data matrix, designated here as cladogram A and cladogram B (Fig. 1). Both have 23 steps, with a consistency index of 0.75 after the noninformative characters were excluded. The successive weighting procedure did not discrimi- nate either of the two trees, which differ only in the placement of F. cyrtandroides and F. pro- cumbens. In cladogram A, F. cyrtandroides is the sister species of the three New Zealand species, while in cladogram B, F. procumbens is the sister species to F. cyrtandroides, F. excorticata, and F. perscandens. In both cladograms, F. excorti- cata and F. perscandens form a clade character- ized by the presence of constricted floral tubes. 520 Annals of the Missouri Botanical Garden other sections 13(1) pas cyrtan 1(1) 201) 3(1) 14(2) 16(1) m— other sections 10) 20) 40) 70) 170) AA+ + В 7(2) 8(1) 7 t 3(0) 14(0) 16(0) 7(2) 8(1) 9(1) 10(1) 11(1) 16(2) 4(0) 7(0) 17(0) 13(1) cyrtan 3(1) 14(2) 16(1) 4. d d " d 14(1) 16(1) + + persca 12(1) 6(1) 6(1) + + excort 901) 100) 11(1) 1802) 4 + + + procum FIGURE 1. The two most parsimonious cladograms (length = 23; consistency index = 0.75 when UE characters are excluded) found from is the sister group of the New Ze rest of the se reversals. aland species Of the 17 characters used, 7 are informative in defining the phylogeny of the section. Only one of these remains as originally coded in both trees (#12, floral tube constriction), with the differences in the two trees determined by the evolution of the other six characters. In cladogram A, one extra step is required in each of three characters: 3 (ovule number), 14 (chlorophyll presence in spongy me- sophyll), and 16 (floral color change), whereas in cladogram B the extra step occurs in characters 4 (flavone sulphates), 7 (sexual system), and 17 (leaf phyllotaxis). Characters 1 and 2 (flavones, blue pollen) are present in all members of the in- group and define the section as a monophyletic group. Seven characters are autapomorphies (5, 6, 8,9 ll, and 13), with three additional autapomorphies in two multistate, nonadditive characters (14 and 15). DISCUSSION The main consequence of our reevaluation of flavonoid data in sect. Skinnera was a reduction the scie of the data matrix. — A. Th — B. The topology in ection. Character state changes are superimposed on the кыы single lines = e b 1 in which F. androides ocumbens is the ка. P group of the e apomorphies, X = which F. in the number of flavonoid characters from seven (four informative) in Wiliams & Garnock-Jones (1986) to four in our analysis (only one of these informative). This was due to a more conservative choice of characters employed in our data matrix. We justify our use of flavonoid classes (instead of individual compounds) as being more congruent with current knowledge of flavonoid evolution, es- pecially when the data are based on general surveys such as in Fuchsia (Gornall & Bohm, 1978; Rich- ardson, 1983; Averett & Raven, 1984). Our ma- trix also reflects character state changes in some the characters that were maintained in bot studies, such as flavones in F. cyrtandroides and flavonol glycosides in F. procumbens and F. per- scandens. These are “presences” of compounds that were not detected by Williams & Garnock- Jones (1986), but were found by Averett et al. (1986). Averett et al. (1986) examined 3-16 in- dividuals per taxon in sect. Skinnera compared to 1—3 per taxon in Williams & Garnock-Jones (1986), which supports the use of larger sample sizes in Volume 77, Number 3 1990 Crisci 4 Berry Phylogenetic Reevaluation of Old World Fuchsia 521 flavonoid surveys to more reliably detect the pres- ence of compounds. Our results differ considerably from the previous cladistic analysis of Williams & Garnock-Jones (1986), whose single shortest cladogram is shown in Figure 2. In their tree, F. perscandens forms a terminal clade with F. procumbens, whereas our results place F. perscandens in a terminal clade with F. excorticata (Fig. 1). Williams & Garnock- Jones used two flavonoid characters, presence of apigenin and loss of flavonols, as apomorphies de- fining their perscandens-procumbens clade, but the study by Averett et al. (1986) showed that both kinds of compounds are in fact found in all members of the section. Their third apomorphy for this clade, lianoid habit, is inappropriate, since it is unlikely that the creeping, barely woody char- acter of F. procumbens is homologous with the woody, scandent habit of F. perscandens. The only character that defines our perscandens-excorti- cata clade is the strongly constricted floral tube, and this is probably associated with pollination by honeyeater birds (Meliphagidae) in these two species (Thomson, 1927; Delph «€ Lively, 1985). Bird pollination, on the other hand, is not known to occur in F. procumbens. The other main difference in our cladistic anal- ysis from that of Williams & Garnock-Jones is the ambiguous resolution of the sister species to the rest of the section (F. cyrtandroides or F. pro- cumbens). With the Tahitian F. cyrtandroides as the sister species, our first hypothesis requires that F. procumbens (a) reverted back to a low ovule number, (b) regained chlorophyll in the spongy mesophyll (green leaf undersides), and (c) lost the derived floral color change. With F. procumbens as the sister species, F. cyrtandroides is required to have (a) lost flavone sulphates, (b) lost male sterility, and (c) reverted back to opposite leaves. This situation indicates that there must be extensive homoplasy in the section, but it is not yet evident from our results in which set of the above char- acters this has occurred. The large number of autapomorphies in F. pro- cumbens underscores how differentiated this species is from the rest of the section, yet does not help us determine if those characters are the result of a fundamental divergence or are a secondar velopment related to an unusual pollination syn- drome or habitat type, for example. Because Ё. procumbens is now so rare in nature and its natural pollinators may have gone extinct, its pollination system remains unknown. The species is distinctive in Fuchsia, however, in occupying a seashore hab- itat, and it is restricted to the northern part of the other sections of Fuchsia F. cyrtandroides F. excorticata > Р. х colensoi Р. perscandens F. procumbens FIGURE 2. The single most parsimonious cladogram obtained by Williams & Garnock-Jones (1986); this differs from our hypotheses mainly in the position of F. procum- North Island, an endemic-rich area of New Zealand (Craw, 1988). Fuchsia’s disjunct distribution between the New World and New Zealand and Tahiti has stimulated several hypotheses about its origin. Croizat, who opposed arguments of chance long-distance dis- persal in Fuchsia and other groups, first considered the presence of the genus on Tahiti as an ancient one that exemplified the “trans-Pacific” track join- ing southeastern Brazil, south-central Chile, and the southwest Pacific (Croizat, 1962, p. 547). Lat- er, Raven (1972, 1979a) suggested that Fuchsia reached New Zealand by trans-Pacific long-dis- tance dispersal from South America, first to New Zealand by the mid-Miocene (to account for the fossil pollen of Fuchsia dated from that time; see Couper, 1960), then secondarily to Tahiti. More recently, Berry (1982) proposed an older, more direct connection from South America to New Zea- land via Antarctica, with subsequent dispersal to Tahiti. This was based on older, Oligocene fossil records of Fuchsia from New Zealand (Mildenhall, 1980) and a newer understanding of the oppor- tunities for direct migration across Antarctica until the Miocene (Raven, 1979b). Most explanations for the presence of Fuchsia sect. Skinnera on Tahiti have been based on hy- potheses of recent, probably bird-mediated dis- persal from New Zealand (Carlquist, 1967, 1974; Fleming, 1976; Godley, 1979; Raven, 1979a; Ber- ry, 1982). This view is based on the isolated oceanic position of Tahiti, its volcanic origin and recent age (less than two million years old; Dymond, 1975), and the fleshy fruit and small seeds of F. cyrtan- droides. As a result of their cladistic analysis show- ing F. cyrtandroides to be the sister species of the rest of the section, Williams € Garnock-Jones (1986) dissented with this view, suggesting that F. 522 Annals of the Missouri Botanical Garden cyrtandroides was derived instead from the com- mon ancestor of the section before the events that produced the different species in New Zealand. Although cur results do not allow us to support one of these hypotheses, phylogenetic studies in other groups of organisms with similar distributions could help clarify the biogeographical relationships within Fuchsia sect. Skinnera. Our results do in- dicate, however, that we should place more em- phasis on the possibility that F. procumbens is the sister species of the rest of the section, in which case its large suite of ipe: may reflect a long history of divergence from the rest of the group. LITERATURE CITED ALLAN, H. H. 1927. pi of wild xo in the New Zealand flora V. Genetica 9: 499-5 New Ze dud "Trees and UN and How to Identify Them. Whitcombe & Tombs, Auck- land. AVERETT, J. E. & P. Н. Raven. 1984. ji eser н nagraceae. Ann. Missouri Bot. Gard. . HAHN, P. E. BERRY & Р. " ii 1986. Flavonoids and flavonoid evolution in Fi а hsia ео Amer. J. Bot. 73: 1525-15 Berry, Р. E. 1982. The systematics and е of е E Fuchsia. Ann. Missouri Bot. Gard. 69: Boum, D E 987. Ve gres flavonoid variation. v (neater 53: -279. е "D. E., . BERRY a Р. H. Raven. 1982. = Mexican Central American species of Fuch- a nagra aceae) except for sect. Encliandra. Ann. Miss a Bae Gard. 34 Camas, $. 967. The biota of long distance dis- persal. V. Plant Pure to Pacific islands. Bull. Torrey Bot. Club 94: 70 ————. 1974. Island A Columbia Univ. Press, New York. Couper, R. A. 1960. New Zealand Mesozoic and Cai- nozoic plant E iia New Zealand Geol. Surv. Paleontol. Bull. Craw, R. 1988. a the synthesis between pan- biogeography, phylogenetic systematics and geology as illustrated by empirical studies on the biogeogra- phy of New Zealand and the Chatham Islands. Syst. Zool. 37: 291-310 Croizat, L. Space, Time, Form: The Biological Synthesis. Caracas, Venezuela. [Published by the au- thor. CROWDEN, R. K., J. Wnicur & J. В. HARBORNE. 1977. Anthocyanins of Fuchsia (Onagraceae). Phytochem- istry 16: 400-402. C. M. LiveLY. 1985. Pollinator visits to floral colour phases of Fuchsia excorticata. New Zealand J. Zool. 12: 599-603 DvMoND, J. 1975. K-Ar ages af Tahiti and Moorea, Society Islands, and implications for the hot-spot model. 3: 236-240. Farris, J. S. 1988. HENNIG86. Version 1.5. Docu- mentation. 1 The retention index and vd rescaled consistency index. Cladistics 5: 417-4 FLEMING, C. A. 1976. New Zealand plants as a minor source of terrestrial plants and animals on the Pacific. Tuatara 22: 30-37. GopLEY, E. J. 1955. Breeding systems in New Zealand plants. 1. Fuchsia. Ann. Bot. (London) 19: 549- 1963. Breeding systems in New Zealand plants. 2. Genetics of the sex apr in Fuchsia procumbens. New Zealand J. Bot 1979. Flower ology in | New Zealand. New Zealand J. Bot. 17: GORNALL, R. J. & B. 1978. Angiosperm flavonoid evolution: a reappraisal. Syst. Bot. 3: 353- 368. GOTTLIEB, O. R. 1975. Flavonols. Pp. 296-375 in J В. Harborne, Т. J. Mabry & Н. Mabry ТА The Flavonoids. Chapman & Hall, London. HARBORNE, J. B. 1975. Biochemical a of fla- vonoids. Pp. 1056-1095 in J. В. Harborne, Т. J. Mabry & H. Mabry мше The ode "Chap- man & Hall, London Mabry (editors), The Flavonoids: Advances in Re- search. Chapman & Hall, London. Mappison, W. P., M. J. DONOGHUE & D. R. MADDISON. 984. Outgroup analysis and parsimony. Syst. Zool. 3-103. Шренк D. C. New эон Late due: us and Cenozoic plant biogeography: a co она шый il Palaeoecol. EN -233 PLATNICK, N. L 1989. An empirical comparison of microcomputer parsimony programs, II. Cladistics 5: 145-161. Raven, P. Н. 1972. Plant species disjunctions: a sum- mary. Ann. Missouri Bot. Gard. 59: 234-246. . 1979a. A survey of Pops biology in the Onagraceae. New Zealand J. Bot. . 1979b. Plate tectonics = Southern Hemi- sphere biogeography. п К. Lars Holm-Nielsen (editors), Tropical ШО, Academie n. TE 1983. Methods of flavonoid data analysis. Pp. 495-499 in J. Felsenstein (editor), Nu- i ASI Series, Volume G1. Sprin SUCKLING, T А. The leaf anatomy of some trees and shrubs growing in the Port prs ibi e d ; Trans. Proc. New Zealand Inst. 4 ЅҮТЅМА, К. J. J. Е. SMITH. ai ue Б mor- phology: comparisons in the Onagraceae. Ann. Mis- souri Bot. Gard 7. -123 Тномѕом, С. M. . The Ж-тан of New Zealand flowers by birds and Раа Trans. Proc. New Zea land Inst. 57: 106-125. — is & Q. Waea 1981. The out-group on method of ы analysis. Syst. Zool. 30: ; WiLLIAMS, c A., J. Н. Fronczyk & J. В. НАК 1983. Leaf flavonoids and other p penne as indicators of parentage in six ornamental Fuchsia species and their hybrids. Pini leri 22: 1953- 1957. . GARNOCK-JONES. 1986. Leaf flavonoids and other phenolic glycosides and the taxonomy and phylogeny of Fuchsia sect. Skinnera (Onagraceae). Phytochemistry 25: 2547-2549. PHYLOGENETIC IMPLICATIONS OF RIBOSOMAL DNA RESTRICTION SITE VARIATION IN THE PLANT FAMILY ONAGRACEAE! Jorge V. Стіѕсі,23 Elizabeth A. Zimmer,‘ Peter C. Hoch,* George B. Johnson,’ Christy Mudd,’ and N. S. Pan** ABSTRACT Phylogenetic relationships i in the family Onagraceae were assessed using ribosomal DNA restriction site mapping. n sites were detected among 11 species of Onagraceae (one each of Circaea, Fuchsia, Hauya, Ludwigia, and six species of Epilobium) using 20 restriction enzymes. Phylogenetic analysis of the (cluster analysis and principal coordinate analysis) were also applied to the data as a analysis. Wagner parsimony analysis generated four equally parsimonious trees (53 ste s, pared nd index 0.76), and Dollo parsimony analysis generated six equally parsimonious trees (59 steps, consistency index 0.67). In all trees Fuchsia-Hauya, and Oenothera and Epilobium form the other branch. Cluster analysis and principal component analysis do not agree with the phylogenetic analyses mainly in the position of Epilobium, suggesting the possibility of different rates of evolutionary change in different lines of the family. Onagraceae are a well-defined family of flow- ering plants, consisting of seven tribes, 16 genera, and approximately 652 species of worldwide dis- cells (Metcalfe & Chalk, 1950; Carlquist, 1961, 1975); (3) the presence of septa dividing the spo- rogenous tissue (Tobe & Raven, 1986); (4) “par- tribution (Raven, 1979, 1988). The family is un- ambiguously included in the order Myrtales (Dahl- gren & Thorne, 1984; Johnson & Briggs, 1984), sharing with all members of the order a number of traits, including a distinctive set of eight em- bryological characters (Tobe 1983). Within the order, Onagraceae are highly distinctive in the following features: (1) a characteristic four- nucleate embryo sac (Tobe & Raven, 1983); (2) the presence of abundant raphides in the vegetative Raven, acrystalline beaded" pollen ektexine (Skvarla et al., 1975, 1976); and (5) viscin threads or ektex- inous strands on the proximal pollen wall (Skvarla et al., 1978; Patel et al., 1984). Each of these features is a synapomorphy within Myrtales, and the combination of them strongly supports the nat- uralness of Onagraceae. The family has been studied in close biosyste- matic detail, and modern taxonomic revisions are available or in preparation for all species. Leaf, study was supported by ie from the ана ы Science Foundation (РЕВ-85 18906) and the John D. and o Peter Н. Кау ! This Catherine T. MacArthur Foundation and support, to Barbara A. Schaal for ae, As PAUP version 3.0 available to us prior to and/or useful discussions: Paul E. eter H. Raven, David L. aay and Alan R. Tem e are grateful to general release. "We thank the following for critical review of the manuscript Berry, Joseph uns Judith Greenlee, Ro i . We also thank W. R. Sykes, and D.S.I.R.-Christchurch for seeds or living material, P. Malesevich and the Missouri Botanical Garden for expert care of plants, M. Crisci for preparation of the drawings, and G. Hoch for typing the ma nuscript. aboratorio de Sistemática y Biologia Evolutiva, Museo de La Plata, 1900 La Plata, Argentina. З > Missouri Botanical Garden, P.O. Box 299, St. ment of ee Louisian Ө Bnet of Biology, Washin gton University, St. Lou Louis, жез аа U.S.A. ouge, Louisiana 70803, U.S.A. , U.S.A. * Present address: Dent of. Botany, Beijing University, Beijing, People’ s Republic of China. ANN. MISSOURI Вот. GARD. 77: 523-538. 1990. 524 Annals of the Missouri Botanical Garden TABLE 1. Tribes and genera of Onagraceae, with estimated numbers of species and geographical distributions of each (Raven, 1988). Estimated number of species Geographical distribution I. Jussiaeeae l. Ludwigia 82 Pantropical, best represented in South America; 3 sections with 21 species in temperate North America, 2 monotypic sections in temper- ate Asia; 7 species endemic to Africa, 1 to tropical Asia, and 2 en- demics common; total of 25 species in Old World, 12 endemic. П. Fuchsieae 2. Fuchsia 105 Andean South America, with 12 species endemic in Mexico and Central America, 2 on Hispaniola, 8 in coastal Brazil, 1 in Tahiti, and 3 in New Zealand III. Circaeeae 3. Circaea 7 North temperate forests and bordering alpine areas, 5 species endemic to Asia and all 7 found there. IV. Lopezieae 4. Lopezia 22 Mostly Mexican, 1 south to Guatemala and a second south to Panama. V. Hauyeae 5. Hauya 2 Mexico to Costa Rica. VI. Onagreae 6. Gongylocarpus 2 l endemic to 2 islands off west coast of Baja California, 1 widespread in Mexico and Guatemala, but not common. 7. Gayophytum 9 Western North America, 1 endemic to temperate western South Ameri- ca and 1 common to both Ө Xylonagra 1 Central Baja California. . Camissonia 61 Western North America, 1 endemic to temperate western South Ameri- ca. 10. Calylophus 6 Central US south to central Mexico. 11. Gaura 21 Southwestern and central US east to Atlantic coast and south to Mexico d Guatemala; centers in Texas 12. Oenothera 123 70 North American, 49 South American, plus 4 common to both, 1 of European origin; far more diverse in North America, especially in west and in Mexico. 13. Stenosiphon 1 Great Plains of central US. 14. Clarkia 44 Western North America, all but 1 in California; 1 endemic to temperate western South America. VII. Epilobieae 15. Epilobium 162 Cosmopolitan at high altitudes and latitudes; Hn species in Eurasia (9 c ica, 6 common to Africa), 46 in Australasia (all endemic), 35 in North America (26 bein 12 in South Amer- ica (10 endemic), 10 in Africa (4 endemic) 16. Boisduvalia 6 Western North America, 1 in western temperate South America, 1 common to both continents. wood, and floral anatomy have been studied in detail; chromosome numbers are known for most taxa; and breeding systems and pollinators, flavo- noids, palynology, and embryology all have been investigated for most of the family. Because of this, Onagraceae are clearly the best-known plant family of their size from many points of view. One surprising conclusion from these studies has been the distinctiveness of most genera of the fam- ily from one another; five of the seven tribes are monogeneric, and another includes two genera (Ta- ble 1). Pollen of Onagraceae has been recognized in the fossil record from the Maestrichtian Age (73-65 MYA) at the end of the Cretaceous Period Volume 77, Number 3 1990 Crisci et al. 525 Ribosomal DNA Restriction in Onagraceae Taxa analyzed, acronyms, and collection data. All vouchers deposited at MO unless otherwise noted. ТАВ ; o MBG = Missouri Botanical Garden Acronym Taxon Collection data Circaea cordata Royle БЕ МВС, Носћ 1904; seed from USSR, Vladivostok, Р. Raven 975. а MBG, Hoch M3248; seed from NEW ZEALAND, S. Island, Banks Peninsula, Wilson & Syk 198 Cultivated MBG, ез M3187; die from USA, California, Marin Co., |, агр їп Cultivated MBG, E M3259; seed from INDIA, Ganderbal, G. Dar 1983. A. с MBG, Hoch M3356; seed from NEW ZEALAND, S. Is- pop. X. B. Cultivated MBG, Hoch M3360: = from NEW ZEALAND, S. Is- C. Cultivated MBG, Hoch pras dd E from NEW ZEALAND, S. Is- pop. Cultivated MBG, Hoch M3292; ans bun NEW ZEALAND, N. Is- Meo MBG, Hoch M3258; seed from INDIA, Sind Valley, G. Dar 8664. Cultivated MBG, Hoch 1903; from Fuchsia *Gartenmeister', Longwood n, Pennsylvania 80/01 Ebi Epilobium billardieria- num Sér. Eci E. ciliatum Raf. Ehi E. hirsutum L. Eme E. melanocaulon Hook. land, Flock Hill, D.S.LR. land, Dry Creek, D.S.LR. р land, Dry Creek, D.S.LR. Epa E. pallidiflorum Sol. ex A. Cunn. land, Gardner 4191. Ero E. royleanum Hausskn. FUC Fuchsia Xtriphylla L. Gar HAU Hauya heydeana Donn. Cultivated MBG, Ho Smith ch M1499, ВИЕ from MEXICO, Chiapas, Breed- love 42080 (CAS LUD Ludwigia peploides H OEN Oenothera biennis L. USA, Missouri, St. Louis City, Forest Park, Hoch 1900. USA, Illinois, Calhoun Co., Hoch & Hoch 1901. (Drugg, 1967; Pares Regali et al., 1974a, b), in- dicating that the very distinctive genera in the family may represent evolutionary lines that di- verged in the remote past and subsequently have evolved separately (Raven, 1988). Because the genera are so well differentiated for the most part, phylogenetic reconstruction may prove to be par- ticularly difficult A major phylogenetic hypothesis within the fam- i is that кешр is the sister group of all other Onagra . This was first suggested by Eyde (1977, 1979, 1981), based on the presence in Ludwigia alone of central vascular bundles in the ovary; massive, highly ovuliferous placentas; ar- chaic nectary position; and 4+ merous perianths. Eyde's hypothesis is supported by data from che- mosystematics (Averett & Raven, 1984) and cy- tology (Kurabayashi et al., 1962; Raven & Tai, 1979; Raven, 1988). Also, in a study of the 40 N-terminal amino acid residues of the small subunit of ribulose bisphosphate carboxylase, Martin & Dowd (1986) found that Ludwigia has a conven- tional terminal sequence common to most other plants, whereas the other ten species of Onagraceae examined share a highly distinctive sequence, ac- counting for at least three nucleotide differences. Despite the large amount of information available for the family, few other unambiguous hypotheses of relationships among the tribes and genera have been possible. In view of the lack of resolution of the evolu- tionary relationships within Onagraceae based on the large base of morphological and chemosyste- matic evidence, the family may be an ideal group on which to use molecular methods to elucidate phylogeny. Results of these techniques will also contribute to the continuing debate within system- atics regarding the best methods and the best type of data with which to reconstruct phylogenetic his- tories (Patterson, 1987). ottlieb and associates have utilized evidence derived directly or indirectly from changes in pro- tein and DNA sequences to approach phylogenetic questions in Onagraceae. However, their work has focused primarily on relationships within Clarkia (Gottlieb, 1982; Sytsma & Gottlieb, 1986a, b) or Fuchsia (Sytsma & Smith, 1988), rather than between genera or tribes. The conservative organization and evolution of ribosomal DNA (Hillis, 1987) make this molecule 526 Annals of the Missouri Botanical Garden well suited for comparative systematic investiga- tions. Previous studies have clearly demonstrated the potential of rDNA for resolving evolutionary relationships because (1) it is mid to highly repet- itive; (2) the rDNA repeat length is within a range that can be examined by restriction-fragment anal- ysis; and (3) rDNA contains both slowly evolving regions (the 18S, 5.85, and 285 ribosomal genes) and more rapidly evolving regions (the transcribed and nontranscribed spacers), so that information from various levels of evolutionary history can be recovered (Appels & Dvorák, 1982). Few phylo- genetic studies using restriction map analyses of rDNA have been reported in the literature, and these are mostly confined to groups of closely re- lated organisms (Sytsma & Schaal, 1985; Hillis & Davis, 1986). Here we present a restriction-site analysis of rDNA as applied to questions of phylogenetic re- lationships among genera of the family Onagra- ceae, applying cladistic methods to analyze the restriction site variation. Some recent studies of this type of variation have applied a phenetic ap- proach (Ovenden et al., 1987; Schaeffer et al., 1987; Yonekawa et al., 1988), and we have ap- plied such approaches to our data as a complement to the cladistic analyses. Two specific questions are addressed in this in- vestigation: (1) What phylogenetic relationships can be deduced from the restriction-site data? an ) Are they congruent with our present knowledge of Onagraceae? MATERIALS AND METHODS TERMINAL TAXA Eleven species of Onagraceae were analyzed; acronyms for these species, collection data, and geographical distributions are listed in Table 2. The taxa represent six of the seven tribes of Onagraceae (cf. Table 1); only Lopezieae were not available for this study. The six species of Epilobium included were chosen to allow assessment of the utility of the rDNA approach at the infrageneric level. We included three species of Australasian Epilobium (Ebi, Eme, Epa), a group thought to be closely related and recently evolved (Raven & Raven, 1976); a species from South Asia, E. royleanum (Ero), which is morphologically similar to the Aus- tralasian group and from a region thought to be the source area of that group; and two more dis- tantly related species from Asia (Ehi) and North America (Асі) with no close affinities to the other species (Seavey & Raven, 1977, 1978). One pop- ulation was sampled for each species except Ё. melanocaulon, for which three populations were examined to test for infraspecific site variation. MOLECULAR TECHNIQUES Total DNA was isolated from 10—50 grams of frozen leaves, stems, and immature inflorescences according to a previously published procedure (Zimmer et al., 1981); for members of the Ona- graceae, most plant extractions required the use of antioxidants in the grinding buffer. Restriction endonuclease digestions, agarose gel electropho- resis, filter hybridizations, and autoradiography were performed as described in other studies of plant ribosomal gene organization (Sytsma & Schaal, 1985; Doyle & Beachy, 1985; Zimmer et al., 1988). For production of rDNA maps, nitrocellulose filters with DNA fragments from single or double digests were successively probed with nick-trans- lated DNA from the plasmids pXBrl, pGmr3, and pGmrl, which contained 5%, 50%, and 100% of a complete soybean repeat unit, respectively (Jupe etal., 1988; Zimmer et al., 1988). DNA fragments containing ribosomal-specific sequences were sized with a nonlinear regression analysis computer pro- gram (Learn & Schaal, 1987). Standard markers used included phage lambda DNA digested with EcoR I, Hind III, Sma I, or combinations thereof, and Zea mays rDNA fragments generated with BamH I. From the fragment sizes, and the posi- tional information gained from the successive hy- bridization experiments, it was possible to generate and align the physical maps of restriction endonu- clease cleavage sites for the Onagraceae ribosomal genes in the same manner described for other plant species (references cited above). A total of 65 re- striction sites for 20 commercially available en- zymes were mapped in the rDNA repeat units of all 11 species. PHYLOGENETIC ANALYSIS The restriction site data were analyzed using two different parsimony methods of phylogenetic anal- ysis: Wagner and Dollo parsimony. The preferred tree with Wagner parsimony (Kluge & Farris, 1969; Farris, 1970) is the one of minimal length in a Manhattan metric (Farris, 1972, 1981), with no "a priori” restriction on the nature of permissible character state changes, i.e., with equal probability of parallel site losses and parallel site gains. In Dollo parsimony (Le Quesne, 1974; Farris, 1977), each character (restriction site) is assumed to have arisen only once on the tree; the preferred tree is the one that minimizes the number of subsequent Volume 77, Number 3 1990 Crisci et al. 527 Ribosomal DNA Restriction in Onagraceae losses. Thus parallel site gains and loss-regains are strictly prohibited. The bootstrap sampling method (Efron, 1979; Felsenstein, 1985) was used to place confidence intervals on monophyletic groups. This technique samples from the character set at random and provides the number of times that each partition of OTUs occurred. One hundred replicate samples were performed using Wagner and Dollo parsi- mony, and a majority-rule consensus tree was con- structed (see review of consensus techniques in Smith & Phipps, 1984). The confidence intervals for each monophyletic group found in the consen- sus tree are given in percentages in the **consensus figures." The consensus tree of the bootstrap meth- od does not necessarily have the same topology as the shortest tree(s) generated by Wagner or Dollo parsimony in the original analysis. This is a limi- tation of the consensus techniques (Miyamoto, 1985) as used in the construction of general clado- grams and classification, but it is not necessarily problematic when used in association with the bootstrap method to reveal which clades are stable, unstable, or ambiguous. The direction of evolutionary change (character- state polarization) of restriction sites was deter- mined by outgroup comparison (Hennig, 1966; Watrous & Wheeler, 1981; Humphries & Funk, 1984). Sites present in both ingroup and outgroup are considered symplesiomorphic; sites present or absent in only part of the ingroup are considered synapomorphic. Whereas the best choice for out- groups would be other families of the order Myr- tales (Dahlgren & Thorne, 1984), no data on rDNA restriction site variation are available for those groups. We performed the analysis using a func- tional outgroup for Onagraceae. The functional outgroup method formalized recently by Watrous & Wheeler (1981) works as follows: if some reliable criterion can be used for dividing the taxa into two or more groups, then these groups can be used as the outgroups of one another. The technique of “functional outgroup" can be applied effectively to Onagraceae because there is strong evidence from floral anatomy (Eyde, 1977, 1979, 1981) and molecular studies (Averett & Raven, 1984; Martin & Dowd, 1986) that Ludwigia (Jussiaeeae) represents a branch distinct from the rest of the family (Raven, 1979, 1988). Therefore we have used Ludwigia as a functional outgroup to polarize the restriction site data. PHENETIC ANALYSIS Cluster analysis and principal coordinate anal- ysis, an ordination technique, were performed on the data. A Manhattan distance coefficient (Sneath & Sokal, 1973) between each pair of the 11 OTUs was calculated for the cluster analysis. The re- sulting OTU x OTU distance matrix was used to calculate a phenogram using the UPGMA cluster- ing procedure (Romesburg, 1984). The cophenetic correlation coefficient was computed as a mea- surement of distortion (Sokal & Rohlf, 1962). Prin- cipal coordinate analysis (Gower, 1966) was per- formed on the OTU x OTU Manhattan distance matrix calculated in the cluster analysis procedure, and the first three factors were extracted. To ex- amine the efficiency of the method, the Euclidean distance (Sneath & Sokal, 1973) between all pairs of OTUs in factor space was calculated using the eigenvector matrix, and the resulting OTU x OTU distance matrix was compared with the original distance matrix using the cophenetic correlation coefficient. The computational work was performed on an IBM-AT microcomputer. Wagner and Dollo par- simony analyses were performed using PAUP (ver- sion 3.0, Swofford, 1988) with the branch and bound algorithm (Hendy & Penny, 1982). The bootstrap analysis was performed using PHYLIP (version 3.0; Felsenstein, 1985); the global option was used to find the shortest tree. Cluster analysis and principal coordinate analysis were performed using NTSYS-pc (version 1.22, Rohlf, 1987). The Dollo program of the PAUP package as- sumes that the ancestor has all states O. This as- sumption will rarely if ever be true with restriction site data, since any outgroup usually has a number of restriction sites present. In considering Lud- wigia as the functional outgroup, we found that it did not have the fewest restriction sites among Onagraceae. In order to guarantee that Ludwigia would remain cladistically outside the functional ingroup, we had to force the monophyly of the remaining Onagraceae using the “СОМ- STRAINTS” option of PAUP. PHYLIP does not have a comparable option, so for the bootstrap runs we included an imaginary character that has O for Ludwigia and | for all the other OTUs. This imaginary character is then weighted 10 times. Although the new character brings 10 extra steps to the tree, these have been subtracted out of the total when reporting the results. RESULTS PATTERN OF RDNA REPEAT UNIT VARIATION The rDNA repeat units from 11 species of On- agraceae were analyzed with 20 restriction endonu- cleases that for the most part cleave at six-base 528 Annals of the Missouri Botanical Garden TABLE 3. Ribosomal DNA restriction site mutations used in phylogenetic and phenetic analyses of 11 species of Onagraceae using 20 enzymes. Acronyms and voucher information for all species are listed in Table 2. Map positions (refer to Fig. 1) were derived primarily from OEN. Characters were scored as follows: 0 = site absence; 1 = site presence; ? = equivocal; — = no sample (? and — scores were treated as missing data in the analyses). Map Site position LUD OEN FUC СК HAU Ес Epa Ebi Ero Ehi Ете l. Xba I 157 l l 1 1 1 1 1 1 1 1 1 2. BamH I 576 1 1 1 1 1 1 1 1 1 1 1 3. Sac I 1363 1 1 1 1 1 1 1 1 1 1 1 4. Bgl I 1422 1 1 1 1 1 1 1 1 1 1 1 5. Sma I 1537 1 0 0 0 0 1 1 1 1 1 1 6. Tth I 1555 1 1 1 1 1 1 1 1 1 1 1 7. EcoR I 1572 1 1 1 1 1 1 1 1 1 1 1 8. BamH I 1629 0 0 ? 0 0 0 0 0 0 0 0 9. Xmn I 1819 0 0 1 1 1 0 0 0 0 0 0 10. Sph I 1893 0 0 0 1 0 0 0 0 0 0 0 11. Bel I 2011 0 0 0 0 0 1 1 1 1 1 1 12. EcoR V 2159 1 1 1 1 1 1 1 1 1 1 1 13. Sac I 2376 1 1 1 1 1 1 1 1 1 1 1 14. Tth I 2841 1 1 1 1 1 1 1 1 1 1 1 15. Sph I 3201 1 1 1 1 1 1 1 1 1 1 1 16. BamH I 3222 1 1 1 1 1 1 1 1 1 1 1 17. Xho I 3227 0 0 0 0 0 1 1 1 1 1 1 18. Sac I 3762 1 1 1 1 1 1 1 1 1 1 1 19. Stu I 3707 0 1 0 0 0 1 1 1 1 1 1 20. Sma I 3795 1 0 0 0 0 0 0 0 0 0 0 21. Bgl I 3798 1 1 1 1 1 0 0 0 0 0 0 22. Bgl II 3932 1 1 1 1 1 1 1 1 1 1 1 23. BamH I 4202 1 1 1 1 1 1 1 1 1 1 1 24. Xmn I 4232 1 1 1 1 1 1 1 1 1 1 1 25. Xho I 4245 0 0 0 0 0 1 1 1 1 1 1 26. Sma I 4359 1 0 0 0 0 0 0 0 0 0 0 27. Xmn I 4662 1 1 1 1 1 1 1 1 1 1 1 28. Tth III 4929 1 1 1 1 1 1 1 1 1 1 1 2 ac I 5121 1 1 1 1 1 1 1 1 1 1 1 30. EcoR I 5373 1 1 1 1 1 1 1 1 1 1 1 31. Sph I 5445 1 1 1 1 1 1 1 1 1 1 1 32. Tth III 5487 1 1 1 1 1 1 1 1 1 1 1 33. Sac I 5578 0 0 0 0 0 1 1 1 1 1 1 34. Stu I 5742 0 0 0 0 0 0 1 0 0 0 0 35. Xmn I 5885 1 1 1 1 1 1 1 1 1 1 1 36. Xba I 5908 0 1 0 0 1 0 0 0 0 0 0 37. EcoR V 5993 0 0 0 0 0 1 1 1 1 1 1 38. Sph I 6083 0 0 0 0 0 1 0 0 1 1 0 39. Bel I 6112 0 0 0 0 0 0 1 1 1 0 0 40. Hind Ш 6115 0 1 1 0 1 0 0 0 0 0 0 41. Stu 6116 1 0 0 1 0 0 0 0 0 0 0 42. Sac I 6120 1 1 0 1 0 0 0 0 0 0 0 43. Nde I 6376 0 0 0 0 0 1 1 1 1 1 1 44. Tth 1 6555 0 0 0 0 0 0 1 1 1 0 1 45. Xmn I 6726 0 0 0 0 0 1 0 0 1 0 0 46. EcoR I 6729 1 0 0 0 0 0 0 0 0 0 0 47. Sac I 6994 0 0 0 0 0 0 0 0 1 0 0 48. EcoR V 7358 1 0 0 0 0 0 0 0 0 0 0 49. Bel I 7888 0 0 0 0 0 1 1 1 1 1 1 50. EcoR V 8069 0 0 0 0 0 1 1 1 1 1 1 51. Tth I 8184 0 0 0 0 0 1 1 1 1 1 1 52. BamH I —1825 1 1 0 1 0 0 0 0 0 0 0 Volume 77, Number 3 Crisci et al. 529 1990 Ribosomal DNA Restriction in Onagraceae TABLE 3. Continued. Map Site position LUD OEN FUC СК HAU Ес: Epa Ebi Ero Еш Ете 53. BstE I —1749 0 0 0 1 1 0 0 0 0 0 0 54. Stu —1702 1 0 0 1 0 0 0 0 0 0 0 55. Bgl I —1174 1 0 1 0 1 0 0 0 0 0 0 56. Nde I —1096 0 0 1 0 0 0 0 0 0 0 0 57. Bgl II —902 0 0 1 0 0 0 0 0 0 0 0 58. BstE I —665 1 0 1 1 0 0 0 0 0 0 0 59. XhoI —500 — 0 1 0 0 0 0 0 0 0 0 60. BstE I —299 1 0 1 0 0 0 0 0 0 0 0 61. Sca I —212 0 0 — 1 1 0 0 0 0 0 0 62. Sph I —176 1 1 1 1 1 0 0 0 0 0 0 63. Tth III —50 0 1 1 1 1 0 0 0 0 0 0 64. Nde I 48 1 1 1 1 1 1 1 1 1 1 1 65. Xmn 1 101 1 1 1 1 1 1 1 1 1 1 1 recognition sites. Sixty-five sites were mapped in the rDNAs of each species (Table 3; Fig. 1). Thirty- four of these 65 sites were in regions coding for the large, mature cytoplasmic ribosomal RNAs (po- sitions 1-8, 14-37, and 64-65). Another буе po- sitions (9—13) occur in the internal transcribed spacer (ITS). All but one of the cleavage sites that were invariant (24/25) occurred in these tran- scribed portions of the ribosomal repeat unit. This pattern is similar to that observed with many other groups of higher plant rDNA repeat units (Sytsma & Schaal, 1985; Zimmer et al., 1988). Extensive variation also was found for the length of inserts in the nontranscribed portion of the in- tergenic spacer region of the rDNAs (Table 4; Fig. 1). Although this variation was not phylogenetically informative, and therefore not used in any of the phylogenetic analyses reported here, there was some indication from the restriction site data that species sharing common inserts are closely related. There were no restriction site differences found among the three population samples of Epilobium melanocaulon, and in Table 3 only one entry is recorded for this taxon. However, there were minor differences in repeat lengths among the samples (Table 4), indicating the presence of some infra- specific variation. The basic data matrix consists of 65 restriction sites scored for each of 11 OTUs (Table 3). Only 34 of these mutations are phylogenetically infor- mative, i.e., shared by two or more OTUs, 25 are constant among the 11 OTUs, and б are autapo- morphies (i.e., unique mutations). The Wagner par- simony analysis resulted in four equally parsimo- nious trees (Fig. 2.1 A-D) requiring a total of 53 steps, with a consistency index of 0.76 (Kluge & Farris, 1969). The calculation of the consistency index includes autapomorphies. When there was ambiguity in reconstructing the sequence of changes (Swofford & Maddison, 1987), gain-loss and par- allel loss sequences were preferred over parallel gain and loss-regain sequences (Templeton, 1983a, b). The different kinds of homoplasious changes in the four trees are shown in Table 5 In the four trees, with Ludwigia as outgroup, Circaea is the sister group of the rest of the family, and the six species of Epilobium form a mono- phyletic group. Inside Epilobium there are two hypotheses of relationships: (1) Eci and Ehi are a monophyletic group and the rest of the species are another monophyletic group; the latter has Eme as a sister species of the trichotomy Epa-Ebi-Ero (Fig. 2A, C); and (2) similar to the first, except Ero forms a monophyletic group with Eci and has Ehi as its sister group (Fig. 2B, D). Aside from Epilobium there are also two hy- potheses: (1) Oenothera is the sister group of Fuch- sia and Hauya, the latter forming a sister group to Epilobium; and (2) Fuchsia is the sister group of Hauya, Oenothera, and Epilobium, with Oe- nothera as the sister group of Epilobium. The majority-rule consensus tree constructed from 100 bootstrap samples (Fig. 2.11) shows that the six species of Epilobium form a highly signif- icant (100%) monophyletic group. Within Epilo- bium, two groups occur with 50% or more fre- quency: Ero-Eci (50%) and Ebi-Epa (52%). The other two groups inside Epilobium have confidence intervals of less than 50%: Ero-Eci-Ehi (46%) and Ebi-Epa-Eme (27%). The group formed by the species of Epilobium plus Oenothera has a 39% confidence level, whereas Fuchsia and Hauya form 530 Annals of the Missouri Botanical Garden 18s (1.8kb) - 1000 - 3000 26s (3.6kb) +— BamH I L— Stu 1 BstE ! ET Sca! ——— Soh 1 h BamH | T hi атн 1 Ecor | ч —— Tth 1 Sph I Xho)? BamH! Stu I Fr $тат Sec! A Bgl ll ате ¡Xho | Xmn | Tth Ml HA Sac | | | — EcoR E ps l Tth 1и ac Stu I Xmn ! Xba І EcoR V Sph I —____— Bcll OT dm Sac | Stu 1 — Nde | P—EcoR | Sac 1 EcoR V nw: =! EcoR V pone Tft jJ TABLE 4. Major repeat-unit lengths (kb) in rDNA of selected taxa of Onagraceae. Taxon Repeat units LUD 6.9, 8.6, 10.6, 11.5 OEN 10.0, 13.0 FUC 9.3 CIR 9.2 HAU 7.7, 11.4, 12.6, 16.5 i 6 Ehi 9.5, 10.4, 11.0, 11.5, 14.7 8 8.3 Eme C 8.1, 8.3 a monophyletic group with a 46% confidence level. All of the OTUs beyond Circaea and Ludwigia form a monophyletic group with a 54% confidence level. The Dollo parsimony analysis resulted in six equally parsimonious trees (Fig. 3) requiring a total of 59 steps, with a consistency index of 0.67. this case, all homoplasious site changes are parallel losses or gain/losses. The six species of Epilobium form a monophyletic group in all trees. In three of the trees (A, C, E), there are км ЕИ іп Epilobium that are also monophyl (((Eme-Ebi-Epa)Ero)Eci)Ehi), А e Ebi-Epa) as a trichotomy. Їп the other three trees (В, D, F), two monophyletic subgroups ((Eci-Ehi)Ero) form a trichotomy with Ebi and Epa, with Eme as out- group to the other five species. Oenothera is the sister group of Epilobium in all trees. In four of the six trees (Fig. 3C-F), Circaea is the sister group of the remaining OTUs (excluding Ludwig- ia), and in the other two trees (Fig. 3A-B), Circaea forms a monophyletic group with Fuchsia and Hauya. Fuchsia and Hauya form a monophyletic group in four of the trees (Fig. 3A-B, E-F), but in the other two (Fig. 3C-D) Fuchsia is the sister group of the group formed by Hauya, Oenothera, and Epilobium. FIGURE 1. Мар of the Onagraceae ribosomal DNA repeat unit oun the location of 65 restriction sites mapped by 20 commercially available Ne TRI RA Map positions were derived primarily from OEN. Map includes 185 and 285 genes, internal transcribed spacer (ITS), and nontranscribed spacer (NTS). The triangle in the NTS indicates the approximate location of length mutations (Table 4). Region length and scale in kilobases (kb). Volume 77, Number 3 Crisci et al. 531 1990 Ribosomal DNA Restriction in Onagraceae RE 2. Wagner parsimony analysis for s species of Onagraceae, using soit yd as ри oe and based on 65 rDNA restriction site mutations. I. A-D, four alternative most parsimonious trees; consistency index = 0.76. II. Majority-rule consensus tree constructed from 100 bootstrap samples s. The. dd indicate the number of times that monophyletic group occurred in the dimid samples. The majority-rule consensus tree constructed РНЕМЕТІС ANALYSIS OF RESTRICTION SITES from 100 bootstrap samples is shown in Figure 4. The Basic dats тані Table 9) of LLOTI x The monophyletic group formed by the six species of Epilobium is highly significant (100%). Within Epilobium none of the four groups occurred with greater than 50% frequency. All the species of 1. Cluster analysis. Figure 5A presents the Epilobium and Oenothera form a monophyletic results of the cluster analysis using the group 47% of the time. Fuchsia and Hauya form procedure. The phenogram has a high cophenetic a monophyletic group 40% of the time, and Cir- correlation coefficient of 0.97. This indicates that caea joins them 30% of the time. the phenogram presents an accurate representation 65 characters was analyzed by cluster analysis and by principal coordinate analysis. 532 Annals of the Missouri Botanical Garden TABLE 5. Homoplasious changes in four trees pro- duced by Wagner parsimony analysis using Ludwigia as the functional outgroup, run with the ACCTRAN option of PAUP; numbers in parentheses derive from the DEL- TRAN option, which favors convergent gain over gain- loss; refer to Figure 2. Conver- Conver- Gain- Loss- gen gent loss gain gain loss Trees A-B 3 (1) 4 (4) 6 (8) 0 (0) Trees C-D 5 (3) 4 (3) 3 (5) 1 (2) of the original distance matrix. There are two well- defined and distant phenetic clusters: the six species of Epilobium, and the remaining OTUs. Inside the Epilobium cluster there are also two well-defined phenetic subgroups: Eci-Ehi and Epa-Eme-Ebi. Ero is an isolated OTU intermediate between the two Epilobium subgroups. Ludwigia is an isolated OTU in the other cluster. Oenothera and Hauya are grouped together, with Fuchsia and Circaea pro- gressively less closely related to them. The re- sults of the ordination method are shown in Figure 2. Principal coordinate analysis. 5B. The percentages of variation explained by each of the three factors are: I = 83.71%, II = 10.04%, and III = 4.39%. The total variation explained by this model is 98.14%, with a high correlation coef- ficient (0.99). These results therefore present an accurate representation of the original distance ma- trix. The principal coordinate analysis results in three basic, phenetically significant groups: (1) the six species of Epilobium; (2) Oenothera, Fuchsia, Hauya, and Circaea; and (3) Ludwigia. The six species of Epilobium form a closely related group, with Ero in a relatively isolated position, two groups of tw OTUs each (£ci-Ebi and Ehi-Eme), and Epa in an intermediate position between Ero an Eci-Ebi. Within the group formed by Oenothera, Hauya, Fuchsia, and Circaea, the genera are much less close to each other than are the six species of Epilobium to each other. Fuchsia is relatively isolated in this group. Factor I is a good discriminator of Epilobium from the rest of the family; factor Il is a good discriminator of Lud- wigia from the rest of the family; and factor III is a good discriminator of Fuchsia from the rest of the family. DISCUSSION Restriction site data on the ribosomal DNA of Onagraceae provide valuable new information on the evolution of a diverse group of genera. Because rDNA contains both highly conserved coding re- gions and variable spacer regions, phylogenetically informative sites are found both within Epilobium, where many of the species are closely related (Ra- ven & Raven, 1976), and between distantly related genera within the family. PHYLOGENETIC ANALYSES Trees A and B (Fig. 2.1) derived by Wagner analysis have Oenothera anomalously as a sister group to Fuchsia, Hauya, and Epilobium, in con- flict with other trees and other phylogenetic infor- mation. In particular, Oenothera and Epilobium share several synapomorphies, including lobed stig- mas, absence of stipules, and presence of interxyla- ry phloem (Carlquist, 1975), none of which are shared with Fuchsia and Hauya. Oenothera also has very specialized chromosome number and structure compared with FUC-HAU (Kurabayashi et al., 1962; Raven, 1979, 1988). In addition, trees А and B have more convergent site gains than do trees C and D (Table 5). Several authors (DeBry & Slade, 1985; Mindell & Sites, 1987; Jansen & Palmer, 1988) have suggested that parallel site gains are much less probable than parallel losses for this kind of data. Trees C and D differ in topology only within Epilobium, and of the two, tree D best fits our hypotheses based on geography and morphology. This groups the three morphologically similar species from New Zealand (Epa-Ebi-Eme) and places the Himalayan species (Ero) with the remaining two species, a position supported by the bootstrap consensus tree (Fig. 2.11). The phylogeny for the other genera in tree D, in which Circaea, Fuchsia, and Hauya are more primitive than Oenothera and Epilobium, does not contradict evidence from floral and veg- etative anatomy and chromosome number and morphology. On the contrary, the position of Epi- lobium as one of the most apomorphic genera is congruent with the specialized chromosome mor- phology and number (Kurabayashi et al., 1962; Raven, 1976), specialized integument structure (Tobe & Raven, 1985), lobed and commissural stigma (Eyde, 1981), and lack of stipules in Epi- lobium, all characters that have been established as apomorphic in Onagraceae using Myrtales as the outgroup. The bootstrap consensus tree (Fig. 2.П) does not have the same topology as any of the most parsimonious trees, but it is very close to tree D, the only difference being that Fuchsia and Hauya form a monophyletic group by themselves in the consensus tree. Volume 77, Number 3 1990 Crisci et al. 533 Ribosomal DNA Restriction in Onagraceae FIGURE 3. Dollo parsimony analysis for 11 species of Onagraceae, using Ludwigia as a functional outgroup and based on 65 rDNA restriction site mutations. A—F, six alternative most parsimonious trees; length = 59, consistency index = 0.67 Morphological data support the position of Oe- nothera as the sister group of Epilobium as shown by the Dollo analysis with Ludwigia as functional outgroup. In three of the six trees (Fig. 3A, C, E), the phylogeny within Epilobium is also supported by morphological data, but the phylogeny in the other three trees (Fig. 3B, D, F) has no such independent support. Morphological data are also less helpful in choosing among the hypotheses for the other genera in the six Dollo trees. Trees C and D have the most conservative phylogeny, with stepwise additions of Hauya, Fuchsia, and Circaea to the monophyletic Oenothera-Epilobium group. Trees E and F differ in making Fuchsia and Hauya a monophyletic group, leaving Circaea as sister group to the rest of the taxa. Tree A, in which the monophyletic group (CIR(HAU-FUC)) forms a sis- ter branch to OEN-EPI, and which includes the favored phylogeny in Epilobium, is nearly coin- cident with the bootstrap consensus tree (Fig. 4), 534 Annals of the Missouri Botanical Garden FIGURE 4. Dollo parsimony analysis for 11 species of Onagraceae using Ludwigia as functional outgroup. Majority-rule consensus tree constructed from 100 boot- strap samples. Percentages are as in Figure 2 differing only in that Eci-Ehi form a monophyletic group in the consensus tree. Several authors, as noted above, argue that Dollo parsimony is a more consistent and efficient estimator of phylogenetic relationships for restriction site data because par- allel site losses are much more probable than par- allel site gains. Based on these considerations, we prefer tree A of the Dollo analysis as the most likely geneological hypothesis for the taxa we have igure 6 shows this hypothesis, with the characters superimposed on the tree; there are no included. parallel gains or loss-gains. Wagner analysis can generate a tree of the same topology that is only 2 steps longer than the most parsimonious ones, but there are at least 300 trees of equal or shorter length. DISTANCE FIGURE 5. Phenetic analysis of 11 species of Onagraceae based on 65 rDNA restriction site mutati anh attan paved ps — B. Principal coordinate analysis. Perspective three-dimensional projection of the OTUs on the axes representing 3.7 the first three factors. The percentages of va riation explained by each of the three factors are: I = 83. 10.04%; and Ш = 4.39%. The total variation explained by the model is 98.14% Volume 77, Number 3 1990 Crisci et al. 535 Sinai DNA Restriction in Onagraceae LUD OEN Есї Epa Ebi Eme Ero Ehi FUC HAU CIR 5(0) = =æ 10(1) 36(0) 53(0) 45(1) 36(0) ji 40(0) Ж 42(0) = 4100) 9250) | 42(0) of 52(0) T 54(0) 3000) Ж 55(0) 4. 63(0) 60(0) "T7 1100 — 17(1) 25(1) 33(1) 210) | S20 = 62(0) "T 3(1) 49(1) 50(1) 9(1) 5100 T 530) 61(1) w 19(1) 41(0) 54(0) 55(0) dz 58(0) 60(0) 20(0) 36(1) 26(0) m yr 46(0) 63(1) 48(0) | FIGURE 6. Phylogenetic hypothesis (опе of six most parsimonious trees) from Dollo parsimony analysis using Ludwigia as functional outgroup and based on variation = reversals. Note that all homoplasious changes are pa excluded from illustration. PHENETIC ANALYSIS The results of phenetic analysis using either cluster analysis or principal coordinate analysis (Fig. 5) demonstrate: (1) the extreme distinctness of Epilobium; (2) a fairly great distinctness of Lud- wigia; and (3) strong clustering of Epilobium species. We can attribute the extreme distinctness of Epilobium within the family in the phenetic analyses to the possibility of different evolutionary rates in Onagraceae. Since the algorithms for phe- in rDNA restriction sites. Restriction site gains (1) and losses (0) are superimposed onto the tree; single line = synapomorphies; double line = parallel or convergent evolution, X rallel losses or gains/losses. Constant characters have been netic techniques are usually rate-dependent (Hillis, 987; Ruvolo, 1987), they are more likely to pro- duce spurious results than are rate-independent methods such as Wagner or Dollo analysis. The groupings of Epilobium species (e.g., close clustering of Eme-Ehi and Eci-Ebi) by principal coordinate analysis are contrary to relationships suggested by both morphological and cytological data (Raven & Raven, 1976). Rohlf (1968) has noted that distances between close neighbors are not well represented by ordination techniques. 536 Annals of the Missouri Botanical Garden CONCLUSIONS compared with the other genera and the greater One limitation of this study is related to the use of restriction maps rather than nucleotide sequence ata. The restriction sites analyzed involve only about 5% of the total base pairs (ca. 390/8000+ bp) in the rDNA repeat unit of Onagraceae, and the only changes that can be detected involve gain or loss of a recognition sequence at a site and major changes in repeat length. Loss of restriction sites can be accomplished by changes at any of the six nucleotides in the recognition sequences, so the acquisition of parallel, nonhomologous losses is rel- atively likely. However, the relative likelihood of various convergent events is fairly well understood (Hillis & Davis, 1986; Templeton, 1983a, b), so certain tree-construction algorithms (e.g., Dollo parsimony) can correct for inherent biases in the data set. Furthermore, the extent of homoplasy within rDNA analysis is limited by the conservative nature of rDNA. Therefore, despite the small sam- ple size of actual base pairs examined, restriction- site mapping provides valuable data for phyloge- notic reconstruction. e lack of restriction site variation among sam- ples of three populations of Epilobium melano- caulon suggests that the consistency of the re- striction site data is high. The six species of Epilobium always cluster together, regardless of the algorithm applied to the data, and when boot- strap methods are applied, the monophyletic Epi- lobium group appears 100% of the time. This gives us confidence in the robustness of restriction site data as taxonomic character It is also possible to ue the relative per- formance of the different algorithms in analyzing our data set. As expected for restriction site data, Dollo parsimony gives better results than Wagner parsimony, when compared with the previous knowledge of Onagraceae. As noted by Jansen & Palmer (1988), however, Dollo parsimony is too rigid because it completely prohibits parallel gains, and although convergent gains have a very low probability, they can occur occasionally. A parsi- mony algorithm that gives differential weights to homoplasious gains versus losses would be pref- erable. The main difference between the phylogenetic analyses and the phenetic analyses rests on the extreme distinctness of Epilobium in the latter. Independent data from morphology, anatomy, and protein sequence analysis strongly support the re- sults of phylogenetic analyses. Therefore, the anomalous results of the phenetic analyses seem to reflect a different evolutionary rate in Epilobium sensitivity of phenetic analyses to this phenomenon. LITERATURE CITED APPELS, R. & J. Dvorák. 1982. Relative rates of di- gene sare Theoret. ay Genet. 63: 361-365 шне x & P. H. Raven. 1984. drea of Ona- e. Ann. Missouri “Bot. Gard. 71: 30-34. 1961. Comparative Plant Ган Holt, t & Winston, New York. Wood anatomy of Onagraceae, with notes on alternative modes of photosynthate move- ment in dicotyledon woods. Ann. Missouri Bot. Gard. 62: 386-424. DAHLGREN, R. & R. F. THORNE. 1984. The order Myr- tales: circumscription, variation, and les 71: 633-699. М. A. SLADE. 1985. Cladistic analysis of restriction endonuclease cleavage maps within a maximum-likelihood framework. Syst. Zool. 34: 21- 34. gra e. Camus. i Rin S J. J. & R. М. BeacHY. 1985. Ribosomal gene riation in soybean (Glycine) and its relatives. Theor, faie Genet. 70: 369-376 Drucc, W. S. 1967. Palynology of the Upper Moreno Formation (Late Cretaceous-Paleocene), Escarpado Canyon, California. Palaeontogr. Abteil. B120: 1- EFRON, B. 1979. Bootstrap methods: another look at the jackknife. Ann. Statist. 7: 1-26. EvpE, К. Н. Reproductive structures and evo- lution in Ludwigia a aceae). I. Androecium, placentation, merism. Ann. Missouri Bot. Gard. 64 644-655 979. Reproductive structures and evolution in Ludwigia (Onagraceae). II. Fruit and seed. Ann. uri чан Gard. 66: 656 “19 сае а structures and evolution їп L us ruis id III. Vasculature, nectar- ies, conclusions. Ann. Missouri Bot. Gard. 68: 470- = E S. 1970. ч 2 computing Wagner s. Syst. Zool. 19: 8 1972. die c M trees from dis- tance matrices. Amer. Naturalist 106: 645-668 Ts debui nA under Dollo's Law. Syst. Zool. 26: 77 —— ———, 1981. Distance d in phylogenetic analysis. Pp. 3-23 in V. А. Funk Brooks (editors), Advances in Cladistics: Proceedings of the First Meet- Ad à че Willi Hennig Society. New York Botanical , New York. шо. J; 1985. Confidence limits on phylogenies: n approach using the bootstrap. Evolution 39: 783- GOTTLIEB, L. D. 1982. Conservation and duplication of isozymes in plants. Science 216: 373-380. GOWER, J. С. 1966. 5оте distance properties et latoni + and Vector Biometrika 53: 325-338. Henpy, М. D. & D. Penny. 1982. Branch and bound algorithms to determine minimal evolutionary trees. Math. Biosci. 59: 277-290 Volume 77, Number 3 Crisci et al. 537 1990 Ribosomal DNA Restriction in Onagraceae HENNIG, W. 1966. е Systematics. Univ. Onagraceae as a model of plant evo- Illinois Press, Urbana, Hius, D. М. 1987. Molecular versus morphological approaches to systematics. Ann. Rev. Ecol. Syst. 18: 23-42. . Davis. 1986. Evolution of ribosomal DNA: fifty million years of recorded history in the à 8 984. Cladistic vd odology. Pp. .32 23- 362 in V. Heywood (editors), Current се. in Plant Taxonomy. a . D. PALMER. 1988. deii ii implications of chloro last DNA restriction site v ation in the tisieae (Asteraceae). Amer. J. Bot. 75: 751-764. JOHNSON, L. A. S. & B. G. Briccs. 1984. Myrtales and —a ee analysis. Ann. Missouri 1: 756. m UN & E. A. ZiMMER. 1988. al RNA genes and algal phylogeny: hn Chlamydomonas example. BioSystems 21: 223- KLUGE 4 С. & J. S. Farris. 1969. Quantitative d letics and the evolution of anurans. Syst. Zool. 18: 1-3 KURABA YASHI, M., Н. Lewis & P. Н. Raven. 1962. A Dy up owe study of mitosis in Onagraceae. Amer. 03- 026. LEARN, . A. SCHAAL. 1987. Population subdi- vision 1 for ribosomal DNA е variants in Clematis fremontii. Evolution 4: 433-438. LE posee W. 1974. The uniquely evolving character pus and its cladistic application. Syst. Zool. 23: 917. eta Р. С. & J. М. Dow. 1986. Phylogenetic studies using protein sequences within the order Myr tales. Ann. Missouri Bot. Gard. 73: 441-448. METCALFE, С. R. & L. CHAL 1950. Anatomy of the Dicotyledons. Clarendon Press, Oxford. MINDELL, D. E & J. ye бё. JR. 1987. Tissue expres- sion patterns of a isozymes: a preliminary e o phylogenetic pci po Zool. 36: 137- 4 М. М. 1985. Consensus cladograms and general classification. кыны 1: 186-189. OVENDEN, С. MACKINLAY & R. Н. 2e 987. Systematics and aloe konda al genome evo- lution of Australian rosellas (Aves: PED и Molec. Biol. Evol. 4: 526-543. Pares REGALI, M. p. S., N. s A. D. S. SANTOS. 1974a. Palinologia dos sedimentos meso- cippo do Brasil (I). Bol. Tecn. enin 17: 177-179. E. dos sedis meso-cenozoicas do Brasil (II). Bol. Tecn. Petrobras 17: 263-301. PaTEL, V. C., J. J. SkvaRLa & P. Н. Raven. 1984. Pollen rem in relation to the delimitation of . Molecules and Morphology i in Evolution: Conflict or Compromise? Cambridge Univ. Press, Cambridge. Raven, P. H. 1976. Generic and sectional delimitation in Onagraceae, tribe Epilobieae. Ann. Missouri Bot. Gard. 63: 326-340. 9 survey of reproductive biology in Onagraceae. New Zealand J. Bot. 17: 575-593 E lut .85-107 i iude аео ue (or "Plant кел Biology. А айыы Honoring G. Ledyard Stebbins. Gear & Hall, ondon. ———— & T. E. Raven. 1976. The genus Epilobium in Australasia: a systematic and pup ed somes in Ludwigia (Onagraceae). Ann. Missouri! Bot. Gard. 66: 862-879. Конік, F. J. 1968. Stereograms in numerical taxono- my. Syst. Zool. 17: 246-255. 1987 S-pc: Numerical Taxonomy and Multivariate Analysis System. Exeter Publishing Co., New York. RoMEsBURG, H. С. 1984. Cluster Analysis for Re- sea oT Lifetime Learning Publications, Belmont, Califor ш M. 1 87. Reconstructing genetic and liguistic s: phenetic and cladistic me P. 193 in . Hoenigswald e (өй 5), Bio- logical Metaphor and а Classifica tion. Univ Pennsylvania Press, Philadelphia. SCHAEFFER, 5. W. ET AL. 1987. Restriction-map vari- ation in the alcohol dehydrogenase region of Dro- = pseudoobscura. Molec. Biol. Evol. 4: 254- EN. a R. € P. H. Raven. 1977. Chromosomal differentiation and the sources of the South American species of Epilobium (Onagraceae). J. Biogeogr. 4: 55-59. 78. Chromosomal evolution in Epilobium sect. Epilobium (Onagraceae). III. Pl. Syst. Evol. 130: eta SkvaRLa, J., P. H. N & J. PRAGLOWSKI. 1975. The evolution of "int tetrads in Onagraceae. Amer. J. Bot. 62: 6-35. А & —— ——. 1976. керш А survey of Onagraceae pollen. Pp. 447-479 i Ferguson & J. Muller (editors), The Pune Significance of the Exine. Linnean Society of London. , W. F. Сніѕѕ0Е & M. SHARP. 1978. An ultrastructural study of = Pag in Ona- Pollen & Spores 2 43. 3. тв. PuiPPs. 1984. к, trees in phenetic analyses. Taxon 33: 586-594. SNEATH, P. H. A. & R. R. Sokar. 1973. Numerical Taxonomy: The Principles and Practice of Numerical Classification. W. H. Freeman & Co., San Francisco. SOKAL, R. R. & F. J. RoHLF. 2. The comparison of dendrograms by objective methods. Taxon 11: 33-40. SwOFFORD, D. L. 1988. PAUP, Phylogenetic Analysis sing Parsimony. Version 3.0. Illinois Natural His- tory Survey, Champaign, Illinois. & W. P. Mappison. 1987. Reconstructing ancestral character states under Wagner parsimony Math. Biosci. 87: 199-229. SYTSMA, K. J. & L. D. GOTTLIEB. 1986a. DNA evidence for the derivation of the erogaura from a species l Clark ia нен е), Prog, Natl. кы 2 0 4-5557. 86b. нө DNA evolution and EE pied ships in Clarkia sect. Per- leia (Onagraceae). Evolution 40: 1248-1262. RL 538 Annals of the Missouri Botanical Garden В. А. SCHAAL. Lisianthius in Panama utilizing DNA restriction agent pr ysis. Evolution 39: 594-608. & J. F. SMITH 1988. DNA and morphology: comparisons in the Onagraceae. Ann. Missouri Bot. 37. ,.1985. анай 9 HR Phylogenetic inference from restriction | endonuclease cleavage site maps with par- ticular reference to the evolution of humans and apes. Evolution 37: 221-244 Convergent evolution and non-para- f. trieti f. t d DN metric inf g sequence data. Pp. 151-179 їп B. Weir (editor), Statistical Analysis of DNA Sequence Data. Marcel Dekker, New York. na H. & P. Н. RAVEN. analysis of My Ann. Missouri a Gard. 70: т P & The histogenesis and evo lution of шшен in Qnagraceae. Ann. Missouri 468 Bot. Gard. 72: 1985, An embryological & ——. 1986. Evolution of P giate anthers in Onagraceae. Amer. J. Bot. 73: 475- 488. WATROUS, L. & Q. WHEELER. 1981. The out-group comparison method of character analysis. Syst. Zool. 30: 1-11. YONEKAWA, H., K. MORIWAKI, O. GOTOH, N. MIYASHITA, Y. MATSUSHIMA, L. Sur, W. S. CHo, X.-L. ZHEN & E а 1988. Hybrid origin of es ice “Mus musculus molossinus”: evidence from restriction analysis of mitochondrial DNA. Molec. 5: 63-78. ZIMMER, E. A., E. R. Jupe & V. E. WALBOT. Ribosomal gene structure, variation in maize and 1136 1988. and inheritance its ancestors. Genetics 120: 1125 м & V. E. WaiBor. 1981. A DNA isolation eue suitable for most higher plant species. Pl. Molec. Biol. Newslett. 2: 93-96. NEW CONSIDERATIONS REGARDING THE CORONA IN THE VELLOZIACEAE! Nanuza Luiza de Menezes? and Joao Semir ABSTRACT Subfamily Barbacenioideae of the Velloziaceae is characterized by sessile anthers that are connected to a corona (or to the upper part of the hypanthium). Barbacenia spir ralis L. B. Smith & Ayens that distinguishes it clearly from the rest of the Barbacenioideae. It has stamens in whic u, however, displays a онен hich the filaments and anthers are totally independent of the corona lobes. This feature is considered important enough to justify the ө миш оа of a new genus, based on this species Menezes (1970, 1973) concluded that the peta- loid appendices found in the genus Barbacenia (Vandelli, 1788) and described as “flattened fila- ments” by the latter author and all later specialists of the family Velloziaceae, including Smith (1962), should in fact be regarded as lobes of a corona, as in Narcissus (Amaryllidaceae). Studies of the floral vascular tube of many species belonging until then to Barbacenia showed the vascuļar bundles of the petaloid appendices to di- verge from petal and sepal traces in the upper part of the hypanthium and to display an inversion (Menezes, 1970, 1973). A similar inversion of tissues was observed by Arber (1937) in the corona petal (or sepal) bundles, and the phloem is internal to the xylem. According to the available evidence, if the peta- loid appendices were “flattened filaments,” then their vasculature should consist of bundles origi- nating from the staminal trace. But this is not the case: not only do the bundles of the corona originate in the perianth, but the staminal trace itself passes through the inside of the corona without any al- teration in its concentric structure before pene- trating the anther (Menezes, 1970, 1973, 1984). Other authors have attempted to explain the nature of these petaloid appendices. On the basis of external morphology, Noher de Halac & Cocucci (1971) interpreted them as staminodes. Smith & Ayensu (1976), while admitting that either the staminode concept of Noher de Halac & Cocucci or the corona concept of Menezes might be correct, adopted the term **coronoid appendices” (Smith & Ayensu, 1976, 19 in partial support of Me- nezes's hypothesis. he present work considers the consequences on generic delimitation within subfamily Barbaceni- oideae Menezes of our new observations on Bar- bacenia spiralis Smith & Ayensu. MATERIALS AND METHODS The following plant material was collected for analysis: Barbacenia spiralis L. B. Smith & Ay- ensu, Brazil. Mato Grosso: Muncipio de Diaman- tina, Curralinho, 13 Nov. 1980, Menezes 1043 F). Pleurostima plantaginea (L. B. Smith) Me- nezes, Brazil. Mato Grosso: Serra do Cipo at km 128, 14 July 1971, Menezes 129 (SPF). Anatomical studies were based on serial sections of material fixed in FAA 50 (Johansen, 1940) and embedded in paraffin. Sections were stained with safranin and fast green (Sass, 1951). RESULTS As the basic pattern of vascularization in the Velloziaceae has previously been described in other species by Menezes (1973), only aspects relevant to Barbacenia spiralis are described here. Figure | shows a flower of Barbacenia spiralis, ' Gratitude is due t o Dr. Simon Mayo (Royal Botanic Gardens, Kew) and Dr. W. Wayt Thomas (New York Botanical Garden) for suggestions and help with English and to Maria Odeti F. Pedrosantti (UNICAMP) for assistance preparing slides. Financial support came from CNPq (Conselho Nacional do Desenvolvimento Cientifico e Tenológico — Proc. 300939-80), FAPESP (Fundação de Amparo à Pesquisa de Sao Paulo — Proc. 81 /0088-2), and WWF (World Wildlife Fund — Project 3311). > Departamento de Botánica, Universidade de Sao Paulo, Caixa Postal 11461 — 05499, Sao Paulo, SP, Brazil. ' Departamento de Botánica, UNICAMP, Caixa Postal 1170—13.100, Campinas, SP, Brazil. ANN. MISSOURI Вот. GARD. 77: 539-544. 1990. 540 Annals of the Missouri Botanical Garden which is sectioned in Figure 2. The filament (ЕІ) is completely free from the corona (Co) lobe, and a descending basal region (Dbr) of the filament can be seen; the anther (An) is basifixed. The auriculate and sessile anther (An) is attached to the corona lobe in Pleurostima plantaginea (Fig. 3). The anther is easily detached (Fig. 4) and is accom- panied by the filament tissues ш which are su- perficially adnate to the coro п Barbacenia spiralis (Figs. 5-16) the petal таи traces (Ps) are situated opposite the petal- corona complex (Cpc), which divides into a branch that enters the petal (P) and to one that runs into the corona (Co). The ventral bundle of the carpel (V) and the stamen-carpel complex (Sc) originate from the same basic bundle, situated at the base of the ovary. The stamen-carpel complex also gives rise to the dorsal bundle of the carpel (D) and the stamen trace opposite the sepal (Ss). In the peduncular region of the flower (Fig. 6) there are 12 fundamental bundles (Fb). Further up (Fig. 7), the locules still have no ovules, but septal nectaries (Sn) appear. At the level of Figure 8 the placentas bear ovules, and nectariferous sacs (Ns) can be observed. The ventral bundles of the carpel (V), the dorsal bundles (D), the traces of the sepalar (Ss) and petalar (Ps) stamens, and of the petal-corona (Cpc) and sepal-corona (Csc) com- plexes are now distinct. The septal nectaries follow the locules right from the base (Fig. 7), and even when the locules (Lo) close (Fig. 11), the nectar- iferous sacs remain wide open (Ns; Figs. 10, 11). In Figure 12 a descending basal region (Dbr) of the filament (less prominent than in Figs. 2 and 5) is seen. This brings the filament (Fl; Figs. 13, 14) close to the style (St). Figure 12 also shows the individualization of two corona lobes (Co). At the level of Figures 14 and 15, the petals, sepals, petalar and sepalar lobes of the corona, and the filaments of petalar and sepalar stamens are all clearly distinct. At a higher plane (Fig. 16) the corona lobes are no longer visible. The anthers are introrsely dehiscent, although three anthers appear to be extrorsely dehiscent due to twisting of the filaments Figures 17-20 show details of the vasculari- zation displayed in Figures 5-16. In Figure 1 divergences (Dv) of the sepalar traces to form the corona traces (Ct) may be seen. Figures 18-20 show a single plane at different scales. The region indicated by two dotted lines in Figure 18 is shown in greater detail in Figure 19; Figure 20 shows a magnified portion of Figure 19. In Figure 20 the corona bundle (Cb) shows inversion of the vascular tissues: the xylem (X) is turned outward to face the xylem of the petal (P), while the phloem (Ph) is internal to the xylem. DISCUSSION Barbacenia spiralis was first described based on a specimen with only a flower bud. For this study, the species was collected in full flower. If there were any remaining doubts as to the true nature of the petaloid appendices in subfamily Bar- bacenioideae, they can now be put to rest, since in this species, the filaments and corona lobes are completely separate. The origin and inversion of the bundles indicate that the corona is an appen- dage of the perianth; there is, furthermore, a com- plete independence of stamen and corona. This feature is unique to Barbacenioideae, which contain some 106 species in three genera. All except В. spiralis have sessile anthers (basifixed-auriculate or dorsifixed) adnate to the corona in Barbacenia, Aylthonia, and Pleurostima sect. Graziela, or to the hypanthium in sect. Pleurostima (Menezes, 1980b) According to the circumscription of Barbacenioi- deae proposed by Menezes (1970, 1971), the ses- sile anther connected to the corona lobe is a fun- damental character. Studies of leaf anatomy (Menezes, 1970, 1975) permitted the addition of a further fundamental character for Barbacenioi- deae: the presence of a double sheath in the vas- cular bundles of the leaf (Menezes, 1980a, b, 1984). In subfamily Vellozioideae, the bundle sheath is simple and each anther is connected to a filament, although this filament may at times be inconspic- uous (Menezes, 1980a, b). -4. Flowers of Barbacenia AEn and Pleurostima атой! nea. 1, 2. В. spir ona lobes simulating a tube. on a “cylindric filament d d is independent of the coro Pet Fic with flexed a (P) and sepals (S), and c adnate filament (Af) free from the corona lobe. A — 1. Flow ‚ Топ gitudinal n [o ower and has ostima plantaginea. — 3. al and corona tie ud. Ae separated fr from the corona, НЫ the previously езй; Hy —hyp nthium; Sg — stigma; St— style. 541 & Semir The Corona in Velloziaceae Menezes Volume 77, Number 3 1990 542 Annals of the Missouri Botanical Garden ——— SS Eee Li UTI, Питт пт ШШ з T3 Volume 77, Number 3 Menezes 4 Semir 543 The Corona in Velloziaceae P i E Б (S e FIGURES 17-20. Details of the vascularization of Barbacenia spiralis flower bud. — trace (Dv) from the sepal-corona comp ex (Csc). The corona bundles (Cb) are not attached to the stamen bundles (arrows). — 18. Cross section of flow ud with corona lobes (Co) independent of filaments (Fl). — 19. Enlargement of the portion of Figure 18 between gm dashed lines. — 20. Enlargement of portion of Figure 19 between dashed lines; the xylem (X) is turned outward and the ica (Ph) is internal to the xylem. Hy — ias P—petal; —sepal; St— style. 17. Divergence of the corona <— o 5-16. Floral vascularization of Barbacenia Pridie See text id explanation of figures. An— anthe o— corona; Cpc— petal-corona complex; Сѕс — sepal-corona complex; D—carpel dorsal bundle; Dir descending basal ж. Fb— fundamental bundle; Fl— filament; p "о le; Ns—n ariferous sacs; P— ; Ps—petalar stamen trace; S—sepal; Sc —stamen-carpel complex; Sg— stigma; Sn—septal nectary; е stamen trace; St— style; V— carpel ventral bundle. ч] 544 Annals of the Missouri Botanical Garden Barbacenia spiralis has a well developed fila- ment and a basifixed anther (without auricle), as in Vellozia (Vellozioideae); the flower, however, has a corona, and the leaf bundle sheath is double (endodermic internally, and externally a sheath derived from a mesophyll) as in the Barbacenioi- deae. Thus В. spiralis, although belonging to Bar- bacenioideae, shows some intermediacy. Probably, free filaments should be considered ancestral to adnation of the filament to the corona seen in all other species of Barbacenioideae. Other intermediate conditions are known: in Pleurostima plantaginea the filament is not deeply fused to the corona lobe, which thus shows a similar but presumably independently derived condition. The filament independent of the corona lobe and auricle-free basifixed anther with distinct dehis- cence (Menezes, 1988) by themselves justify the description of a new genus based on this species, which will be the subject of a forthcoming paper (Menezes & Semir, in prep.). LITERATURE CITED ARBER, A. 1937. Studies in flower structure III. On the corona and androecium in certain Amaryllidaceae. Ann. Bot. (London) n.s. 1: 293-304. JOHANSEN, D.A. 1940. Plant Microtechnique. McGraw- ill, New York. MENEZES, N. L. DE. 1970. Estudos anatómicos e a taxonomia da familia Velloziaceae. Tese de Douto- rado Institu én Sao 1 . 1971 Velloziaceae. Cienc. Cult. : 421-422. ———— . 1973. Natureza dos apéndices petalóides em Barbacenioideae (Velloziaceac). Bol. Zool. Biol. Mar n.s. 30: 713-755. 975. Presenga de traqueides de transfusáo e bainha mestomática em ес oru ceae). Ne Botánica, Univ. Sào Paulo 9-60. ————. 1980a. Evolution in ordinal Bs. special mos to androecium characters. Pp. 117-139 in C. D. Brickell, D. К. Cutter & M. Gregory о Petaloid Monocotyledons Horticultural and Botanic Research. Linn. Soc. Symp. Ser. 8. Academic Prea, London. 1980b. Re-establishment of the genus P ~- rostima Raf. (Velloziaceae). Revista Brasil. Bot. Е 7. 84. Características anatômicas ea filogenia . Evolution of the anther i v the family Velloriapeae. Bol. Bot. Univ. Sào Paulo 10: 33-41. iio DE HaLac, R. I. & A. E. vinti naturaleza de los *apendices p ipae nisle е. Кашаа 6: 265- 269. . Botanical Microtechnique. Iowa State 1962. A synopsis of the American Vel- loziaceae. Contr. U.S. Natl. Herb. 35: 215-292 S. AYENSU. 1976. Arevision of American Velloziaceae. Smithsonian Contr. Bot 181-205. & Velloziaceae Brasiliae II. Bradea 3(14): 105-114. VANDELLI, D. 1788. Fl. lusit. & brasil. spec., Coimbra. ADDITIONAL TRANSFERS OF ASIATIC MACHILUS SENSU NEES, NON DESROUSSEAUX, TO PERSEA MILLER (LAURACEAE) A. J. С. Н. Kostermans' In 1962 I reviewed the genus Machilus sensu Nees in Asia, relegating the species known at that time to Persea because there is not the slightest difference between Machilus Nees and Persea Mil- ler. This can be easily demonstrated by comparing the drawings of the American Persea (Kopp, 1966) and those of Chinese Machilus (Lee, 1979). The name Machilus goes back to the pre-Lin- nean Rumphius, Herbarium Amboinense 3 (1743), a latinization of the vernacular name makilan (which is perhaps the general name for lauraceous trees, like medang in the greater part of Indonesia). Rumphius described initially two species of Machilus, which he called the male and the female. The “male” is depicted on plate 40 and represents Litsea glutinosa (Lour.) C. B. Robinson. Of the “female” a leaf is shown on the same plate 40; it is certainly not the “male.” It might not belong to Lauraceae. Rumphius added two more species on page 70, of which the third one, depicted on plate 41, rep- resents Dehaasia media Bl. The fourth one, called the smallest by Rumphius, which I formerly thought to be a real Machilus, is not lauraceous, as is evident from the presence of five perianth lobes and the shape of the fruit; it might be ех. Merrill's (1917) suspicion that this should be a species of Phoebe is not accurate. Thus Rumphius's Machilus is a mixture of a Litsea, a Dehaasia, and two non-Lauraceae. The Rumphian species were validated under the name of Machilus by Desrousseaux (1791) in Lamarck's Encyclopedie (3: 668). No type species was indi- cated. As the “male” (Litsea glutinosa) is the most elaborately described, we accept it as the lectotype. Litsea Lamarck was published a few pages ear- lier (p. 574) and hence takes priority. If this is not acceptable, Litsea Lamk. should be conserved against Machilus Desrousseaux. Machilus turned up again as a generic name in 1831 as adapted by Nees. Nees's species were based on herbarium material of plants from India, Sumatra, and the Malay Peninsula. He incorrectly added Rumphian names in synonymy; they do not invalidate Nees’s names. Hence, as I have pointed out (1962), Nees's Machilus is a later homonym of that of Desrous- seaux. Whether Nees was unacquainted with Des- rousseaux’s publication is unknown. Machilus sensu Nees was described with four species (no type indicated). The first, Machilus odoratissimus Nees, based on an Indian specimen, should be accepted as the type species of Nees’s genus. Nees’s addition of the synonym Machilus quarta Rumphius is a mistake. The second species, M. macranthus, likewise belongs to Machilus sen- su Nees. The third one, with the Rumphian syn- onym М. “tertia media,” is Dehaasia media Bl. The fourth species, M. incrassatus Nees, was based on Laurus incrassatus Jack and was relegated by Nees first to Persea and later moved to Haasia. Its correct name is Dehaasia incrassata (Jack) Kosterm. Machilus sensu Nees was accepted by Meissner (1864), but wrongly ascribed to Rumphius, where- as Desrousseaux is not mentioned. Baillon (1872), Bentham & Hooker (1880), Pax (1889), Allen (1936), and Chinese students of Lauraceae (Lee, 1979) likewise accepted Machilus sensu Nees, unaware of the existence of Machilus Desrous- seaux. I suggested in 1962 that Machilus sensu Nees was congeneric with Persea Mill. All Asiatic species belong to sect. Machilus Kopp. Persea is very similar to Phoebe Nees, non Mez, but the latter in the fruiting stage has enlarged, thickened, yellowish, erect perianth lobes, clasping tightly the black, ovoid, gradually pointed fruit. In Persea the lobes are not enlarged, not thickened and reflexed or completely deciduous in sect. Machilus; other American species have persistent, indurate not or little enlarged nonerect perianth lobes in the fruiting stage and might be referred to a different genus. ' Herbarium Bogoriense, Jalan Juanda 22, Bogor, Indonesia. ANN. Missouni Bor. GARD. 77: 545-548. 1990. 546 Annals of the Missouri Botanical Garden Persea austroguizhouensis (S. K. Lee & F. N. Wei) Kosterm., comb. nov. Machilus austro- guizhouensis S. К. Lee & F. N. Wei, Guihaia 4(2): 95. 1984. Persea chayuensis (S. K. Lee € H. W. Li) Kosterm., comb. nov. Machilus chayuensis S. К. Lee & Н. W. Li, Acta Phytotax. Sin. 17: 46. 1979. Persea chekiangensis (S. K. Lee) Kosterm., comb. nov. Machilus chekiangensis S. K. Lee, Acta Phytotax. Sin. 17: 53. 1979. Persea chienkweiensis (S. K. Lee) Kosterm., comb. nov. Machilus chienkweiensis S. К. Lee, Acta Phytotax. Sin. 17: 48. 1979 Persea chrysotricha (H. W. Li) Kosterm., comb. nov. Machilus chrysotricha H. W. Li, Acta Phytotax. Sin. 17: 50. 1979, Persea chuanchienensis (S. K. Lee) Kosterm., comb. nov. Machilus chuanchienensis S. K. Lee, Acta Phytotax. Sin. 17: 47. 1979. Persea chunii (Chang ex S. C. Lee) Kosterm., comb. nov. Machilus chunii Chang ex S. С. Lee, Forest Bot. China 175. 1973. Persea cicatricosa (S. K. Lee) Kosterm., comb. nov. Machilus cicatricosa S. K. Lee, Acta Phytotax. Sin. 8: 182. 1963. Persea dinganensis (S. К. Lee & К. М. Wei) Kosterm., comb. nov. Machilus dinganensis S. K. Lee € F. N. Wei, Guihaia 4: 94. 1984. Persea fasciculata (H. W. Li) Kosterm., comb. nov. Machilus fasciculata H. W. Li, Acta Phytotax. Sin. 17: 53. 1979. Persea foonchevii (S. K. Lee) Kosterm., comb. nov. Machilus tege S. K. Lee, Acta Phytotax. Sin. 8: 3. 1963. Persea fragrans (Kanehira ex S. C. Lee) Kos- term., comb. nov. Machilus fragrans Kane- hira ex S. C. Lee, Forest Bot. China 175. 1973. Persea fukienensis (H. T. Chang) Kosterm., comb. nov. Machilus fukienensis H. T. Chang, Acta Phytotax. Sin. 17: 52, table 5, fig. 4. 1979. Persea glaucifolia (S. K. Lee & F. N. Wei) Kosterm., comb. nov. Machilus glauc ifolia S. K. Lee & F. N. Wei, Guihaia 4: 98. 1984. Persea gongshanensis (H. W. Li) Kosterm., comb. nov. Machilus gongshanensis H. W. Li, Acta Phytotax. Sin. 17: 48. 1979 Persea grandibracteata (S. K. Lee & F. N. Wei) Kosterm., comb. nov. Machilus gran- dibracteata S. K. Lee & F. N. Wei, Guihaia 4: 96. 1984. Persea grandifolia (S. K. Lee & F. N. Wei) Kosterm., comb. nov. Machilus grandifolia S. K. Lee & F. N. Wei, Guihaia 4: 100, fig. 3. 1984. Persea kwangtungensis (Y. C. Yang) Kosterm., comb. nov. Machilus kwangtungensis Y. C. Yang, J. West China Border Res., Ser. B 15: 77. 1945. Machilus polyneura H. T. Chang, Acta Sci. Univ. Sun- yatsenia жес 21. 1960; Н. W. Lee, For. Fl. China 823. d a Chun . T. Chang, Acta Sci. v. Sunyatsenia eens 21. 1960; Н. W. Lee, e M China 823. Persea lenticellata (S. K. Lee & F. N. Wei) Kosterm., comb. nov. Machilus lenticellata S. K. Lee & F. N. Wei, Guihaia 4: 97. 1984. Persea lichuanensis (W. C. Chang) Kosterm., comb. nov. Machilus lichuanensis W. C. Chang, Acta Phytotax. Sin. 17: 51, table 5, fig. 2. 1979. Persea litseifolia (5. K. Lee) Kosterm., comb. nov. Machilus litseifolia S. K. Lee, Acta Phy- totax. Sin. 17: 46. 1979. Persea lohuiensis (S. K. Lee) Kosterm., comb. nov. Machilus lohuiensis S. K. Lee, Acta Phytotax. Sin. 8: 184. 1963. Persea longipedunculata (S. K. Lee & F. N. Wei) Kosterm., comb. nov. Machilus longi- pedunculata S. K. Lee & F. N. Wei, Guihaia 4: 93. 1984. Persea melanophylla (H. W. Li) Kosterm., comb. nov. Machilus melanophylla H. W. Li, Acta Phytotax. Sin. 17: 54, table 8, fig. 4. 197 Persea mikweiensis (S. K. Lee) Kosterm., comb. nov. Machilus mikweiensis S. K. Lee, Acta Phytotax. Sin. 17: 52, table 5, fig. 3. 1979. Persea minutiloba (S. K. Lee) Kosterm., comb. nov. Machilus minutiloba S. K. Lee, Acta Phytotax. Sin. 17: 50, table 7, fig. 5. 1979. Volume 77, Number 3 1990 Kostermans 547 Transfers of Asiatic Machilus to Persea Persea monticola (S. K. Lee) Kosterm., comb. nov. Machilus monticola S. K. Lee, Acta Phytotax. Sin. 8: 183. 1963. Persea nakao (S. K. Lee) Kosterm., comb. nov. Machilus nakao S. K. Lee, Acta Phytotax. Sin. 8: 188. 1963. Persea nanchuanensis (N. Chao) Kosterm., comb. nov. Machilus nanchuanensis N. Chao, 9 Acta Phytotax. Sin. 17: 47. 1979 Machilus parabreviflora Chang, Fl. Sichuanica 1: 117. 1981. Persea obscurinervis (S. K. Lee) Kosterm., comb. nov. Machilus obscurinervis S. К. Lee, Acta Phytotax. Sin. 17: 51, table 5, fig. 1. 1979. Persea ovatiloba (S. K. Lee) Kosterm., comb. nov. Machilus ovatiloba S. K. Lee, Acta Phy- totax. Sin. 17: 56. 1979. Persea pedicellata Kosterm., nom. nov. Based on Machilus longipes Н. Т. Chang, Acta Sci. Univ. Sunyatsenia 1960(1): 20. 1960. [The combination in Persea is occupied by Persea longipes (Schl.) Meissn. ] Persea pyramidalis (H. W. Li) Kosterm., comb. nov. Machilus pyramidalis H. W. Li, Acta Phytotax. Sin. 17: 53, table 8, fig. 2. 1979. Persea rufipes (H. W. Li) Kosterm., comb. nov Machilus rufipes H. W. Li, Acta Phytotax. Sin. 17: 55, table 8, fig. 5. 1979 Persea salicoides (S. K. Lee) Kosterm., comb. nov. Machilus salicoides S. К. Lee, Acta Phytotax. Sin. 17: 48. 1979. Persea shiwandashanica (H. T. Chang) Kos- term., comb. nov. Machilus shiwandashani- ca H. T. Chang, Acta Sci. Univ. Sunyatsenia 1960(1): 19. 1960. Persea sichourensis (H. W. Li) Kosterm., comb. nov. Machilus sichourensis H. W. Li, Acta Phytotax. Sin. 17: 51. 1979. Persea sichuanensis (N. Chao) Kosterm., comb. nov. Machilus sichuanensis N. Chao, Acta Phytotax. Sin. 17: 47, table 4, fig. 3. 1979. Persea suaveolens (S. K. Lee) Kosterm., comb. nov. Machilus suaveolens S. K. Lee, Acta Phytotax. Sin. 8: 187. 1963. Persea tenuipilis (H. W. Li) Kosterm., comb. nov. Machilus tenuipilis Н. W. Li, Acta Phy- totax. Sin. 17: 54. 1979 Persea velutinoides (5. К. Lee & Е. N. Wei) Kosterm., comb. nov. Machilus velutinoides S. K. Lee & F. N. Wei, Guihaia 4: 101, fig. 4. 1984. Persea verruculosa (H. W. Li) Kosterm., comb. nov. Machilus verruculosa H. W. Li, Acta Phytotax. Sin. 17: 55, table 2. 1979. Persea versicolora (S. K. Lee & F. N. Wei) osterm., comb. nov. Machilus versicolora S. K. Lee & F. N. Wei, Guihaia 4: 98, fig. 5. 1984. Persea wenshanensis (H. W. Li) Kosterm., comb. nov. Machilus wenshanensis H. W. Li, Acta Phytotax. Sin. 17: 49. 1979. OTHER SPECIES ОЕ MACHILUS SENSU NEES Machilus dumicola (W . W. Smith) H. W. Li, Acta Phytotax. Sin. 17: 49. 1979. — Persea bonii (Lec.) Kosterm Machilus kanehirai (Sphalm.) = Cinnamomum kanehirai C. E. Chang in Fl. Taiwan 2: 416. 1976. — Cinnamomum micranthum Hayata. Machilus macrophylla var. arisanensis Hayata, J. Coll. Sci. Tokyo 30: 243. 1911; Koster- mans, Bibl. Laur. 917. 1964; T. S. Liu, Illustr. Nat. and Introd. Pl. Taiwan. 1960: table 103; C. E. Chang, Bull. Taiwan Prov. Pingtung Agric. Inst. 11: 476. 1970; in Fl. Taiwan 2: 462. 1976; H. W. Li, Tree Fl. China 814. 1973. — Persea thunbergii (S. & Z.) Kos- term. Machilus mushaensis F. I. Lu, Quart. J. Chin. For. 2: 19, table 6. 1969; C. E. Chang, Bull. Pingtung Agric. Inst. 11: 44. 1970; in FI. Taiwan 2: 164. 1976. — Persea zuihoensis Hayata. Machilus nanshoensis Kanehira, Formos. Trees 449. 1917; Kostermans, Bibl. Laur. 919. 1964; T. S. Liu, Illustr. Nat. and Introd. Pl. Taiwan 1: table 106. 1960 (synonym of Per- sea thunbergii), Lu, Quart. J. Chin. For. 2: 21. 1969; C. E. us Bull. Taiwan Prov. Pingtung Agric. Inst. 11: 47. 1970; in Fl. Taiwan 2: 452. 1975. — Persea thunbergii (S. & Z.) Kosterm. Machilus parabreviflora H. T. Chang, Acta Sci. Univ. Sunyatsenia 1960(1): 17. 1960; in FI. Sichuanica 1: 117. 1981; Н. W. Li, Tree FI. China 845. 1973. — Persea nanshoensis (Chang) Kosterm. Machilus pomifera (Kosterm.) S. K. Lee, Acta Phytotax. Sin. 8: 186. 1963; W. Y. Chun, 548 Annals of the Missouri Botanical Garden Fl. Hainan. 1: 270. 1964; H. W. Li, Tree Fl. China 843, fig. 355. 1973. = Persea pomifera Kosterm. Machilus sericea (Nees) Blume, Mus. Bot. Lugd. Bat. 1(21): 330. 1851; Kostermans, Rein- wardtia 8: 110. 1974. = Persea wallichii Long, Notes Roy. Bot. Gard. Edinburgh 41: 518. Machilus gammieana King ex Hooker f. in Hara, Fl. E. Himalaya 3: 42. 1975; Persea gam- mieana (King ex Hooker f.) Kostermans in Hara et al. (editors), Enum. Fl. Pl. Nepal 3: 186. 1982; Long, Notes Roy. Bot. Gard. Edin- burgh 4: 521. 1984 (as a synonym of Persea clarkeana (King ex Hooker f.) Kosterm.); Grierson & Long, Fl. Bhutan 1(2): 264. 1984. LITERATURE CITED J. Arnold Arbor. 17: 376. [See Ann. Missouri Bot. ALLEN, C. K. 19 also, Studies in the Lauraceae. I. Gard. 25: 361-434. 1938.] BaiLLoN, H. 1872. Lauracées. In: Hist. Pl. 2: 429- =н. I ur D. Hooker. 1880. Genera Plantar- e 3, part Duo ^ А. J. 1791. In: Lamarck, Encyel. Ме : 668. КОРР, L. та Mem. New York Bot. Gard. KosTERMANS, A. J. G. H. 1952. A historical b of Lauraceae. I. J. Sci. Res. (Jakarta) 1: 91, 92, 116, i ` 1954. Machilus. In Bibliotheca Laurac. 900- 32 The Asiatic rcu of Persea Mill. 194. (Сайраса Reinwardtia 6: LEE, 5. К. Асїа Phytotax. Sin. 17: MEISSNER, С. К. 1864. Machilus. In: A. э Todo mus 15(1): 39-43. MERRILL, E. D. 917. An interpretation of Rumphius’ Herbarium Amboinense. 234 1831. NEES VON ESENBECK, С. С. In: Wallich, Plantae Asiaticae Rariores 2: 70. Pax, 1889. Machil us. In: Engler E ни Die Natürlichen Pflanzenfamilien. III(2): 1 Rumpuius, E. 1743. Herbarium к, 3: 68- 70. NEW ОК NOTEWORTHY ORCHIDS FOR THE VENEZUELAN FLORA. УШ. NEW SPECIES AND COMBINATIONS FROM THE VENEZUELAN GUAYANA! Germán Carnevali?’ and von Ramírez?’ ABSTRACT In preparing the treatment of the Orchidaceae for the Flora of the Venezuelan Guayana, several new taxa have been detected. These include specimens from recently completed expeditions and earlier herbarium material. Recon- sideration of generic boundaries within several subtribes has e ne comb. nov.; So of the various taxa propose ero p. nov.; Pleurothallis dui sp. nov.; P. pemonum, sp. n bralia oliva-estevae, sp. nov. In addition, comments are MESE aue MI. on the affinities ed. cessary various nomenclatural changes. The . поу.; - rosea f. м € nov.; © unifoliata, telis garayi, The Orchidaceae are the largest of all flowering plant families, with probably more than 25,000 species. Its range is worldwide, but it is especially diverse in the tropics of both hemispheres (Dressler, 1981). It is also the largest family of plants in the Venezuelan Guayana, accounting for about 750 species now recognized; new taxa are constantly being added as new explorations are carried out. Our estimates suggest that this figure will even- tually rise up to 800 or more once the whole area is well botanized. In this article six new taxa are described as the result of recently completed field trips or the study of earlier herbarium material. Furthermore, reconsideration of generic bound- aries within several subtribes have made various nomenclatural changes necessary. CLEISTES The genus Cleistes L. C. Rich. consists of about 30 species (Dressler, 1981), widely ranging through the American tropics and subtropics, with a con- centration of taxa in southern Brazil. This genus is closely related to Pogonia Juss., from which it differs by ie pollen grains coherent in tetrads. Some authors do not recognize both genera as distinct (Foldats, 1969), but most recent authors have preferred to retain the neotropical group Cleistes as a distinct entity. Nine species of Cleistes are known from Venezuela, eight of them occurring in the Venezuelan Guayana; one of these is newly described. Cleistes huberi Carnevali & I. Ramirez, sp. nov. Bolivar: Aparamán- -tepui, 2°07'W, summit of highly eroded sandstone mesa, 22 Mar. 1987, B. Holst 3480 (holotype, VEN; iso- type, MO). Figure 1. Species haec Cleistes strictae C. Schweinf. proxima sed statura vegetativa et florali minore, labello trilobato recedit. Small, erect, wp ai en terrestrial or subter- restrial herbs, 4— igh, AT or growing in small, е Res T ipsoid, 1-1.5 cm long and wide. Stems iced erect, straight ! We a for publication in his Flora of the Venezuelan Gua are indebted to the late J. Steyermark for же support and for the use of drawings originally intended We tha nk B. Manara for the illustrations of diagnostic details of the new taxa and for advice on the Latin © мна Н The curators of the following herbaria kindly allowed us to study their material and other references: AMES, MO, MY, TFAV, VEN. Gustavo A. Romero and F. Oliva Esteva provided logistic support for fieldwork. Angel Carnevali, E. Foldats, C. Luer, J. MacDougal, G. Morillo, and G. . Romero gave valuable suggestions on first drafts of this article. Finally, we are grateful to our reviewers, E. Christenson and L. Garay, E our editor, G. 2 Jardín Botánico de Cara as, Herbario Nacional de Venezuela, Aptdo. Rogers, for their — and patience. 2156, Caracas 1010 A, Venezuela. ! Current address: Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S.A ANN. Missouni Bor. GARD. 77: 549-558. 1990. 550 Annals of the Missouri Botanical Garden Y | FIGURE 1. A, B. Cleistes huberi Carnevali & I. Ra- mirez.— A. Flowering habit. — B. Spread floral parts. or slightly flexuous, glabrous, fistulose, internodes mm long, clothed by strongly fused leaf sheaths. Leaves 4-7 mm long, 1.5- reduced to small, fleshy, elliptic, acute, sheathlike mm wide, blades, concave above, rounded below, margins finely erose. Inflorescences terminal, 1-2(-3)-flow- ered, flowers produced in succession. Floral bracts 2-5 mm long, similar to leaves but smaller. Flowers small for the genus, pale yellow to whitish, with subparallel здес segments. Pedicellate ovary —10 mm long, subcylindric or narrowly ellipsoid, 6-ribbed. Sepals oblong-elliptic, acute, concave, 5-nerved, lateral sepals ob ж'а alcate; dorsal sepal 12-14 mm lon 5 mm wide, lateral sepals 11-13 mm iong, 2. 8-3 mm wide, dorsally carinate. Petals 10-12 mm long, 2.2-2. wide, linear-subfalcate, less concave than sepals, 3-nerved. Labellum 10 mm long, 3.5-4 mm wide, provided at base with 2 pedunculate, 0.2-0.3 mm long, subglobose calli, in outline narrowly oblong- ovate, 3-lobed about the middle; terminal lobe 5.5 mm long, 3.5-4 mm wide, linear-oblong to trian- gular-oblong, rounded or obtuse at apex, margins undulate, erect, making the blade concave; lateral lobes very small, 0.5 mm long and wide, toothlike, acute, porrect; disc with a pair of longitudinal, elevated, oblong, tuberculate callosities, the cal- losity longitudinally sulcate in middle, highest to- ward the base, apically extended into 3 tuberculate carinae occupying the terminal lobe. Column 4.5- 5 mm long, basally attenuate; clinandrium wide with 4 teeth about 1 mm long. Anther subquadrate, with 1.5-mm-long projections. Rostellum obtrian- gular. Capsule 15 mm long, obovoid, olive green. Paratypes. VENEZUELA. BOLÍVAR: Distrito Roscio, cumbre del Yuruani-tepui, al NNW del Cerro Kukenán, vegetación herbáceo-arbustiva = planicies de arenisca en el sector centro-este del te 2,300 m, 5?19'N, 60°51'W, 27 abril 1984, О. nane 9432 cud Ku- kenán-tepui, cumbre del sector mas septentrional, algo o del macizo principal, vegetación herbáceo-ar- e las lagunas, 2,500 m, 5°61'N, 60%48'W, 28 abril 1984, O. Huber 9461 (NY, VEN); summit of Cerro Roraima, parte noreste, inmediata al sur del hito que marca los limites de Guyana, Brazil y Venezuela, 2,750- 2,800 m, 5°12'N, 60%42'W, 26 agosto-2 septiembre 1976, J Steyermark et al. 112604 (VEN); cumbre del Tramén- tepui, en la porción mas nor-occidental del Macizo de Ilú- (Urú) tepui, 2,650 m, 5%27'N, 61?1'W, 23 enero 1985 (fr), O. Huber 10062 (MFY, NY, VEN). Etymology. After Otto Huber, who has collected this new species on several occasions. Cleistes huberi is closely related to C. stricta (C. Schweinf.) Garay & Dunsterv., also from the Roraima Formation Tepui Area, from which it dif- fers in its smaller size and by the lobation of the labellum, which is variable in its development. The degree of lobation correlates with the shape of the terminal lobe. Pronounced lobing is associated with an oblong terminal lobe, while a triangular terminal lobe is correlated with less conspicuous lobation. Plants of C. huberi are 4-15 cm high (usually under 10 cm high) and 1-2(-3)-flowered, while plants of C. stricta are 23-100 cm high and are much more floriferous, producing 2-7 flowers on a longer rachis. The flowers of C. huberi have sepals 11-14 mm long versus sepals 18-20 mm long in C. stricta. Cleistes paludosa Reichb. f. from Surinam and northern Brazil is similar but has larger leaves (1-2.5 cm long), the floral bracts are longer than the pedicellate ovary, and the peri- anth segments are longer and differently propor- tioned (sepals 18-19 mm long, 2- m wide; labellum 15-16 mm long, 6-8 mm wide); Cleistes aphylla (Barb. Rodr.) Hoehne from southern Brazil (Paraná) is apparently related but plants are larger (up to 25 cm high), flowers are white, and the perianth segments are wider. Cleistes stricta and C. huberi have distinct dis- Volume 77, Number 3 1990 Carnevali & Ramírez 551 New or Noteworthy Venezuelan Orchids tributional patterns: C. stricta tends to be more common in the cerros or tepuis of Territorio Fed- eral Amazonas, while C. huberi is restricted to the northeastern part of the Roraima Formation in Estado Bolivar. Their ranges slightly overlap, how- ever, in Chimanta-tepui, which is the northernmost known location of C. stricta. A recent field trip uncovered an unknown form of the common Cleistes rosea Lindley. Cleistes rosea Lindley forma pallida Carnevali amirez, forma nova. TYPE: Venezuela. Territorio Federal Amazonas: Depto. Átures, morichal 1-2 km E de Piedra Tortuga, unos 8 km S de Puerto Ayacucho, 80-90 m, 30 junio 1988, ca. 5?32'N, 67°32'W, G. Car- nevali, l. Ramirez & G. A. Romero 2661 (holotype, VEN; isotypes, AMES, INPA, MO, MY, PORT, TFAV) A Cleistes rosea Lindley forma rosea perigoniis an- Nic pallide viridis, apice roseis vel purpureo striatis differt. This new form has narrower sepals, petals, and labellum than the typical C. rosea; its flowers are held in a more nodding position, and it is partic- ularly noteworthy by its color. While in the typical form all perianth segments are dien d rose, pink, or purple, with darker zones in the labellum, in this new form the sepals are cuba cream, the petals are white, and the labellum is white with apical pink suffusion, with nerves that are apically dull purple. The callus is deep yellow as in the typical form. Populations of this form seem to be restricted to the northern part of Venezuelan Ter- ritorio Federal Amazonas, always at low elevations, and in Trachypogon savannas or at the ecotone between these savannas and **morichales" or for- ests. Paratypes. | VENEZUELA. TERRITORIO FEDERAL бона ym Atabapo, sabana ubicada en el pie gua, junio 1979, O. Huber 3872 (MYF, VEN): Depto. Atina. sabanas y bosques en la región de Rincones de Chacorro, unos 30 km N de Pto. Ayacucho, unos 5 km al NE de Galipero, 80 m, 5?48'N, 67?20'W, 9 mayo 1980, О. Huber 5240 (MFY, VEN) Our recognition of Cleistes Lindley as a distinct genus requires the following new combination. Cleistes unifoliata (С. Schweinf.) Carnevali & . Ramírez, comb. nov. BASIONYM: Pogonia unifoliata С. Seed: Fieldiana, Bot. 28: 171, fig. 27. 1951. TYPE: Venezuela. Bolivar: Ptari-tepui, scrubby forest on rocky open por- tion of plateau on SE-facing slopes, 1,600 m, 1 Nov. 1944, J. A. Steyermark 59630 (ho- lotype, F). This rare species, previously known only from the type specimen and from recent material from Suriname (Werkhoven, 1986), has lately been col- lected several times in the Venezuelan Guayana, always on the Estado Bolivar tepuis, at 950- 1,500 m. сои: specimens examined. VENEZUELA. BOLÍ- A. F 5%55'N, 62?15'W, 27 mayo 1986, Liesner et al. 21149 (MO, VEN); Cerro Guanacoco, cumbre, porción nor-oeste cerca del borde riscoso, sabana rodeada por bosque enano, 1,450 m, 4?46'N, 63°55'W, Steyermark et al. 109724 (VEN); Cerro Sarisarinama, cumbre, porción E, aflora- miento de arenisca con zanjones en formación de bosque achaparrado y árboles enanos, 1,380 m, 4?41'N, 64°13'W, Steyermark et al. 108961 (VEN). SUBTRIBE PLEUROTHALLIDINAE This group, composed of about 4,000 species widely ranging through the Neotropics, is the larg- est assemblage of taxa at the subtribal level in the rchidaceae. The systematics of the Pleurothalli- inae have been recently reviewed by Luer (1986a) and we follow here the generic concepts he pro- posed. This subtribe comprises about 170 species in 16 genera in the Venezuelan Guayana; the larg- est genera are Pleurothallis (ca. 55 species), Oc- tomeria (ca. 30 species), and Stelis (ca. 18 species). Here we propose as new three ecologically similar species. Octomeria romerorum Carnevali & I. Ramirez, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Depto. Átures, carretera a Gavi- làn, ca. 3 km antes del puente sobre el Rio Gavilán, julio 1987, С. 4. Romero, F. Guán- chez & E. Gutiérrez 1324 (holotype, VEN; isotypes, K, MO, TFAV). Figure 2A-D. Species Octomeria gemmulae Carnevali & I. Ramirez similis sed statura vegetativa manifeste minore, folio quam ramicaulo conspicue breviore et labello anchoriformi dif- fert. Diminutive, creeping, epiphytic herbs, forming clumps 3-8(-15) cm long and wide, adpressed to bark. Rhizome ca. 1 mm thick, bearing ramicauls 0.5-1 mm apart, concealed by scarious sheaths. Ramicauls 1 -2 mm long, 0.6-0.8 mm thick toward apex, monophyllous, abbreviated, shorter than leaves, obconic to subcylindric, 1-articulated, sub- erect to horizontal, concealed by scarious sheaths. 552 Annals of the Missouri Botanical Garden FIGURE 2. flattened. —D. ventral and dorsal view. Lateral view of la ER and column. —G. Pollin Leaves 5-6(-7) mm long, 3.2-3.8 mm wide, fleshy coriaceous, thick, adpressed to the substrate, green, concolorous, elliptic to widely elliptic, apex usually obtuse but ranging from subacute to rounded, base rounded to obtuse, abaxially flat or somewhat con- vex with a flat to sulcate midrib, abaxially convex; the margins rounded. Inflorescences 1 -flowered, successive, erect, originating from a mass of sheaths, peduncle 1.5-2.8 mm long, subterete, 1 -articulate, A-G. Octomeria romerorum Carnevali & I. Ramirez. — A, B. Flowering habit. — C. Perianth segments, A — E. Lateral view of column, showing rostellum. — Е. Anther, clothed with tubulose, scarious sheaths. Floral bracts rudimentary. Pedicellate ovary 4-5 mm long. Ped- icel 3-4 mm long, thin, subterete. Ovary 0.8-1.2 mm long, straight or in angle with pedicel, some- what hexasulcate. Flowers large for the plant, white- hyalinous, opening well with flat or somewhat re- curved-reflexed perianth segments. Perianth seg- ments membranaceous, subequal, 3-nerved. Sepals 3-4 mm long, 0.8-1.2 mm wide, free, narrowly Volume 77, Number 3 1990 Carnevali & Ramirez 553 New or Noteworthy Venezuelan Orchids elliptic to narrowly ovate-elliptic, acute, the lateral sepals slightly oblique. Petals 3-4 mm long and ca. 1 mm wide, narrowly ovate-elliptic, acuminate, apically somewhat concave. Labellum long and 1.1-1.3 mm wide, fleshier than the other perianth segments, concave, from a cuneate claw cordate- suborbicular to cordate-broadly ovate, acute, minutely apiculate; the claw 0.2-0.3 mm long; the blade 3- lobate, basal lobes 0.2 mm long, subfalcate-triangular, obtuse, rounded, to sub- acute, retrorse, thus making an anchiroid labellum; disc provided with a —— elliptic, pulvi- nate callus. Column -0.6 mm long, short and thick, foot ca. 3 mm pem Rostellum 0.1 2 mm long, linguiform. Pollinia 8. Etymology. This species is named after Gustavo and Peggy Romero, who first collected it. This description was compiled from live and liquid-preserved material from the type collection. This species seems to be near to Octomeria gem- mula Carnevali & I. Ramirez from the Rio Sipapo drainage, with which it shares the small creeping habit (but not strongly “Peperomia-like” in О. gemmula, see discussion below), thick leaves, long- peduncled and proportionately large white flowers, and more or less similar labella. In both species the flowers open at about 7:00 A.M. and close at about 1:00 P.M. for 3-6 days. Octomeria gem- mula, however, differs by its larger habit with pro- portionately longer ramicauls, purplish abaxial face of the leaves, and antrorse lateral lobes of the labellum. Octomeria romerorum seems to be re- stricted to dense forests where it grows on high branches, while О. gemmula grows low on shrubs in sandy, open scrublands. Octomeria romerorum seems unique in the ge- nus because of its habit, which recalls some of the species of Pleurothallis subg. Specklinia sect. Muscosae (Luer, 1986b) and a few species of Platystele Schltr. All these species have very short ramicauls along a short- to long-creeping rhizome with thick prostrate leaves forming small masses over the surface of the host tree. In habit they resemble some of the species of Peperomia in the Р. rotundifolia (L.) HBK complex. It is interesting to note that in northern Territorio Federal Ama- zonas and in adjoining Distrito Cedeno of Edo. Bolivar, there are several species of Pleurothalli- dinae that share the same **Peperomia-like" habit and that grow more or less sympatrically, namely: Pleurothallis pemonum Carnevali amirez, P. deborana Carnevali & I. Ramirez, P. nanifolia Fold., Platystele ovalifolia (Focke) Garay & Dun- sterv., and this new species. This is an area of heavy rainfall but with a marked dry season be- tween December and March. It is quite probable that this **Peperomia-like" habit minimizes water loss because the stomata-bearing abaxial leaf sur- face is closely adpressed against the substrate, cre- ating a closed, humid atmosphere. Octomeria romerorum does not fit in any of the proposed subsections of sect. Octomeria (Co- gniaux, 1896; Luer, 1986a). Its small vegetative stature would include it in subsect. Pusillae but its free lateral sepals would place it in subsect. Oc- tomeria. The current classification of Octomeria appears artificial and a new intrageneric classifi- cation is needed. Two small, creeping species of Pleurothallis of the affinity of P. nanifolia Fold. have been found in the northern Venezuelan Guayana which have proved to be new to science. Pleurothallis deborana Carnevali & I. Rami- rez, sp. nov. TYPE: Venezuela. Bolivar: Distrito Cedeño, bosque al borde de sabana al E del Rio Parguaza, 125 km al N de Pto. Ayacucho, ca. 100 m, originalmente colectada por Bruce Holst y florecida en el vivero de los colectores, junio 1987, G. Carnevali & I. Ramírez 2317 (holotype, VEN). Figure 3D, E. Species Pleurothallis nanifoliae Fold. affinis, sed pe- talis ellipticis, acuminatis et labello anguste-obovato, Mus rotundato, recurvato recedit (subg. Specklinia, sect. Mus cosae Small, creeping epiphytes, adpressed to sub- strate, 2-6 cm long. Rhizome 1 mm thick, creep- ing, concealed by scarious sheaths, adult ramicauls 1-3 mm distant. Ramicauls 1-2 mm long and about 1 mm thick, subterete, apically thickened. Leaves 5.5-8 mm long, 4.5-6.5 mm wide, fleshy coriaceous, flat, prostrate over the substrate, broadly elliptic to (rarely) broadly ovate, obtuse, apiculate, roadly cuneate. Inflorescences 1-2 origi- nating from the ramicaul apex, racemes 1-3-flow- ered, erect, shorter than subtending leaves, pro- vided with a tubulose, apiculate sheath. Rachis slightly flexuose, subterete. Floral bracts 1 mm base long, applicate, apiculate, green. Pedicel 2-2.3 mm long, subterete, somewhat thickened apically, green tinged with red-purple. Ovary triquetrous, basally attenuate, 0.8 mm long, green with purple markings. Flowers resupinate, subcampanulate, small but large for plant size. Perianth segments subfleshy. Sepals 3.2 mm long, 0.8-1 mm wide, somewhat concave, clear greenish yellow, maroon- tinged within, narrowly ovate-lanceolate; lateral se- pals slightly connate basally, somewhat falcate, dor- sally carinate. Petals 3 mm long, 0.7 mm wide, 554 Annals of the Missouri Botanical Garden 2mm —+ 5mm 3mm FIGURE 3. A-E. арн deborana С yas & I. Ramirez. — A, B. Flowering habit. — C. Perianth segments, is id s abellum, ventral view ellum, dorsal view. F-1. Pleurothallis pemonum Carnevali Ra > lowering it hell. Perianth segments, Шале, —I. Labellum, flattened. Н redrawn from ап ашы а жы le G. C. K. Роси е Volume 77, Number 3 1990 Carnevali 8 Ramírez 555 New or Noteworthy Venezuelan Orchids with the general coloration of the sepals but with a dark purple longitudinal central zone, narrowly ovate-lanceolate, acuminate, slightly falcate, mar- gins finely erose toward apex. Labellum 1.8 mm long, 0.4 mm wide, fleshier than the other perianth segments, dark red-purple with a yellow longitu- dinal streak, the ventral surface finely papillose, narrowly oblong-obovate, apex recurved and acute beneath, the ventral face provided with a longi- tudinal concavity, margins carinate, at the basal Y provided with 2 small lobes 0.3-0.4 mm long, these somewhat falcate, acute; labellar base pro- vided with two small auricles. Column 1.6 mm long, relatively slender, arcuate, alate in the apical half, clinandrium trilobed, О panded. Anther smooth, subglobose. Pollinia 2. mm wide when ex- Etymology. After Debora, our daughter. This description was prepared from living ma- terial from the only plant known. It was collected in the same tree where Pleurothallis holstii Car- nevali & I. Ramirez was originally found. The only flower is preserved as the holotype. Pleurothallis deborana shares the same general habit and similar flowers with P. nanifolia, P. pemonum Carnevali & I. Ramirez, P. pachyphyta Luer, and other species of Pleurothallis subg. Specklinia sect. Muscosae. Perhaps its closest all y is P. pemonum, which has identical habit and pet- als. However, P. deborana differs by its narrowly obovate labellum (vs. ovate-lanceolate) of smaller 2.5-3.2 mm long in P. pemonum), with a rounded, recurved apex (vs. not recurved). Both are easily differentiated from P. nanifolia by their apically attenuate petals (vs. obovate). size (1.8 mm long vs. Pleurothallis pM Carnevali & I. Ra- mirez, 5р. TYPE: Venezuela. Territorio Federal гене Depto. Аїигез, carretera a Gavilán, sobre Parkia pendula, ca. al E del Fundo Dona Juana, asociada con Octomeria romerorum, julio 1987, G. A. Romero 1334 (holotype, VEN; TFAV). Figure 3F-I. isotype, Species dpi nanifoliae Fold. affinis, sed pe- talis ellipticis, acutis et labello subsimplici differt (subg. Specklinia, sect. Muscosa e). Small, creeping epiphytes up to 2 cm high. Rhi- zome mm thick, subscandent, entirely concealed by scarious sheaths, bearing 1-2 rela- tively thick roots at each node. Ramicauls 1-3 mm long, 1-2 mm apart on the rhizome, subterete, noticeably thickened apically, concealed by tubu- lar, scarious sheaths, apically monophyllous. Leaves 5-7(-9) mm long, 3-5 mm wide, very fleshy, usually adpressed to the host's bark but occasion- ally suberect, usually broadly elliptic or broadly obovate-elliptic, rarely elliptic to almost orbicular, apex obtuse to rounded, with a subapical mucro, base obtuse to rounded. Inflorescence 4-7 mm long, erect, originating from the leaf base, race- mose, 1-2-flowered, basally enclosed by scarious sheaths. Floral bracts 0.8 mm long, ovate-elliptic, acute, acuminate, slightly scabrous. Pedicel sub- terete, 1.5-1.8 mm long. Ovary 0.6-0.8 mm long, obconic, subtriquetrous, in angle to pedicel. Flow- ers proportionately large, lateral sepals patent, dor- sal sepal, petals, and lip parallel to labellum. Peri- anth segments subcoriaceous. Sepals trinerved; dorsal sepal 3.8-4 mm long, 0.8-1 mm wide, narrowly oblong, acute; lateral sepals 3.2-3.5 mm long, 1-1.2 mm wide, linear oblong, acute, slightly oblique, free to their bases where they form a -3.7 mm long, 0.8- 0.9 mm wide, margins recae cellular. Labellum 2-2.2 mm long, 0.5- allel to the column, fleshier than the other perianth segments, narrowly ovate-lanceolate, apex acute to subobtuse, basally provided with a pair of small, shallow mentum. Petals mm wide, par- rounded, 0.2 mm long lobules or auricles, ventral face provided with a dense, minute pubescence absent along a longitudinal concavity extending to the apex of the blade. Column 1.8-2 mm long, slightly arcuate, subterete, apically winged and tri- lobed; clinandrium fimbriate. Anther subventral. Paratypes. | VENEZUELA. BOLÍVAR: selva virgen siempre verde a lo largo de la Quebrada Los = 4.5 km al S de Icabarú, 4?20'N, 61?48'W, 1 1976, Steyermark et al. 117784 VEN 450 m, Dunsterville & Dunsterville 1221 (AMES, VEN). Etymology. After the Pemón Indies from La Gran Sabana, Edo. Bolivar, who live where this species occurs. This small, creeping species is closely related to Pleurothallis nanifolia and P. deborana. (See dis- cussion under the latter species.) The flowers of P. pemonum have semitranslucid, deep wine-red se- pals, petals of the same hue but not translucid; the labellum is dark red with the central concavity yellow-cream, basally grandular-pubescent. The column is cream, red-tinged dorsally, ventrally dark red. The anther is creamy red, the ovary is yellow and the pedicel is creamy red. Pleurothallis pemonum has a wide range in the Venezuelan Guayana and with little doubt is present in Brazil and Colombia. The Dunstervilles collected it in several places in Edo. Bolivar: Rio Carrao, 450 m, and Cerro Guaiquinima, 700 m (G. C. K. 556 Annals of the Missouri Botanical Garden Dunsterville, pers. comm.). The authors have found E in the Akaruai River at 750 m (no voucher), in ame state. The holotype collection is somewhat out of range but otherwise is identic We agree with Luer (19862) el jie kind of stigma structure (l-lobed, ventral) found in Apa- tostelis (as defined by Garay, 1980) is only an extreme case in a rather variable character and should not be used as a generic criterion. Hence, the following combination is required: Stelis garayi (Dunsterv.) Carnevali & I. Ramirez, comb. nov. BASIONYM: Apatos stelis M ipe Dunsterv., Amer. Orchid Soc. Bull. 50: 1075. 1981. ТҮРЕ: Venezuela. Bolivar: Sato Para- 250 т, С. С. К. Dun- AMES; isotype, van, Rio Yuruan, ca. sterville 1418 (holotype, N). This interesting little species, related to a group of mainly Central American species (Stelis ciliaris Lindley, S. crescenticola Schltr., Schltr., and others), has recently been collected again, now in the Venezuelan Amazonas. The spec- S. minimiflora imen is: Venezuela. Territorio Federal Amazonas: Depto. Atures, Rio Autana, 200 m, enero 1988, С. A. Romero 1435 (TFAV, VEN). SOBRALIA The genus Sobralia Ruiz Lopez & Pavon, con- taining about 80 species, is mainly a Middle Amer- ican and Andean genus. However, a secondary center of diversity is to be found in the Guayanan Highlands; most of the species occurring there are endemics to that area. The flowers of plants be- longing to this genus are very difficult to study since they are thinly membranous and tend to agglutinate. Because of this problem, many species are ill-defined or poorly known and the genus is in need of a revision. There are 19-20 species of Sobralia in Venezuela, most of them restricted either to the Andean portion of the country or to the Guayana, where 13 species are known to occur. One of them is proposed as new here. Sobralia oliva-estevae Carnevali & 1. Ramirez, sp. nov. TYPE: Venezuela. Estado Bolivar: Par- que Nacional Canaima, La Escalera, bosque nublado enano alrededor del km 125 al S de El Dorado, ca. 1,300 m, 30 agosto-8 sep- tiembre 1987, I. Ramírez, G. Carnevali «€ F. Oliva-Esteva 150 (holotype, VEN). Fig- иге 4. Species Sobralia speciosae C. Schweinf. similis, sed caules ramificantes, subscandentes (vs. simples); folia abaxialia 2-carinata (vs. 3-5-carinata), inflorescentia sem- per uniflora (vs. (2-)3-5-flora), petala apice non undulata (vs. undulata) recedit. Medium-sized to rather large epiphytic to sub- terrestrial, suffruticose herbs. Roots thick. Stems 100-250 cm long, terete, erect, arcuately ascen- dent to subpendulous, sometimes red-tinged, some- what lignified basally, branching and rooting at the upper half, basal internodes 18-25 cm long, nodes somewhat thickened, apical internodes and branch- es 1-8 cm long, leafy. Leaves 5-10.5 cm long, 1-1.9 cm wide, subfleshy when fresh, rigidly co- riaceous when dry, abaxially with 2 raised nerves, decurved in natural position, narrowly ovate-elliptic to ovate-elliptic, long acuminate, articulate with their sheaths; sheaths 0.5-1.2 cm long, tightly clasping the stem. Inflorescences uniflorous, axil- lary from the upper portion of main stem or branch- es, up to б flowers per branch; peduncle 20-30 mm long, somewhat ancipital and apically thick- ened, its basal half enveloped by the subtending leaf sheath, apically produced into a fugacious aborted flower, noticeable only on fresh material. Pedicellate ovary 1.4-1.5 cm long, subterete, twisted. Flowers 9-12 cm natural spread, showy, resupinate, spreading perianth segments, lasting only one day. Sepals and petals purple, paler toward apex. Dorsal with submembranaceous, widely sepal 5 cm long, 1.8 cm wide, narrowly obovate- elliptic, subacute. Lateral sepals 5.2 cm long, 1.7 cm wide, elliptic, acute. Petals 5.3 cm long, 2.3 cm wide, obovate, abruptly acute, somewhat in- curved in natural position. Labellum 5.7 cm long, 4.4 cm wide, rich purple with a white throat and a yellow central carina, this reaching only the open- ing of the throat, hairs on disc pale purple, in outline broadly obovate-elliptic or somewhat rhomboid, broadly rounded apically, emarginate and mucro- nulate, labellar margin undulate-crisped in the api- cal half, disc with 4 hairlike antrorse to suberect lamellae-covered carinae at each side of central nerve, this bearing lamellae at the apical third. Column 2.8 cm long, 4.5-5 mm thick, slender, clinan- somewhat clavate, apically triquetrous, drium trilobate, central lobe larger than laterals. Paratypes. GUYANA: Kaieteur Plateau, 11 May 1944, Maguire & Fanshawe 23380 (AMES, NY). VENEZUELA. BOLIVAR: Uei-tepui, between southeastern slope and summit, between Luepa and Cerro Venamo, vicinity of km 125 S of El Dorado, 1,100-1,300 т, Mar. 1962, Steyermark & Aristeguieta 21 (VEN) Etymology. After Francisco Oliva-Esteva, who participated in the collection of the type specimen. This description was based on pickled material Volume 77, Number 3 Carnevali & Ramírez 557 1990 New or Noteworthy Venezuelan Orchids бст FIGURE 4. A, B. Sobralia oliva-estevae Carnevali & I. Ramirez. — А. Flowering habit. — B. Labellum, flattened. from the holotype collection and on dried material from the paratypes. The flowers of this showy species are almost identical to those of Sobralia speciosa C. Schweinf. from southern Territorio Federal Amazonas (Cerros Neblina, Avispa, and Aracamuni), but in 5. oliva- estevae the petals are marginally flat, and the la- bellum is more obovate and proportionally wider. The pedicellate ovary in S. speciosa is propor- tionally longer than in S. oliva-estevae; in the first species it is 0.75—1.3 times longer than the column (p. ovary 2-4.7 cm, column 3-3.5 cm), while in the new species the pedicellate ovary is only about half as long as the column (p. ovary 1.4-1.5 cm, column 2.8 cm). The most striking differences be- tween both species, however, lie in the vegetative morphology and in their ecological preferences. Sobralia speciosa grows terrestrially in open sa- vannalike tepui associations, usually in boggy con- ditions while 5. oliva-estevae is an epiphyte in cloud forests. In 5. oliva-estevae the stems are freely branching and proliferous in the upper half, rooting at the branching internodes; in this way the plants become subscandent, eventually prostrate or pen- dulous. The stems in S. speciosa are simple and erect. The leaves of S. speciosa have 3-5 nerves raised abaxially, while in 5. oliva-estevae there are only 2. In S. speciosa the inflorescence is a suc- 558 Annals of the Missouri Botanical Garden cessively (2-)3-5-flowered raceme (rarely with 2 flowers open simultaneously), while in the new species the inflorescence is always 1-flowered; sev- eral axillary inflorescences are produced succes- sively toward the apex of each stem or branch. LITERATURE CITED COGNIAUX, A. 1896. Flora Brasiliensis 3: 319-646. mer R. L. 1981. Ба Orchids: Natural History Classification. Har niv. Press, Cambridge. Kilian. E. 1969. laa Pp. 129-145 in Flora de Venezuela 15(1). Edicion especial del Instituto Botánico, Caracas. sa L. A. 1980. Systematics " 2 genus Stelis. . Mus. Leafl. Harvard Univ 167-259. s. E A. 1986a. Icones Pleu a m I. Sys- matics of the Ple ma llidinae (Orchida о Monogr Syst. Bot. Missouri Bot. Gar 1986b. Icones Peso thallidinarum IIl. Sys- tematics x et dara vip aie Monogr. Syst. Bot. Missouri Bot. Gar WERKHOVEN, M. C. M. ur Orchids of Suriname: 214. VACO, Paramaribo. POLLINATION ECOLOGY OF SEVEN SPECIES OF BAUHINIA L. (LEGUMINOSAE: CAESALPINIOIDEAE) Omaira Hokche? and Nelson Ramirez? ABSTRACT Pollination and pic biology of seven species of Bauhinia were analyzed between 1982 and 1983 in different Venezuelan plant с Tylotaea, which Ве. lianas. The sp in two sections: Pauletia, which includes trees, and B. pauletia, B. ungulata) have comparatively ace, ыны flowers, while the species of sect. Tylotaea (В. glabra, of flow В. guianensis, В. rutilans) exhibit different color and could be intermediate between the two sections. The rs and variations in form and color of the upper petal. composition, and secretion tend to be associated with the life form of the two sections of Bau In an ecological context, caesalpinioid legume flowers are less specialized than their mimosoid and papilionoid counterparts. Caesalpinioid flowers are open, usually with exposed pollen and nectar avail- able to specialized and nonspecialized pollen vec- tors. Only in some of the advanced genera are resource conservation and pollinator selection ev- ident (Arroyo, 1981). The Caesalpinioideae exhibit a great variety of pollinating agents and mecha- nisms with an entomophilous trend (Arroyo, 1981). For example, many Cassia species are bee-polli- nated (Delgado et al., 1977). In this sense, orni- thophily and chiropterophily are scarce (Arroyo, 1981 Studies of chiropterophily have paid compara- tively more attention to the legumes of the New World than of the Old World (Frankie & Baker, 1974; Heithaus et al., 1974, 1975; Howell, 1975; Bernhardt, 1982; Ramirez et al., 1984; Prance, 1985). Some neotropical Bauhinia species are bat- pollinated (Heithaus et al., 1974; Ramirez et al., 1984). However, Vogel (1954) reported that Bau- hinia galpinii and B. mucronata are sphingophi- lous, and Arroyo (1981) suggested that many other species of Bauhinia are probably sphingophilous. The flowers of neotropical Bauhinia species ex- hibit great diversity in form, size, and color, which has been poorly studied from an adaptive view- point. The species of Bauhinia are grouped in three sections according to Stuard da Fonseca Vaz (1 de Section Pauletia comprises trees and shrubs; contrast, species of sects. Tylotaea and Schnella comprise climbing plants. For all Bauhinia species studied, the flowering periods occur during the dry season. The following study provides information about the floral biology and pollinator activity of seven species of Bauhinia belonging to sects. Pauletia and Tylotaea found in different plant communities of Venezuela. The chemical composition, secretion, and volume of the nectars produced were analyzed for comparing both sections of Bauhinia. DATES AND METHODS Bauhinia is widely distributed in several eco- systems in Venezuela. The localities for this study ! We thank S. Tillet, W. S. Armbruster, S. Renner, P. Berry, A. Castillo, and P. Ber , H. Farinas, and S. G ? Departamento nhardt for comments on the nd to F. Fernández ER Botánica, Escuela de Biologia, EE. de Ciencias, ИУ Central de Venezuela, Aptdo. 47114, Caracas, Venezuela. Reprint requests to N. Ramire ANN. Missouri Вот. GARD. 77: 559-572. 1990. Annals of the 560 Botanical Garden issouri M epuel “UA Opeisy “See Je”) “3JUOJA] о[әң әр seumo-) ‘(ejanze -U9A әр [епцәг)у pepis N 189.0 J “ет8о[о1{ әр PJANISH ej әр ӯс-81 001°'1-0©6 euejuouraJd ÁX] — AA,£G,99 *М,0%01 001: Unia10quy) aalesay үеої#оо{ Аерү—зос) 'qa4-'des вив 0190]8 `g DIDIOJÁ | ооепсу opejsy “oz "oqeper) Jo JS Wy cT Ajoreur -rxo1dde 'sopeinjeN| serouor) 15910 4 ap eue[ozauaA peperoog ‘sou 6с-сс 008'1-000'T Кіа [291902] M,SZoL9 “N.9S08 SL “erq SOT әр водо[ої!{ uorwisq стерү—оә( 'uef—aoN em pjo]ngun 'g 183.0 4 engeiy opeisy ÞZ-81 OOT‘I-0S¢ aueJuoua1g AIG = A\,O1oL9‘N,9S06 005 “saÁ9y SO] әр иецѕедәс̧ ueg eW- — ^4oN-'deg 29m опәјтра `g ѕе8іе А ojueurej1edo(] 19910 J “[e.19p3 4 оц] ‘severe ӯс-81 00T'I-0SS euvjuour1g Ад] A,t£.99 ‘N,LEOI 8 SOT '19AIy seoP1e)) ayy Buoy слерү—оә( "ue(-'AoN әәд mD142110mu q AISI 159 -10] ogu] jo Wy 09 pue 19910 J Jealjog opeisy “exedy jo 65-66 008'[-000'I Ка [eotdoip — 4,720.99 ‘N.6£o2 001 ЯМ WY 0с ‘Bepa reumeq [3 epuel TJA Opeisy ‘seoesey ‘JUON ojeg әр в®шцүогу ‘(enza -UdA әр PNU peprs1eAmu(] 159.0 J ‘eiBojoig әр ejanosy еү әр ӯс-81 0011-0985 әиешошәг4 AIG М№Җ,Е5,99 ‘NOE 001: — umje1oqry) әлләѕәу [еәтдојој — "AoN.-Apnf K[f-'qo4 әәд seal di MH ШЙ (Də) (шш) ¡adA1 15310 4 9]eutpJoo") (ш) Аще2от ponad poured WLIO} sardadg o1nje [[gjure1 uon Suum J Suuewo|] әрт uono»ag -1eduie] jenuue BATT uray ‘satoads piunjnpg jo says Ápnis pue ‘AZojouayd aanonposdas un; әрт *] 318v x Volume 77, Number 3 1990 Hokche 8 Ramirez 561 Pollination Ecology of Bauhinia Species Continued. TABLE 1. Mean annual Eleva- Temper- ature rainfall (mm) tion (m) Life form Fruiting Flowering Section Forest type' (°С) Coordinate Locality period period Species Imataca Forest Reserve in the 266 8°03'N, 6139 W Humid Tropical 1,800 24 lana — Feb.-Apr. Маг.-Мау B. guianensis Forest Río Grande, Distrito Piar y Roscio (Estado Bolívar), and 2,000-4,000 Henry Pittier National Park, 1,100 10%21'N, 67?41'W Very Humid Pre- Oct.- Mar. Sep.-Oct. liana B. rutilans montane Forest logical Station, Estado Ara- gua ' According to Ewel et al. (1976). were chosen in accordance with the flowering and fruiting periods indicated on specimens in the Her- bario Nacional de Venezuela (VEN) and through field observations (Table 1) in various regions. Field observations were made of the life form and height of plants, and pollination and floral biology were analyzed in 1982-1983 during the flowering and fruiting periods of each species. FLORAL CHARACTERISTICS The floral parts (corolla, pistil, stigma, and petal lengths) were measured using samples preserved in 70% ethanol for 20 flowers from five to ten individual plants of each Bauhinia species. FLORAL BIOLOGY Flowers were observed in situ to record anthesis: inflorescences with buds about to open were marked before anthesis, and progress of anthesis was ob- served every 30 minutes. The pattern of nectar production was measured periodically with micro- capillaries inserted in the hypanthium cavity of bagged flowers. Solute concentration of the nectar was measured with a manual Bausch € Lomb refractometer (range 0-30%). The presence of sugar, proteins, amino acids, lipids, and other com- California, Berkeley, California, U.S.A.). Pollinator activity was observed and recorded during five days for each Bauhinia species. The visiting agents observed were captured with hand nets and mist nets and were examined for pollen load. RESULTS FLORAL MORPHOLOGY The inflorescences of Bauhinia are axillary and / or terminal. The sect. Pauletia species have com- paratively large, white flowers; the stamens are dimorphic: five are large and five short (Table 2). However, this trend was not clear for all species of this section. In Bauhinia pauletia short stamens are represented by five staminodes. The flowers of short and long pistils were found on the same tree of В. aculeata. The short pistil flowers are not located in the inflorescence. In 100 flowers of five individuals the large /short pistil ratio was 15 : 1. Flo- ral dimorphism is associated with pistil length, with a significant difference between the two morphs (t;s = 12.26; P < 0.0005). The short pistils are as- sociated with reduction of the gynophore (А = 1.12, SD = 0.24 in large-pistil flowers and А = 0.54, SD = 0.09 in short-pistil flowers), of style Annals of the 562 Missouri Botanical Garden (2) ystua913 Mo[[oÁ (00:0) so'o (40:000£€0 (#00) 0 — — (8000250 (100060 — — (0) яша (gc0)091 (0z0)061 € кир] cg (Z) мо[әќ 1ea]9 (00:0) S0'0 (v0'0)9€0 (£00) #0 — — (80°0) Z8°0 (S0'0) РОТ ==. = (0) әнчм (ZOO) ETS (e00) 97S 5 sisuauping `g (с) seu e[dund цим (00:0) so'o (z0'00670 (z00) *€'O — — (с0`0) 80 (000070 — — (Damya (cr'0)68 I (ct'O £81 S 014918 `g (seur) n2njojXT (900) 810 (IZ'O)ZUI (80°O) ZTT (220)091 (€S5'0)08'% (80) ZU? PUO) SUI (£) paz 0c (20°0) €2°0 (so'00£6'1 (ceo) ser (9+0) сс (S0) 1673 (6co TUF (210) €2'T (с *°1)әнчм (230) СТІ (S9'0) 99+ #6 (G)vmmn2un 'q (SUO) 4€ 0 (670) g£€€ (STO) 401 (950) 22°€ e(I20) 182 (IO DOZZ (03'0) 051 (сәм — — (vUDS8 SI S опәртра `g (£10)820 (cc'O) z6€ (810) FES (580) 08+ (vOc)OUO (0271) 862 (6S'0) FO'E (c'D 9m (670) 82°6 (€6°0) ESOT S D1042u1jnui ^g (000) ОГО (810)80'1 (О1`0)0/`0 (600) *S'O (6t0)867 (1E0) 86€ (800) ZF'0 £I (00:0)0z'0 (0z00)0*'z (600)90' 1 (vcO)cll (220) 86°% (szo) ort (ITO) ҒО (c'D*em^ (6g0)0UZ (0g0)6£€ 25 DIDINID `Y (ѕәәц) 017әјпо (as) X (as) Y (as) Y (as) Y (а5) Х (а5) Y (а5) Y Jojo) (ds) Y (ds) Y N satoadg eurdug ejas Á1eA() e1oqdouá^) 11O0US Zuo] umiyued4y PPM yı3uə] RES yi3ua] [nstg yi3ua| иәше1с 19MO[] (+861) Te 19 2әлшеу woaz (с) ә[415 120цѕ (f) ‘ѕәрошшедѕ (c) ‘seed ¡edioutid 10 saddn (g) :spejed peiye] pue 10uoju[ (1) (шә ш) saloads murynbg цәләз Jo sonsuejo?ieqo елор 1цәлә}тр jo sjueurainseo] "с 318V[ Hokche & Ramirez Pollination Ecology of Bauhinia Species Volume 77, Number 3 1990 (Х = 2.40, SD = 0.20; X = 1.08, SD = 0.18, respectively). The number of ovules per ovary is similar in both morphs but the ovules of short-pistil flowers are abortive. In Bauhinia multinervia there were no floral variations (Table The flowers of both sections are zygomorphic, the petals are different in form and size. In general, the inferior and lateral petals are similar in form, while their areas are slightly different (Table 3). The flowers of sect. Tylotaea are smaller than those of sect. Pauletia; the average floral length varies from 1.83 to 2.26 cm. The gynophore and hypanthium are also shorter than in sect. Pauletia. In Bauhinia pauletia, В. multinervia, and В. ungulata, one or two flowers open per night per inflorescence (Table 4). Anthesis, petal expansion, occurs at dusk between 1700 and 1900 hours, and the process is quick and synchronic (Fig. 1). Bauhinia aculeata has nocturnal anthesis but is comparatively asynchronous (2100-0300) and showed two peaks during anthesis (Fig. 1). In the species of sect. Pauletia anthers dehisce before anthesis. In the species of sect. Tylotaea (lianas), anthesis is diurnal, occurring approximately between 073 and 1130 hours. Petal expansion is slower, and the number of opened flowers per inflorescence per day is higher than in tree species of sect. Pauletia ig. 1). The pink flowers of Bauhinia rutilans showed two peaks during anthesis (Fig. 1). Al- though we did not record anthesis in Bauhinia guianensis, it occurs in the morning between 0900 and 1100 hours and is probably similar to that in B. rutilans (pers. obs.). NECTAR SECRETION In the arborescent Bauhinia species nectar is produced and accumulated in the hypanthium of flowers. Nectar production starts immediately after anthesis; however, in B. multinervia there was a little nectar before anthesis. The average volumes were high for B. multinervia (102.42 ml) and P. pauletia (47.32 ml); nectar concentration was rel- atively low and similar in the species studied (Table 5). The nectar of B. aculeata was produced during day and night; the average nocturnal production is ema: less than the diurnal production (t, = 8.5 < 0.0005). This difference was asso- ciated with a diurnal floral activity higher than nocturnal. The solute concentration of nectar in- creased from the first hours after anthesis until midnight in B. pauletia and B. multinervia, while in B. aculeata the higher concentration of nectar occurred during the night period. The volume pro- length, A — width. Morphological characteristics of petals in six Bauhinia species. L TABLE 3. Upper petals Lateral petals Inferior petals Area L/A Area L (cm) Area Section Form ratio (cm?) (cm) (cm) Form ratio cm) — (cm) cm?) Form ratio (cm) (cm?) Species Pauletia (trees) 3:1 lanceolate 7.19 5.80 1.95 331 8.07 6.10 2.12 341 6.15 2.15 8.23 B. aculeata narrowly narrowly elliptic lanceolate elliptic lanceolate 7.92 8.90 1.40 as] 10.21 9.70 1.45 3:1 1.60 11.43 10.55 B. multinervia 6:1 very narrowly 6:1 narrowly 0.37 3.10 0.18 narrowly- l 6:1 narrowly 0.66 9.60 0.08 6: 9.20 0.08 0.61 B. pauletia oblanceolate oblanceolate oblanceolate Tylotaea (lianas) 0.58 1.65 0.50 3:1 oblanceolate narrowly 1.62 0.75 2:1 0.69 l narrowly 80 1.67 0.90 2: 0. B. glabra obovate suborbiculate 6:1 very narrowly 0.58 1.80 0.40 2:1 1.75 1.25 1.36 2.00 1.30 1.5:1 very elliptic 1.74 B. guianensis elliptic 3:1 oblanceolate 0.66 1.90 0.60 narrowly 3:1 0.84 1.95 0.65 87 2.00 0.70 3:1 narrowly 0. B. rutilans elliptic elliptic Annals of the 564 Missouri Botanical Garden (p) seed Punos nZp1dossoj^) (y) 1eude y 1070951p smuojsojXud oepueurojso[uq (D *191doiÁq;) E 81 [поо 000 OOI 1 0€£81-0021 6с таманты ч ($) CT) sima vydsowng (y) CT) 2D9sn1q0] vydiowng эвр:8шҷас (y) (&oxiq) озәирГ ‘гел 2D12u211 vikuno Jewel) asiu ршәт Joules) D217D]s DSS11YA y (p) epog әрризло]ә soo1up oepueotq (s) 9epini2oN (p) ‘ds sno44210d4 (v) (&1es20]q) ns0o122ds nwanjny (p) 7] paafigow sid aepidy ‘qeg ю]ю14фш1/ ndoo0]KY (y) ‘ds (ndo»ojKxoax) vodos0 Xx aepuoydoy1uy (£) t1a1douaurÁy (p) ‘ds sna1ky209n97] snyinbsow $1йшът]о$Х гут) PAPAL чир (с) звәшлтодродү чч 9T pu? [Pun ION 980 STI -I 00£0-00Ic 0€ ppmv 0]£xo0a3y) vdos0] ky aepuoydoy1uy (£) tiaidousu Ap “Y vc (0 €9'I Ise 2-1 0060-0€S0 GE D1QU]2 `g DIDIOJA | (£) t1a1idousuky (v) ѕеред »u12170s ngv1dossoj«) (y) Jouse pp 20]02s1p snui0jsopAuqd IPP BULOISO[AY A (1) взәзйолАчл) "EI Poe IcO SOT GI 0061-—0Є/1 LI vnəmvd `g әерпләЧзәҢ (€) exardopida7 `1 DUDUILNS n220uÁg aepidso A (p) 1€] DID]¡amos ‘iea Diafyjau sidy (p) "т »4ofigjou sidy aepidy (£g) t1a1idousukp usngp SIULOYIIDYJ his dd uic! ^ (с) ѕәшлоуроду sjuoge SUnIsIA Кџлә8иој Ayandasai as ү əZuey smoy smoy вәтдәәЧс 19M0|4 [клор jo eur] sisoyuy поцел иоцэә$ Кер /sa9u39saJopyut _ /S19M0g pauado 19540 jo Joquinyy 'penuguo) "р ATAV], Missouri Botanical Garden Annals of the 566 "ds <тләиХроиә1$ aepidsa A (s) o?urureuin'g unana si1juaara nf рио. (p) ‘ds snquog aepidy (y) ds vbdos0, Ax aepuoydoy1uy (£) вләјіоиәшАң “| 104098 59151025 Sepo F, (с) sauopipody (zja0] gq) s1ugisu1 DIISON ¡aang sadijjpd viqA]odo1975 (f) “Т Ршюиып$ n220uXg (f) $1]ртиәртәәо DiQA]04 aepidsa А oepnoqeH (f) IUIN 29 e1noJq 27Dur]jdui02 414208207) aepuoydoy1uy (f) әѕәш DuDr¡m тио8з] aepidy £) t1ardousurÁy — 6 aepipmqoo1] (с) seuuoyipody “TY pz pumq = um uq 6 pumq — — = OOTI-O€F0 0011-0060 smoy 9I sunpna `Я ZI sisuauDIng ^g smoy вәтдәйс sjua3e ZUNISIA ÁyaoBuo] Aynandaoa1 as X әЗиеу о CIR Хер /ѕәэиәәѕәлорш /ѕләмор pauado jo 19quinN sısəy}uy UONBA - 1954) uonoag "penunuo?) "pP AVL Volume 77, Number 3 1990 Hokche & Ramirez Pollination Ecology of Bauhinia Species duced can exceed the volume of the hypanthium cavity, and without pollinating visits, nectar starts dripping down or out. Bauhinia aculeata produced nectar for 19 hours, whereas in В. pauletia and B. multinervia, production lasted approximately 13 hours. The secretion rate, estimated as the volume produced per time unit, was highest in B. multinervia (X = 7.88 ml/hr.), followed by B. pauletia (X = 3.61 ml/hr.) and B. aculeata (X = 0.27 ml/hr.). The first two species are char- acterized by nocturnal secretion. Significantly in B. aculeata no difference between diurnal (0.17 ml /hr.) and nocturnal (0.22 ml/hr.) secretion rates was found. The species of sect. Tylotaea produced less vol- ume of nectar with a higher sugar concentration than those of sect. Pauletia (Table 5). In Bauhinia glabra the volume was less than the minimal ca- pacity of microcapillaries and only a sticky sap at the base of stamens and pistil was detected. In B. rutilans nectar secretion took place during seven hours and the rate was 0.27 ml/hr., similar to that in B. aculeata (sect. Pauletia), with a total of 1.88 ml per flower. NECTAR COMPOSITION The nectar of six Bauhinia species contained proteins, amino acids, phenols, and alkaloids, but no lipids were detected in the nectar of any of the species studied (Table 6). Alkaloids appeared in low quantities only in the nectars of B. aculeata and B. rutilans. Only traces of protein were detected in the nec- tar of B. multinervia (sect. Pauletia), while in B. glabra and B. rutilans (sect. Tylotaea) it was detected in low quantities. The proportion of sugar in the nectar of sect. Pauletia showed that sucrose is dominant in B. aculeata, while B. multinervia and B. ungulata were hexose-dominant with sim- ilar proportions of glucose and fructose. The pro- portion of glucose was similar to sucrose in B. glabra but there was a higher proportion of both sugars than of fructose (Table 7). There is a temporal change of the proportion of sucrose, glucose, and fructose in the nectars during the secretion period, with a decrease in the pro- portion of sucrose over time in the tree species. At the beginning of nectar secretion, the proportion of sucrose was 2.74 times higher than 12 hours later in B. pauletia, B. multinervia (both sect. Pauletia), and B. rutilans (sect. Tylotaea): when sucrose decreases, fructose and glucose increase in quantity and the sucrose/glucose + fructose ratio decreases (Table 7 100 7 ] 8. glabra " N=118 100+ 7 B. rutilans ] N* 18 so- 1004 - B. pauletia о | N= 40 Ф 504 = | o ] ü 7] e 1007 Ф 4 В. multinervia а } М= 49 O 3504 $e o] 1004 4 B. aculeata ] ме 112 507 PD. _—— _— A 07:00 1008 зоо 16:00 19:00 e200 окоо Time (hr) FIGUR Anthesis expressed as the percentage of open fowers е half-hour intervals. C] % cumulative, Ш % non-cumulative The proportions of 20 amino acids analyzed differed in the nectar of different Bauhinia species, and varied from 3 to > 10, using a relative scale (Table 6). The scale from 1 to 10 is based on standard concentrations of histidine. A value of 10 is equivalent to 3.9 mg histidine/ml, and each successive unit below 10 represents a halving of concentration (9 — 1.95 mg/ml; 8 — 0.975 mg/ ml; and so forth) (Baker & Baker, 1975). The analyses showed that there was no difference in the presence of essential amino acids. Proline was dominant in all species except B. multinervia. Only one species studied, B. rutilans, had lysine in the nectar (Table 8) POLLINATOR ACTIVITY In bat-pollinated species of Bauhinia the flowers are exposed on the top of the foliage, and petals are separated, leaving the anthers exposed. The pollen adheres to the head and ventral part of the animal. The feeding activity and hence pollination last only seconds with the maximum occurring at dusk and in the first night hours. During the visit, Phyllostomus discolor and Glossophaga soricina seize the flowers and inflorescences so that the branches are bent down by the weight of the animal. Nocturnal visits were made by Sphingidae and but- 568 Annals of the Missouri Botanical Garden TABLE 5. Average volume and solute concentration of nectar for five Bauhinia species. TS Total volume of nectar (ml/flower) Solute concentration (%) Section - = Species Range X SD Range X SD Pauletia (trees) B. aculeata 3.96-5.81 5.03 0.95 12.0-28.0 19.35 4.85 B. multinervia 60.37-138.01 102.42 36.41 16.4-21.0 19.34 1.62 B. pauletia 14.18-87.44 41.32 37.12 13.0-18.0 16.23 2.07 Tylotaea (lianas) . glabra — «0.24 m = > 30.00 = В. rutilans 1.06-2.36 1.88 0.57 26.0-30.0 28.25 1.81 terflies, and phyllostomatid bats were observed on B. multinervia and B. pauletia flowers. Phyllosto- mus discolor and Glossophaga soricina bats were considered effective pollinators because they car- ried pollen on their heads and bodies (Table 4). During the day, the flowers of B. multinervia and B. pauletia were visited by butterflies, wasps, and hummingbirds, though there was little nectar and pollen. In addition, B. multinervia was visited be- ween 0530 and 0800 hours by Phaethornis aguti (Trochilidae) to take nectar of open flowers from the previous night. A different behavior was shown for Bauhinia aculeata (sect. Pauletia): during the night, flowers were visited by Sphingidae (Eumorpha labruscae and FK. vitis) and infrequently by Noctuidae. But- terflies, wasps, bees, and hummingbirds were abun- dant during the day. The bees, Xylocopa (Neoxy- locopa) sp., Apis mellifera, Eulaema speciosa, and several species of Pieridae (e.g., Anteos clorin- dae, Ganyra menciae) carried pollen of В. acu- leata. These insects inserted their heads inside the flower and imbibed nectar. The hummingbird Chry- solampis mosquitus took nectar from the flower in the morning between 0600 and 0730 hours and at dusk between 1600 and 1830 hours. The species of sect. Tylotaea have compara- tively small flowers, which were visited by a great variety of bees, wasps, butterflies, and humming- birds during the time of stigma receptivity (Table 4). Bauhinia glabra was visited by Apis mellifera and Pseudaugochloropsis sp. These insects ar- rived at the flowers posing on the inferior and lateral petals, introducing their body into the flower, while Bombus sp. inserts only its head into the flower. The most frequent visits were by А. mellifera and Bombus sp. during the morning. Bauhinia guianensis was visited by Xylocopa sp. and Synoeca surinama in the morning and afternoon. These bees carried pollen on the legs and head. A species of butterflies (astra insignis, Hesperiidae) took nectar about noon, but it was not a pollinator. In addition, one unidentified hum- mingbird was observed for a long time visiting the flowers. Bauhinia rutilans was visited by bees, wasps, and hummingbirds; Xylocopa sp. and Bom- LE 6. Proportion of organic compounds in nectar of six Bauhinia species (arrows indicate temporary trends TAB after anthesis). Section Species Amino acids' Phenols Alkaloids Proteins Pauletia (trees) B. aculeata tr slightly + ND B. multinervia 5 > =10* = ++ Мр їг В. рашейа з—-7 гэ + Мр Мр B. ungulata 6 tr ND ND Tylotaea (lianas) B. glabra 6 > >10 + ND tr В. rutilans 2T кок RE slightly + + tr = traces; * = suggests piae lodi the night; ND = not detected; ' = scale from 1 to 10 of relative quantities; + = moderate; ++ = abur Volume 77, Number 3 1990 Hokche 8 Ramirez 569 Pollination Ecology of Bauhinia Species Proportion of nectar sugars and their temporal variation (variations in time are in the direction of the ) AB arrows after anthesis). Proportion of sugars Sucrose/ Section glucose + Species Melezitose — Maltose Sucrose Glucose Fructose fructose Pauletia (trees) B. aculeata — — Р ; | .2798 B. multinervia 0.005 > 0.111 — 0.167 ~ 0.211 0.329 > 0.446 0.368 — 0.460 0.203 ~ 0.413" B. pauletia — — 0.188 ~ 0.516 0.279 – 0.463 0.206 > 0.349 0.246 ~ 0.153" B. ungulata 0.016 0.006 0.185 4 0.233" Tylotaea (lianas) B. glabra 0.007 — 0.330 0.38 . 0.499: B. rutilans — — 0.365 ~ 0.558 0.234 — 0.369 0.208 – 0.224 0.573 ~ 1.261: * sucrose dominant, " hexose dominant, bus sp. were abundant collecting pollen during the middle of the day. The flowers were visited fre- quently by Schistis geoffroyi (Trochilidae) and the pollen was collected on their bills. In addition, the flowers were perforated externally at the base by an unidentified nectar-robbing species of hum- mingbird. DISCUSSION The morphology, color, and scent of flowers are associated with size and behavior of pollinators. Chiropterophilous flowers are often white, exposed above the foliage, nectar continuously, show noc- turnal anthesis, and have a disagreeable smell (e.g., Heithaus et al., 1974; Sazima & Sazima, 1975, 1978; Voss et al., 1980; Howell & Schropfer Roth, 1981; Ramirez et al, 1984). Entomophilous species, including those of Bauhinia, have flowers of smaller size, of varied color, fragance, diurnal anthesis, and low nectar production. In addition, bee flowers often have dense inflorescences (e.g., Bolten & Feinsinger, 1978; Frankie et al., 1983). Such floral characteristics as flower size and time of anthesis of the studied Bauhinia species can be related to their different pollination systems. The white-flowered Bauhinia pauletia and B. multi- nervia are chiropterophilous, and B. glabra and B. guianensis are entomophilous, while the pink flowers with red bracts of B. rutilans were visited frequently by hummingbirds, which carry pollen. In most chiropterophilous species, anthesis seems to occur at dusk (1800-2000 hours) Meere et al., 1974; Sazima & Sazima, 1975; Gould, 1978; Lack, 1978; Ramirez et al., 1984). juris of Bauhinia pauletia and B. multinervia occurred at similar evening hours. The flowers last one night, similar to Markea neurantha (Solanaceae) (Voss “ intermediate between a and b. et al., 1980), Lafoensia pacari (Lythraceae) (Sa- zima & Sazima, 1975), and Bauhinia ungulata (Ramirez et al., (Passifloraceae) anthesis occurs between 0100 and 0200 hours, with a duration of less than 12 hours (Sazima & Sazima, 1978). The nocturnal flowers of Bauhinia can be considered as synchronic in anthesis because more than 50% of the flowers open within 30 minutes. The total process occurs in two and one-half hours. Anthesis of Bauhinia glabra and B. rutilans is diurnal, unimodal, and 1984). In Passiflora mucronata asynchronous, the peak of flower opening involving less than 40% of the flowers. In Bauhinia aculeata, anthesis is nocturnal and asynchronous, with two peaks of less than 40% each; this asynchronic anthesis could promote cross- fertilization; the flowers are visited by a variety of pollinators. Bauhinia aculeata showed a combi- nation of floral features: the floral morphology, nectar chemistry, timing of anthesis, and the pat- tern of nectar production cannot be placed with the other species studied. Bauhinia aculeata could be intermediate between nocturnal and diurnal pol- lination because a great number of specialized and unspecialized diurnal and nocturnal floral visitors and pollinators are associated with this species. In bat-pollinated plants, higher production of nectar has been reported than in hummingbird- and butterfly-pollinated plants, and nectar produc- tion is continuous (Cruden, 1976; Baker, 1978). Nectar production in Bauhinia pauletia and B. multinervia is higher than in Ochroma, Parkia, Chiranthodendron, and Lafoensia pacari, which produce 5 to 20 ml/flower or inflorescence (Heit- haus et al., 1975; Sazima & Sazima, 1975; Voss et al., 1980). Flowers visited by bees frequently produce low nectar quantities. Frankie et al. (1983) found dif- 570 Annals of the Missouri Botanical Garden TABLE 8. Amino acid composition in nectar of six species of Bauhinia. Amino acids Section Species Ala* Argo Asp Aspt Суз Glu Glut Gly* Histo [sol Lewo Lyso Meto Pauletia (trees) В. aculeata + + + ? + + + + + + + ND + B. multinervia + + + ? + + + + E + + Мр + В. pauletia + + ++ + + + + + ? + + Мр + B. ungulata + + + ND + + + + ND ND ND ND + Tylotaea (lianas) B. glabra + + + + + + + + Мр ? ? ND + B. rutilans + + їг ND ND + Мр + Мр їг tr + ? 2 — essential amino acids for insect nutritio — amino acids in nectar of hummingbird- salinasi plants tr = traces ND = not detected ++= + = goo + = moderate ferent flower sizes associated with the daily nectar more frequent at the first hours after anthesis. In production. These authors defined moderate nectar contrast, nectar concentrations of melitophilous- production as 1.0 to 8.0 ul/day and high nectar ornithophilous species (В. glabra and В. rutilans, production as on average higher than 8.0 ul/day. respectively) increased at the midday hours. The In addition, bee plants with elevated concentrations increase of nectar concentrations and the higher of solutes have small flowers and low nectar pro- pollination activities could be related to the tem- duction (Hainsworth & Wolf, 1972; Baker, 1975). perature elevation during midday hours and con- However, Bauhinia glabra and B. rutilans differ comitant evaporation from the nectar. Bees and from this expectation since they produce higher hummingbirds prefer nectar up to 20% or 40% of volumes of nectar than those reported by Frankie sugar concentration (Percival, 1974; Baker, 1975). et al. (1983). In addition, the visits and pollen load The flowers of sect. Pauletia produce nectar on Schistis geoffroyi (Trochilidae) suggest the im- during approximately 12 hours. Heithaus et al. portance of birds in the pollination systems of B. (1974) reported a rate of nectar secretion of 0.5 rutilans at the canopy level, so this species cannot ml/hr. for the first hours of production (from 1800 be considered as a strictly melitophilous species. to 2300) in B. pauletia; however, the total rate The solute concentration of bat-pollinated flow- of nectar secretion was 3.16 ml/hr. Bauhinia ers is frequently low (Howell, 1975, 1978; Baker, pauletia and B. multinervia showed a higher rate 1978; Steiner, 1983), and an increase of solutes than B. ungulata (Ramirez et al., 1984). The from early to later hours after anthesis has been difference could be associated with the greater flow- reported, e.g., in Lafoensia pacari from 6.8% to er and hypanthium cavity sizes of the first two 11.0% (Sazima & балта, 1975). By contrast, species. Ramirez et al. (1984) showed in Bauhinia un- Nectar has a variety of nutritional compounds gulata a higher solute concentration immediately (Percival, 1965) and elements with a selective before anthesis, which then decreased from 15.4% function (Baker & Baker, 1975). The alkaloids in to 12.0%. Bauhinia aculeata (sect. Pauletia) and B. ruti- Percival (1965) found an increase in solute con- lans (sect. Tylotaea) probably reflect a selective centration with flower age in species of sect. Ty- force at pollination level. The high diversity of lotaea and B. aculeata (sect. Pauletia). The in- — visiting agent species in both plant species could crease in nectar concentration can increase the be selected by deterrent compounds. The absence exploitation efficiency in flowers with low quantities of nectar proteins in the species of sect. Pauletia of nectar (Hainsworth & Wolf, 1976). The nectar is related to bat pollination because some pollinating concentration of Bauhinia multinervia and B. bat species eat insects (Heithaus et al., 1975) and pauletia flowers decreased with flower age. In the pollen as a protein source (Alvarez & Quintero, chiropterophilous species the visits are probably 1969; Howell, 1974). In contrast, the nectars of Volume 77, Number 3 1990 Hokche & Ramirez 571 Pollination Ecology of Bauhinia Species TABLE 8. Continued. Amino acids Pheo Pro* Ser* Tyr Treoo Tripo Valo + ++ + + + + + + + + ? + + + + + + + + + + + ++ + ND + ND + + ++ + + + ? + Мр їг tr ND tr + the species of sect. Tylotaea have some proteins; the insects that visit these species presumably ob- tain their nitrogenous requirement mainly from nectar and pollen, while hummingbirds obtain their nitrogenous requirement from nectar. Flowers pollinated by butterflies and humming- birds are reported as rich in sucrose, while nectar of bat flowers tends to be rich in hexose (Baker, 1978), and the nectar of bee flowers has no definite pattern in sugar proportions. Bauhinia multiner- via, B. pauletia, and B. ungulata are hexose- dominant chiropterophilous species and have noc- turnal nectar secretion. In these species, sucrose decreases with time, and glucose and fructose in- crease simultaneously. This pattern suggests a breakdown of sucrose, and then the sucrose/glu- cose + fructose ratios decrease. The breakdown of sucrose can be considered an advantage for pollination because bats cannot assimilate sucrose (Harborne, 1977). This pattern has been found in B. rutilans but was associated with hummingbird and bee pollination. Bauhinia glabra is rich in sucrose and glucose and has an entomophilous pol- lination system e flower — floral biology, pollinator species, nectar composition, and timing and amount of secretion are associated with life form and subge- neric designation of the Bauhinia species studied. The species of sect. Pauletia are trees or shrubs, frequently pollinated by bats. In contrast, the species of sect. Tylotaea are lianas, pollinated by insects and birds. The pollinator specificity among Bau- hinia species with similar pollinators is achieved basically through their geographic distributions. Sympatric distribution and overlapping flowering periods were found only for Bauhinia species of different sections. In this sense, the most important attribute is floral morphology. The floral charac- teristics and pollination biology provide additional characters for Bauhinia systematics. The agree- ment among reproductive and taxonomic proper- ties could be related to evolutionary patterns at the sectional level. LITERATURE CITED ALVAREZ, T. & L. GONZÁLEZ QUINTERO. 1969. Análisis polínico del contenido gástrico de murciélagos Glos- O cs = Mexico. Anales Esc. Nac. Ci. Biol. 18: ARROYO, М. T K 1981. Breeding systems and polli- nation biology in Fa т=н Pp. 723-769 in R. M. Polhill & P. H. Raven (editors), Advances in egume Systematics. Royal Botanic Gardens, Kew BAKER, н. С. 1975. Sugar concentrations in nectars from hummingbird flowers. Biotropica 7: 37-41 1978. Chemical aspects of the Pag biology of woody plants in the tropics Tomlinson & M. H. Zimmerman (editora. "Tropical Trees as Living Systems 3: 57-82. Cambridge Univ. Press, England. I. BA 973. Amino acids in nectar and their evolutionary significance. Nature 241: 54 1975. Studies of nectar consti- ollinator-plant coevolution. Pp. 100-140 in L. E. Gilbert & P. H. Raven (editors), Coevolution of ш and Plants. Univ. Texas Press, Austin, Tex tution and p велин, Р. 1982. аа аа of Australian Acacia. Pp. 85-101 in E. G. Williams, R. B. Knox, J. Н. "Gilbert & P. ек (editors), Pollination 82. School of Botany, Univ. of Melbourne, Australia. BoLTEN, А. B. & D. FEINSINGER. 1978. Why do hum- mingbird flowers secrete dilute neci? Biotropica 10: 7-309. Crupen, R. W. 1976. Intraspecific variation in pollen- ovule ratios and nectar secretion; preliminary evi- dence of 2 adaptation. Ann. Missouri Bot. Gard. 63: 89. DELGADO, S., a NSO & MARIO SOUSA SANCHEZ. 1977. Biología floral del género Cassia en la región de las Tuxtlas, Veracruz. Bol. Soc. Bot. México 37: 5-45. EweL, J. J., A. MADRID & J. A. Tost, JR. de vida p Venezuela. cre, Caracas e de Investigaciones Agropecuarias, Cara- BUE С. W. & H. С. BAKER. 1974. The importance of pollinator behavior in the reproductive biology of tropical trees. Anales Inst. Biol. Univ. Nac. Autón. éxico, Ser. oo 45: 1- а. Р. А. OPLER & К. S. Bawa. 1983. al and нее of the large Handbook of инн: Pollination Biology. trand Reinhold, or GouLp, E. 1978. Parasite behavior of oo nectar feeding bats. Biotropica Hainsworty, F. R. & L. L. Worr. 1972. Energetics of nectar extraction in a small, high rage tropical ае сое flammula. J. Comp. Physiol. 80: & i s ctar an i RES selection by ба ыы. Oecology 25: 572 Annals aia ios Garden иш: ч В. 1977. try. Academic dir Haro E R., P. A. OPL H. С. Baker. 1974. at activity and pollination of Bauhinia pauletia: plant pollinator coevolution. Ecology 52: 412-419. ———, Т. Н. FLEMING & P. A. OPLER. 1975. For- aging patterns and resource utilization in seven species " a seasonal tropical forest. Ecology 56: 84 Howe t, D. ^ 1975. dodi loving bats, bat-loving plants. Nat. Hist. 85: 52-5 ——. 1978. Flock жы in nectar-feeding bats: advantages to the bats and to the host plants. Amer. Naturalist 114: 23-49, . SCHROPFER RorH. 1981. Sexual repro- The ecology of the flowers of savanna tree Maranthes polyandra and their poda with m referenc e to bats. J. Ecol. 66: 287-295. PERCIVAL, M. S. 1965. Animal food in flowers, nectar and nectaries. /n: Floral Biology 5: 81-99. Perga- mon Press, London 74. Floral ecology B PUT scrub in south- east Jamaica. Biotropica 6: -12 PRANCE, С. T. 1985. The тен оѓ Amazonian jeta to Ecological Bio- plants. Pp. 166-191 in С. Т. Prance & Т. E. Love joy (editors), Key Environments: Amazonia. Perga- mon Press, London RAMIREZ, N., SOBREVILA, М. X. DE ENRECH & Т. RUÍZ-ZAPATA. 1984. Floral biology and breeding system of Bauhinia ungulata A icut ani | bat-pollinated tree in Venezuelan “llanos.” Amer Bot. 71: 273-280 SaziMa, M. & I. Sazima. 1975. араа ет Lafoensia pacari St. Hil. d na Serr Cipó, Minas Gerais. Ci. Cult. 27: 416. а Bat ll of the pas- sion flowers, Раз siflora mucronata, in southeastern Brazil. үү 10: 100-109. 1983. Pollination of Mabea СР (Euphorbiaceae) in Panamá. Syst. Bot. 8: 10 STUARD DA FONSECA Vaz, A. М. 1979. os sobre a oca do género Bauhinia L., sect. Dina Vogel (Leguminosae- Caesalpinioideae) do Brasil. Rodriguesia XXXI: 127-234. VocEL, S. 1954. Blütenbiologische Typen als Elemente der Sippengliederung. Bot. Stud. 1 338. Voss, R., M. TURNER, К. INOUYE, M. FisHER & R. Conr. 1980. Floral TEE of Markea neurantha Hemsley (Solanaceae), a bat- rrr epiphyte. Amer. Midl. Naturalist 103: 262-2 ADICIONES A LAS PAPILIONADAS DE LA FLORA DE NICARAGUA Y UNA NUEVA COMBINACIÓN PARA OAXACA, MEXICO! Mario Sousa S.? RESUMEN Se describen dos especies nuevas para la ciencia. Se hace dos nuevas combinaciónes y se tipifican dos especies en la familia Leguminosae para la Flor exico. Estas corresponden a las Papilionoideae. de Wi due caudatum) y otra a la de las a de dern rd d se p una nueva combinación para la as nu Millettieae к ош bicolor). Las nuevas de Oaxaca, una pertenece a la tribu de las Sophoreae combinaciones son as espect una en las Sophoreae (Styphnolobium conzattii) y otra en Aeschynomeneae (Diphysa americana). Se tipifican a dos Millettieae: Lonchocarpus macrocarpus Benth. y L. parviflorus Benth. ABSTRACT o new species are described, two new combinations are made, and two species are typified within the family for the Flora de Nicaragua. Also a the Papilionoideae, of which Serius cauda belongs to the tribe Millettieae. Of the tw Diphysa americana to the Aeschynomeneae. Two macrocarpus Benth. and L. parviflorus Benth. ew combination is made tum belongs to the tribe rcnt and Lonchocarpus bicolor o combinations, Styph species belonging to the Millettieae are typified: Lonchocarpus for the Flora de Oaxaca, Mexico. These belong to nolobium conzattii belongs to the Sophoreae, and A partir de la publicación “Adiciones a las le- guminosas de la Flora de Nicaragua" (Sousa, 1986 [1987 ]) se han hecho nuevas colectas que añaden dos especies más al grupo de géneros de leguminosas que se nos invitó a participar en el proyecto Flora de Nicaragua, tambien ha sido necesario hacer nuevas combinaciones y tipificar taxa descritos con elementos heterogéneos. А pesar de tratarse de adiciones a la Flora de Nicaragua, se hace una nueva combinación de una especie endémica de Oaxaca, Mexico; se tomó esta libertad, dado que de otra forma no se comprende adecua- damente el concepto y la circunscripción del género Styphnolobium para México y Mesoamérica. Es tambien de notarse que a Diphysa se le ubica en la tribu de las Aeschynomeneae en vez de las Ro- binieae; ésto es debido a los recientes estudios que viene realizando Matt Lavin (1987) y que parecen justificar este cambio. SOPHOREAE Styphnolobium caudatum M. Sousa & Rudd, sp. nov. TIPO: Nicaragua. Esteli: Salto de Es- tanzuela, 6 km al S de Esteli, 13%02'N 6? G. Davidse, P. P. Moreno & A. Grijalva (ho- lotipo, MEXU; isotipos, BM, HNMN, MO) Figura 1. Arbore бе» cortex squamosus placoideus irregularis, cinereus. Folia 9-13-foliolata; foliola stipellis minutis oblongis, iet lanceolata, 4-9 c к ga, (1.3-)2- 2.7 cm lata, apice longiacuminato vel с to, margine шык. Fuiruetescentae parentes, а. 12-17 cm үш I , torulosum, homogeneu 9-1 m longum, 2. 32 .7 cm diametro, dense ferrugi- neo- BE. 1, 2(-4)-seminale. Arbol de más de 30 m de alto; corteza escamosa placoide irregular, con tonos metálicos, con abun- ! Se agradece por aceptar colaborar con el autor a la Dra. V. E. Rudd; ui baa todo lo necesario para llevar a cabo este trabajo, al Dr. Sardi las diagnosis al latin; al Sr. Daniel Cobian C., quien logró una muy gh a un procesador de textos; al Dr. incorpor ar el manuscrito tevens, editor y ejecutor del proyecto Flor icaragua; al iang por ámina; a Mario jon dem or . Chater por RUD nar copia de la lámina de Plukenet; y a los curadores de BM, BR, F, G, K, MEXU, MICH, MO, y US por facilitar el material кыы. рага su estudio México 2 Herbario Nacional, Instituto de Biología, U.N.A.M., Apartado Postal 70-367, Coyoacán, 04510, México D.F. ANN. MISSOURI BoT. GARD. 77: 573-577. 1990. Annals of the Missouri Botanical Garden FIGURA 1. Styphnolobium caudatum. —a. Ram DANIEL COBIAN С ma con infrutescencia. — b. з de estipelas y pulvinulos. —c. 1298 Legumbre monosperma.—d. Legumbre tetrasperma. Tomado de Sousa Volume 77, Number 3 1990 Sousa S. 575 Papilionadas de la Flora de Nicaragua dantes lenticelas prominentes; ramas jóvenes mi- nutamente velutinas pronto esparcidamente, gris claro; lenticelas orbiculares, cremas. Hojas 9-13- folioladas; estipulas caducas, no vistas; peciolo 2.5- 5.5 cm de largo; raquis 7-15 cm de largo; foliolos estipelados, estipelas hasta 8 mm de largo, oblon- gas, pronto caducas; láminas cartáceas, lanceola- das, 4-9 cm de largo, (1.3-)2-2.7 cm de ancho, base e Ageramento anaes en оса- siones cuneada, apice a cau- dado, acne: planos, glabros en al haz y espar- cidamente sericeos en el envés, nervaduras secundarias 10-13, inconspicuas. Infrutescencias axilares, péndulas, racemosas, simples o ramifi- cadas, 12-17 cm de largo, pedicelos 5-9 mm de largo. Legumbre torulosa, homogénea, estipitada, 5-14 cm de largo, 2.3-2.7 cm de diámetro, den- samente ferrugineo velutina, estipite hasta de 4 cm de largo; 1, 2(-4) semillas por fruto (semillas ma- duras no fueron vistas). Distribución. Solo de la localidad tipo en Es- telí, de un sólo árbol, en una cañada húmeda, en la vega de un rio, a 900 т de altitud. Con frutos inmaduros durante mayo. Muestras adicionales examinadas. NICARAGUA. ESTELÍ: Salto de Estanzuela, 6 km al S de Esteli, 13%02'N, 86°22'0, 11 mayo 1985, Р. P. Moreno 25655 (MEXU, ). Styphnolobium conzattii (Standley) M. Sousa & Rudd, comb. nov. Sophora conzattii Stand- ley, Contr. U.S. Natl. Herb. 23: 436. 1922. Calia conzattii (Standley) Yakovl., Proc. Len- ingr. Chem.- Inst. 21: 45. 1967. TIPO: México. Oaxaca: C. Conzatti 3171 (holotipo, US; isotipo, herbario Conzatti).* Especie hasta ahora conocida del Cerro Espino en la Sierra Madre del Sur, muy característica por tener las flores de gran tamaño. Hasta hace poco fue colectada en fruto, lo que permitió su ubicación taxonómica en Styphnolobium. A esta especie Rudd (1971) le asignó en Sophora una sección propia, Oresbios, distinguiéndola por sus caracteres ve- getativos y NS desconociendose en ese mo- mento su frut Por otro ua Yakovlev (1967, 1968) revivió al género Сайа de Terán & Berl., incluyendo en él a cinco especies americanas de Sophora, com- prendiendo a C. conzattii, de la cual comenta: *parece sólo existir un ejemplar, el cual nosotros no contamos, es posible que esta especie sea una subespecie de S. secundiflora, pero hay que es- tudiar mejor este asunto.’ Calia es separada de Sophora y géneros afines por contar con un cáliz bilabiado y fruto seco, caracteres ausentes en C. conzattii y mucho menos puede considerarse coespecifica de S. secundiflora. Muestras adicionales examinadas. MÉXICO. OAXACA: Cerro Espino, Finca Montecristo, 23 abril 1976 (fr), M. Sousa 5598 (MEXU); 5 abril 1984 (fr), Torres C. 4901 (MEXU); 31 mayo 1984 (fl), Torres 5231 (MEXU). MILLETTIEAE Lonchocarpus bicolor M. Sousa, sp. nov. TIPO: Nicaragua. Zelaya: near Lago Siempre Viva, km SW of Bonanza, 14 May 1978, D. Neill 4038 (holotipo, MEXU; isotipo, MO). Arbores foliis deciduis; cortex interior ubi incisus suc- cum resinaceum emittens; rami juniores dense ferrugineo- velutini. Folia 7-foliolata stipulis deciduis orbiculatis vel ligulatis 2.5-3 mm longis; petiolus elongatus; foliola epunctata, elliptico- oblonga, 11-15 cm bue 5-6 cm lata, apice acuminato, valde bicoloria. Inflorescentiae pe- dunculatae, florescentia praecoci; pedunculi florales ad 2 tata, Батя erecto еї concavo; antherae lo ovarium tenuata, apice rot 1 velutinum, sutura vexillari anguste alata, ala ad 1.5 mm ta. 5 Arboles de 20 т de alto; caducifolios; la corteza interior con fluido resinoso al corte; ramas jóvenes acostilladas, densamente ferrugineo velutinas, len- ticelas poco visibles por la densa pelosidad, pos- teriormente glabrescentes. Hojas estipuladas, es- tipulas pronto caducas, erectas, liguladas, 2.5-3 mm de largo; peciolo 4.5-8.5 cm de largo; raquis foliar velutino como las ramas jóvenes, 7-11 cm de largo; hojas 7-folioladas; fo- orbiculares a líolos cartáceos, epunteados, eliptico-oblongos, 1 1— 15 cm de largo, 5-6 cm de ancho, base ligeramente cuneada, ápice acuminado, marcadamente bico- loros, verdosos y glabrescentes en el haz y blan- quecinos y densamente canescente sericeos en el envés, nervaduras laterales 9-11. Inflorescencias pedunculadas, 2.5-6 cm de largo, floración precoz; pedúnculos florales hasta 2 mm de largo; pedicelos hasta 2.5 mm de largo; bractéolas subopuestas a alternas, de la mitad al tercio superior del pedicelo, filiformes, 0.9-1.1 mm de largo. Flores 9.5-11 * Herbario privado de la familia Conzatti que aun se conserva en la Cd. de Oaxaca, México. 576 Annals of the Missouri Botanical Garden mm de largo; cáliz ciatiforme, epunteado, 3-3.5 mm de largo, 4.5-5 mm de ancho, moderadamente sericeo, casi trunco, sólo presentes los 3 dientes carinales, el medio 1 mm de largo; corola epun- teada, todos los pétalos canescente sericeos; estan- darte erecto, su lámina cóncava, orbicular, 8 mm de ancho, moderadamente pelosa en las venaciones de la superficie externa; anteras glabras; ovario sericeo, 4-ovulado. Legumbre (sólo se vió inma- dura) aparentemente indehiscente, coriacea, de eliptica a oblonga, atenuada en la base, redondeada y rostrada en el ápice, lateralmente compresa; 1 – 2 semillas por fruto, 7-8 cm de largo, hasta 1.6 cm de ancho, amarillo velutina, sutura vexilar an- gostamente alada, ala hasta sutura carina mm de ancho, E NE duras no fueron vistas). Distribución. Solo conocida del Departamen- to de Zelaya en Nicaragua, en selvas húmedas, de altitudes entre los 40 a m. Florece en marzo, e inicia la fructificación a partir de mediados de mayo. Nombres vernáculo. Cerizo, palo de cera. Muestras adicionales examinadas. NICARAGUA. ZELAYA: sur del Río Wawa, 60 km NO de Puerto Cabezas, 14?19'N, 83%56'0, E. L. Little 25117 (MO); Francia Sirpi, 30 km SO de Waspán, 14?28'N, 84%03'0, Little 25303 (MO). Lonchocarpus macrocarpus Benth., J. Lin Soc., Bot. 4 (Suppl.): 91. 1860. TIPO: «Mé. xico” [Nicaragua]. Herb. Pavón s.n. [J. M cino | (lectotipo, С, aqui redesignado). Aqui se lectotipifica a Lonchocarpus macro- carpus Benth. con el ejemplar del Herb. Pavón (G), a pesar de pss Pittier (1928) lo hizo con el e А. Fendler 1861, pero en forma mecánica, habiendolo oes por ser el ejemplar de Vene- zuela, área que él estudiaba. Simplemente trans- corresponde a L. hedyosmus Miq. s más antiguo que el de L. macrocarpus е акаң el cual, pasaría como sinónimo. Por otro lado, el escoger al del Herb. Pavón se evita crear otro nuevo nombre y se estaria en mayor concordancia con el sentido que Bentham le dió al epiteto, dado que el material del Herb. Pavón en efecto cuenta con el fruto de mayor tamaño que el resto del material que Bentham vió. Esta especie ha sido confundida con L. costaricensis Pittier. Lonchocarpus parviflorus Benth., J. Linn. Soc Bot. 4 (Suppl.): 89. 1860. TIPO: [Nicaragua]. Segovia: A. S. Oersted 2 (lectotipo, K, aqui designado). Lonchocarpus orotinus Pittier, Contr. U.S. Natl. Herb. 20: 74, 75, pl. 5А, fig. 25. 1917. TIPO: Costa Rica Guanacaste: Salinas Bay, A. Tonduz 2731 (holotipo, US; isotipos, В е whitei Lundell, Wrightia 1(2): ; Ri 155. 946. T TIPO: Nicaragua. Chinandega: along Rio Aco me, S. White & C. L. Gilly 5385 do MICH Lonchocarpus parviflorus fue descrito por Bentham en base a elementos heterogéneos sin definir al holotipo, así tenemos que el sintipo de C. Jurgensen 219 (K) corresponde a L. a Benth.; el de “New Spain” Herb. Pavón (G, fragmento) a L. minimiflorus J. D. Smith, y a de Oersted 2 (К) y 18 (К) a L. orotinus Pittier. Habiendo otro ejemplar de Oersted 64 (K), el cual, por estar estéril no nos es posible precisar su iden- tificación. En vista de lo anterior, al escoger al lectotipo se tomó en cuenta el como Bentham dis- tinguió a este binomio de las especies afines que él disponía, asi hace hincapié en lo ancho del fruto, y las pocas semillas que contiene, caracteres que corresponden a L. orotinus Pittier, el cual a su vez pasa a sinonimia, dado que se trata de un nombre posterior. Aqui se ubica a Lonchocarpus whitei como sinónimo de L. parviflorus, ya que, el holotipo designado por Lundell en fruto corresponde a esta especie, el idus (paratipos) en flor a L. mini- mith. miflorus J. D. © AESCHYNOMENEAE Diphysa americana (Miller) M. Sousa, comb. México. Veracruz: (enviado en 1730), W. Houstoun s.n. (holotipo, BM). Diphysa robinioides Benth. in Benth. et Oerst., Vidensk. Meddel. Dansk Naturhist. Foren. Kjoebenhavn. 1853. 11, 12. 1854. TIPO: Nicaragua: Volcán Mom- bacho, cerca de Granada, 4. S. Oersted 36 (lecto- tipo, K, aqui designado). A pesar de que Bentham vió los ejemplares de Houstoun (están anotadas en puno y letra por el al pie de cada ejemplar) no usó el epiteto de Miller debido a que los identificó como Diphysa cartha- genensis Jacq., error que hasta nuestros dias ha oscurecido la identidad de Colutea americana. En el Museo Británico existen dos ejemplares de W. Houstoun anotados como Colutea americana, pero P. Miller en su publicación aclara que se trata *'del Volume 77, Number 3 1990 Sousa S. 577 Papilionadas de la Flora de Nicaragua enviado a él de Veracruz, en Nueva España, en el ano de 1730." Asi en tinta en manuscrito (no en máquina) su ejemplar tiene anotado “Уега Cruz, Houstoun 1730" y otro *e Vera Cruce 1731," siendo el primero el que consideramos al tipo, lo que es afortunado, ya que, se trata de un ejemplar con hojas, flores, y frutos; en cambio, el segundo està depauperado e inclusive su identidad especifica es dudosa. Miller (1768) citó a un polinomio de Plukenet tab. 165, fig. 3), sin basar su nombre en él, aunque si ci- tandolo como la misma entidad, sin embargo como Britten & Baker (1897) ya nos aclararon, no se trata de lo mismo, es más la figura citada es una Colutea verae crucis vesicaria (1692, hoja de una mimosoidea, que pudiera ser un Pithe- cellobium. Colutea americana, como Miller nos informo, se basa en Colutea americana, vesicules oblongis compressis del manuscrito de Houstoun, además del material botánico traido a el en 1730. Bentham identificó erróneamente al ejemplar de Seemann de Panamá (K) como D. carthaginensis [carthagenensis ] Jacq., error que desde hace tiem- po ha sido anotado por varios autores. LITERATURA CITADA BRITTEN, J. & E. С. BAKER. de Houstoun's Central American Legumin osae. J. Bot. 35: 225-234 Lavin, M. 7. A cladistic т of the tribe Ко- binieae (Papilionoideae, Leguminosae). Pp. 31-64 in Advances in im System- w. 8. epee a la dendrologia de nezuela. Arboles y ария s del orden de las Le- > minosas. Il rab. Mus. Com. Venezuela tuj Se R.R.E. E. no. 4- 7) 4: 179- PLUKENET, L. 1692. Almagestum Botanicum, tab. 121- 50. London. xe id E. 1971. Studies in the я (Legu- ae) I. Phytologia 21(5): 3 SOUSA AS. M. 19 987]. гу кона a las leguminosas e la flora de Nicaragua. Ann. Missouri Bot. Gard 14: 722-7 YAKOVLEV, G. P. 1968. The genus sen Terán & peda (Sophoreae) in America. Proc. Leningr. Chem.-Phar Inst. 26: 104-112. [In Rus ы” A NEW COMBINATION IN DIOCLEA KUNTH (FABACEAE-DIOCLEINAE) FROM THE CLARIFICATION OF D. GLABRA BENTHAM, FLORA BRASILIENSIS' Richard H. Maxwell? ABSTRACT Comparing the Dioclea glabra oran cited by Bentham in Flora Brasiliensis in 1859, with the collections cited in his original descriptions of D. g D. coriacea in 1837, r eveals that the 1859 D. glabra collections include bu species: D. glabra Benth, pu D. coriacea Benth. (here ea ose and D. scabra (Rich.) radios b comb. nov. (here described and assigned a neotype). Dioclea scabra var. rownii and var. schultzii, both ne AEN are also described. The three species are placed in their майк ше В sections. In 1837 Bentham described 12 new species of Dioclea, including D. glabra and D. coriacea, and a new section, Pachylobium Bentham. Bentham's 1837 descriptions of Dioclea glabra, which he placed in sect. Pachylobium, and of D. coriacea, which he placed in sect. “Eudioclea” (sect. Dio- clea), lacked descriptions of fruit characters. When he described Dioclea glabra in Flora Brasiliensis in 1859, Bentham was able to include fruit characters from new collections. The 1859 description of D. glabra, however, contained ele- ments of three separate taxa, D. glabra, D. co- riacea, and D. scabra. He also moved D. glabra to his new sect. Platylobium Bentham and erro- neously omitted D. coriacea. My dissection of Pohl 1578 (W), the lectotype of Dioclea glabra Benth. (1837), and study of Bentham’s syntypes and other collections indicate that Bentham’s original placement of D. glabra in sect. Pachylobium was correct, and that D. co- riacea is in sect. Platylobium along with the new combination D. scabra. The invalidly published name D. elliptica Maxwell has been used in the literature (Kavanagh & Ferguson, 1981; van Roosmalen, 1985). KEY TO VARIETIES OF DIOCLEA SCABRA la. Leaflets with primary lateral veins in 6-8(-10) pairs, upper lamina mostly smooth. 2a. Flowers 2.3-3 cm long ............. la. var. scabra 2b. Flowers ca. 2 cm lon . var. brownii lb. Leaflets with primary lateral veins in 10-12 pairs, upper lamina rugose le. var. schulzii I. Dioclea scabra (Rich.) Maxwell, comb. nov. TYPE: Guyana: Essequibo, Pomeroon River, 17-24 Dec. 1922, J de la Cruz 3090 (neotype, UC; isoneotypes, F, MO, NY, US). Dolichos scaber Rich., Actes Soc. Hist. Nat. Paris 1: 111. 1792 la. Dioclea scabra var. scabra. Figure 1. Nonsynonymous names applied to this species. Dioclea ag auct. div.: Pulle, Enum. 233. 1906; Huber . Mus. Paraense Hist. Nat. 4: 407. 1909; Ducks, i h. Jard. Bot. Rio de Janeiro 1: 42. 1915, 4: 95, 330. 1925; Amshoff, Meded. Bot. Mus. Herb. Rijks Univ. Utrecht 52: 69,70. 1939(a), Flora of Dioclea elliptica Maxwell var. elliptica, nom. inval., The Genus Dioclea (Fabaceae) in the New World. Doctoral Dissertation. Southern Illinois Univ., Car- bondale, Illinois. 1969. Pomeroon River, 17-24 D Cruz 3090 (holotype, UC; isotypes, F, MO, NY, US). ' [ am grateful to the late Dr. Julian Steyermark for the opportunity to work on the Flora of the Venezuelan Guayana. The reviewers’ helpful comments on no dir ectors and curators of A, BM. BRG, F, G UC, US, and VEN for loans and, i ‚ GH, СОЕТ, IAN, K, MG, MO, n many instances, hospitality in their herbaria. Thanks are due to Lewis Johnson menclature are acknowledged with gratitude. I also thank the NY, P, PORT, RB, S, SI, SIU, U, for assistance with the illustration. A Grant-in-Aid of research from Indiana University Southeast is acknowledged. * Indiana University Southeast, 4201 Grant Line Road, New Albany, Indiana 47150, U.S.A. ANN. Missouni Вот. Ganp. 77: 578-583. 1990. Volume 77, Number 3 1990 Maxwell 579 A New Combination in 7 Dioclea Lianas to 30 m tall, woody vines, or shrublets; stems terete, twining, occasionally with tendrils, usually glabrous, with raised, elliptic lenticels. Leaves trifoliate, the leaflets brittle or coriaceous, mostly elliptic, occasionally ovate or broadly lan- ceolate, 8-12(-24) x 4-9(-12) cm, the surfaces glabrous, the lower surface brownish, the apices mostly with elongate drip tips to ca. m long, the bases rounded, the primary lateral veins in 6— 8(-10) pairs; petioles to ca. 10 cm long, the rachis ca. 2 cm long, both glabrous; stipules nonproduced, acute, to ca. 3 mm long, mostly persistent; stipels not seen, probably lacking. Inflorescence axillary, single, 20-50(-80) cm long, unbranched, ferru- ginous puberulent, becoming glabrate, flowering to са. % its length, the rachis usually strongly angular, frequently with sections swollen and inhabited by ants; tubercles subsessile, the stalks stout, ascend- ing, the heads incurved; bracts ovate, ca. 2 mm long, glabrous, semipersistent; bracteoles suborbic- ular to ovate, ca. 1.5 mm long, persistent. Flowers ca. 2.5 cm long, the pedicels 5-12 mm long, the calyx tube 7-12 mm long, sparsely ferruginous puberulent, the lobes 4, strongly upcurved, velu- tinous inside, the upper lobe obtuse, entire, ca. 6 X ]O mm, the lateral lobes falcate, acute to lan- mm, the lower lobe lanceolate, ca. 12 mm long; standard reflexed, broadly oblong to somewhat orbicular, ca. 20 mm long with a claw a. 5 mm long, entire or slightly emarginate api- cally, usually purple, lighter with age, bicallose, ceolate, ca. 10 x yellow or whitish in the center, somewhat carnose, glabrous; wings obliquely oblong to obovate, to ca. 15 x 10 mm with a claw 8 mm long; keels semior- bicular, to ca. 10 mm long with a claw 7 mm long, the upper margin basally auriculate, unlobed, the lower margin rising ca. 12 mm, culminating in a narrow or obtuse beak; stamens 10, pseudomon- adelphous, the base of the vexillary free ca. 3 mm, mostly glabrous, the vexillary and inner alternate anthers of the staminal sheath imperfect, ca. 1.5 mm long, the 5 perfect anthers oblong, ca. 1.5- late or somewhat sigmoid, 2.0 mm long; pistil g rising distally ca. 12 mm, the ovary ca. 7 m long, short-stipitate, canescent villous, invariably 2-ovulate, the style with lower part hirsute, then swollen, somewhat triangular, narrowing to a flat, truncate apex, the upper part glabrous ca. 3.5 mm, the stigma subterminal. Fruit flat, twisting at dehiscence, dry, mostly obovate or oblanceolate, ca. 17 cm long, 2.2-3.0 cm wide basally to ca. 5 cm wide apically, the base rounded, the upper the exocarp mostly smooth, glabrate, the upper suture raised, with 2 close parallel ribs, the lower margin appearing flanged, 2-seeded; seeds flat, soft, dark, suborbicular, diameter 20-30 mm, 4-7 mm thick, the hilum oblong, 6-7 mm long. Selected specimens examined. BRAZIL. AMAPÁ: Rio Oiapoque, Irwin et al. 48038 (U, US); 2 km SE of Clevelândia, Maguire et al. 47 115 (NY, US). AMAZONAS: Rio Negro, between Manaus and Sáo Gabriel, Alencar et al. 358 (NY); Black 48-2759 (NY, U, US, VEN); Ma- naus, Chagas 356 (MG); 3 Feb. 1941, Ducke 673 (F, MO, NY, SI, UC, US); 7 Dec. 1927, Ducke RB 20423 (ВВ, S, U); Rio Cauaburi, Holt & Blake 535 (К, NY, US); Rio Negro, Kuhlmann 1030 (RB); Rio Negro, Prance et al. Ed (JEF, NY); Manaus, Prance et al. 3178 (NY, ea PARA: еа Cid et al. 2469 (NY); Aram y, da Costa 237 (F); Santarém, Duarte 7231 (RB, TOM {е Тарай, Мїззао Cururú, Egler 825 (МС); i s & Silva 4394 (NY); Obidos, Oct. naus], June 18 | CH River Comté, аар ў 360 (US); Gourdonville, Benoist 1514 (Р); Cayenne, L.C. Richard s.n. (Р); Sinnamary, route to Ste. Elie, Sue 6026 (US). GUYANA. ESSEQUIBO: Cuyuni River, Aitken 1070 (S); Macouri Creek, Archer ; Ess pa hr River, Atkinson 83 (BM); Kartabo, ; 4 mi. above Kaieteur Falls, Cowan & Versteeg 274 (U); Upper Litanie River, Versteeg 401 (U); Tapanahoni River, Versteeg 662 (U). VENEZUELA. AMAZONAS: Cerro eblina, Rio Mawarinuma, Án- derson 13337 (NY); km. 11 NE of San Carlos de Rio Negro, Davidse & Miller 26536 (MO, NY); O to 0.5 km SE of San Carlos de Rio Negro, Liesner 4019 (JEF, ; Río Cunucunuma, Maguire et al. 29498 (NY); Alto Orinoca, Ll. Williams "15235 (F. VEN); forest of Orinoco, Esmeralda, Ll. Williams 15510(G, US, VEN); Casiquiare River, Ll. Williams 15672 (F, US, VEN); Rio Orinoco, frequent just above Tama-Tama, Wurdack & Adderley 43113 (СН, NY, US, VEN). BOLÍVAR: Reserva Forestal Imataca, Stergios et ai. 2769 (PORT). lb. Dioclea scabra var. brownii Maxwell, var. nov. TYPE: Venezuela. Territorio Federal Ama- zonas: Dept. Atabapo, SE bank of middle ‚рап of the Caño Yagua at Cucurital de Yagua, 8 May 1979, Davidse et al. 17450 ia МО; isotypes, MYF n.v., NY). Flores ca. 2 cm longi; pedicellis ca. 4 mm longis; vexillo obovato-orbiculari, valde n vexillari filo pubescenti ad basem. Legumen ignot Leaflets elliptic, 9-12 х ca. 5 cm, both sides dull, glabrous; petioles and rachis glabrous. Flowers Annals of the 580 Missouri Botanical Garden Volume 77, Number 3 1990 Maxwell A New Combination in Dioclea to ca. 2 cm long, the pedicels ca. 4 mm long, the calyx tube ca. б mm long, the lobes slightly up- curved, the upper shallowly bifid or entire; standard obovate-orbicular, strongly reflexed, bicallose; an- thers dimorphic, the base of the vexillary stamen pubescent; pistil deeply bent, almost sigmoid. Fruit unknown. This variety shows some characters of Dioclea ruddiae Maxwell and D. macrocarpa Huber, but androecium and gynoecium characters are shared with D. scabra. І expect fruit and seed to be similar to var. scabra but smaller. Known from type lo- cality only. am naming this variety after H. E. Brown, who realized that Bentham’s Dioclea glabra of 1859 was not the same as Bentham's D. glabra of 1837. Brown wrote on an envelope affixed to Gleason 330 [ var. scabra ] (NY), “This species = Jenman 625 [ var. scabra] & 984 [n.v.] and un- numbered specimen of Schomburgk [n.v.] which have been named Dioclea glabra Benth. But (as I noted in the herbarium in 1880) it is quite distinct from Bentham's type of D. glabra and requires a new name.” le. Dioclea scabra var. schulzii Maxwell, var. nov. TYPE: Guyana: Essequibo, Potaro, brown tough rope from crown of tree in Kakaralli clump Wallaba forest on red laterite soil, 7 Mar. 1949, Atkinson 116 (holotype, BM; is- otypes, NY, US). Record No. 6025, Forest Dept. No. 2878. [**D. B. Fanshawe” is on the NY sheet.] Foliola elliptica, infra manifeste reticulata, glabrata, supra rugosissima, glabra, abrupte acuminatis apicibus ca. 2.5 cm longis, 10-12 venis; stipulis lanceolatis. Leaflets elliptic, distinctly reticulate below be- coming glabrate, strongly rugose above, glabrous, the apices abruptly acuminate, the drip tips about 2.5 cm long, the primary lateral veins in 10-12 pairs; stipules lanceolate, mostly exceeding 6 mm. nown from the type locality only. Named after J.P. Schulz (Dienst's Lands Bosbeheer, Surinam (1968)). This variety is similar in flower characters (and I assume fruit) to var. scabra. DISCUSSION AND LECTOTYPIFICATIONS Bentham’s new sect. Pachylobium of 1837 in- cluded Dioclea glabra Benth., which is lectotyp- ified here. Dioclea glabra Benth., Comm. Legum. Gen.: 69. 1837. TYPE: Brazil. Goias (?): ad San Izidro, Pohl 1578 (lectotype, W, photo at M, F neg. no. 32009); ad San Izidro, Pohl s.n. (isolec- totype? K, photo F, photo S of questionable isolectotype at K, NY photo neg. series 2479). Of the syntypes cited by Bentham, I believe “Ad San Izidro” Pohl s.n. is Pohl 1578 (W) and the collection number was added later. I selected Pohl 1578 (W) as lectotype because of notations (es- pecially concerning localities) on the herbarium sheets, the preservation of Bentham’s original in- tent in 1837 as to section placement and descrip- tion, and the type photos in current usage. My dissection of Pohl 1578 (W), the lectotype of Dioclea glabra Benth. (1837), revealed 8-9 ovules and calyx characters that do not fit Ben- tham’s sect. Platylobium. Study of D. glabra seed characters, especially the linear, half-encircling hilum, indicates the original placement in sect. Pachylobium was correct. In 1859 Bentham de- scribed sect. Pachylobium as having 2-3 (rarely 4) ovules. I believe the number of ovules in this section is much more variable Bentham’s 1837 sect. “Eudioclea” [sect. Dio- clea ] included Dioclea coriacea Benth., which is lectotypified here. Dioclea coriacea Benth., Comm. Legum. Gen. 69. 1837. ТҮРЕ: Brazil. Amazonas: Goias: Congo do Padre, Pohl 1996 (lectotype, W); Congo do Padre, Herb. Mus. Vind. 1837, Pohl s.n. (isolectotypes?, K, NY, photo of NY specimen at S, photo of K specimen at US, NY photo neg. series 2480). Of the syntypes cited by Bentham, I believe the “Congo do Padre” Pohl s.n. is Pohl 1996 (W). I selected Pohl 1996 (W) as lectotype because no- — FIGURE 1. Dioclea scabra var. scabra. — А. rachis section with tube D. Pistil showing ovule positions (Prance et al. 317 Leaflets, part of stem (Prance et al. 16007, JEF).— rcles and bracts (Prance et al. : 6007 eds anthers (Wurdack & Adderley 43113, P — B. Inflorescence — C. Inflorescence section (Liesner 4019, MO). — ‚ JEF). 5). —E. Staminal sheath with seudomonadelphous stamens —G. Petals: . Flower (Irwin et al. 55379, VEN).—I. Flower bud (Irwin et al. 55379, VEN).—J. Fruit, 2-seeded (LL Williams 15672, VEN); seeds (Ducke, 7-12-1927, RB ) No. 20423, RB 582 Annals of the Missouri Botanical Garden tations (including *'tipse") on the herbarium sheet, and the accurate type photos of the presumed isolectotype at I believe Bentham's other Dioclea coriacea syn- types (1837) included heterogeneous elements, which I have determined as follows: Ega Amazo- num, Poppig [= Poppig 2886] (paratype? W, 3 sheets with very immature inflorescences = D. coriacea or D. macrocarpa Huber?); In margine sylvarum prope Para, Martius [= Martius 2716] (M) = D. glabra sensu Bentham (1837). Bentham's 1859 description of Dioclea glabra in Flora Brasiliensis is based on the following collections: “Habitat in silvis prov. Paraénsis, Piau- hiensis et do Alto Amazonas: M[ artius ], pres in prov. Pernambucensi: Gardner n. 2823.; in prov. Goyazensi: Pohl; et in prov. Mato Grosso secus flumen Paraguay: Weddell.” I have sorted out these Flora Brasiliensis collections into three sep- arate taxa as follows. The Martius collections (M) are paratypes of D. glabra (1837). The Spruce puis Barra do Rio Negro, June 1851, Spruce 3? (P); Óbidos, Spruce s.n. (P) are D. scabra. PAR 1139 (M), San Gabriel da Cachoeira, Rio Negro, Jan., Aug. 1852, is D. glabra? Of the Spruce collections from between Santarém and Barra do Rio Negro, Oct. 1850, Spruce s.n. (W) is D. coriacea; Oct. 1850, Spruce 1190 (M) is D. glabra (1837). Gardner 2823 (BM, K) is D. coriacea. The Weddell collections from Paraguay R., Weddell 3269 (F, P); April-May 1845, Wed- dell 3269 (P), 1848, Weddell 3269 (P) are all D. glabra Benth., sensu 1837. Again, I believe the number “3269” was added later. Bentham's 1859 citation “in prov. Goyazensi: Pohl;" is I believe “ad S. Izidro" Pohl 1578 (W), the lectotype of Dioclea glabra Benth. (1837). The other “ad S. Izidro; Pohl” citation in Flora Brasiliensis is Dioclea latifolia Benth., but the type is unequivocal as Bentham (1837) cited only one collection, Pohl 1565 (K, W). Pohl's itinerary included **Corgo (sic) do Padre” in Goiás, accord- ing to Urban (1906), and it is possible Bentham included the lectotype of D. coriacea (1837) among the collections under D. glabra (1859), although D. coriacea was not cited in synonymy. Bentham (1859) also gave a northern distri- bution for Dioclea glabra, “Crescit etiam in Guy- ana anglica et gallica." However, D. glabra, sensu 1837, has never been found in Guyana or Frenc Guiana; this part of Bentham's distribution refers to D. scabra. Pulle (1906) followed the concept of Dioclea glabra as used in the Flora Brasiliensis. Huber (1909) noted discrepancies between specimens de- termined D. glabra and Bentham's description in Flora Brasiliensis. Work by Ducke (1915, 1925a, b) on the Amazonian flora established the concept of D. glabra in sect. Platylobium rather than sect. Pachylobium, at least in the New World. Ducke described D. leiophylla in sect. Pachylobium in 1925a, and I (1969) placed that binomial in syn- onymy under D. glabra Benth., sensu Bentham 1837); this treatment was followed by Lewis (1987). Pittier (1944) also noted that Bentham's 1859 calyx and fruit description of D. glabra did not match the common D. glabra (= D. scabra) of Esmeralda, Alto Orinoco, Venezuela. Dioclea co- riacea Benth. became lost in Ducke's (1925, 1949) “forms” of D. glabra and D. bicolor Benth. Amshoff (19392) cited Richard’s description (1792) of Dolichos scaber and noted, **When this is really a Dioclea species, the description agrees very well with D. glabra Benth." This was D. glabra Benth., sensu 1859 (= D. scabra). Amshoff (19392) further stated that no speci- mens determined Dolichos scaber could be traced in the Paris herbarium. Amshoff (1939b) in Pulle's Flora of Surinam added under Dioclea glabra, " indicating she might ~ “— vs. Dolichos scaber... have seen dried specimens. New species described by L. C. Richard (1792) were based on specimens collected by Leblond in Cayenne and are now housed at G according to Stafleu & Cowan (1983). A search for the type of Dolichos scaber was undertaken by Dr. A. Charpin at Geneva. A holotype was not found in G and G-DC collections. Article 37 of the International Code of Botanical Nomenclature (1988) states citation of a type is not necessary prior to 1958. However, since no types have been found and Richard's description (1792) was extremely brief, I have selected J. S. de la Cruz 3090 as the neotype. I have taken up the epithet scabra as a valid name and have placed Dioclea elliptica in syn- onymy. The three taxa, Dioclea glabra, D. coriacea, and D. scabra, more or less share Bentham's de- scription of small, nonproduced (probably lacking in D. glabra) stipules, an overall glabrous aspect, as well as ovate or elliptic, coriaceous leaflets. How- ver, D. glabra Benth. (1837) is a sizable liana distributed throughout the southern Amazonian re- gion of Brazil and farther south along the rivers and gallery forests into the planalto and Mato Gros- so. It is often collected. The fruit was illustrated by Ducke (1925a, pl. 5) as D. leiophylla. Dioclea scabra, the northernmost element of sect. Platylobium, is found in Amazonas and Bo- Volume 77, Number 3 Maxwell 583 1990 A New Combination in Dioclea livar of the Venezuelan Guayana, the Guianas, and 1925a. Plantes nouvelles ou peu connues de in Атара, Pará, and Amazonas in Brazil. Dioclea coriacea is sympatric with D. scabra to the north, but extends farther south in Brazil into the planalto and campos. Both grow as vines or shrublets. Dioclea scabra differs from D. coriacea by possessing a stout, ridged peduncle and rachis, ascending stout tubercles, larger flowers, sharply upturned calyx lobes, the lower lobe tip hooked or cupped over the upper lobe in the bud, standard broadly oblong to somewhat orbicular, and larger ruit. In contrast, Dioclea coriacea has a slender in- florescence, long-stalked tubercles extending out from the rachis, smaller flowers, the largest to ca. 2.3 cm long, calyx lobes somewhat straight or the lower lobe upturned, standard mostly obovate, and smaller fruit, the largest to ca. 12 cm long, 3 cm wide basally to ca. 4 cm wide apically. LITERATURE CITED AMSHOFF, С. J. Н. 1939a. On South American Papil- ionaceae. = Bot. Mus. Herb. Rijks Univ. Шем . 19306, и In: Pulle, Flora of Suri- nam. 2(2): 2 ji G. tr Commentationes de Leguminosa- m Generibus. 68-78. Vienna. 59. Papilionaceae. In: Martius, Flora Brasi- inai 15(1): 1 Ducke, A. 1915 Phantes nouvelles ou peu connues de la région Amasonienmë. Arch. Jard. Bot. Rio de Janei- ro l: 7-59. la région Amazonienne. Arch. Jard. Bot. Rio de Ja- neiro 4: 1-208. 1925b. As Leguminosas Estado do Pará. Arch. Jard. Bot. Rio de Janeiro 4: 211-341. ———. 1949. As Leguminosas da foe Brasileira, 2nd edition. Bol. Técn. Inst. Agron. N. 18: 3-248 GREUTER, W. ET AL. (editors). 1988. International Code of Botanical Nomenclature. Regnum Veg. 118 . Materiaes para a Flora Amazonica. Bol. Mus. a Hist. Nat. 4: 407. КАУАМАСН, T. & I. К. FERGUSON. 1981. Pollen т. and Бет of the subtribe Diocleinae (Leguminosae: Papilionoideae: Phaseoleae). Rev. laeobot. Palynol. 32: 317-367. Lewis, G. P. 1987. Legumes of Bahia. Royal Botanic Gardens, Kew. MaxwELL, R. Н. 1969. The Genus Dioclea (Fabaceae) in the New World. Ph.D. Dissertation, Southern Il- linois Univ., Carbondale, Illinois. PrrriER, Н. . Leguminosas de Venezuela ilionaceas. Bol. Técn. Minist. Agric. 5. Editorial Elite, Caracas, Venezuela. 1906. An Enumeration of the Vascular 1792. Catalogue des Plantes de Cayenne, envoyees par Leblond. Actes Soc. Hist. 11. . & R. S. Cowan. 1983. Taxonomic URBAN, 1. 1906. Vitae Itinaraque Collectorum Botan- icorum. Їп: Martius, Flora Brasiliensis. 1(1): 78-82. VAN ROOSMALEN, М. С. М. 1 Fruits of the Guianan Flora. Institute of Systemic Botany, Utrecht Univ., the Netherlands. NOTES NEW TAXA OF DIOCLEA KUNTH (FABACEAE-DIOCLEINAE) FROM THE VENEZUELAN GUAYANA Study of new collections and reexamination of old collections for the Flora of the Venezuelan Guayana have necessitated the validation of names for three taxa. Eudioclea Benth. Dioclea sect. Dioclea [sect. 1. Dioclea holtiana Maxwell, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Boca del Vichada, 13 Jan. 1930, Holt & Gehriger 224 (holotype, US; isotype, VEN). [Dioclea holtiana Pittier, Bol. Técn. Minist. Agric. No. о: 84, fi 1944, nom. inval. publ. sine descr. latin. Based on: Venezuela. Territorio Federal Amazonas: Boca del Vichada, Alto Orinoco, Holt & Gehriger 240 (VEN n.v.).] Foliola elliptica-ovata, vel late lanceolata. Inflorescen- tiae 40-60 cm longae. Flores 2.5-3.0 cm longi; bracteolis parvis, acutis, caducis; calycis tubo pubescenti, calycis supero lobo tubum multo superanti; vexillo elliptico-ob- longo, emarginato; alis sine calcari. Ovarium ca. 10-ovu latum. Vines, climbing in thickets and woods; stems terete, with erect, velutinous, rufous or somewhat canescent pubescence. Leaves trifoliolate; leaflets oval, Ae ovate or broadly lanceolate, 4.0-8.5 x (1-)2 O cm, the upper surface sparsely pubescent, dense along the midrib, the lower sur- face villous with diffuse somewhat curly pubes- cence, the apices mostly obtuse, occasionally acute or rounded, the bases oblique, mostly rounded, occasionally cuneate or slightly cordate, the pri- mary lateral veins in ca. 7 pairs; petioles 2.5-4.5 cm long, the rachis 1 -3(-9) mm long, this, petiole, and pulvinules with dense erect pubescence; stip- ules not produced below insertion, lanceolate, ca. 2.5 mm long, caducous?; stipels setaceous, 1-2 m long, apparently persistent. /nflorescences ax- Шагу, single, 40-60 cm long, with appressed, as- cending, short pubescence, flowering to ca. Y2 its length; tubercles clavate or somewhat elongate, the lower occasionally long-stalked, the nodes distant; bracts lanceolate, ca. tent; bracteoles ovate, ca. 2 mm long, caducous; pedicels 3-5 mm long. Flowers 2.5-3.0 cm long, the buds straight or slightly downcurved, the calyx mm long, mostly persis- tube ca. б mm long, sparsely pubescent, canescent velutinous inside, the upper lobe lanceolate, entire, ca. 10 mm long, entire, the lateral lobes 6-8 mm long, the lower lobe somewhat linear-lanceolate, ca. 10 mm long; petals with claws ca. 4 mm long, the standard reflexed, elliptic-oblong, emarginate, ca. 20 x 14 x 8 mm, the auricle sharp-pointed, without a spur, mm, the wings oblanceolate, ca. 23 the keels obliquely oblong or somewhat oblanceo- late, 16 х 6 mm, weakly auriculate, the upper middle margin serrate; stamens 10, glabrous, pseu- domonadelphous, the anthers perfect; pistil straight for ca. 16 mm then geniculate, rising distally ca. 5 mm, the ovary ca. 10 mm long, short-stipitate, with a disc collar, ca. 10-ovulate, canescent villous; style uniform, distally glabrous ca. 8 mm; stigma terminal, capitate. Fruit compressed, oblong, straight or slightly downcurved, to ca. 11.0 x 1.5 cm, dehiscent, the exocarp with appressed, white to ferruginous pubescence, the upper suture with a shallow parallel rib ca. 1.5 mm on each side, the lower suture lacking ribs, 8-1 1 -seeded; seeds hard, oval or ellipsoid, to ca. 9 х mm, the hilum linear, ca. 7 mm long. VENEZU JELA. : Atures, cuenca del Additional specimens examined. TERRITORIO FEDERAL AMAZON Cataniapo, Guanches 1555 ( MO): Atures, alrededores de Puerto Ayacucho, Huber 1335 (US), Huber & Cerda 1448 (US), Stergios 3225 (PORT) Pittier in 1945 again cited Holt & Gehriger 240 as the Dioclea holtiana type. My selection of the type is based on the assumption that Killip at US selected Holt & Gehriger 244 as the type with Pittier’s knowledge. Pittier corresponded with Killip during this period regarding Dioclea. Many VEN herbarium collections of Dioclea have invalid ANN. MISSOURI Bor. Garp. 77: 584-587. 1990. Volume 77, Number 3 90 Notes 585 names written on the genus covers. “Tipo” is writ- ten on several sheets with these same names. There are sheets with illustrations of floral dissections attached, probably indicating intended publication and certainly indicating that Pittier’s work was left unfinished. Few of these names warrant species rank, but D. holtiana is an exception. he Guayana collections cited date from 8 De- cember to 26 January. Collectors report the calyx dark purple or dark red-purple with the corolla rose-whitish, pallid purple, blue-purple to whitish, or white. Collectors further report the species common in fields, woods, and thickets, from 90 to 150 m altitude. Flowers of the cited Guayana specimens are similar to those of the closely related Dioclea albiflora Cowan (1958). Dioclea albiflora has leaf- lets with denser pubescence, longer flowers, short- clawed wings with spurs, somewhat oblong keels, thinner fruits, and an almost-glabrous calyx except for the upper and lower midribs. While I consider D. albiflora a restricted Guayana endemic, holtiana ranges from Guayana into Vichada, Co- lombia, and north into other Venezuelan states. Dioclea virgata (Rich.) Amshoff, On South Amer- ican Papilionaceae. Meded. Bot. Mus. Herb. Rijks Univ. Utrecht 52: 69. 1939. 2. Dioclea virgata (Rich.) Amshoff var. cren- ata Maxwell, var. nov. TYPE: Brazil. Атара: region of Calcoene, Calgoene River, coastal region, creeping liana, frequent in shrubby savanna, 20 Aug. 1962, Pires & Cavalcante 52528 (holotype, U; isotypes, NY, SP, US). Flores ca. 2-3 cm longi; bracteolis ca. =, supero, medio margine carinae minute cren | de tato. Legumen eue persistenti, canescenti, velu- tina vel tomento 4 mm; Stems twining, climbing. Leaflets mostly ovate occasionally broadly lanceolate to elliptic, 6-11.5 3.5-5.5 cm, sparsely pubescent above and be- low, the apices abruptly acute or somewhat acu- minate, the bases rounded to somewhat cordate, the primary lateral veins in ca. 7 pairs; petioles .9-4 cm long, the rachis to ca. 2 mm, this and petiole sparsely pubescent; stipules not produced; stipels setaceous. Inflorescences axillary or ter- minal, usually single, long, sparsely pubescent; tu- bercles mostly elongate, upcurved; bracts acute, persistent; bracteoles oval or orbicular, ca. 6 X 4 mm, occasionally larger; pedicels ca. 5 mm long. Flowers 2-3 cm long; keels nearly oblong, the upper middle margin usually with ca. 5 shallow dentate or crenate teeth, the wings obliquely ob- ovate, usually without a spur; stamens 10, anthers perfect; pistil usually somewhat sigmoid, the ovary canescent. Fruit canescent pubescent, nearly to- mentose, lanate or erect-velutinous. Additional specimens examined. BRAZIL. AMAPÁ: coastal region, Pires & Cavalcante 51998 (NY, RB, VEN), 52026 (NY). AMAZONAS: Barcelos, А.Р. Duarte dnd (STU); кеш of Santarém, Ginzberger s.n. (W). RÁ: Oriximiná, Rio Paru do Oeste, Cid et al. 2306 (NY). Oriximiná, 17 Sep. 1910, Ducke MG 10996 (BM); Montealegre, 24 Apr. 1916, Ducke MG 16048 (BM, RB); Santarém, Silva & Souza 2213 (NY). SURINAM: gios & Aymard 4164 (PORT); Bajo Casiquiare, Stergios & Aymard 7328 (MO); Rio Casiquiare, Stergios et al. 9772 (JEF, PORT). Dioclea virgata var. crenata is separated from var. virgata by having smaller flowers, pedicels, and bracteoles, and the keels lack fimbriations, and the fruit exocarp has short (mostly canescent) pu- bescence rather than the stiff, erect ferruginous hairs of var. virgata. An excellent illustration of var. virgata is found in the Flora Brasiliensis (plate 44, as Dioclea lasiocarpa Mart. ex Benth.). Dioclea aff. sect. Macrocarpon Amshoff, On outh American Papilionaceae. Meded. Bot. Mus. Herb. Rijks Univ. Utrecht 52: 69. 1939. фә . Dioclea steyermarkii Maxwell, sp. nov. TYPE: Venezuela. Territorio Federal Amazo- nas: Dept. Atures, 5?21'N, 66?15'W, 550 m, savannas, situated in a region of hills and ridges S and SE of Cerro Camani, Huber 4476 (holotype, US; isotypes, MYF n.v., NY). Fig- ure 1. rutex ad ca. 1 m altus vel scandens. Foliola rigida- coriacea, lanceolata-ovata-elliptica, ad 14 x 10 cm, sparse гае stipulis поп productis. Inflorescentiae axil- cm longae; tuberculis clavatis; bracteolis longi, calycis lobis tubum quasi ae biculari-ovato; alis oblanceolatis cum calcaribus; carinis triangularibus Vines or shrublets, to 1 m tall; glabrescent. Leaves trifoliolate, the leaflets rigid stems terete, coriaceous, the lamina elliptic to broadly ovate, to ca. 14 x 10 cm, the upper surface glabrous, the lower surface glabrescent, the apices rounded, then abruptly acute, the bases somewhat rounded to cordate, the primary lateral veins in 7-9 pairs; rachis and petioles sparsely pubescent; stipules acute-acuminate, not produced below insertion, ca. 4 mm long; stipels not seen. /nflorescences axillary or terminal, single, to ca. 60 cm long, mostly ferruginous tomentose, becoming glabrescent, flow- Annals of the 586 Missouri Botanical Garden Volume 77, Number 3 1990 Notes 587 ering for % or more of their lengths; tubercles long-stalked, each ca. 5-flowered; bracts not seen; bracteoles ovate, ca. 2 X 1.5 mm, persistent; ped- icels ca. 2-3 mm long. Flowers to 1.5 cm long, usually smaller, the buds + straight, the calyx tube to ca. б mm long, sparsely pubescent, sericeous inside extending up the lobes, the upper lobe ob- tuse, entire, the lower lobe longest, lanceolate, 5— 7 mm long; petals with claws ca. 5 mm long, the standard strongly reflexed, orbicular or broadly ovate, mm, the wings л to obliquely oblong, longer than the keels, 7 х usually with a spur, the keels triangular to obliquely oblong, 7 X 6 mm, the upper margin basally au- riculate, entire, the beak obtuse or slightly out- curved; stamens 10, pseudomonadelphous, the base of the vexillary filament free ca. perfect; pistil somewhat sigmoidal, the ovary ca. 4 5 mm, mm, the anthers mm long, substipitate, hirsute, 2—4-ovulate; style glabrous distally ca. 5 mm, swollen; stigma appar- ently terminal. Fruit oblong to oblanceolate, com- pressed, ca. 9 cm long, ca. 1 cm wide proximally to ca. 2.6 cm di stally , dehiscent, the ex p den ly ferruginous pübescents 2-4-seeded; socia ovoid, 4 3 x 3.5 mm, the hilum oblong, smooth, ca. ca. 4 mm long. Additional specimen examined. VENEZUELA. TERRI- RAL AMAZONAS: savannas, basin of Río Manapi- are, Huber 1201 (NY, US). The ten perfect anthers and the oblong fruit with seeds evenly distributed suggest placing this species in sect. Macrocarpon; however, the habit and general characteristics show a close affinity to Dioclea coriacea Benth. (1837) of sect. Platylobi- um Benth. (1859). Dioclea coriacea has the five perfect plus five imperfect anthers and is invariably two-ovulate, producing an oblanceolate, two-seeded fruit. This species is named in honor of the late Dr. Julian Steyermark. I thank Dan H. Nicolson and Velva E. Rudd for helpful comments. I also thank Lewis Johnson for assistance with the illustration. For the oppor- tunity to study Dioclea material, I thank the di- rectors and curators of the following herbaria: BM, MO, NY, PORT, RB, SIU, SP, U, US, VEN, and W LITERATURE CITED AMSHOFF, С. J. Н. 1939. On South American Papili- onaceae. Meded. Bot. Mus. Herb. Rijks Univ. Utrecht 2: 69. 5 BENTHAM, C. 1837 Ro ag de leguminosarum generibus. Sollinger, Vien 1859. In: Martius, Flora Brasiliensis 15(1): 154. Cowan, R.S. 1858. In: The Botany of the Guayana i ork Bot. Gard. m 150. 945. Catalogo de la Flora mnie 1: па PITTIER, E 1944. 1. T ^ | —Richard Н. Maxwell, Indiana University South- east, 4201 Grant Line Road, New Albany, In- diana 47150, U.S.A. FIGURE 1. Dioclea steyermarkii. inflorescence, Huber 4476 monadelphous stamens, Huber 1201 ke F. Petals: рана} wing, and keel, Huber 4476 (NY).— — A. Leaflet attached to stem, Huber 4476 (NY). — NY). E Pistil showing positions of ovules, Huber 1201 (NY). — NY). — Е. Stigma, — С. Flower, Huber 4476 (US). — H. Flower bud, Huber 4476 — В. Upper portion of — D. Staminal sheath upper style, and anther, Huber 1201 (NY). — (US). —I. Portion of inflorescence with fruit and interior of mature fruit, Huber 1201 (NY). —J. Seeds, Huber 1201 (NY) NOTES AND NEW COMBINATIONS IN HAWAIIAN PANICUM (POACEAE: PANICEAE) The genus Panicum (excluding Dichanthelium sensu Clark & Gould, 1978) is quite diverse in the Hawaiian Islands, with 11 native species (five pe- rennial, six annual) recognized by me in the account of the genus for the Manual of the Flowering Plants of Hawaii. Hitchcock (1922) also recog- nized 11 species, but the circumscriptions in my account differ from those of Hitchcock for many of the species. St. John (1987) published 20 new species and one new variety while my account was in press. This included two later homonyms: P. furtivum St. John, Phytologia 63: 369. 1987, non Swallen, Contr. U.S. Natl. Herb. 29: 421. 1950; and P. simplex St. John, Phytologia 63: 371. 1987, non Willd. ex Spreng., Syst. Veg. 1: 318. 1825, nec Кош. ex Trin., Gram. Panic. 216. 1826. I have studied all but one (P. sylvanum) of the types of these names, and as a result, reduce all but one (P. lineale) to synonymy. The synonyms l attribute to P. fauriei are listed under the three varieties recognize; all the others are treated in the Manual. Two of the varieties need new combinations. For P. lineale, a species І independently recognized as new, I provide an amplified description and other additional information. — PANICUM FAURIEI COMPLEX This is a distinctive coastal species characterized by an annual habit; short stature; usually branch- ing, puberulent culms; loosely involute blades, pu- berulent below and velutinous above; and small panicles with appressed, puberulent, or shortly pu- bescent branches. Before St. John's (1987) paper, this group had in recent years been recognized as five species. I believe that three varieties are rec- ognizable primarily on the basis of spikelet differ- ences as summarized in the following key: la. Spikelets 1.5-2.3 mm long, ain acute, some- times acuminate, glabr - r. fauriei lb. Spikelets 1.8-4.2 mm ds on 2a. Spikelets 2-4.2 mm long, acuminate to acute, shortly pubescent except for prom- inent s of hairs toward the tips id the glum ar. latius 2b. Spikelets 1.8-2.3 mm long, acute, sere var. carteri pubesc ANN. MISSOURI Bor. САВР. Panicum fauriei A. Hitchc., Mem. Bernice Pau- ahi Bishop Mus. 8(3): 182, fig. 71. 1922. TYPE: U.S.A. Hawaii: [ Molokai], Halawa, June 1909, Faurie 1318 (holotype, US; isotype, BM) The length of the glumes in plants with predom- inantly acuminate spikelets often varies from acu- minate to shortly acuminate or acute within the same inflorescence; thus, plants assigned to var. latius are very variable in spikelet length. Plants that display the most strongly expressed charac- teristics of var. latius occur on Lanai and Oahu. On other islands (Kaui, Molokai, Maui, Kahoolawe, and Hawaii), as well as in other populations on Oahu and Lanai, the tufts of hairs tend to be shorter and the spikelets tend to be less acuminate and more often acute. Since stem, leaf, and inflores- cence morphology are virtually identical in the complex, and since the shape of glumes is demon- strably variable within and among populations and individuals, the only reasonably useful character is glume pubescence, which ranges from absent to shortly pubescent, sometimes with increasingly prominent apical tufts of long hairs. I believe this variation is indicative of varieties. Panicum fauriei var. fauriei Р. e d Mitt. Bot. Gart. Mus. Berlin-Dahlem 128, fig. 1. 1953. TYPE: U.S.A. Hawaii: Hawaii, X of СДА Light House, Hawi, in prostrate tufts b. 1952, Degener 21807 (holotype, B not seen; isotypes, K, MO, US). Р. moomomiense St. John, Rhodora 78: 543, fig. 1. 1976. .S.A. Hawaii: Molokai, Moomomi, limestone san ш dune, 1 mi. SW of beach, 21 Jan. 1973, Pekelo, Jr. 18 (holotype, BISH not seen). P. sylvanum St. John, Phytologia 63: 372. 1987. TYPE: U.S.A. Hawaii: Maui, Maliko Bay, Sylva s.n. (BISH not ew. The holotype of P. degeneri is described and illustrated as having a line of short hairs along the midnerve. Such hairs are not evident in all spikelets of the cited isotypes and when present are so sparse and small that the spikelets are essentially glabrous. Although I have not seen the holotype of P. 77: 588-590. 1990. Volume 77, Number 3 1990 Notes 589 moomomiense, Ї have studied the two paratypes (Pekelo 8, 303, both BISH) and one topotype (Herbst & Pekelo 2948, BISH). They represent plants with spikelets 1.8-2 mm long, which is in the lower size range for the variety, but they are in no way significantly different from the variation pattern of the variety. I also have not seen the type of P. sylvanum, but the illustration and description leave no doubt that this specimen belongs to var. fauriei because of its glabrous spikelets 1.7-2 mm long. Panicum fauriei var. latius (St. John) Davidse, stat. nov. Р. nubigenum Kunth var. latius St. John, Phytologia 47: 376. 1981. TYPE: U.S.A. Hawaii: Kahoolame Island, on sea cliff of rocky coast W of Waikahalulu Bay, E of Hanakana- ea Cove (Smuggler's Cove), ca. 100 ft., 21 Apr. 1980, Cuddihy & Char 349 (holotype, BISH). P. nubigenum sensu Hitchcock (1922), Es Kunth (1833). P. ninoleense St. 1987. TYPE: U Ha 949, St. John, pin & Morton 23954 э ы BISH). The name P. nubigenum Kunth was misapplied by Hitchcock (1922) to P. fauriei var. latius. Kunth's name was a nomen novum for P. mon- tanum Gaud., non Roxb. However, the type of P. montanum Gaud. [Insulis sandwicensibus, 1829, Gaudichaud s.n. (holotype, P not seen, fragment, US; isotype, К) ] actually is a depauperate specimen of P. torridum Gaud. This is evident from the flat rather than loosely involute blades, spikelets with hairs mostly 2-2.5 mm long, borne mainly from just below the middle of the glumes to the beginning of their acuminate tips (rather than primarily at the tip), and more pilose branches. Mature spikelets of the holotype of P. nino- leense are 2-2.6 mm (not 1.5-2 mm as in the description) and have the pubescence pattern of var. latius. Panicum fauriei var. carteri (Hosaka) Davidse, comb. et stat. nov. Р. carteri Hosaka, Occas. Pap. Bernice Pauahi Bishop Mus. 17: 67, fig. 1. 1942. TYPE: U.S.A. Hawaii: Oahu, Mokolii Islet, off coast of Oahu, rocky ledge on north end of islet, rare, 10 ft., 6 Nov. 1941, Hosaka & Maneki 2611 (holotype, BISH not seen; isotypes, K, US). P. annuale St. John, Phytologia 63: 368. 1987. TYPE: U.S.A. Hawaii: West Maui, Wailaku Dist., Kahaku- loa, mouth of Makamukaole Stream, right side, coastal ы 15 ft., 7 July 1978, Sylva & Clarke n. (holotype, BISH). P. kukaiwanense St. John, Phytologia 63: 370. 1987. : U Hawaii: East Molokai, N coast, Ku- мА in bare dirt just above sea cliff, 29 Aug. 1984, Hobdy 2184 (holotype, BISH P. malikoense St. John, Phytologia 63: 371. 1987. TYPE: U.S.A. Hawaii: Maui, 4% Haiku of Maliko Gulch, sea headland, 50-150 ft., open vegetation, 19 Dec 1976, Sylva s.n. (holotype, BISH). — The types of Р. annuale, Р. kukaiwaaense, and P. malikoense are very similar, but they differ slightly in spikelet size. The lower glume is very sparsely pubescent in all three specimens, which is the reason they are assigned to var. carteri. One of the practical effects of this reclassification of P. fauriei is that the legal status of P. carteri may require reevaluation since it has been legally declared as endangered (Cook, 1981; Arnett, 1983). I interpret the Mokolii Island population, P. carteri, sensu stricto, to represent one of several populations of var. carteri spread over at least two other islands (Maui, Hawaii). Panicum lineale St. John, Phytologia 63: 370. 1 TYPE: U.S.A. Hawaii: Kauai, ridge 1,300 ft. S of Kulanaililia, 1,400 ft., steep ledges in openings in moderately wet forest, bunchgrass 1, ft. tall, 4 Mar. 1978, Chris- tensen 324 (BISH). Densely caespitose perennial bunchgrass, 55- 126 cm tall; culms erect, unbranched, glabrous. Leaves primarily basal, often strongly distichous; sheaths glabrous or puberulent toward the apex; ligule a rudimentary membrane with cilia 0.4-0.6 mm long; collar densely velvety puberulent, es- pecially on the sides; blades 59-85 cm long, 4-9 mm wide when unrolled but mostly involute, gla- brous below, densely velvety pubescent above at the base behind the ligule and the lower portion, otherwise glabrous, the tip long-acuminate. Pani- cles 20-35 cm long, 6-10 cm wide; rachis gla- brous; branches spreading at acute angles, very sparsely scaberulous, the pulvini pubescent. Spike- lets 3.8-4.8 mm long, linear-lanceolate, acumi- nate, glabrous; rachillar internodes prominent; glumes unequal, the lower 1.7-2.6 7-nerved, acute, the upper 3.4-4.5 mm long, 9- nerved, acuminate; lower floret sterile; lower lem- mm long, 3- ma as long as the spikelet, similar to the upper glume, 7—9-nerved; lower palea ,—7, as long as the lower lemma, narrow, acuminate; upper floret 2.5-2.9 mm long, 0.7-0.8 mm wide, linear-ellip- tic, cartilaginous, smooth, shiny, acute, the lemma and palea equal in length; anthers 3, 1.7-2.4 mm 590 Annals of the Missouri Botanical Garden long, orange; styles 2, separate; stigmas plumose, purple; caryopsis 1.6-1.7 mm long, 0.7 mm wide, oblong-elliptic, the embryo 0.8 mm long, the hilum -0.5 mm long, basal, elliptic. Additional specimens examined. U.S.A. HAWAII: Kauai, Kalahu, W rim of Kalalau Valley, precipitous slope at 3,000 ft., single specimen seen, 1 Feb. 1950, Degener & Hatheway 20444 (US); Kauai, Na Pali Coast, between Waiahuakua and Hoolulu Valley, ca. 600 ft., large bunch- grass growing beside trail with Eragrostis variabilis, 10 Apr. 1980, Corn ESP 190 (BISH); Kauai, along a steep, wet ridge running up the W slope of Wainiha Valley to SW of Kulanaililia, ben of di Eugenia, etc., bunchgrass, anth e, quite common, 400 m, 20 Mar. 1976, Fay, Mariarity & Robinson 588 (BISH, K, MO, PRE, SI, US). Panicum lineale is so far known only from the moist, northwestern windward side of Kauai. Al- though it is a prominent grass, it must be ver localized, judging from the paucity of collections. J. Fay (pers. comm.) collected this species at the margin of a pasture. Fay also noted that clumps of this grass pulled out of the ground readily, and this may indicate that it does not persist long in secondary areas that are subject to grazing. his species is amply distinct from all other native Panicum (s. str.) species by its short lower glume and long, narrow spikelets. All others have a lower glume as long as or slightly shorter than the spikelet. Although this species probably belongs to subg. Panicum sect. Panicum (sensu Hsu, 1965), a good understanding of its precise relationships eludes me. Anatomical studies of its leaves, lemmas, and lodicules would probably make such a determina- tion easier. Specimens of the Fay et al. 588 collection have been distributed with the unpublished name P. na- paliensis I ane John J. Fay, U.S. Fish and Wildlife Service, Endangered Species Office, for access to his excellent collections and for observations about P. lineale. LITERATURE CITED ARNETT, С. В. 1983. Endangered and threatened wild- d v р iu to in pta каке (Carter’s ss) as a red s and determine spec n Wer dn habitat Federal | Register 481 1 98): 46328- 6332. CLARK, С. A. & F. W. Gourp. 1978. Dichanthelium subgenus Turfosa (Poaceae). Brittonia 30: 54-59. Cook, R. S. Proposal to list Panicum carteri (Carter’ s panicgra ss) as an endangered species and determine its critical habitat. Federal Register 46(20): 9976-9979 HrrcHCOCK, A. S. 1922. The grasses of Hawaii. Mem. Bernice Pauahi Bishop Mus. 8: 101-230. Hsu, С. С. 65. The classification of Panicum (Gra- mineae) and its allies, with special reference to the characters of lodicule, style-base and lemma. J. Fac. 150. Diagnoses of Panicum species (Gramineae): Hawaiian plant studies 49. Phytologia 95. —Gerrit Davidse, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. THREE NEW SPECIES OF FERNS FROM MESOAMERICA The following new species are described as part of my work on the pteridophyte volume for Flora Mesoamericana. Tectaria acerifolia К. С. Moran, sp. nov. TYPE: anama. Panamá: El Llano to Carti road, 200 т, Churchill & de Nevers 4200 (holotype, MO; isotypes, PMA, UC). Figure la, b. Petiolus castaneus, о = aequans. Lamina pal- matiloba; soris seriebus 3-8 i venas principales dis- positis; indusiis reniformibus vel circularibus, haud pel- tatis. Rhizome erect; sterile and fertile leaves dimor- phic; petiole of sterile leaves ca. equaling the lam- ina, castaneous throughout, puberulent, scaly at wide, lanceolate, spreading; lamina 14—29 cm long, pal- mately lobed; basal lobes or pinnae 12—18 cm long, falcate, entire but with 1—2 short basal basiscopic lobes; rachis and costae castaneous towards the base, the scales 1-2 mm long, 0.2-0.5 mm base, tan distally, sparsely puberulent, the hairs ca. 0.1 mm long, inconspicuous; laminar tissue glabrous on both surfaces, drying dull green; pet- ioles of the fertile leaves 2—3 times longer than the lamina; the fertile laminae 15-20 cm long; sori in 3-8 rows between the principal veins; indusia non- peltate, attached laterally, reniform or circular. Additional specimens examined. Costa Rica. ALAJUELA. Upala, Bijagua El Pilón, de la intersección del camino San Miguel a El Pilón con el Río ern murria, 600 m, Herrera 2061 (MO). PANAMA. CANA E: Parque Nacional Soherama, camino del Ole lbs p 239 12 mi . PANAMÁ: Cerro Jefe, 6 mi. past to Altos Pacora, 800 m, Sytsma & D Асу 3680 (МО); 5-9 ті. an-American Hwy. оп El Llano-Carti road, 200-250 ompson 4631 (CM). SAN BLAS: forest SE of Puerto Obaldia, Croat 16842 (MO). This species grows in wet forests at 200-800 m. It can be distinguished from all other tectarias in the New World by the combination of palmately divided leaves, dark axes, sori in 3-8 rows between the lateral veins, and fertile leaves with longer and more erect petioles than those of the sterile ones. It resembles T. heracleifolia, a species which dif- fers by its peltate indusia, sori in 2-3 rows between the principal veins, stramineous axes, and usually l-pinnate leaves. Bolbitis simplex R. C. Moran, sp. nov. TYPE: Panama. Darién: Parque Nacional del Darién, ridge between N & S branches of Rio Pucuro, in forest N of old village of Tacarcuna, ca. 18 km E of Pucuro, terrestrial on slopes along small stream, 600-800 m, Hammel et al. 16467 (holotype, MO; isotypes, F, PMA, UC, US). Figure lc, d Folia simplicia, integra, anguste oblanceolata ad an- guste elliptica, apice gemma prolifera munita; venis ar- eolas venulis inclusis formantibus; rhachidi dimidio prox- imali abaxialiter anguste alata. Terrestrial; rhizome 4-6 mm wide, short-creep- ing, horizontal, scaly, the scales 2-4 x 0.5-1 mm, blackish, opaque; sterile and fertile leaves dimor- 23-50 cm long, 3.5-5 cm wide, simple, entire, narrowly oblanceolate to nar- phic; sterile lamina rowly elliptic, cuneate, glabrous, with a subterminal bud; petiole 1-5 cm long, alate, scaly, the scales lanceolate, brown, spreading, nonclathrate; rachis with a narrow abaxial wing ca. 0.5 mm wide, best developed in the proximal half, parallel (not per- pendicular) to the lamina; veins anastomosing, the areoles with recurrent free veinlets; fertile leaves 30-50 cm long, 0.7-1.1 cm wide, long-petiolate. This new species resembles Bolbitis panduri- folia (Hook.) C. Chr. because both have simple leaves, subterminal buds, and areoles with included veinlets. The two species differ in several obvious characteristics of the sterile leaves. The petioles on sterile fronds of B. simplex are 1-5 cm long (about ¥,, the length of the lamina), whereas those of B. pandurifolia are 16-27 cm long (about / the length of the lamina). The sterile lamina of B. simplex is 3.5-5 cm wide, narrowly oblanceolate to narrowly elliptic, and cuneate basally, whereas that of B. pandurifolia is 6-12 cm wide, elliptic, oblong, or ternate with a single pair of basal pinnae, and short-decurrent, subcordate, or auriculate ba- sally. The rachis of B. simplex has an abaxial, green, herbaceous wing, whereas that of B. pan- durifolia lacks such a wing. Aside from morphol- ogy, the geography also differs: B. simplex is known from eastern Panama, whereas В. pandurifolia is known from central Ecuador to central Peru. Salpichlaena thalassica Grayum & К. С. Mo- ran, sp. nov. TYPE: Costa Rica. Heredia: forest ANN. Missouni Вот. GARD. 77: 591-593. 1990. 592 Annals of the Missouri Botanical Garden 15 cm FIGURE l. а,Ь. a o : leaves, note longer petiole of the fertile leaf. — с, d. Bolbitis simplex (Hammel et al. 16467 MO).— c. Sterile and fertile leaves. —d. Adaxial surface of the sterile leaf. Tectaria acerifolia (Vásquez 239 MO). — a. Abaxial surface of fertile leaf. — b. Sterile and fertile Volume 77, Number 3 1990 Notes 593 1800, e ө .. 1600 N í N e. 1400) \ ee . Ne ° 1200+ ^ ee е \ Е >. > a e N с 1000р e by o ее ее М e = ° ° = е N > 800% e x = e 6 " M Б: eee воо ee ee ө е 400} e ° ° *e 200+ . оь . . ® ее Stalk length of basal fertile pinnule (mm) E 2. Elevation versus stalk length of the basal fertile pinnule in Salpichlaena thalassica (upper right; leaves blue-green) and S. volubilis (lower left; leaves dark green). Specimens measured were from Costa Rica and Panama between Rio Peje and Rio Sardinalito, Atlantic slope of Volcán Barva, 10?17'N, 84°4.5'W, 800-1,000 m, Grayum & Chazdon 6833 (holotype, MO; isotype, CR). Species haec ab 5. volubili differt stipite longiore (9- mm) pinnulae basalis шш foliis thalassicis, distri- butione altitudinali superio Additional specimens examined. NICARAGUA. ZELAYA: Cerro La Pimienta, bosque enano, 1,000- 1, 200 m, Grijalva үе (СВ); costado SW de Cerro El Horm 1,000 m, Grijalva 470 (CR), 489 (CR): miguero, dense, virgin elfin forest, 1,100- 1,183 m, Pipoly 5187 (MO). Costa RICA. CARTAGO: El 12. К); Tapanti, porn m, Berro S ez 122 (CR), 723 (CR); ca. m S of Tapanti along the new road, on the E slop E Rio Grande de Orosi near P concrete bridge, 1, P500 m, Burger & wa 6807 (MO, ); mountains ca. 5 mi. S of Cartago, Maxon 512 (NY). HEREDIA: jue Carillo Na eal "Park, 1,215 m, Hennipman et al. 6893 (MO). PUNTARENAS: Monteverde, Veracruz River Valley S of reserve, 1,300-1,500 m Haber ex Be llo & Care а PANAMA. CHIRIQUÍ: trail w from Fortuna Dam Camp to La Fortuna, broad- leaved ded forest, 1,300 m, Hampshire & Whitefoord 370 (BM). DARIÉN: Mamey, ridge, 450 m, Whiteford & ei 357 (CR); Cerro Mali base camp, Colombian border, ,400 m, Gent try & Mori 13754 (CR, MO). COLOMBIA. a Fontino, road to Murri, 15 km W of Nutibara (Altos de Cueva), 1,850 m, Brant & Martinez 1386 (MO, UC). BOYACÁ: carretera Chinquinquira a Pauna, 1,900 m, Jaramillo et al. 3541 (МО). MAGDALENA: Sierra Nevada de Santa Marta, Cerro Raton, Serrania de san Javier, 1,900 m, Forero & Kirkbride 672 (MO). Sierra de la Macarena, Central Mts., N ridge, 1,500 m, Philipson & Idrobo 1838 (BM). NARIÑO: Ricdurte, 1,300 m, von Sneidern A.603 (MO); Reserva Natural La Plan- m de Chuncunés, 1,800 m, de Benavides 9013 This species was first noted by Dr. Michael Gra- um, who observed that the leaves were blue-green rather than dark green, as in S. volubilis. His extensive fieldwork in Costa Rica has shown that the blue-green-leaved plants generally grow at higher elevations than S. volubilis, an observation supported by elevational data on herbarium spec- imens (Fig. 2). Further study of herbarium speci- mens found a quantitative character: the length of the stalk of the basal fertile pinnule was longer in the blue-green-leaved plant than in S. volubilis (Fig. 2). The lengths of the spores and character- istics of the stem scales were also checked, but these showed no differences with those of S. vol- ubilis. It was decided to recognize as a new species the plants with blue-green leaves because elevation and length of the basal fertile pinnule stalk correlated well, overlapping only slightly with those features of S. volubilis. Because of the overlap, it could be argued that S. thalassica is best treated as a sub- species or variety; however, Michael Grayum and I decided to treat it as a species to bring it to the forefront of pteridologists’ attention. I first noted that these species were undescribed while I was a National Science Foundation post- doctoral fellow for the Flora Mesoamericana proj- ect (BSR-8614880). I thank John Dwyer and Roy Gereau for checking the Latin descriptions. — R. C. Moran, Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. TWO NEW SPECIES OF ELAPHOGLOSSUM (ELAPHOGLOSSACEAE) FROM AMAZONAS, VENEZUELA Elaphoglossum is one of the most complex and taxonomically difficult of all fern genera with over 600 species, three-fourths of which occur in the New World. Ninety-nine species were treated in Smith’s Pteridophytes of Venezuela (1985). ‘Twenty-nine of those species were described as new (Mickel, 1987), and I here describe two additional species. Elaphoglossum longicaudatum Mickel, sp. nov. TYPE: Venezuela. Territorio Federal Ama- zonas: Dept. Rio Negro, 1-2 km E and SE of San Carlos de Rio Negro, 01%51'N, 67%03'W, 11 Nov. 1987, Liesner & Car- nevali 22993 (holotype, NY; isotypes, MO, VEN). ll 1A-C. | 21.5 + 1 | Ab ЕЁ. inaque apice caudata е Rhizome compact, 3-6 mm diam.; rhizome scales linear-lanceolate, lustrous, heavily indurated, dark red-brown, 2-3 mm long, margin with sparse, weak, hairlike teeth, apex twisted; phyllopodia inconspic- uous; fronds clumped, 12-32 cm long, 1.9-3.4 cm broad; stipe 1-3 cm long, /,-/ the sterile frond length, densely clothed with spreading orange scales 3-5 mm long; blade narrowly oblanceolate, papyraceous, apex caudate, base narrowly cu- neate; veins indistinct, free, simple or l-forked, ca. | mm apart, at 75° angle to costa; hydathodes lacking; blade scales orange, dense but not strongly overlapping, linear-lanceolate, 1—2 mm long, most- ly skeletonized with long hairlike teeth, the teeth twice as long as the width of the scale body, some scales reduced nearly to stellate hairs; fertile fronds ca. Y2 the sterile frond length, 11 cm long; stipe 3 cm long; blade linear, 8 cm long, 0.7 cm broad, apex cuspidate, base cuneate, densely imbricate- scaly above, costa densely scaly below; intersporan- gial scales lacking. Epiphytic in and taller forest” а iie between bana 20 m elevation. Additional specimen exami VENEZUELA. AMAZONAS: 2 km E and SE of San Sm de Rio Negro, 01%51'N, 67°03'W, 12 Nov. 1987, Liesner & inb ali 23003 (NY). Elaphoglossum longicaudatum differs from Ё. auricomum (Kunze) Moore in having dark (rather than pale) rhizome scales and caudate (rather than acuminate) blade apex. Elaphoglossum parvulum Mickel, sp. nov. TYPE: Venezuela. Territorio Federal Amazonas: Dept. Atures, forested area along stream, on plateau N of unnamed (1,760 m) peak, 9 km NW of settlement of Yutajé, 4 km W of Rio Coro- Coro, W of Serrania de Yutaje, 05?41'N, 66?10'W, 7 Mar. 1987, Liesner & Holst 21772 (holotype, NY; isotypes, MO, VEN). Figure 1 Elaphoglosso horridulo similis laminae forma hyda- thodisque conspicuis sed pilis multis glandulosis differt. tal compact, ascending, ca. 2 mm diam.; les linear, orange, entire, 2-3 mm longs jus lacking; fronds clumped, 2-5 cm long, ).5-0.7 cm broad; stipe slender, /,-Y the sterile frond length, clothed with spreading, orange, hair- like, subulate scales ca. 2 mm long and with abun- dant 3-5-celled erect glandular hairs; blade linear- oblong, papyraceous, apex obtuse, base narrowly cuneate; veins obscure; hydathodes present and conspicuous; blade scales hairlike, subulate, den- ticulate, castaneous, 2-3 mm long, with abundant erect glandular hairs; fertile fronds ca. as long as the sterile fronds; stipe ca. % the fertile frond length; blade smaller than on the sterile fronds, spatulate; intersporangial scales lacking. Epiphytic, 1,050-1,300 m elevation. Elaphoglossum parvulum resembles E. horri- dulum (Kaulf.) J. Smith in blade shape and con- spicuous hydathodes but has many glandular hairs. It is probably more closely allied to E. siliquoides (Jenm.) C. Chr. species (fronds 2-5 cm long vs. 17-50 cm). but is much smaller than that LITERATURE CITED Mic Pus т Т. 1985. ке изен Pp. 78-107 in . Smith (editor), Pteridophytes of Venezuela, an el List. Published a author w species of Plaphomlossun (Ela- phoglossaceae) Кош northern South America. Brit- 339. tonia 39: 313 —John T. Mickel, New York Botanical Garden, Bronx, New York 10458, U.S.A ANN. Missouni Bor. Garp. 77: 594-595. 1990. Volume 77, Number 3 1990 Notes 595 EA IGURE 1. A-C. Elaphoglossum Elaphoglossum parvulum.—D. Habit. Po T Т pum t A х) FX qu El E Ll flent A SS S f AN D S a AN longicaudatum.— A. Habit.—B. Rhizome scales. —C. Blade scales. D-F. — E. Blade scales. —F. Rhizome scales. A NEW COMBINATION IN LUDWIGIA (ONAGRACEAE) Jussiaea dodecandra has been described by De Candolle (1828) based on material collected in Guyana. This taxon has been considered as a va- riety of Jussiaea affinis DC. (Munz, 1942) or Ludwigia affinis (DC.) Hara (Munz, 1965) of sect. Seminuda, based on its minutely puberulent pu- bescence, which differs from the pilose indument of Ludwigia affinis. Recent relevant differential data indicates that the chromosome number for this taxon is n = 24 (this paper). This is different from Ludwigia affinis with n = 32 and 40 (Raven & Tai, 1979) and has so far been reported in this section for two other unrelated species (Raven & Tai, 1979; see also key below). As a result, we consider this taxon to be best treated at the specific level, for which the new combination is needed. Ludwigia dodecandra (DC.) Zardini & Raven, comb. nov. Jussiaea dodecandra DC. Prod- romus 3: 53. 1828. Jussiaea affinis var. do- decandra (DC.) Munz, Darwiniana 4: 261. 1942. Ludwigia affinis var. dodecandra (DC.) Munz, N. Amer. Fl., ser. 2, 5: 36. 1965. TYPE: Guyana. Demerara, Parker in 1826 (holotype, G-DC, photos POM and F 7897; isotypes, K, U) 24, counted by Hongya Gu: French Guiana. Saúl, en- trée du village, Granville et al. 9062 (CAY, MO). Voucher for chromosome number, n = Additional specimens examined. Costa Rica. GUANACASTE: El Arenal, Standley & Valerio 45417 (F). PANAMA. BOCAS DEL TORO: Chiriquí Lagoon, Wedel 1884 (MO); Changuinola River, Dunlap 391 (F). CANAL ZONE: Barro Colorado Island, north shoreline of Gigante Bay, Portobelo, Kennedy 2238 (МО). COLOMBIA. CHOC di, Ordoñez et al. 072 (MO). FRENCH GUIANA: St Lau rent-Paul Isnard, between PK 100 and 120, Granudle 5198(CAY, МО); Granville 5197 (CAY, MO); Montagne Bellevue de l'Inini, Granville et al. 8074 (MO), 7852 (MO); Saul, trace ORSTOM Belvedere Est PK 2, Gran- ville 3137 (CAY, MO); Saul, Carbet Mais Trail, vicinity of Crique Popote, Prance 28108 (MO); River Approuage, between “saut” and crique Couata, Oldeman T-6 1 (CAY ANN. Missouni Bor. GARD. 77: 596. 1990. winiana 4: 179-285, pl. 1 ——— ——. 1965. МО); region of contact between Sommet Tabulaire and Massif des Emerillons, Granville 3747 (CAY, MO); Monts Bakra, 3 km W of Pic Coudreau, Granville 4138 (CAY, MO). GuYANA: Mbura Hill Maas et al. 5899 (MO). Ludwigia dodecandra can be distinguished from the other species of sect. Seminuda by the following key: Key TO DISTINGUISH THE SPECIES OF LUDWIGIA SECT. SEMINUDA la. Flowers 4-merous 2a. Sepals 2.5-4.5 mm long, plants finely pu- berulent, n = 24 UE ees ee africana (Brenan) Hara 2b. aa 6-10 mm plants pilose, n = Каен Ludwigia ado (Mich.) Hara ; fae 5-6-merous. 3a. Leaves lanceolate, sepals 5-8 mm long, disc м: m A Y iip width of the seed body, — c КЎ ке» leptocarpa (Nutt.) Hara 3b. Leaves o ovate, sepals 3.5-5 mm long, disc elevated, raphe уи V4 the width of the seed body. 4a. Plants pilose, n = 3 ANN са. ae (pc ) Hara 4b. Plants minutely puber rulent, n — . Ludwigia бобат (DC.) Zardini & Raven We are grateful to Jean-Jacques de Granville for providing materials for chromosome counts and to Hongya Си for counting them. LITERATURE CITED De CANDOLLE, A. P. 1828. Prodromus Systematis Na- turalis Regni Vegétabilis 3: 53. Munz, P. 1942. Studies in Onagrac eae— XII. vision of the New World кя of Jussiaea. ы Опаргасеае, А. on Fl., ser. 2, 5: 1-2 : ш: = eis & W. Tai. 1979. Observations of chro- es in Ludwigia i aac Ann. Missouri 879 Шш. “Ca rd. 66: 862- —Elsa M. Zardini and Peter Н. Raven, Missouri Botanical Garden, P.O. Box 299, St. Louis, Mis- sourt 63166, U.S.A PSYCHOTRIA BERRYI, A NEW NAME FOR Р. DAVIDSAE Steyermark named Psychotria “davidsae” af- ter his co-collector (Davidse, male) of the type specimen, not noticing that the International Code of Botanical Nomenclature, Art. 73.10, stipulates that the epithet must be corrected to davidsei, which makes the name an illegitimate homonym. As all three collectors of the type specimen already have Psychotria species named after them, I re- name the species as follows, after the first-named co-collector of a paratype. Psychotria berryi Wingfield, nom. nov. based on P. davidsei Steyerm. (as “davidsae”), Ann. Missouri Bot. Gard. 71: 1177. 1985, non P. davidsei Dwyer, Ann. Missouri Bot. Gard. 67: 368. 1980. —Robert Wingfield, Apartado 7357, Coro, Fal- cón, Venezuela; herbario “Coro,” Dpto. de In- vestigación, Instituto Universitario de Tecnolo- ^ 29 gia “Alonso Gamero. ANN. Missouni Вот. GARD. 77: 597. 1990. NEW SPECIES OF GUATTERIA (ANNONACEAE) FROM THE GUAYANA HIGHLAND In preparing the treatment of Guatteria Ruiz & Pavon for the Flora of the Venezuelan Guayana by J. A. Steyermark and collaborators, we found two species from the Flora region to be apparently undescribed. Formal descriptions for these taxa, Guatteria liesneri and G. atabapensis, are pro- vided below. An additional species from nearby Guyana, G. clusiifolia, is described as well. Guatteria liesneri D. M. Johnson & N. A. Mur- ray, sp. nov. TYPE: Venezuela. Territorio Fede- ral Amazonas: Dept. Rio Negro, along Rio Baria (= Rio Mawarinuma) just upstream from base camp, SW side of Cerro de la Neblina, 140 m, 15 Feb. 1985 (fl, fr), Nee 30864-a (holotype, NY; isotypes to be distributed). Species forsan Guatteriae modestae Diels proxima et ramulis glabris, foliis nitidis glabris conspicue reticulatis, et venis secundariis divaricatis similis, sed lamina нө 14.6-22.1 сш longa et 5.1-7.0 cm Bee oblonga ve oblongo-elliptica et monocarpiis 9-10 mm longis et 4-5 mm latis stipite 15-20 mm longo оа diversa. Tree 4-20 m tall with a DBH of 10 cm in one individual. Twigs glabrate. Laminas of larger leaves 14.6-22.1 cm longa et 5.1-7.0 cm lata oblonga vel ceous, oblong or oblong-elliptic; base broadly cu- neate to rounded, short-decurrent; apex cuspidate or short-acuminate, the acumen 13-20 mm long; glabrate or with sparse minute hairs abaxially; mid- rib plane or slightly m adaxially, raised abaxially; secondary veins per side, di- o midrib, brochidodromous, 5 mm from the margin, raised on verging at 60- both surfaces of the blade; higher-order veins raised and conspicuous on both surfaces of the blade. Petiole 8-13 mm long, 1.5-2.2 mm wide, winged, involute toward base, glabrate. Pedicels 1-2 per leaf axil, occasionally from axils of fallen leaves, 1.2-2.7 cm long, 0.7-1.2 mm wide, articulate 2— 7 mm above the base, with minute caducous brac- teoles below articulation, glabrate. Sepals 3-4 mm long, 2.5-3 mm wide, free and contiguous at base, chartaceous, triangular to ovate, acuminate at apex, sericeous abaxially, sparsely sericeous or glabrate adaxially. Largest petals 14—19 mm long, 8-10 mm wide, subcoriaceous, oblong, elliptic, or widest above middle, adaxially pubescent with the hairs densest at base, abaxially pubescent with a glabrous patch at base on inner petals. Stamens ca. 1.4 mm long, oblong, connectives truncate at the apex, papillate. Carpels ca. 100-120; ovaries sericeous, a. 0.7 mm long; styles coherent, ca. 1.2 mm long, clavate, pubescent on sides, more densely so at apex. Torus 4-5 mm diam., with a rim between stamens and carpels, glabrate. Pedicel in fruit 1.8- 2.] cm long, ca. 1.5 mm thick, articulate 3-5 cm above base, glabrate, with a few small lenticels; torus in fruit depressed-globose, expanded or not, sparsely pubescent. Monocarps up to 60, the seed- containing portion 9- 10 mm long, 4-5 mm wide, ellipsoid, apiculate at apex, stipe 15-20 mm long, О mm wide, glabrate or with a few sparse hairs, pericarp 0.1 mm thick. Seed 8 mm long, 5 mm wide, ellipsoid, orange-brown, smooth, shiny, the exostome paler than seed coat, slightly exsert- Additional specimens examined. BRAZIL. AMAZONAS: Rio Negro, Jucabi (near mouth of Rio Curicuriari) and vicinity, 21-25 Sep. 1947, Schultes & Lopez 8856 NY). VENEZUELA. TERRITORIO FEDERAL AMAZONAS: Dept. (MO, NY); Dept. Rio Negro, from Los arbore of Co- misión de Limite to Cario Erubichi on Rio Baria, 2 Apr. 1984 (fl), Liesner 17120 (MO, NY); Dept. Atures, road Pte. Cataniapo-Gavilán, 20-25 km cho, 4 Nov. 1980 (fl), Maas & Huber 5106 (VEN). DEPT. RÍO NEGRO: Neblina Base Camp on the Rio Mawarinuma, 17 July 1984 (bud), Davidse & Miller 27447 (MO); Cerro Marea between base camp and “Puerto Chimo" n o Mawarinuma, 26 Apr. 1984 (fl), Gentry & Stein 46949 (MO, NY), 46983 (MO, NY); 1 to 4 km E of Cerro de La Neblina Base Camp on Rio Mawarinuma, 10 Feb. 1984 (fl), Liesner & Funk 15287 (MO, NY); along Río Baria (= Rio Mawarinuma) just upstream from base camp, SW side of Cerro de la Neblina, 27 Jan. 1985 fl, fr), Nee 30576 (NY, undistributed duplicates); 16 km E of San Carlos de Rio Negro on road to Solano at Cano de Cholo, 7 Apr. 1984 (fl), Stein et al. 1472 (MO). >= This species is named for Ron Liesner, in rec- ognition of the many excellent collections of Guat- teria that he has made in the Venezuelan Guayana, which have often helped clarify confused taxono- my. This species is placed in sect. Pteropus R. E. Fries by virtue of its leaves decurrent at the base ANN. Missouni Вот. Garb. 77: 598-600. 1990. Volume 77, Number 3 1990 Notes 599 and tendency of the flowers to occur in fascicles. It resembles most closely Guatteria modesta Diels, described from Peru, in the glabrous shoots, the thin leaves with raised secondary and higher-order veins, and the wide-spreading secondary veins, but differs in its oblong to oblong-elliptic leaf blades 14.6-22.1 cm long and 5.1-7.0 cm wide and the monocarps with the seed-containing portion 9-10 mm ns and 4—5 mm wide on stipes 15-20 mm long. Guatteria modesta, in contrast, has lanceo- Бы leaf blades 10-16 cm long and 3.5-5 cm wide, and monocarps with the seed-containing por- tion 15 mm long and 8 mm wide on stipes 12-16 mm long. The character of the flowers is very similar, and additional collecting may prove them to be conspecific. Guatteria atabapensis Aristeguieta ex D. M. Johnson . A. Murray, sp. nov. TYPE: Ven- ezuela. Territorio Federal Amazonas: Dept. Atabapo, Rio Atabapo, at margin of Sabana Cumare on right bank of Cario Cumare (20 km above San Fernando de Atabapo), 125- 140 m, 3 June 1959 (fl, fr), Wurdack & Adderley 42759 (holotype, NY; isotype, F). Species folis coriaceis ad basim г: et mono- carpus parvis brevistipitatis Guatteriae maguirei R. Fries similis, se are elon 9.8-15.9 cm pieds et 4.5-5.2 cm lata ad apicem acuminata, pedicello brevi, 0.4-0.8 cm longo, et connectivo antherarum papillato absimilis. Treelet or shrub 3-6 m tall. Twigs pubescent, at length glabrate. Laminas of larger leaves 9.8- 15.9 cm long and 4.5-5.2 cm wide, subcoriaceous to coriaceous, lanceolate to ovate, occasionally ob- long; base rounded; apex acuminate, the acumen 9-15 mm long; glabrate adaxially, densely pubes- cent but soon sparsely pubescent to glabrate abax- ially; midrib slightly impressed adaxially, raised abaxially; secondary veins 11—14 per side, diverging at 60—90° to midrib, brochidodromous, loops clos- ing at 5-7 mm from the margin; secondary and higher-order veins raised on both surfaces of the blade, more strongly so abaxially. Petiole 5-11 mm long, 1.7-2.5 mm wide, canaliculate, glabrate. Pedicels 1—2 per leaf axil, occasionally from axils of fallen leaves, 0.4-0.8 cm long, 1-1.5 mm wide, articulate 2-3 mm above the base, pubescent, sometimes with tiny persistent amplexicaul brac- teoles below articulation. Sepals 3.5 mm long, 2.5- .5 mm wide, free or connate at base, subcoria- ceous, ovate, acuminate to obtuse at apex, ap- pressed-pubescent abaxially, sparsely so or glabrate adaxially. Outer petals 13-16 mm long, 5-5.5 (-9) mm wide, subcoriaceous, oblong or oblanceolate, rounded at apex, adaxially pubescent, hairs densest at base, abaxially pubescent, with a glabrous patch at base; inner petals 14-21 mm long, 8-12 mm wide, oblong or obovate, rounded at apex, pubes- cence as in outer petals. Stamens 1-1.2 mm long, long-trapezoidal, connectives truncate at the apex, papillate. Carpels ca. 40; ovaries (immature) ca. 1.1 mm long, sericeous; styles (immature) 0.8 mm long, clavate, pubescent. Torus hemispheric, 3.5- 4 mm diam., with rim present between stamens and carpels, glabrate. Pedicel in fruit 0.7-0.8 cm long, ca. 1.8 mm thick, articulate above base, glabrate; torus in fruit depressed-globose or unex- panded, glabrate. Monocarps up to ca. 15, the seed-containing portion 7-9 mm long, 4-5 mm wide, ellipsoid to pyriform, apiculate at apex, stipe 4-5 mm long, 0.5-0.7 mm wide, glabrate or with a few sparse hairs, pericarp 0.2-0.3 mm thick. Seed 7-9 mm long, 4-4.5 mm wide, ellipsoid- fusiform, dark brown, sulcate and somewhat pitted, shiny, the exostome oblique, paler than seed coat. Additional specimens examined. VENE EZUELA. nearly opposite Cucurital de Caname, 2 May 1979 (fl, fr), Davidse et al. 17066 (MO); Dept. Atabapo, Caro Caname, sabanas de Cucurital, Apr.-May 1979 (bud, fr), pend et e died Mena Dept. Atabapo, sabana abierta a un al S del medio Caño Caname, 10 Mar. 1980, por 5146 (VEN); Dept. Atures, sabana ubicada en la ribera izquierda (S) del Río Guayapo medio, 27 July 1980 (fr), Huber & Tillett 5509 (VEN). This taxon was twice proposed as new, but, as far as is known, neither Fries nor Aristeguieta ever San Fernando de Atabapo, where it is found in shrub islands, sometimes on white sand, at 95- . It is readily distinguished from G. ma- guirei, i, which i is sympatric in similar habits, by the shorter pedicel, papillate anther connectives, and larger leaves. It may also be confused with large- leaved forms of G. schomburgkiana C. Martius, which differs by having setulose anther connec- tives, and monocarps with the seed-containing por- tion 7-10 mm long and 5- : mm wide on Np that occasionally reach 5 mm long ( ly 0.5-2 mm long). Guatteria clusiifolia D. M. Johnson & N. A. Murray, sp. nov. TYPE: Guyana. Upper Ma- zaruni River Basin, Mt. Ayanganna, NE side of mountain, 800-900 m, 2 Aug. 1980 (8), Tillett, Tillett & Boyan 45009 (holotype, NY). 600 Annals of the Missouri Botanical Garden Inter species Guatteriae sectionis Mecocarpi forsan G. durae R. E. Fries arctissime similis, sed ab ea foliis glabratis ad apicem truncatis et ко zt iid cuspidatis (3 mm), petalis 0.6-0.9 mm crass ocarpiis 22-27 mm longis et 13-14 mm ven а ? mm longis sustentis divergens. Tree 25-35 m tall, diam. 40-80 cm at base. Twigs initially with tightly appressed pubescence, then glabrate. Lamina of larger leaves 12-15 cm long, 6.8-7 ceous, obovate; base cuneate and decurrent; apex ‚7 ст wide, subcoriaceous to coria- truncate, emarginate or short-cuspidate, the cusp to З т appressed pubescence abaxially, becoming sparsely m long; surface even, initially with tightly pubescent to glabrate; midrib impressed adaxially, raised abaxially; secondary veins 13-17 per side, diverging at 60-70? to midrib, brochidodromous, loops closing at 3-4 mm from the margin; sec- ondary veins plane, slightly raised or slightly im- pressed adaxially, raised abaxially; higher-order Petiole 14-20 mm long, 2.5-3 mm wide, with a prominent involute wing, sparsely veins indistinct. pubescent to glabrate. Pedicels axillary, 1.4—1.7 mm wide, obliquely articulate 3-5 mm above the base, with persistent tightly ap- em long, 1.5 pressed pubescence, bracteoles caducous from be- low articulation. Sepals 4-6 mm long, 4-6 mm wide, connate at base, subcoriaceous, broadly tri- angular, apex revolute, appressed-pubescent abax- ially, sparsely so or glabrate adaxially. Outer petals 18-21 mm long, 7-12 mm wide, subcoriaceous, elongate-rhomboid or oblong, rounded or obtuse at apex, adaxially pubescent, hairs densest at base, abaxially pubescent, with a glabrous patch at base; inner petals 14-23 mm long, 6-9 mm wide, oblong or oblanceolate, rounded or obtuse at apex, pu- bescence as in outer petals. Stamens 1.2 mm long, long-trapezoidal, connectives truncate at the apex, papillate. Carpels 13-16; ovaries sericeous. Torus hemispheric, 4 mm diam., glabrate. Pedicel in fruit 1.9 cm long, 3.5 mm wide, longitudinally wrinkled and with a few small lenticels. Monocarps up to 5, the seed-containing portion 22-27 mm long, 13- 14 mm wide, ellipsoid to elongate-ellipsoid, apex rounded, slightly apiculate, stipe 4 mm long, 3- wide, ellipsoid-fusiform, dark brown, sulcate, ru- gulose, shiny, the exostome paler than seed coat. Additional specimen examined. GUYANA: Upper Mazaruni River Basin, along Kako River, 25 Sep. 1960 (fl), Tillett & Tillett 45531 (NY) This distinctive species, a large tree with cori- aceous paddle-shaped leaves and large thick-walled monocarps, is known from riverine and lower mon- tane forest up to 800-900 m on and near Mt. Ayanganna in western Guyana. It may perhaps be expected in the eastern Venezuelan Guayana as well. It is most similar to Guatteria dura R. Fries, but that species has leaves pubescent be- neath, the leaf apex rounded with a cusp 0.5-1.5 cm long, thicker petals, and monocarps with the seed-containing portion 13-17 mm long and 8- 10 mm wide and with stipes 9-12 mm long. We thank the staff of the Missouri Botanical Garden, in particular Paul Berry and Bruce Holst, for the opportunity to examine the Guatteria ma- terial from the Venezuelan Guayana. We also thank the curators of F, MO, NY, US, and VEN for making specimens available to us. Rupert Barneby corrected the Latin diagnoses. —David М. Johnson and Nancy A. Murray, De- partment of Botany-Microbiology, Ohio Wesley- an University, Delaware, Ohio 43015, U.S.A. A NEW COMBINATION IN CHASMANTHIUM (POACEAE) Yates (1966a, b) treated Chasmanthium Link as a genus distinct from Uniola L., based primarily on differences in embryo structure, leaf anatomy and epidermal micromorphology, and chromosome number. He recognized five species within Chas- manthium, C. latifolium (Michaux) Yates, C. lax- um (L.) Yates, C. sessiliflorum (Poiret) Yates, C. nitidum (Baldwin) Yates, and С. ornithorhynchum (Steudel) Yates. During preparation of a treatment of this genus for the Manual of North American Grasses, it became evident to me that two of these species, С. laxum and С. sessiliflorum, were mor- phologically vey similar and showed significant overlap in several quantitative features, including leaf length and width, panicle length, number of florets per spikelet, and spikelet length and width. The only consistent qualitative differences between them are the pubescent leaf collars and more or less divergent panicle branches of С. sessiliflorum as contrasted with the glabrous leaf collars and appressed panicle branches of axum. In ad- dition, the two taxa are almost completely sym- patric; they extend from New Jersey to Texas and Oklahoma. Given their similarities and sympatry, I feel that these two taxa are best treated as sub- species of C. laxum rather than as distinct species. The new combination for C. sessiliflorum is pre- sented here. Chasmanthium laxum (L.) Yates subsp. lax- um. Holcus laxus L., Sp. Pl. 2: 1048. 1753. TYPE: U.S.A. Virginia: Clayton 589 (lecto- type, here designated, LINN, microfiche con- sulted; isolectotype, BM). Uniola laxa (L.) Britton, Sterns & Pogg., Prelim. Cat. 69. 888 Uniola procs Michaux, Fl. Bor. Amer. 1803. in umbrosis sylvarum, a ee ad See (holoty e, P not seen). Chasmanthium acile аа ра, Hort. Berol. 1: 159. sge Uniola ge B шщ sh, Fl. Amer. Sept. 1 82. 1814, nomen nu Uniola infor Benke, Rhodora 31: 148. 1929. TYPE: U.S.A. Tennessee: Memphis, 1928, Benke 4874 (holotype, F not seen; isotype, BM). Chasmanthium laxum (L.) Yates subsp. ses- siliflorum (Poiret) L. Clark, comb. et stat. nov. Uniola sessiliflora Poiret in Lam., En- сус]. 8: 185. 1808. TYPE: “Cette plante n'a ete communiquee par M. cueillie dans la Caroline" Chasmanthium sessiliflorum (Poiret) Yates, Southw. Naturalist 11: 426. 1966. id P va е Bull. Torrey Bot. Club 21: 5 U; Georgia: De Kalb Co., А Stone ioe 1893, Small s. n. (lectotype, here designated, US). LITERATURE CITED Yates, Н. O. 1966a. Morphology and cytology of Uni- ola (Gramineae). Southw. Naturalist 11: 145-189. . 196 Revision of grasses traditionally re- ferred to Uniola, II. Chasmanthium. Southw. Nat uralist 11: 415-455 —Lynn С. Clark, Department of Botany, lowa State University, Ames, lowa 50011, U.S.A. ANN. Missour!I Вот. Garb. 77: 601. 1990. THE SOUTH AMERICAN EREMODRABA (BRASSICACEAE) Eremodraba O. E. Schulz was originally de- scribed as a monotypic genus (Schulz, 1924) that was considered to be closely related to Stenodraba O. E. Schulz, Alpaminia O. E. Schulz, Pelagatia O. E. Schulz, and Weberbauera Gilg & Muschler (Schulz, 1936). The last four genera have been critically evaluated by Al-Shehbaz (1990); there 1 concluded that all four constitute а well defined, monophyletic genus recognized under the earliest name, Weberbauera. The recognition of Eremodraba as a distinct genus is strengthened by the discovery of E. schulzii. The genus consists of glabrous annuals with sag- ittate-amplexicaul cauline leaves, conspicuously flattened fruits, yellow flowers, and filaments with papillose, dilated bases. In my opinion, Eremodra- ba is unrelated to Weberbauera. The latter in- cludes cespitose, usually pubescent perennials with petiolate or sessile leaves that are neither sagittate nor amplexicaul, terete or slightly flattened fruits, white flowers, and glabrous filaments. Eremodraba apparently has no close relatives. It resembles only superficially the Peruvian monotypic Dictyo- phragmus O. E. Schulz that differs from Eremo- draba in having conspicuously nerved septa, broadly winged seeds, and accumbent cotyledons. Eremodraba was erroneously reported as suf- frutescent herbs (Schulz, 1924; Macbride, 1938). There is a poor representation of the genus among the holdings of the major herbaria consulted. Both species of Eremodraba are apparently very rare and are restricted to mid altitudes in the deserts of nothern Chile and southern Peru. The following account aims to provide the basis for a better understanding of this very rare genus. SYSTEMATIC TREATMENT Eremodraba О. E. Schulz, Pflanzenreich IV. 105(Heft 86): 362. 1924. TYPE: E. tissima (Philippi) O. E. Schulz. intrica- Glabrous, somewhat fleshy annual herbs; stems much branched above. Cauline leaves auriculate to sagittate-amplexicaul. Inflorescences ebracteate, corymbose racemes, elongated considerably in fruit. Sepals oblong, glabrous, nonsaccate at base, erect to spreading. Petals yellow, spatulate. Stamens 6; filaments dilated and papillose at base; anthers ob- long. Nectar glands confluent, subtending the bases of all stamens. Fruits oblong-lanceolate, dehiscent, flattened parallel to the septum, glabrous, straight or falcate; septum complete; stigma capitate, much broader than style. Seeds oblong, uniseriate to sub- biseriate; cotyledons incumbent. KEY TO THE SPECIES OF EREMODRABA Fruits straight; fruiting pedicels divaricate, straight, 6-8 mm long; rachis of infructescence stra aig tou E. schulzii Fruits falcate; fruiting pedicels reflexed, caray curved, 2-3.5(-5) mm long; rachis of infructescenc strongly geniculate „1 2. E. intricatissima 1. Eremodraba schulzii Al-Shehbaz, sp. nov. TYPE: Peru. Depto. Arequipa: Yura, 2,500 m 18 May 1957, R. Hirsch P508 base GH). Figure 1. Herba annua glabra; folia basales pinnatisecta, breve petiolata; folia caulina о integra vel dentata, sagitta- to-amplexicaula; racemi ebracteati; sepala oblon nga, erec- ta, 2.5-3 mm longa; pe ea flava, spathulata, 3-4 mm longa; filamentae a basi papillosae; pedicelli fructiferi di- varicati, recti, 6-8 mm longi; siliquae anguste oblongo- lanceolatae, compressae, rectae, 11-17 mm longae, 2- 2 m latae; stylus 0.2-0.3 mm chy semina oblonga, 1.1-1.2 mm longa, 0.6-0.7 mm lata Annual herbs, glabrous gud Stems erect, branched above, 1-6 dm . Basal leaves not rosulate, petiolate, ule. des 7 cm long; lateral lobes oblong to linear, 0.5-2 cm long, 0.5-2 m wide. Upper cauline leaves narrowly linear, strong- ly sagittate-amplexicaul at base, entire to dentate or rarely pinnatisect, 2-3.5 cm long, 0.5-1.5 mm wide. Inflorescences ebracteate, corymbose ra- cemes, elongated considerably in fruit; rachis of infructescence straight. Sepals oblong, erect, scar- ious at margin, glabrous, 2.5-3 mm long, 1-1.: mm wide. Petals yellow; spatulate, attenuate to clawlike base, 3-4 mm long, Filaments erect, dilated and papillose at base, 2- 2.5 mm long; anthers ovate, 0.5-0.6 mm long. Fruiting pedicels divaricate, straight, glabrous, 6- 8 mm long. Fruits narrowly oblong-lanceolate, flat- tened parallel to septum, straight, 11-17 mm long, m wide; valves glabrous, obscurely nerved, mm wide. acute-acuminate at apex, obtuse at base; septum complete; style 0.2-0.3 mm long; stigma capitate, ANN. Missouri Bor. Garb. 77: 602-604. 1990. Volume 77, Number 3 1990 Notes 603 FIGURE 1. ibn sc id — a. Portion of plant. —b. Lower cauline leaf. — c. Petal. —d. Stamen. — e. Fruit. Scales a, b = 1 cm; c-e = . Drawn from the holotype except b, which was drawn from Sandeman 3944. 604 Annals of the Missouri Botanical Garden much wider than style. Seeds oblong, somewhat biseriately arranged, 1.1-1.2 mm long, mm wide; cotyledons incumbent. Additional specime ned. PERU. DEPTO. ARE- QUIPA: southern slope of Chachani Mt., 3,050 m, Hinck- ley 70 (B, GH); near the old road from Arequipa to Mollendo, Sandeman 3944 (OXF). Eremodraba schulzii, named in honor of Otto Eugen Schulz (31 Oct. 1874-17 Feb. 1936), an outstanding student of the Brassicaceae, was first named by Schulz (1936) as “E. ? hinckleyana.” However, the latter name was invalidly published because it was not accompanied by Latin descrip- tion or diagnosis. Furthermore, Schulz treated Ere- modraba as monotypic and was doubtful in his assignment of E. hinckleyana to the genus because I have de- scribed the species as new and chose a type with fruits and seeds, rather than to validate Schulz’s name, which was based on a flowering material. the material examined had no fruits. 2. Eremodraba intricatissima (Philippi) O. E. Schulz, Pflanzenreich IV. 105(Heft 86): 363. 1924. Draba intricatissima Philippi, Anal. Mus. Nac. Chile Bot. 8(2): 5. 1891. Sisym- brium intricatissimum (Philippi) Reiche, Fl. Chile 75. 1895. Hesperis intric к E lippi) Kuntze, Revis. Gen. Pl. 3(2): 5. TYPE: Chile. [Region I] Tarapacá e prov- ince): between Mocha and Guavina, C. Rah- mer s.n., 12 Mar. 1885 (holotype, SGO 63194 seen). Annual herbs, glabrous m Stems erect, much branched above, 1-5 igh. Lowermost leaves not seen; middle and upper cauline leaves somewhat fleshy, oblong to linear, auriculate to sagittate-amplexicaul at base, usually entire, 0.7— 5 cm long, 1-4 mm wide. Inflorescences ebrac- teate, corymbose racemes, elongated considerably in fruit; rachis of infructescence conspicuously ge- niculate. Sepals glabrous, oblong, erect to spread- ing, caducous or persistent, scarious at margin, 1.8-2 mm long, 0.9-1 mm wide. Petals spatulate, yellow, 2.5-3 mm long, 0.5-0.8 mm wide. Fila- ments dilated and papillose at base, 1.5-2 mm long; anthers ovate, 0.6-0.7 mm long. Fruiting pedicels strongly curved, reflexed, 2-3.5(-5) mm long. Fruits oblong-lanceolate, falcate, flattened 10(-14) mm long, 1.5-2 mm wide; valves glabrous, obscurely nerved; parallel to the septum, (3-)5- septum complete; style 0.1—0.2(-1) mm long; stig- ma capitate, much broader than style. Seeds ob- long, uniseriate to subbiseriate, 1.2-1.3 mm long, 0 7 mm wide; cotyledons incumbent. Additional specimens examined. CHILE: [Región I] Tarapacá (as province), Pica, 1,400 m, Werdermann 751 (E, G, GH, MO, UC); Depto. Tarapacá о. province Iquique), Noasa-Mamina, 2,700 m, Werdermann 1573 (NY); a Philippi s.n., 1888 (B) There is considerable variation in the orientation and duration of sepals and in the length of fruits of Eremodraba intricatissima. Because of the scarcity of material at my disposal, I have refrained from recognizing these variants formally. LITERATURE CITED AL-SHE “HBAZ, I. A. A revision of Weberbauera 21-250. MACBRIDE, J. Field ae Nat. Hist. bas 13(2): 937-983. ScHuLz, O. К. 1924 ciferae-Sisymbrieae. /n: m Pflanzenreich TV. 105(Heft 86): 1-38 i Crucifer ms (editor), Die Nath "Pflanzenfamilien, 2nd edition. 17B: 227 — [hsan A. Al-Shehbaz, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166- 0299, U.S. A. BOOK REVIEW Yungjohann, John С. (edited by С. Т. Prance, with an epilogue by Yungjohann Hillman). White Gold, the Diary of a Rubber Cutter in the Amazon 1906-1916. Synergetic Press, P.O. Box 689, Oracle, Arizona 85623, U.S.A. 1989. Frontispiece + 103 pp., illus., ISBN 0-907791-16-6. Retail price: $7.95. Not a diary in the sense of a day-by-day log, White Gold is an account of the experiences of New York native John Yungjohann's ill-advised plunge into the upper Amazon as a rubber tapper. Ghillean Prance edited the first-hand text lightly, and augmented it with an introduction that lends perspective. Using a glossary and notes, he defined terms, identified plants and creatures, and clarified obscure points. Numerous black and white pho- tographs taken by Dr. Prance and a map enrich the book. John Yungjohann made his way up the Amazon and Rio Purus to the mouth of the Rio Xapuri in far-western Brazil. Here a series of surprises began when he was trapped into the purchase of excess supplies at inflated prices and teamed up with six "a couple of hours" up the river turned out to be 45 days up the river. After teaching the team the basics, the guide disappeared. Before long all seven adventurers were malarial, with the outcome that the hero, himself near death, variously disposed of the remains of his less fortunate companions. He survived and more or less flourished until beriberi forced him to abandon a subterranean treasure trove of rubber and return to New York, where he resumed his interrupted career as a tilesetter. This is no place to summarize the plot. Suffice it to note that Mr. Yungjohann described rubber tapping, malaria, the experience of being discov- ered living alone by the local Indians (who turned out to be most hospitable), a system of enslaving rubber tappers through debt, wildlife encounters, and a series of incredible adventures. The book reads like a novel, and there is even a (very brief) romance, in which boy did not get girl. The only flaw is that the story is intriguing and sketchy at the same time, leaving the reader hun- gry for more. It is thoroughly pleasing, easy read- ing, and would make a good supplement to a high school or college course that deals with economic plants or the Amazon, especially with the ecology and economy of the region in the news, and with the tragedy of Chico Mendes a fresh memory. Any person interested in these subjects or neotropical botany, rubber, tropical diseases, human nature, or adventure will find the book a delight. Plan on reading it in one sitting.— George K. Rogers, Mis- souri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. Present address: Cox Arboretum, 6733 Springboro Pike, Dayton, Ohio 45449, U.S.A. ANN. MISSOURI Bor. GARD. 77: 605. 1990. Volume 77, Number 2, pp. 225-425 of the ANNALS Ei THE MISSOURI BOTANICAL GARDEN was published on May 1, 1990. Rolf Dahlgren (1932-1987) SYSTEMATICS AND EVOLUTION OF THE MONOCOTYLEDONS The following seven papers were part of the symposium Systematics and Evolution of the Monocotyledons, which was sponsored by the American Society of Plant Systematists and the Systematics Section of the Botanical Society of America and was held at the American Institute of Biological Sciences meetings at the University of Massachusetts, Amherst, August 1986. The publication of these proceedings is supported in part by National Science Foundation grant BSR- 9005141. The symposium papers are dedicated to the memory of Rolf Dahlgren, noted student of monocot systematics and evolution. nn New Publications from the Missouri Botanical Garden Thesaurus Dracularum, 3 Carlyle A. Luer, with German translations by Fritz Hamer. Watercolor paintings by Stig Dalstróm. 1990. Thesaurus Dracularum is a popular monograph of the orchid genus Dracula. Each of the 80-odd known species 18 illustrated by a watercolor painting accompanied by descriptions, discussions, and distributions in both English and German, with distribution maps and black and white line drawings. In elephant format. Standing orders available. The third of six fascicles (fascicles 1 and 2 are also available at the price listed here). U.S. orders, $40 each, postage paid; non- u S. orders, $42, postage paid. Index to Plant Chromosome Numbers 1986-1987 Edited by Аан Goldblatt and Dale E. Johnson. Monograph in Systematic Botany from the Missouri Botanical Garden, Number 30. 1990. The Index to Plant Chromosome Numbers is compiled by an international committee and collated and edited by Peter Goldblatt and Dale E. Johnson, Missouri Botanical Garden. It is indispensible to those engaged in a variety of botanical studies, ranging from systematics and evolution to plant breeding, forestry, and horticulture. Counts for all plant groups—algae, cual bryophytes, pteridophytes, and spermatophytes —are included and arranged alphabetically by family, genus, and species. Each count includes the name of the taxon as used in the original report, the number reported, and аа to ME original report. References are contained in a bibliography that contains citations at the level of about 500 per year. 243 pages. $15, plus postage. Botanical Research and Management in Galápagos Edited by J. E. Lawesson, O. Hama nn, G. Rogers, G. Reck and H. Ochoa. Monograph in Systematic Botany from the Missouri Botanical Garden, Number 32. 1990. This volume contains the proceedings of the UR pad Workshop. on PESEN Research and Management in Galápagos, organized by the Charles Darwin Research Station and the Galápagos National Park Service, held in April 1987 at the Charles Darwin Research se Sant "Cia. Galápagos, Ecuador. The 34 papers contained in this — discuss various aspects of conse g the Galápagos flora and Metus: 297 + pages. Soft cover. illustrated, 990, plus postage. кын Polyglottum Doka, A multilingual "dua for С: КС by R. E. Magill. Monograph in Systematic Botany from the Missouri Botanical Garden, Number 33. 1990. as er eee definitions for nearly 1,200 terms in four languages — English, French, German, and lish terms are зене into Japanese, Latin, and Russian, This volume will be relire : о паь te , and students as a means of learning the specialized vocabulary of bryology and goose the v varying usages of terms. 291 pages. Paper cover, $12, plus postage, ийе an order, vend check or money order in U.S. би; BITE TE U.S. shipments: odd“ one book and $.50 for each additional book; non-U.S. shipments: add $2.50 for one book, and $.50 for co book. Orders should be > prepaid a ы $1. 00 {ее will be added to ə orders topire. Бой. No ^r Ei Є" until payment i is denen M Y. М. : R oe y P. 0. Box 299 ^ ek Louis, MO 63166.0299, USA a Please send the Publications circled above Mc ; 2 t 2 Payment enclosed. . e o (100 fee il be + : added to tot А кыш: бошпу, уз Ш E Е 2 as 118) | Prices ARE Sumecr To Cuance нот Norice CONTENTS. "The Publication of the Flora of China Will Be a Great Contribution to the Scientific Circles of the World Zheng-Yi Wu ... T р - 42 Systematics of Lapeirousia (Iridaceae— -Ixoideae) i in раа Aftica Peter Goldblatt ... 43 ў Léon Croizat’s Plant Collections from the Franco-Venezuelan Expedition to the Headwaters Же of the Río Orinoco Bruce К. Holst & Carol A. Todaiü oa Ел КЫ 4 б. А Phylogenetic Reevaluation of the Old rara E of Fuchsia ia Onagraceae) Jorge 2 : V. Crisci & Paul E. Berry .... AI $l _ Phylogenetic Implications of балаа DNA TES ; Site Variation in the Plant Family я m Onagraceae Jorge V. Crisci, Elizabeth A. Zimmer, Peter C. Hoch, petis &, ae Johnson, Christy Mudd & N. S. Pan -ai | 52 Же А. „ы Neo Cai irding the Corona i in the СИ УСО Мата Luiza de Men af Oot nezes & ‘Joao бейш, EIA ыза С AAA анаа IL > Additional Transfers. E Asiatic Machilus Soria Nees, non Desrouesenus, to Persea Miller _ (Lauraceae) | A. J. С. Н. Койегіане nani par rindo лел: New or Noteworthy Orchids from the Venezuelan Flora. уш. New Species ind Combinations .. from the Venezuelan Guayana. Germán Carnevali & Іобп Ramirez .. = -Pollination Ecology of Seven Species of Bauhinia de бороне Caesalpinioideae) : - Omaira Hokche & Nelson Ramirez ... e Мы i -Adiciones a las Papilonadas de la Flora de Nicaragua) ya una S NM Combinacíon Para EA Oaxaca, México | Mario Sousa jocis A New Combination in Dioclea Kunth (Fabaceae-Dioeeinae) froin the Clarice ior of oe ets Flora В Brasiliensis . Richard. H. Maxwell - —— Now Tai of Dioclea Kunth D abaceae-Dioelenae) from the Venezuelan Guay ota Richard H. Maxwell .. : ES ‚ and New Combinatior Annals of the Missouri Volume 77 Number 4. Volume 77, Number 4 Fall 1990 Annals of the Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, con- tributed from the Missouri Botanical Garden, St. Louis. Papers originating outside the Garden will also be accepted. Authors should write the Editor for information concerning arrangements for publishing in the ANNALS. Instructions to Authors are printed in the back of the last issue of each volume. Editorial Committee Marshall R. Crosby -Pato > ; Missouri Botanical Garden Amy Scheuler _ Managing Editor, | Missouri Botanical Garden | Glenda Nau | Magdalen Lampe Publications Staff Gerrit Davidse Missouri Botanical Garden John D. Dwyer Missouri Botanical Garden & Saint Louis University Peter Goldblatt | Missouri Botanical Garden : Dale E. Johnson Missouri Botanical Garden Henk van der Werff — Missouri Botanical Garden _ For subscription information contact Department Eleven, Р.О. Box 299, St. Louis, MO 63166. Sub- - scription price is $75 per volume U.S., $80 Canada . and Mexico, $100 all other countries. Four issues - : S per volume. — Missouri Botanical Garden, 2345 Tower Grov enue, St. Louis, MO 631 10. Second class ae paid at St. Louis, MO and additional mailing offices [THIS PUBLICATION 15 PRINTED ON ACID-FREE PAPER. The ANNALS OF THE MISSOURI BoTANICAL GARDEN (ISSN 0026-6493) is published quarterly by > POSTMASTER: Send address changes to фина" THE Missouri BOTANICAL GARDEN, /. Eleven, Р.О. Box 299, St. Louis, MO 63160. EA E II A L Volume 77 Number 4 1990 Annals of the Missouri Botanical Garden NZ PHYLOGENY AND CLASSIFICATION OF IRIDACEAE! Peter Goldblatt? ABSTRACT A cladistic analysis of Iridaceae, a family of some 1,630 species and ca. 77 genera, and the closely allied Geosiris and sometimes accorde and /sophysis, both monotypic a family status, suggests a phylogeny in which there are four major lineages, recognized as subfamilies. Characters used in the analysis include vegetative and floral morphology, anatomy, embryology, pollen ultrastructure, chromosome cytology, and ауар and amino acid chemistry. Iso- physidoideae, with a superior ovary, inclu Aristea, three a ids, and inflorescences. Thre e only the Tasmanian Isoph sis. N the Afro-Madagascan ia, and Ceosiris The last- mentioned, a sapro re recognized in Ixioideae and four in Iridoideae in both of which some subtribal groupings are suggested. Described formally here are Nivenioideae and Pillansieae. Iridaceae are a relatively large family of petaloid monocots (Liliiflorae sensu Dahlgren et al., 1985) comprising over 1,630 species in ca. 77 genera. Although distributed worldwide, the family has a marked concentration on the southern continents and the major center of radiation in Africa south of the Sahara. Iridaceae are easily сы among uitant leaves, the monocots by I having 1 flowers with three stamens, and, with the exception of the monotypic Tasmanian /sophysis, an inferior ovary. Isophysis (Fig. 1A) has in the past been assigned че ы to Isophysidaceae, Hypoxida- ceae, or Liliaceae-Melanthioideae as well as to Iridaceae (Goldblatt et al., 1984). Such is the mor- phological distinctness of Iridaceae that there is virtually no controversy over their status and cir- cumscription, except for the treatment of Isophysis and the monotypic saprophyte Geosiris from Mad- agascar, both of which have been regarded as sep- arate families (Jonker, 1939). ' Supported by grants DEB 78-10655 and DEB 81-19292 from the U.S. National Science Foundation. I thank mes Walker or encouraging me to undertake this study; Bruce L. Stein for assistance with the cladistic analysis; Раша Rudall, кулш Williams, and В. L. Burtt for their helpful comments on the manuscript and in its preparation; and Margo Branch Myers for the original illustrations used her ohn 2 B. A. Krukoff Curator of African Botany, Missouri Botanical Garden, P. О. Box 299, St. Louis, Missouri 63166, S.A. ANN. Missouri Вот. GARD. 77: 607-627. 1990. 608 Annals of the Missouri Botanical Garden Iridaceae are usually assigned to the order Lil- iales (the definition of which varies considerably) but have also been treated in Iridales, either alone (Hutchinson, 1934, 1973), or with Geosiridaceae, Burmanniaceae, and Corsiaceae (Takhtajan, 1969). The unanimity of opinion concerning the status and circumscription of Iridaceae has largely ob- scured any critical appraisal of the relationships of this family. Colchicaceae (Liliaceae- Melanthioi- deae in part) have been suggested as immediately allied to Iridaceae (Takhtajan, 1969, 1980), and Takhtajan (1980, 1985) has allied Iridaceae with Tecophilaeaceae in suborder Iridineae, a relation- ship first espoused by Hutchinson (1934) but dis- counted by Goldblatt et al. (1984). Dahlgren & Rasmussen (1983) have allied Iridaceae with the Pacific family Campynemataceae and the two to Colchicaceae. New evidence (Goldblatt et al., 1984; Goldblatt, 1986b) now makes this also seem un- tenable. Infrafamilial phylogeny and classification of Iri- daceae, the emphasis of this paper, have varied with each major treatment of the flowering plants (or monocots alone), but most systems have con- sistently accepted the existence of one of the main subfamilial categories, subfamily Ixioideae (tribe Ixieae of Bentham & Hooker, 1883; Diels, 1930). There has been little agreement, however, over the circumscription and rank of the other half of the family, here treated as subfamilies Isophysidoideae, Nivenioideae, and Iridoideae. Bentham & Hooker (1883) admitted two more tribes (their major in- frafamilial category), Moraeeae (now lrideae) and Sisyrinchieae, the latter including genera here re- ferred to subfamilies Iridoideae and Nivenioideae. The rather different treatment of Pax (1888) rec- ognizes the two major subfamilies admitted here, Ixioideae and Iridoideae (the latter including Niv- enioideae as a tribe) and a third, Crocoideae, for four acaulescent genera, Romulea, Syringodea, Crocus, and Galaxia, now confidently referred either to Iridoideae or Ixioideae (Goldblatt, 197 1). The treatment of Iridaceae by Hutchinson (1934, 1973) is similar to that of Bentham & Hooker but admits 11 tribes, essentially recognizing each of Bentham & Hooker's three tribes and subtribes at the same rank. Hutchinson was, however, the first to include /sophysis in Iridaceae, as the only mem- ber of Isophysideae. Recent treatments of Takhtajan (1980: 310) and Thorne (1983) include Geosiris in Iridaceae as a subfamily, Takhtajan also including /sophysis and Campynemataceae (see Dahlgren et al., 1985) in Iridaceae as subfamilies. The recent and thor- ough studies of the monocots by Dahlgren and co- workers recognize Geosiridaceae as distinct from Iridaceae, the latter (Dahlgren et al., 1985) in- cluding five subfamilies, Isophysidoideae, the no- men nudum Aristeoideae (Nivenioideae here), Sis- yrinchioideae, Iridoideae, and Ixioideae. This paper represents the first objective phy- logenetic classification of Iridaceae using modern cladistic methods. It is also the first to use data extensively from fields other than morphology, adapting information from chromosome cytology, flavonoid and amino acid chemistry, anatomy, and pollen morphology. The results confirm that /so- py is шау best included in Iridaceae in a mily Isophysidoideae and suggest that Соз be теү: in Iridaceae—Nivenioideae. METHODS OF ANALYSIS Cladistics affords the most objective and critical method of assessing phylogeny, and the results of my cladistic analysis of Iridaceae are presented in the following Aon in several cladograms. The cladograms (Figs. 3-5) were constructed manually folowing concepts of groping: by "ai derived y Hennig (1966) together with the principle of parsimony as adapted by several botanists (Bremer, 1976; Funk, 1982; Dahlgren & Bremer, 1985; Goldblatt, 1985). The manually generated cladograms were tested using the PAUP Program (Swofford, 1985), which produced the same results. The polarities of the characters were determined either by outgroup comparison or following widely accepted general character trends in the monocots or flowering plants. The characters used for the cladistic analysis are discussed in detail below, and are outlined in Table 1. Anatomical data are taken from studies by Chea- dle (1963) and Rudall (1983, 1984, 1986) and pollen data from the extensive studies of Schulze, summarized in 1971 (Schulze, 1971). Flavonoid data have only recently become available and in- formation is adapted from a wide survey by Wil- liams et al. (1986). Other important phytochemical data are from studies by Larsen et al. (1981, 1987). Chromosome cytology is comparatively well known for Iridaceae, but the unusually variable karyolog- ical data are of limited use above subtribal levels owing to the difficulty in establishing basic numbers for higher ranks The characteristics of the families and orders of the Liliiflorae are taken largely from the several recent works of Dahlgren and his co-workers (Dahl- gren & Clifford, 1982; Dahlgren & Rasmussen, 1983; Dahlgren et al., 1985; Dahlgren & Bremer, 985). Volume 77, Number 4 1990 Goldblatt 609 Phylogeny and Classification of Iridaceae CIRCUMSCRIPTION OF TRIBES Given the present state of our knowledge, it is impractical to deal with individual genera of Iri- daceae for this type of analysis. Instead, I have chosen apparently natural generic groups (some- times corresponding to tribes or subtribes) as work- ing taxonomic units. It is clear that three of the four tribes of Iridoideae— Irideae, Mariceae, and Tigrideae—are natural and monophyletic groups, while the nature of the fourth, Sisyrinchieae, is less certain. Although it may be an unnatural alliance, for the purposes of this study it is regarded as monophyletic, and the data so far accumulated support this. However, an important question still remains as to whether Orthrosanthus and Libertia (both Pacific and South American) and Bobartia (South African) are in fact related to the core genera of Sisyrinchieae (all New World). Nivenioideae may also be found to be unnatural, but no convincing data have yet come to light that indicate an alternative treatment. In Ixioideae I had no preconceived ideas about tribal groupings when I undertook this study, but Watsonieae have emerged as a tribe distinct from Ixieae, which still comprise a large number of genera that may in the future be found to be better treated in more than one tribe. It also became clear from the pre- liminary analysis that the monotypic Pillansia is very isolated in Іхіоійеае and probably the sister group of the rest of the subfamily (Fig. 5). It is referred to a monotypic tribe, a decision that could not be avoided. Geosiris has all the derived states of Nivenioi- deae as well as some of its own, and according to the data available, is the sister genus of Aristea. Isophysis merits its own subfamily, and in this analysis emerges as the sister taxon of the rest of the Iridaceae. I must, however, emphasize that the embryology of both Isophysis and all Nivenioideae except Geosiris (Goldblatt et al., 1987) is unknown, and has been assumed to correspond with Geosiris (presence of a parietal cell in the ovule; secretory tapetum; successive microsporogenesis; helobial endosperm development) and not the embryologi- cally better known Iridoideae and Ixioideae, which appear to be specialized in having simultaneous microsporogenesis and nuclear endosperm forma- tion (the polarity of these two characters is un- certain). Geosiris and Isophysis have not been examined for free amino acids. CHARACTER ANALYSIS The features of systematic and phylogenetic sig- nificance are presented below under headings of Morphology, Embryology, Cytology, Anatomy, and Phytochemistry. the more important char- acteristics are discussed at length. The reasons for assigning character polarities are indicated unless they are self-evident. The data are summarized in Table 1 and are used in the construction of the cladograms (Figs. 3-5). MORPHOLOGY Rootstock. The basic type of rootstock in Iridaceae is almost certainly a creeping persistent rhizome; bulbs and corms are independently derived from a rhizome. All members of Tigrideae have a bulb covered by dry brownish tunics. Elsewhere in the family, bulbs occur in /ris subgenera Scorpiris, Xiphion, and Reticulata, and they may have evolved independently in each (Mathew, 1981, and pers. comm.). Corms are basic in Ixioideae and also occur in the several African genera of Irideae subtribes Homeriinae and Ferrariinae (Lewis, 1954; blatt, 1976). In Ixioideae (Fig. 8) the corms have a distinct stele, produce roots from the base, and are usually composed of several internodes (Lewis, 54; de Vos, 1977; Goldblatt, 1982a). Devel- opment proceeds in two different ways. In Wat- sonieae and Pillansieae the corm is formed entirely from an axillary bud (Watsonia-type sensu de Vos, 1977) at the base of the terminal flowering stem, and the corm’s shoot primordium is apical. In Ixieae, by contrast, the corm develops at least partly from the base of the flowering stem (/xia-type of de Vos, 1977), and a bud in the axil of the upper node of the corm produces the following season’s growth. In the Watsonia-type the new corm is attached to the base of the flowering stem and thus lateral to it, and in the /xia-type the new corm is basal to the flowering stem, and situated below its insertion. In both cases the flowering axis is terminal and the growth pattern is sympodial and analogous to the sympodial growth of a rhizome where the flow- ering stem is terminal and buds lateral to it continue the next growth flush. The /xia-type of corm ap- pears to be derived and is here regarded as the major synapomorphy uniting the genera assigned to Ixieae. In Irideae the corms have a diffuse stele (de Vos, 1977), produce roots (Fig. 6) from the apical bud (Goldblatt, 1987), and, except in Ferraria (only genus of Ferrariinae), are of a single inter- node. The Homeria-type corm is a major synap- omorphy for this large subtribe of Irideae. Leaves. The basic leaf type is ensiform and iso- bilateral (Arber, 1921) with an open sheathing 610 Annals of the Missouri Botanical Garden TABLE 1. Characters used in the cladograms (Figs. 3-5). The derived сонар ante states are listed first followed by the presumed ancestral aa condition. Characters believe Although characters 31 mi 45 did not ultimately contribute to the are indicated ouble lines in the cladograms. o have evolved independently (parallelisms) cladograms, they were included in the study and hence remain in the table. Apomorphic states Presumed ancestral conditions MORPHOLOGICAL CHARACTERISTICS 1. Rootstock a bulb 2. Rootstock a corm of several internodes Corm partly apical in origin and basal to the flow- ering stem Leaves ensiform and equitant osed Inflorescence consisting of one or more rhipidia, each enclosed in large spathes г -з3 © cB = O @ — Ф [7 x O £e + ssile Inflorescence a spike Rootstock a persistent creeping rhizome Rootstock a persistent creeping rhizome Corm axillary in origin and lateral to the flowering stem Leaves bifacial Leaves flat Leaf sheaths open The i pee condition is uncertain and will remain a specific outgroup has been identified тат нога to few-flowere Rhipidia single Flowers ra. Inflorescence not a spike Flowers lasting at least two days owers shades of yellow to orange Tepals not clawed Tepals free Stamens six in two whorls Filaments supporting free anthers Exine reticulate Stem h Ovary su Style er ог r divided above the anthers and each branch stigmatic along its entire length Style arms parallel to and appressed to the stamens Style branch apices symmetrical and uniformly stigmatic Style branches slender and symmetric Style branches slender and symmetric Style branches undivided or apically forked Nectaries septa Nectaries located on inner and outer tepals Р. Nectaries located on inner and outer tepals Nectaries жаі. only sugars Nectaries presen ANATOMICAL CHARACTERS 11. Flowers fugacious 12. Flowers blue 13. Tepals clearly divided into a limb and broad claw 14. Tepals united in a tube 15. Stamens three (outer whorl absent) 16. Filaments weak, anthers adhering to the style branches 17. Pollen grain exine micropunctate (punctitegillate) 18. Stem unbranched 19. Ovary inferior 20. Style deeply three-forked to below the base of the anthers and ih division conduplicate and ter- minally stigm 21. Style arms el between the stamens 22. Style branch apices produced into erect paired ap- pendages (crests) above, stigmatic only below 23. Style branches thickened and somewhat com- pressed radially 24. Style laren flattened and petaloid (compressed tangentially 25. Style do deeply forked 26. Nectaries perigonal 27. Perigonal nectaries restricted to the base of the outer tepals 28. Perigonal nectaries most strongly developed on the inner tepals 29. Nectaries secreting oil from special elaiophores 30. Nectaries lacking 31. Styloid crystals in vascular bundle sheaths and other tissues 32. Vessel perforations simple 33. Primary thickening meristem M reduced and pericyclic vasculature sca 34. Distinct midrib in leaf v 35. Leaf margins with ie Eo sclerenchyma 36. Mesophyll cells elongated across the horizontal axis 37. Epidermal cells with sinuous walls 38. Epidermal cells with two-few papillae Raphides present and styloids absent Vessel perforations scalarifor PTM and pericyclic ae more or less extensive No single major leaf vein pres Marginal subepidermal ida not developed Mesophyll cells + isodiametric or longitudinally elongat- ed Epidermal cells with + straight walls Epidermal cells with one or no papillae [There is no character 39.] Volume 77, Number 4 Goldblatt 611 1990 Phylogeny and Classification of Iridaceae TABLE l. Continued. Apomorphic states Presumed ancestral conditions PHYTOCHEMICAL CHARACTERS FLAVONOIDS Flavonols absent 41. Flavones present 42. Mangiferin (xanthone C-glycoside) often present 43. Flavonol O-glycosides predominant (to exclusive) 44. Flavone C-glycosides absent 45. Flavone O-glycosides accumulated 46. Flavonol sulfate present 47. Biflavones absent Flavonols abundant Flavonols present Mangiferin absent Flavone C-glycosides p Flavone C-glycosides pre Flavonol O-glycosides dern or flavone C-glycosides pres- ent Flavonol sulfate absent Biflavones present FREE AMINO ACIDS AND PEPTIDES 48. Free meta-carboxyphenylalanine and carboxyphe- nylglycine present 49. Gamma-glutamyl peptides produced [There is no character 50.] These compounds absent Absent EMBRYOLOGICAL AND KARYOTYPIC CHARACTERS 51. Microsporogenesis simultaneous 52. Endosperm formation nuclear =7 Microsporogenesis successive Endosperm formation helobial = 10 base. Depending on the choice of outgroup for the family, the equitant leaf may or may not be derived. Among possible immediate ancestors of Iridaceae, several genera of Melanthiaceae (Melanthiales sen- su Dahlgren et al., 1985) have similar leaves, no- tably Tofieldieae, but dorsiventral leaves are basic for the core families of Liliales (e.g., Dahlgren & Rasmussen, 1983). Important modifications of the basic equitant leaf are the pleated e leaves of Tigridieae, the synapomorphy that unites the tribe, together with the bulbous diera Plicate leaves also occur in some Ixioideae, notably Babia- na and most Crocosmia species, and appear to have evolved repeatedly in this subfamily. Sec- ondarily bifacial and dorsiventral leaves have evolved in several genera, most significantly in the largely African Irideae- Homeriinae (Fig. 6). This leaf type and an apically rooting single-internode corm define the subtribe. Similar bifacial leaves may also be basic for /ris subgenera Scorpiris, Xiphium, and Reticulata, all of which also have bulbs. In Ixioideae, Crocus and the closely related southern African Syringodea have dorsiventral leaves. Closed leaf sheaths are a feature of Ixioideae and are scattered in Iridoideae but are probably not basic in the latter Inflorescence. The basic inflorescence is probably the distinctive so-called rhipidium, a specialized monochasial cyme. There can be no doubt that its distinctive structure comprises at least one (as treated here) and perhaps three separate synapo- morphic states if the spathelike sheathing bracts and the collapsed axis are considered separately from the arrangement of the flowers. All Iridoideae have rhipidia either terminal on the main and lat- eral branches in a variously elaborated paniculate arrangement, or there may be only one or a few rhipidia clustered terminally. The rhipidia consist of two large bractlike sheathing green spathes en- closing a few or sometimes several pedicellate flow- ers, these essentially attached at a single point. Each flower has a basal bract which also encloses all the younger buds. The flowers are raised suc- cessively out of the spathes over a period of a week or more depending on their number. The two rhi- pidal spathes are probably best interpreted as the outer representing the subtending bract of the in- florescence and the inner the bract of the first ower. Variously fused rhipidia, usually paired (binate) (Weimarck, 1939), characterize all genera of Niv- enioideae. In /sophysis (Fig. 1A) the flower is sol- itary but enclosed in what appear to be the opposed spathes typical of an iridaceous rhipidium. The single-flowered state is presumed to be a derived condition for the genus. In Ixioideae the flowers are always sessile and are subtended by an outer bract and an inner bracteole, the latter a double structure like the flower bracts in the rhipidia else- where in the family. Each flower and its bracts are here regarded as the homologue of a rhipidium. In most genera of Ixioideae, notably excluding Pil- lansia, the flowers are arranged along a straight or slightly flexuose axis, thus constituting a spike 612 Annals of the Missouri Botanical Garden RE 1.— А. Isophysis tasmanica (Hook.) Т. Moore (Isophysidoideae) (from Cooke, 1986). Ecklon ex Baker (Nivenioideae), showi ovary. Habits x14, details life size or ui enlarge (Fig. 8). In Pillansia the flowers are arranged in a panicle, the presumably ancestral condition. Pil- lansia is unspecialized in several ways in relation to other Ixioideae (Fig. 7). A few Ixioideae have a reduced spike, sometimes a single flower, notably Romulea, Syringodea, and Crocus, where the flowers are solitary, or solitary on each branch of the flowering stem The flower. The iridaceous flower is variable in color, size, and form, and it is difficult to treat the considerable variation in formal primitive or ad- vanced states for higher categories. The flower is trimerous with a major synapomorphy for the fam- ily the loss of the inner whorl of stamens. Except for Isophysis, the ovary is inferior (Fig. 1A), and it seems reasonable to regard this specialization as —. =. df: Y, == М = —— ЙЫ, «qe — „ Xe =, —B. Nivenia dispar ing woody stem, compound inflorescence, flower, fruiting branch and detail of ed. fundamental to the success and radiation of the rest of Iridaceae. However, with the inclusion of Isophysis, the inferior ovary is not a family syn- apomorphy. The inferior ovary is assumed to have evolved only once and it is the sole synapomorphy uniting subfamilies Nivenioideae, lridoideae, and Ixioideae. Except in Ixioideae and the specialized Diplarrhena (Iridoideae), the flowers are radially symmetric. The tepals are united in a developed tube (not merely basally contiguous at the apex of the ovary) in some genera or species of Nivenioi- deae (Patersonia, Nivenia, Witsenia) and Iridoi- deae (Iris, Moraea sect. Tubiflora, Olsynium (incl. Chamelum, Phaiophleps)), and are always united in Ixioideae. In Ixioideae the flowers are often medianly zy- Volume 77, Number 4 1990 Goldblatt 613 Phylogeny and Classification of Iridaceae gomorphic but the basic state for the subfamily is radial symmetry. Zygomorphy has evolved inde- pendently in several lines in Ixioideae. A perianth tube is characteristic of all Ixioideae, and the flow- ers are long-lived, nondeliquescent, and last at least two days, a contrast with most Iridoideae and Niv- enioideae, which have fugacious and usually deli- quescent flowers. Notably, /sophysis also has long- lasting flowers, and this state is probably the basic one for the family. According to the cladogram generated here, fugacious flowers evolved inde- pendently twice, in Nivenioideae and Iridoideae. An alternative, that it is a shared development in the line leading to Iridoideae, Nivenioideae, an Ixioideae, means that a reversal for the condition must have occurred in the ancestral Ixioideae. A cladogram with this conformation is less parsimo- nious by at least three steps. Nectaries. Perigonal nectaries are considered a derived condition for the families of Liliales (Dahl- gren & Rasmussen, 1983), and septal nectaries are generally accepted to be basic for the monocots. Nivenia and Klattia and probably Witsenia, all Nivenioideae, have septal nectaries but other mem- bers of the subfamily apparently lack nectaries altogether, as does /sophysis (Cooke, 1986). When nectaries are present in Iridoideae they are located on the tepals, either near the base or variously placed on the tepal surface. In Iridoideae perigonal nectaries are apparently lacking or uncommon in the least specialized Sisyrinchioideae (this requires careful checking in living material) but are present at the base of all the tepals at least in Bobartia paniculata G. Lewis. The nectaries are restricted to the base of the outer tepals in some specialized genera or species of Irideae (regarded as apo- morphic for the tribe although absent in the rel- atively unspecialized Dietes), but some specialized species have nectaries on both inner and outer tepals (a secondary condition?). In contrast, nec- taries are best developed on the inner tepals in Mariceae and Tigridieae, where they are often partly concealed by folds in the tepal lamina. In the latter two tribes the nectaries secrete oil (Vogel, 1974) from special club-shaped glands (elaio- phores). Similar glands may also be present on the filaments in Sisyrinchium. Septal nectaries are present in all of the many genera of Ixioideae (Daumann, 1970) so far ex- amined and this presumably represents the basic state for the subfamily. Stamens. An important synapomorphy for Mari- ceae and Tigridieae is the weak filaments that are shared by several genera of both tribes. These slender filaments do not support the anthers, which are lightly attached to the upper part of the style branches. (Some specialized genera of Tigridieae have typical sturdy filaments.) The filaments are united, either partly or completely in several Irideae and Tigridieae, but the condition is not basic in either tribe and has probably evolved more than once in each. Pollen grains are basically monosulcate with a tectate-reticulate exine in Iridaceae (Schulze, 1971), thus conforming to the probable ancestral condition the monocots. Ixioideae, including the taxonom- ically isolated Pillansia (Goldblatt & Stein, 1988), iffer significantly in having micropunctate and usually microspinulate exine (Schulze, 1971; de Vos, 1982; Goldblatt & Stein, 1988). Nonaper- turate grains occur in the specialized Syringodea and Crocus. Bisulcate grains are found in many Tigridieae, and may be a synapomorphy for sub- tribe Tigridiinae (Goldblatt, 1982b) although they also occur in some Cipurinae (Rudall & Wheeler, 1988). Syncolpate grains occur frequently in Irid- oideae but are basic only to certain genera or species groups. Acolpate or anomatreme grains are recorded in some sections and subgenera of [тїз (Schulze, 1971); they are not characteristic of any generic groups. Style. Basic style and stigma structure is presumed to be the condition found in /sophysis and Niv- enioideae, in which the branches or lobes are rel- atively short and stigmatic along the entire upper surface. Many Iridoideae differ in having long style branches (Fig. 2A), often exceeding the style, the margins of which are convolute so that the style branch is tubular and thus stigmatic only termi- nally. This modified structure is further specialized in Irideae, Mariceae, and several (presumably bas- al) genera of Tigridieae, apparently in a similar way. The apices of the style branches are produced above as a pair of flat, erect, nonstigmatic ap- pendages, while the stigmatic surface is confined to the lower part (Fig. 6A) In Irideae the style branches are, in addition, flattened and petaloid, and the stigmatic surface is often a small lobe restricted to the median ventral surface. In Mariceae and many Tigridieae the style branch is substantially thickened, and the stigma lobe often has a second pair of erect appendages similar to but smaller than the appendages at the upper apices of the style branches. Accompanying this are the weak filaments that do not support the anthers, which are lightly attached to the upper part of the style branch below the stigmatic surface. An elaborate style branch organization is not found 614 Annals of the Missouri Botanical Garden FIGURE 2. о == nchium arenarium Correa, 1969). Habits x in all genera of Irideae and Tigridieae. The branch- es may be secondarily reduced and simplified in sometimes, as in Eleutherine and Calydorea (Tigridieae) and Roggeveldia (Irideae), to long filiform arms, or completely divided into some genera, paired and usually slender lobes (Nemastylis, Alo- phia, Tigridia—Tigridieae; Hexaglotti deae). A possible synapomorphy for Sisyrinchieae is the orientation of the long filiform style branches between, rather than opposite, the stamens (Fig. 2A), the latter being basic for the family. Although this may be a weak character, it is treated here as the synapomorphy that unites this tribe. The only other specialized condition that the genera appar- d tribes pce and Tigridieae. — A. Libertia chilensis (Mol.) Gunkel, habit and detail Póppig, umi ^i Тш. habit and details of flower, fruit, and stamens and gynoecium (A, В from з Beals life-size or variously enlarged. habit and det С ail of stamens and Бу noecium. — ently share is the absence of flavonols, another weak synapomorphy, and one also found in Marice- ae and Tigridieae. Fruit and seed. Loculicidal capsules, presumably derived from septicidal capsules, are almost uni- versal in lridaceae. Septicidal capsules occur in most Melanthiales and Colchicaceae of Liliales and are probably basic for the order. lridaceae may therefore be specialized in their possession of loc- ulicidal capsules. The seeds of Liliiflorae have been extensively studied by Huber (1969), who found several features in the seeds of Iridaceae that dis- tinguish generic groups. No seed characters have been used in this study as it seems likely that the basic seed type in the family is unspecialized and Volume 77, Number 4 Goldblatt 615 1990 Phylogeny and Classification of Iridaceae BLE 2. Embryological data known for Iridaceae. Abbreviations: su = successive; si = simultaneous; he = s helobial; nu = nuclear. Unless specific references are given, the data are from Schnarf (1931) Micro- Endo sporo- Parietal Parietal sperm genesis cell tissue formation References ISOPHYSIDOIDEAE Unknown NIVENIOIDEAE Geosiris aphylla su + + he Goldblatt et al. (1987) IRIDOIDEAE Tris tenax — + + пи Smith € Clarkson (1956) І. stylosa = + + = І. pseudacorus pa = = пи . fulva — + — — Riley (1942) I. hexagona subsp. giganticaerulea = + = = pi (1942) Sisyrinchium striatum si + х пи Lakshmanan & Phillip (1971) S. californicum si + х пи Lakshmanan & Phillip (1971) Gelasine azurea - x x nu Kenton & Rudall (1987) Eleutherine latifolia = x x nu Kenton & Rudall (1987) IXIOIDEAE Crocus sieberi = — = nu Rudall et al. (1984) Romulea columnae — + пи R. bulbocodiu — t nu R. rosea var. suum = + х пи Steyn (1973a, b) Tritonia crocata — — — nu In dde simultaneous microsporogenesis has been тер in the following: Sisyrinchium bushii (Farr, 1922); . ber ; С. aurea (as ES re aethiopica (as S. striatu mudiana; Crocosmia crocosmiiflora tholyza); Gladiolus cunonius (as Antholyza), Ixia peri ula n- esia refracta ta, 1. coccinea; 1. culata, and Free (Guignard, 1915а); and Tigridia pavonia; Crocus sativus; C. vernus; Gladiolus x iis (Guignard, 1915b). almost identical with that in several other families of Liliales, notably Colchicaceae and Campyne- mataceae. Huber did, however, record the pres- ence of a lipid layer in the inner layer of the outer integument of the testa as a family characteristic (thus presumably a synapomorphy). Seed storage products include protein, oil, and hemicellulose, the latter present in the thick cell walls of the endosperm. EMBRYOLOGY Embryology of Iridaceae is poorly known (Table 2), so the generalizations made below for the family are tentative. Microsporogenesis, so far as is known, is reported to be simultaneous in the few genera of Iridoideae and Ixioideae examined: Iris, Sisy- rinchium, Tigridia, Crocus, Freesia, Romulea, Crocosmia, Ixia, Chasmanthe, Gladiolus, incl. Anomalesia (Guignard, 1915a, b). Until recently no Nivenioideae nor /sophysis had been investi- gated embryologically, but Rübsamen (pers. comm.; Goldblatt et al., 1987) has found successive mi- crosporogenesis in Geosiris. This suggests that the latter condition may be basic for Iridaceae, as it is for other families of Liliales and Melanthiales (Dahlgren & Bremer, 1985). Unlike most Liliales, but typical of Melanthiales, the archesporial cell produces a parietal cell in most members of the family (Iris, Romulea, Cro- cus, Crocosmia, Sisyrinchium) and, except in Sis- yrinchium where it degenerates, one or two layers of parietal tissue are formed (Wunderlich, 1959). In the two species of Tigridieae (Gelasine and Eleutherine) examined, the archesporial cell func- tions as the megaspore mother cell and the ovules are thus tenuinucellate. The embryo sac is of the Polygonum-type, basic for the monocots. Endo- sperm formation is nuclear (Wunderlich, 1959) in the few genera of Iridoideae (Iris, Sisyrinchium) and Ixioideae (Romulea, Tritonia) so far examined but helobial in Geosiris (Rúbsamen, pers. comm.), the only species of Nivenioideae known in this 616 Annals of the Missouri Botanical Garden respect. By extension, helobial endosperm forma- tion may be basic in Iridaceae if the condition is ancestral to nuclear endosperm formation, as some authorities believe (Dahlgren & Bremer, 1985). Embryological data are urgently needed for more genera of Nivenioideae and for /sophysis. In Mel- anthaceae endosperm formation is also helobial. CYTOLOGY Chromosome number, size, and morphology are variable in Iridaceae, and the basic number for the family is uncertain, particularly as /sophysis is unknown cytologically (Goldblatt, 1971. 1979a, 1982b; Kenton & Heywood, 1984). The most likely basic number for Nivenioideae, Iridoideae, and Ixioideae is x = 10, judging from the distri- bution of base numbers in the less specialized gen- era. A derived basic number is known for Tigri- dieae, which are unusually uniform cytologically and have x = 7 (Goldblatt, 1982b). The closely related Mariceae probably have x = 10. The base number of x = 7 may reasonably be treated as an autapomorphy for Tigridieae even though base numbers for other tribes are not certain. Irideae also possibly have x = 10, but the situation in Iris and its immediate allies is so complex that this is very speculative and 10 is in any event possibly basic for the family. For each tribe of Ixioideae the most likely base number is also x = ANATOMY Several aspects of the anatomy of Iridaceae have recently been examined by Rudall (1983, 1984, 1986), and as a result a number of anatomical features can be used with some confidence for phylogenetic xm The mary thickening meristem (PTM Р dues well devel- oped in the Plate of Isophysis, Aristea, and Patersonia (Nivenioideae) and Orthrosanthus and Libertia (Sisyrinchieae); consequently these genera have rhizomes with substantial pericyclic vascu- lature and usually a prominent endodermis. Irideae and Mariceae have a reduced PTM and little peri- cyclic vasculature. The endodermis thickening is reduced in many Irideae and absent in Mariceae. Similar reduction is characteristic of the corms of Ixioideae. The latter, excepting Pillansia, have a distinctive leaf epidermis, regarded as derived here, in which the cells have sinuous walls and two or more papillae. The mesophyll in many Ixioideae, but not Pillansia, is also modified in being elon- gated along the horizontal axis of the leaf (Rudall, pers. comm.). Secondary growth occurs in most Nivenioideae. Klattia, Witsenia, and Nivenia are woody shrubs while Patersonia has woody underground stems (Rudall, 1986). The apparently less specialized Aristea and Geosiris, however, do not have sec- ondary growth and the condition is unlikely to be a synapomorphy for the tribe as a whole. Secondary growth may have evolved only once in the subfam- ily, for Patersonia shares with the shrubby Cape genera several specializations in leaf anatomy and similar flavonoid characters not present in Aristea (Rudall & Burns, 1989). Leaf margins also provide characteristics that seem significant. These include subepidermal scle- renchyma, a synapomorphy for lrideae and for Isophysis. Subepidermal marginal sclerenchyma is an important xeromorphic feature and the condi- tion is developed independently in several genera. In many Ixioideae the subepidermal marginal scle- renchyma is associated with a vein, which will be useful later in defining natural generic groupings and phylogeny in the large and diverse Ixieae, not attempted in this paper. Mariceae have marginal epidermal cells heavily thickened and radially elon- gated but lack subepidermal sclerenchyma (Rudall, pers. comm.). А similar epidermis occurs in a few genera of Ixieae, in which it may have evolved more than once. The presence of a distinct leaf midrib is most likely a derived condition in Irida- ceae and is present in all Watsonieae and Ixieae (Ixioideae) but not in Pillansieae. In Iridoideae, Mariceae also have leaf midribs, but the related Tigridieae have plicate leaves with more than one major vein, this probably directed related to the foliated nature of the leaf. Most genera of Irideae and Sisyrinchieae lack leaf midribs. It has been known for several years that vessels are confined to the roots of Iridaceae except Sis- yrinchium (Cheadle, 1963). The vessel perfora- tions are scalariform (the plesiomorphic condition) in Isophysis and Nivenioideae (including Geosiris) (Goldblatt et al., 1987). Elsewhere in the family the vessel perforations are simple or predominantly simple (Cheadle, 1963). This is regarded as a syn- apomorphy for Iridoideae and Ixioideae and is ap- parently a major synapomorphy (Fig. 3) uniting the two subfamilies. Vessels are present in the stems and leaves of Sisyrinchium. Too few species and genera of Sisyrinchieae have been examined for vessels, and their presence in the stems and leaves may be extended to the closer allies of Sisyrin- chium. However, Libertia, Orthrosanthus, and Bobartia of the tribe have vessels only in the roots. Styloid calcium oxalate crystals are a family haracteristic (Goldblatt et al., 1984) for Iridaceae. Long styloids are found in most tissues of all genera, Volume 77, Number 4 1990 especially in the vascular bundles. Styloids are, however, lacking in Sisyrinchium (Goldblatt et al., 1990) and its close relatives (e.g., Olsynium, So- lenomelus). In Liliales no other el has styloids, but this type of crystal occurs in several genera of Melanthiaceae (tribes Mila ке Narthe- cieae) where they are confined to the vascular sheaths eee 1980). Several genera of these tribes have, in addition, raphide bundles in paren- chyma tissue. The polarization of the styloid char- acter for Iridaceae largely depends on what family is regarded as the immediate ancestor or sister group of Iridaceae. Styloids, reported previously as lacking in Geosiris (from the scalelike leaves) by Goldblatt et al. (1984), have now been found in the stems and rhizomes of the genus (Goldblatt et al., 1987), and this is taken as important evi- dence for the placement of Geosiris in Iridaceae. PHYTOCHEMISTRY Flavonoids. Iridaceae have an unusually wide range of flavonoid compounds compared with other petal- oid monocot families (Williams et al., 1986). Basic to the Liliiflorae are most likely O-glycosides of the flavonols quercetin and kaempferol (Williams, 1975), and perhaps flavone C-glycosides, the latter of wide distribution in the monocots although rare in Liliiflorae apart from Iridaceae and Orchidaceae. Basic for Iridaceae are flavonol O-glycosides and flavone C-glycosides, as well as biflavones. The last- mentioned are unusual in the flowering plants but have now been recorded in a few Lilüflorae: /so- physis and Patersonia in the Iridaceae and an unconfirmed report in Lophiola (Melanthiaceae). The distribution of biflavones in the petaloid mon- ocots is inadequately known at present, but it seems likely that their presence may be basic for Liliiflorae as a whole rather than a specialization in Iridaceae. If plesiomorphic, they have been lost in Iridoideae and Ixioideae. The possible biflavones provisionally identified in Iridoideae are pu C-methylated flavonoids (C. Williams, pers. com Important flavonoid features for Тйен consideration (data from Williams et al., 1986 include the absence of flavone C-glycosides in Pil- lansieae and Watsonieae (both Ixioideae) (provi- sionally regarded as apomorphic states for these tribes) and the accumulation of flavones (appar- ently an important synapomorphy) and flavone O-glycosides, including unusual myricetin deriva- tives and only traces of or no flavone C-glycosides. Goldblatt 617 Phylogeny and Classification of Iridaceae s Р : : = E E y S г 9 + 20 style Mant tubular 48 free m-carboxy | а/асіаѕ 11 flower fugacious 52 nuclear endosperm І 51 — P simultaneous 47 w/o api 42 m angifer 32 vessel poca simple 8A rhipidia binate 12 perianth blue 11 flower fugacious 1A et han 30 w/o nectaries 19 ovary inferior =“ € A 15 stamens three 7 rhipidium 50 styloids ? 4 leaf equitant IRIDACEAE FIGURE 3. Cladogram of the major lineages of Iri- daceae, here treated as subfamilies. Parallelisms are in- assuming lridaceae belong to Liliales. See Table 1 for explanation of characters used here. Sisyrinchieae, Mariceae, and Tigridieae stand out in largely or entirely lacking flavonols. A significant synapomorphy for Iridoideae and Ixioideae appears to be the frequent, but not universal, accumulation of the xanthone C-glycoside mangiferin. Free amino acids and peptides. The unusual free amino acids carboxyphenylalanine and caboxy- phenylglycine have a distinctive distribution in Iri- daceae (Larsen et al., 1981), occurring regularly in most species examined of Iridoideae, but these compounds are notably absent from Sisyrinchium (its immediate allies have not been surveyed), yet recorded in Libertia, Bobartia, and Orthrosan- thus, also Sisyrinchieae. The presence of these distinctive amino acids, not reported elsewhere in the monocotyledons, is regarded as a major syn- apomorphy for Iridoideae, united otherwise by hav- ing a fugacious flower (a parallelism with Niven- ioideae) and convolute style branches. А second unusual group of compounds, y-glu- tamyl peptides, have been found to be common in Irideae, and also present in about half the species of Mariceae examined. They have not been re- ported in Tigridieae, in which only three species ave been surveyed. Gamma-glutamyl peptides are absent in all Ixioideae and Nivenioideae (known for Annals of the 20 style branches tubular 48 free m-carboxy a/acids IRIDOIDEAE FIGURE 4. Cladogram of subfamily crac ч he major lineages treated as tribes. Parallelisms a dicated by double lines. See Table 1 for о ‘of characters used. only two species of the latter) examined (Larsen et al., 1981 7). The presence of this class of compounds is treated as a synapomorphy for the line including Irideae, Mariceae, and Tigridieae. It seems likely that y-glutamyl peptides will be found in Tigridieae when more species are surveyed, and their absence cannot at present be regarded as reversal for the tribe. These peptides are not in- variably present in Irideae, and are sometimes ab- sent in some species in a genus in which they have been found, or even lacking in some genera. Fur- ther sampling is needed. CLADISTIC ANALYSIS The results of the cladistic analysis are presented in three diagrams (Figs. 3-5), the first for the major lineages of Iridaceae, and the second and third for the large subfamilies Iridoideae and Ixioideae. The manually generated cladograms were tested using the PAUP Program (Swofford, 1985), which pro- duced the same results. There is good evidence that Iridaceae are monophyletic and defined by several synapomorphies, the number depending on the chosen outgroup, which for this study is the Melanthiales. There are four major lineages (Fig. 3), one for Isophysis alone, and the other three united by the possession of an inferior ovary. These four primary lineages are accorded subfamily rank. Missouri Botanical Garden Ф ES Ф > 5° d о? o > v > e? . КУ KJ М © à S "d $ e 4 i e g SE PS 53 7 5 leaves plicate 1 bulb 44 w/o flavone O-glycosides "^ 4t Tevanes preson! 3 16 filaments weak -T- 3 corm apical 24 style petaloid | 25 style branches forked Г р 23 style branches thickened Ps outer t4 pal песми 28 inner tepal nectaries marginal 'sclerench ma 10 spike Я у 29 oil glands ds 46 flavonol sulphate 36, 37, 38 derived leaf | 13 tepals clawed 44 w/o flavone O-glycosides anatomy 22 style apex crested 34 leaf midrib 21 style branches alt. anthers 26 perigonal nectaries 2 corm 33 PTM reduced 40 w/o flavonols | ы : 6 leaf sheaths closed 49 G-glytamyl peptides ara 17 exine micropunctate 14 perianth tube 11 flowers fugacious А 9 flowers sessile IXIOIDEAE Cladogram of subfamily Ixioideae with lin- FIGURE 5. dou- eages treated as tribes. Parallelisms are indicated by ble lines. See Table 1 for explanation of rapid Ixioideae are distinguished by five important apo- morphic states. Iridoideae and Ixioideae share five pad The xanthone, mangiferin, occurs many genera of both subfamilies, and specialized vessels with simple perforations have been found in all genera so far examined. This clade also lacks biflavones and is specialized embryologically in hav- ing simultaneous microsporogenesis and nuclear endosperm formation, although too few genera have been studied embryologically. Biflavones are basic in Iridaceae but absent in Ixioideae and probably absent in Iridoideae since the unidentified DK/DK ested to be biflavonoids in a few 1986) are most likely pidia (usually binate). Their derived fugacious flow- ers with blue perianths appear to be basic for the subfamily The a Asan for Iridoideae (Fig. 4) also has four major lines, accorded tribal status, and cor- responding closely to currently accepted tribal groups in the family. The status and composition of Sisyrinchieae, however, is still open to question. The core genera of the tribe, around Sisyrinchium and its close allies in South America, are clearly monophyletic, but it is not yet established with any onfidence that the southern African Bobartia, or the Australasian—South American Libertia and Or- thosanthus also belong here. Their present position Volume 77, Number 4 1990 Goldbl 619 Sal and Classification of Iridaceae rests to some extent on inadequately investigated phytochemical data and on a subjective interpre- tation of the nature of style branch structure and orientation. In Ixioideae there seem to be three distinct lines (Fig. 5) and each is accorded tribal rank. monotypic Pillansieae are isolated from the line leading to Watsonieae and Ixieae and are clearly ancient and primitive relicts. Watsonieae and Ixieae are less well defined but nevertheless appear to comprise natural groups of equal rank. Ixieae com- prise almost half the genera and species of the family and it would be most satisfactory if this tribe could be broken down into a few large groupings. However, at present this does not seem possible, although some 6-8 small generic groupings (see discussion below under Radiation and Composition of the Subfamilies—Ixioideae) are apparent. RADIATION AND COMPOSITION OF THE SUBFAMILIES ISOPHYSIDOIDEAE Consisting only of the Tasmanian /sophysis (Fig. 1A), the subfamily is typical of Iridaceae in lacking the three stamens of the inner whorl, and in having styloid crystals (Goldblatt et al., 1984), extrorse anthers, and distichously arranged equitant leaves. It differs markedly, however, in having a superior ovary, which isolates /sophysis from the rest of the family. It is also somewhat discordant in its flavonoids: apart from traces of flavone C-glyco- sides, it has biflavones only (Williams et al., 1986), compounds also found in some Nivenioideae. The basic inflorescence of /sophysis is interpreted here as a rhipidium. It consists of a single terminal flower, but its arrangement in two large opposed spathelike bracts that enclose the bud and pedicel is very like that of the rhipidium that is probably basic in Iridaceae. /sophysis apparently lacks nec- 6). taries (Cooke, 1986) NIVENIOIDEAE Including one Australasian and five Afro-Mad- agascan genera, the largest of which is the wide- spread Aristea (ca. 50 spp.), Nivenioideae com- prise some 83 species. Nivenia, Witsenia, and Klattia, all shrubby (Fig. 1B) and with strongly developed secondary growth of the monocot type, are endemic to the southwestern Cape. Geosiris, an achlorophyllous saprophyte with scalelike leaves, is restricted to moist forests in eastern Madagascar. It has numerous ovules and fine seeds but in other respects appears to conform with Nivenioideae (Goldblatt et al., 1987). Most notable are the styloid crystals in the stems and in the cormlike rhizomes, and the vessels with scalariform perforations (the basic condition in lridaceae), and found in /so- physis and the genera of Nivenioideae. The non- African genus, Patersonia, is mainly Australian (Cooke, 1986) and occurs on New Guinea, Su- matra, and Borneo. Nivenioideae are specialized in their fugacious and blue flowers (almost certainly derived in the family), either binate or variously fused rhipidia, and unusual flavonoid pattern with a predominance of flavonol O-glycosides rather than flavone C-glycosides, which are more common elsewhere in Iridaceae. Aristea exhibits little specialization except in capsule and seed morphology. The shrub- by Cape genera all have flowers with a well-de- veloped perianth tube, a derived condition. Wit- senia is specialized for sunbird pollination and has yellow, green, and black flowers. Nivenia is un- usual in its heterostyly, probably basic in the genus and found in five of the nine species. Patersonia has a specialized floral morphology: the tepals are united in a long slender tube, the inner tepals are reduced to scales or are absent, and the stigma lobes are more or less fringed. The roots are woody and have secondary growth (Ru- dall, 1986). Woodiness in Patersonia and the shrubby Cape genera may have had a common origin, as these genera share a similar leaf anatomy and flavonoid profile (Rudall & Burns, 1989). How- ever, the Cape genera share with Aristea the de- rived basic chromosome number, x — 16 (Gold- blatt, 1971), while Patersonia may have x = 11 (Goldblatt, 197792). IRIDOIDEAE Iridoideae share with Ixioideae vessels with spe- cialized simple perforations and the arguably sig- nificant phytochemical character of having, in at least a few species of each, the xanthone mangif- erin, an unusual compound in the monocots. All genera of both subfamilies also lack biflavones. Possibly these three features arose independently in each subfamily, and certainly they appear to share no significant morphological specializations while differing in a large number of features. Iri- doideae are united by having fugacious flowers (a parallelism with Nivenioideae), perigonal nectaries, and an unusual style morphology in which the branches are long and divide below the level of the anthers while the margins are conduplicate so that each branch is stigmatic only apically. This style 620 Annals of the Missouri Botanical Garden TABLE 3. Classification of the Iridaceae, with the genera assigned to subfamily and tribe. Numbers in parentheses indicate the total genera and species for tribe or subfamily or total species for a genus. Important synonyms are indicated. Total 77 genera, 1,630 species. 1. Subfamily Mi ed a M (1980) (1: 1) sophysis (1)— Tasm 2. Subfamily Nivenioideae ерше о ex Goldbl.' (6 : 83) Aristea (50)— Africa, Madagasc Geosiris (1) — Madagascar Nivenia (9)— Cape Region, South Africa Klattia (2)— Cape Region, South Africa Witsenia ч Cape Region, South А Paterson ON. Borneo, uai New Guinea 3. Subfamily Iridoideae (42 : 690) Tribe Sisyrinchieae Baker (1878) (8 : 124) Bobartia (12)— southern Africa Libertia (ca. 9)— Australasia, South America Diplarrhena (2)—southeastern Australia, Tasmania Orthrosanthus (9)— Australia, South and Central America Sisyrinchium (ca. 80)— North and South America Olsynium (incl. Phaiophleps, needs Ona) (11)— North and South America Solenomelus (2)— Chile, Argent Tapeinia de ене еш Chile at Argentina Tribe Irideae (13 : 405) Dietes (6)— East and southern Africa, Lord Howe Island Moraea (120) — sub-Saharan Africa Homeria (incl. Sessilistigma) кыйан Africa Rheome (3)-western Cape, South Afr Hexaglottis (6)— western Cape, Sud: abus Roggeveldia (1)— Karoo, South Africa Barnardiella (1)— western Cape, South Africa Gynandriris (9)—southern Africa, Mediterranean, Middle East Galaxia (14)—South Africa Pardanthopsis (1-2)— eastern Asia Belamcanda (1)— southeastern Asia Iris (ca. 210)— Europe, Asia, North Africa, North America Hermodactylis (1)— Mediterranean Middle East Tribe Mariceae Hutchinson (1934) (3 : 40) Trimezia (ca. 20)—South and Central America Pseudotrimezia (6-8)— Brazil omarica (ca. 12)—South and Central America Tribe Tigridieae Baker (1878) (18 Cypella (ca. 20)—South and Central America Cipura (ca. 6)—South and Central America, West Indies Eleutherine (2)— South and Central America, West Indies Ennealophus P edd America Gelasine (4)—So erica Calydorea bri c Cardoso, Саша, Itysa) (ca. 10) —temperate South America Ainea (2) —M Nemastylis (B) Central America, Mexico, southern U.S.A. desa (ca. 6)— temperate South is southern U.S.A. ra (1)—temperate South Am s (1)— temperate South is Mastigostyla (ca. 16)— South America Cardenanthus (ca. 8) —South America Tigridia (ca. 35) —South and Central America Sessilanthera (3)— Mexico and Central America Alophia (ca. 5)— tropical South and Central America, southern U.S.A. Fosteria (1) — Mexico Cobana (2)— Guatemala, Honduras Volume 77, Number 4 Goldblatt 1990 Phylogeny and Classification of Iridaceae 621 TABLE 3. Continued. 4. Subfamily Ixioideae Klatt (1866) (as subordo Ixieae) (28 : 860) Tribe Pillansieae Goldblatt? (1 : 1) Pillansia (1)—southwestern Cape, South Africa Tribe Watsonieae Klatt (1882) (5 : 99 Lapeirousia (ca. 36)—tropical and southern Africa Savannosiphon (1)—south tropical Africa Micranthus (3)—southwestern Cape, South Africa Thereianthus (7)—southwestern Cape, South Africa Watsonia (52)—southern Africa Tribe Ixieae Dumortier (1822) (22 : 760) Ixia (45)—South Africa Dierama (44)— tropical and southern Africa Sparaxis (including Synnotia) (12)—southwestern Cape, South Africa Freesia (11)— southern Africa Anomatheca (6)— south tropical and southern Africa Crocosmia (9)— south tropical and southern Africa evia (1)— southern Africa Chasmanthe (3)—southwestern Cape, South Africa Tritonia (28)— south tropical and southern Africa Duthieastrum (1) — southern Africa Geissorhiza (82)—South Africa Hesperantha (ca. 62)—tropical and southern Africa Schizostylis (1)— southern Africa Melasphaerula (1)— southern Africa Gladiolus (incl. Homoglossum, Anomalesia, Oenostachys) (ca. 195)— Africa, Madagascar, Eurasia Radinosiphon (1-2)— south tropical and southern Africa Babiana (incl. Antholyza) (64)— southern Africa, Socotra Tritonopsis (incl. Anapalina) = South Africa Zygotritonia (4)— tropical Afric Romulea (ca. 90) —Africa, Mis Syringodea (8) — southern Africa Crocus (ca. 80)— Europe, Asia, North Africa ' Nivenioideae Schulze ex Goldblatt, subfam. nov. Plantae foliis sine costis medianis, caulibus subterraneis vel aeriis et anapa ре» inflorescentiis binatis, floribus actinomorphis usitate caeruleis, tepalis plus minus liberis vel in tubum co 2 Pillansicae Goldblatt, trib. nov. Planta foliis sine costis medianis, inflorescentia paniculata, floribus sessilibus actinomorphis, tepalis in tubum connatis, stigmatibus plus minus integribus type seems basic to the subfamily but is variously modified in specialized lines. Another significant specialization in Iridoideae is the presence of the free amino acids meta-carboxyphenylalanine and glycine (Larsen et al., 1981, 1987). Iridoideae comprise four important lines, which are treated here as tribes. The least specialized of these is the Sisyrincheae (Fig. 2A, B). The tribe may be heterogeneous, and an absence of apo- morphic features makes it difficult to deal with. Possibly it is united by having the long style branch- es extending between the stamens rather than op- posite them (this feature may be basic for the subfamily). All the species so far examined for flavonoids are apparently derived in not accumu- lating flavonols (a reversal). Libertia and Orthro- santhus occur in Australasia and South America, the latter also in Central America, while Diplar- rhena is restricted to southeastern Australia and Tasmania. The large and diverse Sisyrinchium and its close allies, including Olsynium (incl. Phaio- phleps and Ona), Tapeinia, and Solenomelus, all small genera, are solely New World and centered in South America. Bobartia appears unusual phy- togeographically in being southern African (Strid, The floral morphology of Bobartia is virtually identical with other Sisyrinchieae but it has two unusual features: pubescent pedicels and scattered fibers in the phloem (Rudall, 1983). The latter is a constant attribute of several African genera of Irideae, including Dietes and Moraea, while pu- bescent pedicels are known elsewhere in Iridaceae only in Dietes. If Bobartia is misplaced in Sisy- rinchieae, it may have acquired its apparently sim- ple floral morphology by a reversal to a more 622 Annals of the Missouri Botanical Garden HANNS мак NW yp FIGURE 6. Iridoideae tribe Irideae.—A. Moraea atropunctata Goldbl., habit and detail of the stamens and gynoecium. — B. Corm of Moraea with tunics removed showing apical origin of roots. —C. Iris afghanica Wendelbo, habit (from Hedge & Wendelbo, 1972—@ Crown copyright. Reproduced with the permission of the Controller of her Britannic Majesty's Stationery Office). — D. Hexaglottis namaquana Goldbl., habit, flower, and detail of the stamens and gynoecium. — E. Galaxia luteo-alba Goldbl., showing acaulescent habit. Habits x4, details life-size or variously enlarged. Volume 77, Number 4 1990 Goldblatt Phylogeny and Classification of Iridaceae primitive condition, such as is likely to have oc- curred in Eleutherine—Tigridieae and Roggevel- dia—Irideae, for example (Goldblatt, 1979b). Al- ternatively, the tribes as delimited here require reevaluation. The flavonoids and other phytochem- ical features of Bobartia are poorly sampled and further investigation may resolve the conflict. The three other important assemblages in Iri- doideae form a clade united by having flowers in which the tepals are differentiated into a limb and claw, and in which the upper apices of the style branches are produced into nonstigmatic paired appendages (crests). The unusual y-glutamyl pep- tides are present in most members examined (but notably not yet reported in Tigridieae, of which only four species have been sampled). Irideae, cen- tered in the Old World, have the style branches and crests flattened and petaloid (Fig. 6A), nec- taries restricted to the base of the outer tepals (at least in the less specialized genera), and, in all genera, a characteristic subepidermal marginal sclerenchyma. The largest genus is Iris, mainly Eurasian but also in North America, and often treated as several genera (Rodionenko, 1961), in- cluding /ridodictyum, Xiphium, and Scorpiris, all of which have bulbs, and Junopsis, which has swol- len roots and a vestigial rootstock. The phylogeny of [ris s.l. needs careful study before the several segregates are finally accepted. In sub-Saharan Africa there are Dietes, with one species on Lord Howe Island, and several genera with apically root- ing corms, notably Ferraria, Moraea, Homeria, and Galaxia. All but Ferraria also have a sec- ondarily bifacial dorsiventral leaf. Moraea, the largest genus with some 120 species, has radiated extensively in southern and east tropical Africa (Goldblatt, 1977, 1986a). The other genera, all apparently segregates of Moraea (Goldblatt, 1987 and in prep.), are centered in the winter rainfall area of the southwestern coast of southern Africa. Three subtribes have been proposed for Irideae (Goldblatt, 1976): Iridinae (Iris, Belamcanda, Hermodactylis, and Pardanthopsis); Ferrariinae (Ferraria); and Homeriinae (Moraea, Homeria, Galaxia, Hexaglottis, Gynandriris, and other genera). The disposition of Dietes is somewhat un- certain although in the past I have aligned it with Iridinae (Goldblatt, 1981). In the New World, Mariceae and Tigridieae share an apparently identical flower with thickened style branches (in contrast to the flattened branches of Irideae), nectaries best developed on the inner tepals, and unusual slender and weak filaments so that the anthers are supported by being attached to the style branches. The nectaries also produce oils (Vogel, 1974) secreted by clavate elaiophores. FIGURE 7. us. Inflorescence х 6, rootstock and leaf x M. Ixioideae. Pillansia templemannii L. Bol- Mariceae, comprising Trimezia, Neomarica, and possibly Pseudotrimezia, have leaves with distinct midribs and they apparently lack flavonols (a par- allelism with Sisyrinchieae). Tigridieae, a larger leaves, a distinctive type of bulb, and the derived basic chromosome number of x = 7. Tigridieae have radiated extensively in temperate and Andean South America, and in Mexico. Major specializa- tions are all floral and focus on the style and sta- 624 Annals of the Missouri Botanical Garden FIGURE 8. Ixioideae. — A. Watsonia spectabilis L. Bolus, habit. — B. Detail of corm of A with tunics removed showing basal origin of the roots. —C. Hesperantha luticola Goldbl., habit and flower. — D. Geissorhiza mathewsii . Bolus, habit and detail of flower and leaf section. — E. Lapeirousia lewisiae B. Nord., habit and flower. Habits X 12, details life-size or variously enlarged. Volume 77, Number 4 1990 Goldblatt 625 Phylogeny and Classification of Iridaceae mens. Several new and mostly monotypic genera have recently been described by Ravenna (1981, 1983, 1986), and it seems clear that the current systematics, based on the minor variations in the shape or disposition of the style branches and the degree of fusion of the filaments, is unsatisfactory and is leading to the chaotic proliferation of mono- types. Careful cladistic analysis may point the way to a more acceptable taxonomy, reflecting as far as possible, natural relationships. Two subtribes have been proposed in Tigridieae (Goldblatt, 1982b), Cipurinae, with a base number of x = 7, monosulcate pollen grains, and generally simple, thickened style branches, and Tigridiinae, with x usually = 14, bisulcate pollen grains, style branches deeply divided into filiform arms, and stamens united. IXIOIDEAE Ixioideae (Figs. 7, 8) are remarkably distinct from other Iridaceae in numerous features. The subfamily is centered in southern Africa, with the specialized genera Romulea and Gladiolus also in urasia, where the large genus Crocus has ra- diated. Synapomorphies for Ixioideae (Fig. 5, Table l) are presence of perianth tubes, basal rooting corms, inflorescences of sessile flowers arranged in a panicle or spike (Fig. 8), micropunctate pollen exines, and closed leaf sheaths. All Ixioideae so far examined have septal nectaries (Daumann, 1970), which may be the basic condition in the family. Ixioideae comprise over 860 species, about half the family, and some 760 of these belong in Ixieae. The monotypic Cape Pillansia is taxonomically isolated in the subfamily, and is the only genus with a panicle (Fig. 7), presumably ancestral to the spike, which is basic to the other two tribes. These also have distinct leaf midribs and specialized leaf anatomy (Rudall, pers. comm., see earlier discus- sion of leaf anatomy). Pillansia also lacks flavone C-glycosides (a parallelism with Watsonieae). This genus is apparently unique in Iridaceae in having a flavonol sulfate. Despite the undesirability of monotypic groups, it seems necessary to treat Pil- lansia as the only member of a tribe Pillansieae. Watsonieae comprise some 99 species in 5 gen- era, which share unusual deeply divided style branches as well as the absence of flavone C-gly- cosides (as does Pillansia). The tribe includes the rousia, spread through sub-Saharan Africa; and the monotypic south tropical African Savannosi- phon. Watsonia is extremely variable in its floral morphology, and bird-pollinated flowers (e.g., Fig. 8A) have evolved independently in at least three lines in the genus (Goldblatt, 1989). Ixieae have corms that develop partially from the apical bud (explained in detail earlier), an un- usual array of flavones not found in the rest of the family, and flavone O-glycosides. Some eight sub- tribes have been recognized in Ixieae as here cir- cumscribed (Lewis, 1954; Goldblatt, 197 1), all rel- atively small monophyletic assemblages, each of which has a common basic chromosome number and unique character combination. These subtribes include Ixiinae (/xia, Dierama, Sparaxis), x = 10, membranous and usually dry bracts, filiform style branches, and fibrous corm tunics; Tritoniinae (Tritonia, Crocosmia, Chasmanthe, Duthieas- trum), x — 11, short subherbaceous bracts, orange flower colors, and fibrous corm tunics; Gladiolinae (Gladiolus including Homoglossum, Anomalesia, and Oenostachys; Radinosiphon), x — 15, long herbaceous bracts, wiry corm tunics, spathulate style branches, and winged seeds; Hesperanthinae (Geissorhiza, Hesperantha, Schizostylis, and pos- sibly Melasphaerula), x = 13, more or less her- baceous bracts, woody corm tunics, filiform style branches; Babianinae (only Babiana, x — 7, subfi- brous tunics, plicate leaves, spathulate style branches, and herbaceous bracts; Romuleinae (Crocus, Romulea, Syringodea), possibly x — 13, woody corm tunics, flowers solitary on the branches of the flowering stems, often divided style branches. Additionally, Freesia and Anomatheca, both x — 11, and Tritoniopsis, apparently x = 16, are allied and have been placed in Freesiinae and Anapalinae respectively. Their wm to other genera of Ixieae are uncertain, but similarities in leaf margin epi- dermis (Rudall & Goldblatt, in press) suggest links with Tritoniiae. The relationships of the tropical African Zygotritonia are likewise uncertain. There are probably several old and taxonomi- cally isolated lines in Ixieae. However, genera with- in each line are often not well defined and may be of comparatively recent origin. Zygomorphic flow- ers have evolved independently in most lines (Fig. 8A, E), sometimes, as in Geissorhiza, Babiana, and Gladiolus, more than once in each. Large bird- pollinated flowers with red to orange perianths, long dimorphic tubes, and exserted anthers and M branches have also evolved independently in least three lines with Gladiolus (Goldblatt & de Vos, 1989), in Babiana (B. ringens), Chas- manthe, and Tritoniopsis, as well as in Watsonia (Watsonieae). Convergence in floral form has ob- scured the natural relationships of many Ixioideae (Lewis, 1954), and vegetative morphology and chromosome cytology (Goldblatt, 1971) are now 626 Annals of th Missouri mica Garden accepted as more important for phylogenetic con- siderations above the species level than most floral features. LITERATURE CITED AMBROSE, J. D. 1980. A re-evaluation of the gee thioideae (Liliaceae) using numerical analys 6 1 in C. D. Brickell et al. (editors), *Petaloid Monocotyledons. Academic Press, London. ARBER, A. R. 21. The leaf structure of the Iridaceae, considered in relation hs the phyllode theory. Ann. Bot. (London) 35: 30 аш: С. 1878. ы: Iridearum. J. Linn. Soc. ‚ 16: б вт С. OOKER. 1883. Genera Planta- & J. D. H m 3(2): 681-710. London fimus " nis bs сзи Rethania (Compositae). per CHEADLE, ү. L iron ouch in Iridaceae. Phytomorph. 13: 245-2 Cooke, D. A 86. Iridaceae. Flora К иа 46: 1-66. Australian Government, Сап Бае i di 1969. Flora Patagonica IL. TNTA, Bue- бш ae n & K. BREMER. 1985. иј clades of the angiosperms. Cladistics 1: 349-3 & Н. Т. CLIFFORD. 1982. The EM dons: А С omparative Study. Academic Press, Lon on & F. RASMUSSEN. 1983. Monocotyledon evo- lution: characters and phylogenetic states. Evolu- tionary Biology 16: 255- Sig ,H. T. Ач P. Yeo. 1985 . The Families of the o Mas Verlag, Berlin DAUMANN, E. 1970. Das Blütennektarium der Mono- cotyledonen unter besonderer saa TST sei- ner deutung Feddes Repert. 80: 463- DE Vos, M. P. Knol ЕТЕ И һу sommige genera van die Iridaceae pe die systematiese posisie. Tydskrif Natuurwetensk. 17: 9-19. 1982. The African pa Tritonia e Gawler (Lidacnnel: Part 1. J. S. Afric Pu Bot. 48: 105-163. Diets, L. 1930. Iridaceae. In: ngler & К. Prantl (editors), Die Natürlichen чин имин 2nd edi- tion, 15a: 463-505. DuMoRTIER, B. C. 1822. Commentationes Botanicae. Tournai. Farr, C. Н. 1922. Quadripartition by furrowing in Sisyrinchium. Bull. Torrey Bot. Club 49: 51-61. Funk, V. A. 82. The systematics of Montanoa (As- teraceae, Heliantheae). Mem. New York Bot. Gard. 1-133. GOLDBLATT, Р. 1971. Cytological and morphological studies in the southern African Iridaceae. J. S. Af- rican Bot. 37: 317-460 1976. The genus Moraea in the winter rainfall region of southern Africa. Ann. Missouri Bot. Gard. 63: ee 786 77. Systematics of Moraea Leese in a Africa. Ann. Missouri Bot. Gard. 6 295. 9 Preliminary cytology of Australasian Iridaceae. Ann. Missouri Bot. Gard. 66: 851-855 979b. Roggeveldia, a new genus of a African Iridaceae-lrideae. Ann. Missouri Bot. Gar 66: 839- 2 19 Systematics, phylogeny and evolution in n dr Ann. Missouri Bot. Gard. 6 131-152. . Corm ie id in Hesperantha (даса, чаре. and a proposed iaa taxonomy . Missouri Bot. Gard. 370-378. 1982b. hromosome cytology in у. to ал чеш systematics of neotropical Iridaceae. Syst. Bot. 7: 186-198. —. 1985. Systematics of the southern African genus Geissorhiza (Iridaceae). Ann. Missouri Bot. Gard. 277-447. a. The moraeas of southern Africa. Апп. mb лыг Bot. . 14. 19 "deinen and relationships of the bigeneric Pacific family — nemataceae die liales). Bull. Mus. Natl. Hist. Nat., , 4° sér., sect. B, Adansonia 8: 117- 132. Systematics of the southern African genus Sige lee de vcnit Iridoideae). Ann. Mis- souri Bot. Gard. 69. А The nal African genus Watsonia (Iridaceae). Ann. Kirstenbosch Bot. Gard. 19. & B. A. STEIN. 1988. Pollen morphology of ар : Mu ннн Ann. Missouri Bot. Gard. "P a DE А 1989. The reduction of Oenostachys, Homoglossum and Anomalesia, pu- tative sunbird pollinated genera, in Gladiolus L. (Iri- sur d Ixioideae). Bull. Mus. Natl. Hist. Nat. 4° sér., . B, Adansonia 11: 417-428. LL HENRICH & P. RuDaLL. 1984. Occur- rence of crystals in Iridaceae and allied families and their eis EE . Ann. Missouri Bot. Gard. 71: 20 iR ш & J. E. HENRICH. 1990. T депега of the Sisyrinchium alliance (Iridaceae- Irid- oideae): phylogeny and relationships. Syst. Bot. 15: 497-510. ‚ —_, VI. CHEADLE, L. J. Dorr & С. A. WiLLIAMs. 1987. Affinities of the Madagascan en- demic Geosiris, Iridaceae or Geosiridaceae. Bull. Mus Hist. Nat., 4° sér., sect. B, Adansonia 9: 239- 248. GUIGNARD, L. 1915a. Sur la formation du pollen. Compt. end. Hebd. Séances Acad. Sci. 160: 428-433. 9 Nouvelles observations sur la for- chez certain SAU 623-625 mation du pollen ompt. Rend. Acad. Sci. 161: HEDGE, I. С. , WENDELBO. не засна іп the flora of aga XIII: various new tax rec- ords. Notes Roy. Bot. Card. "Edinburgh 31: ЕИ HENNIG, "Ww. Phylogenetic Systematics. Univ. 1966. e Press, Urbana, Illinois. (Translated by D. D. Zangerl. па н. 1969. Die Samenmerkmale und Verwandt- Se, дег M Mitt. Bot. Staats- ml. München ши 1. 4. The Faris of pit tag Plants, lst ee Volume 2. MacMillan, Lon 1973. The Famili es of Flowering Paoi 3rd editi on. отет Press, Oxford. JONKER, F. P. 1939. Les Geosiridacées, une nouvelle Volume 77, Number 4 1990 Goldblatt Phylogeny and Classification of Iridaceae 627 famille de Madagascar. Recueil Trav. Bot. Neerl. 36: 73-479. Kenton, A. & C HEYwoop. 1984. Cytological studies in South ia Iridaceae. Pl. Syst. Evol 146: 87-104. DALL. 1987. An unusual case of com- plex heterozygosity i in Gelasine azurea Man cmi and its implications for reproductive biology. Evol Trends Pl. 1: 95-103. 1866. Revisio iridearum: conclusio. Lin- naea 34: 537-689. 882. Erganzungen und Berichtigungen zu Baker's Systema Iridearum. Abh. Naturf. Ges. Halle 15: 337-404. LAKSHMANAN, К. K. & V. J. PHiLLIP. 1971. A contri- bution to the embryology of Iridaceae. Proc. Indian Acad. Sci. 73: 110-116. Biochem. Syst. & Ecol. 9: 313.323. SUN NDAHL, F. : WIECZORKOWSKA & P. GOLDBLATT. tionship between subfamilies, tribes and genera in Ted y 4 f. ¡A 1 DUC Bio- chem. Syst. & Ecol. 18: 575-579. Lewis, С. J. 1954. Some aspects of the morphology, phylogeny and MEME of the South African Iri- African Mus. 40: 15-113. The Iris. Universe Books, New Pax, F. 1888. Iridaceae. In: A. Engler & K. Prantl (editors), Die Natürlichen Pflanzenfamilien 2(4): 137- 157. RAVENNA, P. . Kelissa, a new genus of Iridaceae from south Brazil. Bull. Mus. Natl. Nat. Paris, 4° 10. tila and Onira, two new genera of a South Ameriüsn Iridaceae. Nordic J. Bot. 3: 197- 986. Itysa and Lethia, two jd ed of neotropical Iridaceae. Nordic J. Bot. -588. RiLey, Н.Р. Development of the ам sac of Iris fulva and I. hexagona var. giganticaerulea. Trans. Amer. Microscop. Soc. 61: 328-335. RODIONENKO, С. I. 1961. The Genus /ris. Academic Press, Moscow. [In Russian RupaLL, P. 1983. Leaf aiy 2 ноа of Dietes A equ Nordic J. Bot. 3: 471-478. Taxonomic and evolutionary implica- tions = ЕВЕ structure and se ary thickening in Iridaceae. Bot. Gaz. (Crawfordsville) 145: 524- 86. Taxonomic significance of leaf anatomy in Australasian Iridaceae. Nordic J. Bot. 6: 277- 289. — & P. Burns. 1989. Leaf anatomy of the woody South African Im ун Bull. 44: 525-532. & Р. GOLDBLATT. Leaf anatomy and phylogeny of Ixioideae (Iridacose). Bot. J. Linn. Soc., Bot. (in press). A. WHEELER. 1988. Pollen е іп bue (ийсе) Kew Bull. 43: 693- =S . Y. KENTON. 4. “Em еъ е Pus paire systems in Crocu s Calas ceae)— dy in causes of chromosome iE Pl. Syst. Evol. 148: 119-134 erlin. 71. Beitráge zur Pollenmorphologie der Iridaceae und ihre Bedeutung für die Taxonomie. Feddes Repert. 82: 101-124 SMITH, F. Н. & Q. D. CLARKSON. 1956 о studies of interspecific hybridization in /ris, su tion Californicae. Amer. J. Bot. 43: 582-588. STEYN, E. 1973a. by: embryology of Romulea rosea Eckl. var. reflexa Bég. 1. The structure, psp and fertilisation of Кл ovule. J. S. African Bot. 113-121 1973b. The embryology of Romulea rosea Eckl. var. reflexa Bé Development of the dosperm, embryo and seed coat. J. S. African Bot. 39: 235-243 STRID, A. K. 1972. A taxonomic irc. of Bobartia L. (Iridaceae). Opera Bot. 37: 1- SworrorD, D. L. PAUP, те ИЗ Analysis Using Parsimony. Version 2.4. Illinois Natural His- tory Survey, Champaign, Illinois. TAKHTAJAN, A. 1969. Flowering Plants: Origin and Dispersal. Oliver & Boyd, Edinburgh. [Translated by effre 1980. Outline of the cl tion of flowering plants оне Bot. Ror. 46: 225-359. Anatomia Seminum Comparativa, I. Шера seu Moncotyledones. Nauka, Leningrad. THORNE, R. F. 1 Proposed new E in the angiosperms. Nordic J. Bot. 3: 85-117. VocEL, S. 1974. Olblumen und Olsammelnde Bienen. Trop. & Subtrop. Pflanzenw. 7: 1-267. WEIMARCK, Н. 1939. Types of inflorescences in Aristea and some allied genera. Meddeland. Lunds Bot. Mus. as 40: 616 WILLIAMS, C. 1975. Biosystematics of the Mono- e E patterns in leaves of the Lil- iaceae. Biochem. Syst. col. 3: 229-244 , J. B. BLATT. 1986. Cor- wai between phenolic patterns and tribal clas- sification in the family баак. Phytochemistry 25: -2 WUNDERLICH, К. | 1959. Zur Frage der ея der Endospermtypen bei den Angiospermen. Oesterr. Bot. Z. 106: 203-293. EVOLUTION AND PHYLOGENY OF THE ARACEAE! Michael Н. Grayum? ABSTRACT Araceae are exceptionally diverse by nearly all Ба. criteria. The two most widely accepted infrafamilial The classification systems, those of E p 5 placentation: anatropous, crassinucellate ovules; elongate stamens w 7 or 14; and monosulcate, reticulate, binucleate pollen grains lacking starch. These and in a character analysis o atures of the classification derived from a cladogram reflecting this analysis include: the Monsteroideae, and the dissolution of his Calloideae; a drastic internal rearrange ment of the Colocasioideae; and the recharacterization of the Aroid mosome number of x = bes алети have been employed Significa Engler’s subfamilies Pothoideae and e vegetative and floral morphology, respectively, an empted Philodendroideae and the assimilation of the subfamily Pistioideae (Pistia stratiotes L.) and the tribe Thomsonieae (formerly included in subfamily Lasioideae). The following suprageneric nomenclatural novelties are created: . Grayum, stat. nov.; Grayum, subtrib. nov.; and Scaphispathinae Grayum, subtrib. nov. tideae Grayum, trib. nov.; Protarinae (Engl.) Cerces- Remusatiinae Grayum, subtrib. nov.; Jasarinae The Araceae, or aroid family, comprise about 107 genera and 2,500 species of monocotyledon- ous herbs and vines. Epiphytes and climbers ac- count for approximately 70% of the total species (French, in press). Although the family occurs nat- urally in every continent except Antarctica, it is predominantly tropical: only about 10 genera ex- tend into northern temperate zones (see Table 1), and no species occur in truly temperate regions of the Southern Hemisphere. Most aroid genera occur in tropical Asia and the New World tropics. Al- though Asia boasts a few more genera, the Neo- tropics are perhaps 50% richer in species (Croat, 1979). Roughly half of the species in the family are contained in just two of the 107 genera, Ап- thurium and Philodendron, both of which are exclusively neotropical. About 20 genera occur naturally in tropical Africa, though many of these are isolated and aberrant. The Araceae are exceptionally diverse by nearly all phenetic criteria, as will be abundantly detailed in the present paper, and the family is certainly a comparatively old one. Nonetheless, it is apparently a natural and monophyletic group, as far as can be concluded from presently available evidence. This statement requires a footnote, however: the temperate, oligotypic genus Acorus, traditionally included in Araceae, is strongly discordant within this family; its inclusion seriously weakens the cir- cumscription of Pis fam piy and probably even ren- ders the A tic. Grayum (1984, 1987) a Tillich келү have detailed the reasons for removing Acorus to its own family, Acoraceae, which Walker (manuscript) ia isolated in its own infraclass, Acoriflorae (envisioned by Walker as a sister group to Ariflorae and Typhiflorae). Recent studies of restriction site variation in chlo- roplast DNA support separate family status for Acorus (French & Kessler, 1989). It has thus become clear that Acorus is extralimital to Arace- ae, hence the genus will not be dealt with further in the present paper. This work is derived from a portion of a doctoral thesis submitted to the Graduate School, University of Massachusetts, Amherst. I am indebted to James W. Walker for his guidance and critical comments. T wa, Margaret E. B. Bigelow, Josef Bogner, J. C. French, Peter Goldblatt, an 1 кошт acknowledge all of their helpful advice. I also especially thank ulating discussion was reviewed, in total or in part, by K. S. Baw David L. Mulcahy, and R. D. Josef Bogner, J. С. French, and Thom y for stim e manuscript ns and for unpublished manuscripts. Financial support was provided by an Albert Delisle боралар (University of Massa: рено) and NSF Grants BMS75-02883 and DEB80-10893 to James ? Missouri Botanical Garden, P. О. ur 299, St. Louis, Missouri 63166-0299, U.S.A. ANN. MISSOURI Bor. GARD. 77: 628-697. 1990. Volume 77, Number 4 1990 Grayum 629 Evolution and Phylogeny of Araceae Minus Acorus, the family Araceae may be char- acterized by the following suite of probable syn- apomorphies (Grayum, 1984, 1987): a bractless, spicate inflorescence; extrorse anther dehiscence; a unipistillate gynoecium; a berrylike fruit; amoe- boid anther tapetum; helobial (i.e., haustorial) en- dosperm development; and the (probably) universal presence of tannins. To this list may be added the following features which, though presumably ple- siomorphic, help characterize the family: bifacial leaves; presence of raphides; lack of ethereal oil cells; hypogynous, actinomorphic flowers; basifixed anthers; successive pollen mother-cell cier absence of perisperm; absence of silica; and a sence of tricin. All of the above characteristics are universally present, as far as is known, in all genera of Araceae; furthermore, no other monocot taxon possesses them all in combination. As regards most other characters, however, the Araceae are far from uniform, presenting a broad and in some cases (e.g., pollen exine sculpturing) all but bewildering range of variation. Among the major goals of this paper are the summarization of this variation for all important character fields; the polarization of character states for as many differ- ent characters as possible for which there are re- liable and extensive data; and the clear indication of important gaps in the existing data base. In the second half of the paper, the results of a manual cladistic analysis, based on the foregoing conclu- sions, are discussed as they pertain to aroid taxa accepted a priori as natural; a preliminary intra- familial cladogram, reflecting these considerations, s offered; and a new infrafamilial classification bod on this cladogram is proposed. HISTORY OF CLASSIFICATION OF THE ARACEAE The history of internal aroid classification has been reviewed by Nicolson (1960a, 1987), from which much of what follows is abstracted. All familiar classifications of the Araceae can be conveniently reduced to two origins: the Schottian school and the Englerian school. The classification of Schott (1860) was the first in which the Araceae were treated as a single unit substantially equivalent with that recognized today. This highly subdivided system was based primarily on floral morphology and employed a comparatively narrow generic con- cept. Only the curious aquatic genus Pistia, of the genera then known, was not yet included in the family. Authors prior to Schott did not necessarily recognize an Агасеае coextensive with the one accepted today, and some (e.g., Lindley, 1847) segregated the genera then known into two or three different families, attributing substantially different affinities to eac Engler’s (1876) first critical classification of the Araceae was already substantially the same as his now familiar scheme, except that in 1876 10 subfamilies were recognized instead of the present eight. Pistia was included as an aroid for the first time, and “Lemnoideae”” was included as a subfam- ily of Araceae. Hooker (1883) adapted and modified Schott’s system, adopting many of Engler’s generic con- cepts. Hooker divided the Araceae into 11 tribes; he included Pistia in his tribe Arinae but excluded the “Lemnoideae” from the family. Engler’s own system was refined in subsequent publications, culminating in his final major treat- ment of the family, partly with Krause, which appeared in Das Pflanzenreich from 1905 to 1920 (see Engler, 1920b, for a summary of the system). It was updated and slightly modified by Bogner in 1978 and miscellaneous subsequent papers (see especially Bogner, 1985b; Bogner et al., 1985), in which form it appears in Table 1 (with suprageneric nomenclature adjusted according to Nicolson, 1984b). Engler’s classification scheme is consid- erably different from those of the Schottian school and places a strong emphasis on vegetative mor- жы and anatomy. ittle-known system of Hotta (1970) is a Bact of Engler’s system, in which three of En- gler’s subfamilies are submerged: Monsteroideae into Pothoideae, Colocasioideae into Philodendroid- eae, and Calloideae partly (Orontieae) into Lasioi- deae and partly (Calla) into Philodendroideae. One new subfamily (Acoroideae s.s.) is recognized, for a total of six subfamilies. Hutchinson’s (1973; Appendix 1) system rep- resents the most modern incarnation of the Schot- tian school of aroid classification, and, as such, is based mainly on floral morphology. An extension and elaboration of Hooker’s system, Hutchinson’s classification utilizes 18 tribes as the major sub- divisions. Although Hutchinson’s system is still occasion- ally employed in certain contexts (e.g., by Raven & Axelrod, 1974), virtually all major contempo- rary aroid workers are of the Englerian school. The most recent refinement of the Englerian system is that of Bogner & Nicolson (in press), in which the tribe Acoreae is elevated to subfamilial rank, and a few lesser taxa are transferred from one subfamily to another. Most basic Englerian concepts, how- ever, remain intact. Engler’s system (Table 1) has also been used as a convenient framework of ref- 630 Annals of the Missouri Botanical Garden TABLE 1. The aroid classification of Engler'. Taxa Numbers of species Ranges Placements in this paper Subfamily Pothoideae (properly Acoroideae)’ Tribe Potheae? Pothos L. Pothoidium Schott Pedicellarum M. Hotta Anadendrum Schott Tribe Heteropsideae Heteropsis Kunth Tribe Anthurieae Anthurium Schott Tribe Culcasieae Culcasia P. Beauv. Tribe Zamioculcadeae? Zamioculcas Schott Gonatopus Hook. f. Tribe Acoreae Acorus L. Gymnostachys R. Br. Subfamily Monsteroideae Tribe Monstereae Rhaphidophora Hassk. Epipremnum Schott Amydrium Schott Scindapsus Schott Alloschemone Schott Stenospermation Schott Rhodospatha Poeppig Monstera Adans. Tribe Spathiphylleae Spathiphyllum Schott Engl. Holochlamys Subfamily Calloideae Tribe Symplocarpeae (properly Orontieae) Lysichiton Schott a Sch ta Salisb. x Nutt. tia L. Tribe Calleae Calla L. Subfamily Lasioideae Tribe Lasieae Subtribe Lasiinae (properly Dracontiinae) 11 C ie aiu, Griffith? a Lou үеге Schott Podolasia N. E. Br. Urospatha Schott Dracontioides Engl. Echidnium Schott Dracontium L 700-1,000 [21] 0 1 1 9 12 3 2 1 20 1 (2) 15 Southeast Asia Southeast Asia Neotropics Neotropics Africa Africa Africa North Temperate Australia Asia, Africa Southeast Asia Southeast Asia Asia South America Neotropics Neotropics, Southeast Asia Southeast Asia North Temperate North Temperate North America North Temperate Southeast Asia Southeast Asia India Southeast Asia Neotropics South America South America Neotropics Pothoideae Pothoideae Pothoideae Pothoideae Philodendroideae Pothoideae Pothoideae excluded from Araceae Pothoideae Pothoideae Pothoideae Pothoideae Pothoideae Lasioideae Lasioideae Lasioideae Philodendroideae? Lasioideae ynonym of Dracontium* Lasioideae Volume 77, Number 4 1990 Grayum Evolution and Phylogeny of Araceae 631 TABLE 1. Continued. Numbers Taxa of species Ranges Placements in this paper Subtribe Pycnospathinae Pycnospatha Thorel ex Gagnepain 2 Southeast Asia Lasioideae Tribe Pythonieae (properly Thomsonieae) Pseudohydrosme Engl. 2 Africa oo Anchomanes Schott 10 Africa hil oideae Plesmonium Schott (2) South Asia Synonym of Amorphophallus’ msonia Wallich (2) South Asia Synonym of Amorphophallus Pseudodracontium . E. Br. 7 Southeast Asia Aroideae Amorphophallus Blume ex Decne. 100 Asia, Africa, Australia Aroideae Tribe Nephthytideae Nephthytis Schott 7 Africa Philodendroideae Cercestis Schott 12 Africa Philodendroideae Rhektophyllum N. E. Br. (2) Africa Synonym of Cercestis* Tribe Montrichardieae Montrichardia Crüger 3 Neotropics Philodendroideae Subfamily Philodendroideae Tribe Philodendreae Subtribe Homalomeninae Furtadoa M. Hotta 2% Southeast Asia Philodendroideae Homalomena Schott 140 Southeast Asia, Neotropics Philodendroideae Diandriella Engl. (1) New Guinea Synonym of Homalomena* Subtribe Schismatoglottidinae Schismatoglottis Zoll. oritzi 100 Southeast Asia, South America Philodendroideae Bucephalandra Schott = Southeast Asia ilodendroideae Phymatarum M. Hotta 2 Southeast Asia Philodendroideae Aridarum Ridley 7 Southeast Asia Philodendroideae Heterardarum M. Hot 1 Southeast Asia Philodendroideae Hua Bogner & Nicolson 4 Southeast Asia Philodendroideae Piptospatha N. E. Br. 10 Southeast Asia Philodendroideae Subtribe Philodendrinae Philodendron Schott 350-600 Neotropics Philodendroideae Tribe Anubiadeae Anubias Schott 14 Africa Philodendroideae Tribe Aglaonemateae Aglaonema Schott 21 Southeast Asia Philodendroideae Aglaodorum Schott 1 Southeast Asia Philodendroideae Tribe Bognereae? Bognera Mayo & Nicolson 1 South America Philodendroideae Tribe Dieffenbachieae Dieffenbachia Schott 30+ Neotropics Philodendroideae Tribe Zantedeschieae Zantedeschia Sprengel 6 Africa Philodendroideae 632 Annals of the Missouri Botanical Garden TABLE 1. Continued. Numbers Taxa of species Ranges Placements in this paper Tribe Typhonodoreae Typhonodorum Schott 1 Africa Philodendroideae Tribe Peltandreae Peltandra Raf. 2 North America Philodendroideae Subfamily Colocasioideae Tribe Colocasieae (properly Caladieae) Subtribe Steudnerinae Steudnera K. Koch 8 Southeast Asia Colocasioideae Remusatia Schott 2 South Asia, Africa Colocasioideae Gonatanthus Klotzsch 3 South Asia Colocasioideae Subtribe Hapalininae Hapaline Schott 5 Subtribe Caladiinae Jasarum Bunting 1 Scaphispatha Brongn. ex Schott 1 Caladium Vent. 8 Aphyllarum S. Moore 1 Chlorospatha Engl. 16 Xanthosoma Schott 45 Subtribe Colocasiinae Colocasia Schott 8 Subtribe Alocasiinae Alocasia (Schott) on 70 Xenophya Schott (2) Tribe Syngonieae Syngonium Schott 33 Tribe Ariopsideae Ariopsis Nimmo 1 Subfamily Aroideae Tribe Stylochaetoneae Southeast Asia South America South America Neotropics Southeast Asia Southeast Asia Southeast Asia Neotropics Southeast Asia Africa Madagascar Madagascar Madagascar South America Stylochaeton Lepr. 21 Tribe Arophyteae Carlephyton Jum. 3 Colletogyne Buchet 1 Arophyton Jum. T Tribe Asterostigmateae (properly Spathicarpeae) Mangonia Schott 2 Taccarum Brongn. ex Schott 5 Asterostigma Fischer & С. Meyer 7 Gorgonidium Schott 3 Synandrospadix Engl. 1 Сеагит N. E. Br. 1 Spathantheum Schott 2 Spathicarpa Hook. 7 South America South America Colocasioideae Colocasioideae Colocasioideae Colocasioideae Colocasioideae Colocasioideae Colocasioideae Colocasioideae Colocasioideae Synonym of Alocasia* Colocasioideae Aroideae Lasioideae Philodendroideae Philodendroideae Philodendroideae Philodendroideae Philodendroideae Philodendroideae Philodendroideae Volume 77, Number 4 633 Grayum 1990 Evolution and Phylogeny of Araceae TABLE 1. Continued. Numbers Taxa of species Ranges Placements in this paper Tribe Protareae Protarum Engl. 1 Seychelles Colocasioideae Tribe Callopsideae Callopsis Eng). 1 Africa Philodendroideae Tribe Zomicarpeae Zomicarpa Schott 3 South America Colocasioideae Zomicarpella N. E. Br. 2 South America Philodendroideae? Filarum Nicolson 1 South America Philodendroideae Ulearum Engl. 2 South America Philodendroideae Tribe Areae Subtribe Arinae Агит L. 24 Eurasia Aroideae Dracunculus Miller 2 Mediterranean Aroideae Helicodiceros Schott 1 Mediterranean Aroideae Theriophonum Blume 5 ndia Aroideae Typhonium Schott 25 Southeast Asia, East Asia, Australia Aroideae Sauromatum Schott 6 South Asia, Africa Aroideae Eminium (Blume) ott 7 West Asia, Central Asia Aroideae Biarum Schott 16 Mediterranean Aroideae Subtribe Arisarinae Arisarum Miller 3 Mediterranean Aroideae Subtribe Arisaematinae Arisaema C. Martius 150 Africa, Asia, North America Aroideae Subtribe Pinelliinae (properly Atherurinae} Pinellia Ten. East Asia Aroideae Subtribe Ambrosininae Ambrosina Bassi 1 Mediterranean Aroideae Subtribe Cryptocoryninae Lagenandra Dalz. 12 South Asia Aroideae Cryptocoryne Fischer ex Wydler 50 Southeast Asia Aroideae Subfamily Pistioideae Pistia L. 1 Pantropical Aroideae ' (1920b; as modified by Bogner (1978) and miscellaneous subsequent papers). Species numbers and geographic range data are mainly from Croat (1979). Only the most general indication of the geographic range is given for each genus. * See Nicolson (1984b); Philodendroideae sensu this paper (including Calla) is properly Calloideae. * African and South American species formerly included in Cyrtosperma are now assigned to Lasiomorpha Schott (1 sp.) and Anaphyllopsis A. Hay (3 spp.), respectively (Hay, 1988). *See Bogner (1985b) 5 See Bogner et al. (1985). * See Hotta (1985). * See Hay (in press). ? J. C. French (pers. comm.) reported secretion tubes from a second species of Zomicarpella recently discovered in Brazil by J. Bogner. Pollen of this species, as judged from unpublished SEM micrographs, more closely resembles pollen of Zomicarpa than that of Filarum or Ulearum. Thus, Zomicarpella may belong in Colocasioideae. Annals of the Missouri Botanical Garden erence in the present paper; however, this should not be taken to imply prejudice in favor of that scheme. The vindication of one or another estab- lished system of classification has not been an ob- jective of the present study. Rather, the phylogeny of Araceae is emphasized, with a view to deduce a defensible cladogram which can ultimately be- come the foundation for a new classification. But it is clear from the outset that many more of Engler's than of Hutchinson's tribes are natural groups, although, as will be seen, Engler’s subfam- ilies are not so robust. The suprageneric aroid taxa referred to in the text that follows are those of Engler (1920b), as modified by Bogner (1978), and include nomen- clatural changes mandated by Nicolson (1984b). In Tables 2 and 3 aroid genera have been assigned to subfamilies according to the new system (the linear order within these subfamilies being of no great significance, although accepted tribes are maintained intact). Table 1 provides a cross-ref- erence between the new system and the more fa- miliar neo-Englerian scheme. FossiL HISTORY OF THE ARACEAE The fossil record of monocots has been reviewed by Daghlian (1981), and the macrofossil record of Araceae partially reviewed by Gregor & Bogner ( . Macrofossils clearly show that the Araceae were already fairly diverse and well advanced by the early Tertiary. At least four modern subfamilies were represented by the Oligocene. At least two modern genera (Philodendron and Peltandra) are known from the Eocene, and Gregor & Bogner (1984) listed six others (Pistia, Arisaema, Lysichi- ton, Stenospermation, Orontium, and Calla) known from unspecified Tertiary epochs. There is one tentatively acceptable aroid fossil from the Paleo- cene of Central Asia, which would be the earliest record attributable to the family. Crepet (1978) suggested that the Araceae radiated in the late Cretaceous or Paleocene. Dilcher & Daghlian (1977) presented evidence of the presence of Philodendron subg. Mecono- stigma in the Eocene of Tennessee, based on fos- silized leaves; Mayo (1986), however, suggested these might belong to Peltandreae. Bogner (1976b) attributed a spadix preserved in Eocene amber to the tribe Monstereae. Madison & Tiffney (1976) surveyed the fossil record of seeds attributed to the Monstereae, and concluded that some fossils could be accommodated in that tribe but assigned others to the Pothoideae or Lasioideae; a few were interpreted as not being aroids. These records were mainly from the Oligocene or earlier. The pollen fossil record, according to the review of Muller (1981), does not reveal much about the family Araceae. Pollen (Liliacidites) representing a type known only in monocots extends back to the very early Cretaceous (Aptian) (Doyle, 1973; Wolfe et al., 1975; Walker & Walker, 1984), yet pollen assignable to extant monocot families does not appear until the late Cretaceous (Maestrichtian) (Muller, 1981). Most Cretaceous monocot pollen is of a general type (reticulate, monosulcate) known from many extant monocot families, and until an extensive survey of modern monocot pollen is avail- able, little use can be made of it at the family level (Daghlian, 1981). he only aroid pollen that can be recognized in the fossil record is the distinctive striate Spathi- phyllum type (Muller, 1981). This type of pollen is known from Mexico, Colombia, and Palau (all within the contemporary range of the genus), only as far back as the Miocene. It is perhaps significant that the easily identifiable pollen of Spathiphyllum has not turned up in earlier deposits, and that other distinctive types of aroid pollen, such as the retic- ulate, forate Anthurium type, have not yet been reported in the fossil record. CHARACTER VARIATION AND POLARIZATION GEOGRAPHICAL DISTRIBUTION The distribution of Araceae was briefly outlined in the introduction and has been covered in detail by Croat (1979), Grayum (1984), and Bogner (1987). Table 1 provides a brief description of the range of each aroid genus. The Araceae are overwhelmingly tropical, with only a few truly temperate genera—all of these in the Northern Hemisphere. The poleward atten- uation of the aroid flora is well exemplified by floristic data from South America: whereas the flora of Venezuela contains nearly 200 species of Arace- ae (Bunting, 1979), there are only 17 in the flora of Argentina (Crisci, 1971), and only the cosmo- politan Pistia stratiotes occurs in Chile (Marticore- na & Quezada, 1985). he Asian tropics have the largest number of genera (44); the Neotropics have about 36 genera, but a higher percentage of these are endemic, and the Neotropics have significantly more species (Croat, 1979). The African aroid flora, with just 20 genera, is relatively impoverished, as is fre- quently the case with other tropical wet-forest taxa. Aroid genera tend to be restricted to major phy- togeographical realms. With Hay’s (1988) recent fragmentation of Cyrtosperma, the monotypic, cosmopolitan and artificially spread Pistia remains Volume 77, Number 4 1990 Grayum 635 Evolution and Phylogeny of Araceae as the only pantropical aroid genus. Arisaema oc- curs principally in temperate to subtropical regions ut has extended sparingly into central Africa, Southeast Asia, and Mexico. Calla, Symplocar- pus, and Lysichiton occur in temperate or boreal regions of North America and Eurasia. Few genera are shared between any two tropical realms. Spathiphyllum, Homalomena, Schisma- toglottis, and perhaps Scindapsus are common to Southeast Asia and the Neotropics, and are absent from Africa. Rhaphidophora, Sauromatum, Po- thos, Amorphophallus, and Remusatia occur in Africa and Eurasia (and the latter three in Australia as well); however, in the cases of Remusatia and Sauromatum (and of Pothos in Africa) we are clearly dealing with recent dispersal of a single, presumably highly vagile species. Epipremnum and Typhonium occur naturally in Eurasia and Aus- tralia. No genera are shared solely by Africa and Australia, nor by Africa and the Neotropics. Only two natural tribes, Monstereae and Las- ieae, are pantropical (though represented in Africa by just two and one species, respectively). In both cases, the African species are more closely related to Asian than American elements (Grayum, 1984). The Englerian subtribe Arinae, a closely knit group of eight genera, has a strikingly Tethyan distri- bution: from western Europe and the Canary Is- lands, throughout the Mediterranean region, and eastward to Hupeh, the Philippines, New Guinea, an Suesasland. hical regions harbor con- centrations of ia sra taxa. For example, Anaphyllum, Theriophonum, Lagenandra, and two sections of Amorphophallus are endemic to southern India and Sri Lanka. Pedicellarum and five genera of subtribe Schismatoglottidinae are endemic to Borneo, and a single genus each to Sumatra (Furtadoa) and New Guinea (Holochla- mys). Pinellia and Amorphophallus sect. Dysa- morphophallus are strictly East Asian. Pseudohy- drosme, Anchomanes, Nephthytis, Cercestis, An- ubias, and two sections of Amorphophallus are restricted to tropical West Africa. Typhonodorum and all three genera of the tribe Arophyteae are Madagascan (the former also occurring, perhaps adventively, in East Africa and Mauritius). Pro- tarum is unique in being endemic to the Seychelles, Gymnostachys to Australia. In the New World, Orontium and Peltandra are confined to eastern North America. The tribes Spathicarpeae and Zomicarpeae occur in a zone fringing the Amazon basin, from Bahia on the east coast of Brazil, southward across northern Para- guay and Bolivia, and northward along the Andean foothills. The New World members of tribe Lasieae show vestiges of a similar distribution. Jasarum is the only aroid genus restricted to the Guayana Shield. Additional biogeographical details relevant to particular genera can be found in the second half of this paper. The geographic distribution of the Araceae is complicated and difficult to interpret. Willis (1949), on the basis of his pre-plate tectonics biogeographical analysis of the family, suggested that the Araceae are polyphyletic. Raven & Axelrod (1974), with the benefit of plate tectonic theory (though perhaps handicapped by their use of Hutchinson's tribal classification), aptly described the distribution of Araceae as “West Gondwanalandic-Laurasian." No Araceae can definitely be traced to an East Gon- dwanalandic origin (Gymnostachys and Protarum being perhaps the best candidates). The pantropical distribution of certain indisputably natural aroid taxa—e.g., the tribes Lasieae and Monstereae— strongly suggests that the family was well estab- lished in tropical West Gondwanaland prior to the separation of Africa from South America. The al- ternative hypothesis, that South America and Af- rica were colonized separately from Laurasia fol- lowing the breakup of West Gondwanaland, seems highly unlikely in light of the long-standing sepa- ration of North and South America (from the Ju- rassic to about the Oligocene; see Raven, 1979). The “relict” distribution patterns of such phenet- ically primitive groups as Spathicarpeae and Las- ieae in South America, and Zamioculcadeae in Africa, imply a Gondwanalandic origin for these taxa as well. It seems probable, then, that Araceae were pres- ent in West Gondwanaland at an early date and radiated there, with some of these elements ulti- mately reaching tropical Asia by way of Africa. That African members of Monstereae and Lasieae appear more closely related to their Asian than neotropical counterparts may be evidence of this. These migrations may have taken place during the late Cretaceous or, more likely, in the Paleocene or Eocene, at which times the two land masses appear to have been connected (Raven, 1979). The African aroid flora was presumably severely depleted due to the aridity of Neogene and Qua- ternary times (Raven & Axelrod, 1974), with some groups (Spathiphylleae, Philodendron, Peltan- dreae, Colocasioideae) pushed out or extirpated and the remainder restricted to refugia or subjected to heavy selective pressures. But the early presence of Araceae in West Gon- dwanaland does not establish а Gondwanalandic 636 Annals of the Missouri Botanical Garden origin for the family. Though extant north-tem- perate Araceae are few, phenetically (and presum- ably cladistically) primitive taxa are disproportion- ately represented among them: notably the four genera of Engler’s subfamily Calloideae, plus Acor- aceae, a putative sister taxon to Araceae. Also noteworthy in this connection is the absence of Araceae in the truly temperate Southern Hemi- sphere; i.e., if Araceae originated in tropical West Gondwanaland and later extended into the north- temperate realms, why did the family not simul- taneously extend southward? Thus it is perhaps reasonable to propose that the Araceae originated in temperate or subtropical Laurasia, then reached West Gondwanaland via Eurasia (the North Amer- ican route having been unavailable) at an early date (late Cretaceous?) and underwent a major radiation in West Gondwanaland. Some elements of this Gondwanalandic aroid flora then migrated secondarily into Eurasia, a few (Philodendron, Pel- tandra, Orontium?) even reaching North America via the North Atlantic route—this having been the main route prior to the Eocene (Raven & Axelrod, 1974; Wolfe, 1975), which is the age of the Ten- nessee Philodendron fossil discussed earlier. The present confinement of Peltandra and Orontium to the eastern United States perhaps reflects the barrier imposed by a north-south epicontinental seaway in the Great Plains region up to the Pa- leocene (Wolfe, 1975), and subsequent climatic changes. The Tethyan distribution of Engler’s subtribe Arinae may reflect a proximal Laurasian origin for this group. Presumably, the ancestral Araceae did not migrate en masse from Laurasia into West Gondwanaland; some remained behind. These pri- mal Laurasian elements, later isolated from their Gondwanalandic counterparts by the Tethys Sea, may then have undergone a radiation of their own in the warm, subtropical forests that prevailed, by the Eocene, along the northern margin of that sea from Western Europe to Indomalaysia (Wolfe, 1975). In addition to the Arinae, groups such as Pinellia, Arisaema, Ariopsis, Arisarum, Ambro- sina, Pistia, and the tribe Thomsonieae are im- plicated in this scenario. Evidence for this inter- pretation includes the present diversity of these groups (especially Arisaema and Amorphophal- lus) in Laurasian areas, as well as their poor rep- resentation in Southeast Asia and Africa and com- plete absence from the Neotropics. Though the proposed Tethys radiation could have involved taxa newly emigrated from Africa, the putatively close interrelationship of the Tethyan taxa and the ab- sence of such groups as Monsteroideae and Phil- odendroideae undermine that scenario. Arisaema clearly reached North America from Eurasia (rather than vice versa), and rather re- cently at that; the Bering Straits route was at relatively low latitudes and became important dur- ing the Eocene and Oligocene, after the closure of the North Atlantic route (Raven & Axelrod, 1974; Wolfe, 1975). Calla, Symplocarpus, and Lysi- chiton may be subject to the same consideration. VEGETATIVE MORPHOLOGY AND ANATOMY Extensive summaries of aroid vegetative mor- phology may be found in Engler (1877, 1920b) and Grayum (1984), and of vegetative anatomy in Solereder & Meyer (1928), Grayum (1984), and, most recently, French (in press). Only the most basic and phylogenetically significant features will be discussed here. provides overviews of important anatomical features. Germination. Germination in most aroids is cryp- tocotylar, or hypogeal (Boyd, 1932; Dahlgren & Clifford, 1982), although in Arum and Philoden- dron it is perhaps epigeal. Seedling morphology in Araceae is highly variable, and has recently been surveyed by Tillich (1985; see discussion in Gray- um, in press b). The cotyledon is generally similar to a foliage leaf, and in a few genera (Eminium, Sauromatum) may even be stalked—an extremely rare circumstance among monocots (Tillich, 1985). The first leaf of the plumule is usually a cataphyll (a leaf-sheath lacking a petiole and blade) but may be a foliage leaf in several genera (Engler, 1920b). The hypocotyl becomes converted into a corm or tuber in Pinellia, however the latter organ is of epicotylar origin in Arum, Arisaema, Arisarum, and Dracunculus (Tillich, 1985). As is typical of monocots, the radicle is short-lived, subsequent roots arising adventitiously from the stem (French, in press). Shoot development and morphology. Aroid shoot development has been freshly interpreted and sum- marized by Ray (1987b). Growth of the main axis in Araceae, at least up to the time of the initial flowering episode, is always monopodial. After flow- ering, growth is sympodial, with four exceptions: in Pothos, Pothoidium, Heteropsis, and presum- ably Pedicellarum, monopodial growth continues exclusively. The inflorescences in these genera are terminal (as in all Araceae) on determinate side shoots, and not axillary as incorrectly described in some previous interpretations. Volume 77, Number 4 1990 Grayum 637 Evolution and Phylogeny of Araceae Ray (1987b) recognized three major categories of sympodial growth in Araceae: anisophyllous, in which the articles (a stem section produced from a single meristem, from its inception to its termi- nation in a developed or aborted inflorescence) have a variable number of leaves, whether foliage leaves or cataphylls; homeophyllous, in which the articles have a fixed number of leaves; and intermittent homeophyllous, in which the above two types of i Maur growth alternate. since the homeo- intermittent | growth is ишү associated with active flowering, this cat- egory will be discussed in a later section. Anisophyllous sympodial growth occurs in a wide range of aroid subtaxa (listed in Ray, 1987b) and is probably the most primitive growth type within the family. Its distribution shows some taxonomi- cally intriguing patterns; anisophyllous growth is, for example, restricted in Anthurium to sect. Poly- phyllium, and in Philodendron to subg. Pter- omischum. Some species of the latter taxon exhibit proleptic renewal—i.e., the renewal shoots develop from a bud that has rested, relative to the parent shoot (Ray, 19872). (Earlier incorrect interpreta- tions of this phenomenon [ Madison, 19782; Blanc, 1980] characterized growth in subg. Pteromis- chum as monopodial [Ray, 1987b].) Sympodial growth in all other Araceae is sylleptic (the bud does not rest relative to the parent shoot), with the apparent exception of a few species of Monstera (Ray, 1988). Various subtypes of homeophyllous sympodial growth may be recognized in Araceae, depending upon the number of leaves (whether foliage leaves or cataphylls) borne by each article (Ray, 1987b). Monophyllous growth is exclusively associated with the development of *'second-order" inflorescences and is discussed later. Diphyllous sympodial growth is characteristic of Philodendron (except subg. Pteromischum) and also occurs in a slightly dif- ferent form in Symplocarpus. Triphyllous sym- podial growth is the rule in Anthurium (except sect. Polyphyllium) and is also known from a few other unrelated genera (Ray, 1987b). Tetraphyl- lous growth is exclusive to Orontium, and penta- phyllous growth to Stenospermation, where it is only imperfectly developed. In the vast majority of aroids, each renewal shoot during sympodial growth arises in the axil of the penultimate leaf (whether foliage leaf or ca- taphyll) below the spathe, except in monophyllous homeophyllous growth. In Symplocarpus and per- haps Lysichiton, however, the renewal shoot reg- ularly arises in the axil of the ultimate leaf, and in Gymnostachys it arises in the axil of the antepen- ultimate leaf before the *'spathe" (interpreted here as a "mesobracteole"; Ray, 1 ). Sympodial growth predominates among mono- cots according to Holttum (1955), who believed it to be more primitive than monopodial growth. Moore & Uhl (1982) also believed sympodial growth to be primitive within the palms, and this is certainly the case for Araceae as well. Permanent mono- podial growth may well represent a neotenic re- tention of this normally juvenile condition (Grayum, 1984), as also suggested by Hay (1986) and Bar- abé (1987). The Araceae exhibit a great diversity of growth habits (see Grayum, 1984; Bogner, 1987), includ- ing caulescent (e.g., Dieffenbachia) and acaules- cent (Spathiphyllum) herbs, pachycaulous, ar- borescent plants to 5 m or more in height (Montrichardia, Xanthosoma), monopodial, shrubby vines (Pothos), and root-climbing, sym- podial vines (Monstera, Philodendron, Syngon- ium). True epiphytes (i.e., with seeds germinating on the host plant) are rare in Araceae. y Ап- thurium and Stenospermation consist largely of epiphytic species. Pistia stratiotes is unique in ing a floating aquatic. The two most common growth habits in Araceae, at the generic level, are the rhizomatous and tu- berous (or cormose) forms. These types are in general easily distinguished, although short and thick and even erect (as in Symplocarpus and Podolasia) so-called rhizomes may blur the dis- tinction. In general, rhizomatous growth habits tend to be associated with wet habitats (Cryptocoryne, Jasarum), tuberous growth with seasonally dry or arid habitats (Biarum, Eminium). Yet both types sometimes occur in the same genus (Stylochaeton, Arophyton, Arisaema) or even species, as in Ar- isarum vulgare (Galil, 1978) and A. simorrhinum (Herrera, 1988). The development of **tubers" and in Araceae is little understood, as is the case for monocots in general (Tomlinson, 1970). Aroid or- gans т inip in one or another of these two categ sarily homologous. Hotta (1971) ы Tuch (1985) distinguished two types of tuber development in the Araceae but did not review their distribution comprehensively. What might the growth habit of the hypothetical aroid ancestor have been? The question is difficult but one can begin by eliminating the ly specialized forms: tuberous geophytes, ано, floating aquatics, monopodial vines, and probably even sympodial vines (Hotta, 1971; Ray, 1989). [13 ы 99 rhizomes 638 Annals of the Missouri Botanical Garden Workers on Liliales (Schlittler, 1955; Conover, 1983) have concluded that nonrhizomatous forms with much-branched, leafy stems (resembling the modern-day Philesiaceae) represent the primitive type from which arborescent and ultimately rosette forms were derived via condensation of the axis and other changes. Tomlinson (1962), by contrast, believed that arborescent forms are primitive in Zingiberales, and that herbaceous and vining forms arose partly via neoteny. This agrees with Corner’s ' which dic- tates that a “stout pachycaul stem with large com- (1954) controversial “Durian Theory,’ pound leaves was the ancestral growth form of the See Hay (1986) for a full-blown appli- cation of the Durian Theory to aroid phylogenetics. I believe that the rhizomatous, or perhaps erect caulescent, growth habit is basic to Araceae. These Arales.” appear to be the most “flexible”” growth forms in the family, i.e., those from which all of the more extreme forms might have been derived in the most parsimonious manner. The rhizomatous habit is particularly widespread and occurs in many osten- sibly primitive groups—e.g., Gymnostachys, the Orontieae, Lasieae, and others. The derivation of tuberous from rhizomatous forms seems but a short step, and Li (1980) proposed such an interpretation for Arisaema. Leaf morphology. Aroid leaves are always alter- nate and when fully developed typically consist of a sheath, petiole, and lamina. Ray (19872) recently reviewed aroid leaf types and presented a detailed classification based on morphology and function. His three main morphological types are: foliage leaves, which are unreduced (generally with all three of the above parts represented); reduced leaves, 10-70% the size of foliage leaves on the same plant and usually with the lamina very small to obsolete; and cataphylls, reduced to less than 10% and generally consisting of just a sheath. The first leaf on a new vegetative axis is called the prophyll, which is a cataphyll in all Araceae except Gymnostachys, Orontium, and Symplo- carpus, in which it is a foliage leaf (Ray, 1988). The leaf subtending an inflorescence within a com- pound inflorescence is termed a bracteole; brac- teoles are apparently always cataphylls. Prophylls and bracteoles are morphologically indistinguish- able (Ray, 1987a); both have a two-keeled struc- ture, as is typically the case for monocots in gen- eral, whereas сек ia in other positions generally have a single The leaf н following the prophyll is termed a mesophyll; by analogy, the leaf imme- diately following the bracteole is called a meso- bracteole (the latter are known definitely only in Gymnostachys, where they are cataphylls). The mesophyll is frequently a cataphyll or reduced leaf. In such cases, additional cataphylls or reduced leaves may follow; these are also referred to as mesophylls. Prophylls, bracteoles, mesophylls, and meso- bracteoles may be described as either “sylleptic” or “proleptic” these leaf types may differ depending on the type of growth involved. For example, proleptic pro- phylls are usually tiny, and the proleptic mesophylls numerous; sylleptic prophylls, on the other hand, are comparatively large, and the sylleptic meso- phyll is generally solitary and often a foliage leaf (Ray, 1987a) The leaf subtending the inflorescence, whether developed or aborted, in sympodial growth is termed a sympodial leaf. Although sympodial cataphylls occur, this is usually a foliage leaf in which the sheath is rudimentary and the petiole base does not encircle the stem. In species of Philodendron with diphyllous sympodial growth (i.e., the entire genus apart from subg. Pteromischum), all of the foliage leaves beyond the juvenile monopodial phase are sympodial leaves, most of which subtend abort- ed inflorescences. The same is true of those sections of Anthurium with triphyllous growth (i.e., all ex- cept sect. Polyphyllium), since both the prophyll and mesophyll are always cataphylls (Ray, 1988). Any leaves between the mesophyll(s) and the sympodial leaf are termed monopodial leaves. These inasmuch as the characteristics of are almost always foliage leaves, usually with a well-developed sheath and a base encircling the stem. The morphological differences between sym- podial and monopodial leaves facilitate the differ- entiation of juvenile from adult plants in many aroid genera, even in the absence of developed inflores- cences. Other, more specialized leaf types, including re- duced leaves or cataphylls associated with dispersal shoots (“flagelles””) and dormancy, were discussed by Ray (1987a) Intravaginal squamules, characteristic of the Zosterales, are known in just a few genera of Arace- ae: some Philodendron species (Engler, 1912; Rit- terbusch, 1971; Blanc, 1977) and apparently also Cryptocoryne and Lagenandra (Engler, 1920a). Phyllotaxy in Araceae may be either distichous or spiral (helical). Spirally arranged leaves almost always diverge at 2/5. Distichous leaves occur almost exclusively in the subfamilies Pothoideae, Monsteroideae, and Calloideae (Grayum, 1984). Outside these subfamilies, only Heteroaridarum (Hotta, 1976) and a single species of Aridarum Volume 77, Number 4 Grayum 639 1990 Evolution and Phylogeny of Araceae (Bogner, 1981b), both in Philodendroideae, are Ptyxis in Araceae is supposed to be exclusively known to possess distichous leaves. Distichy, in monocots, is characteristic of such groups as Gramineae, Orchidaceae, Zosterales, and Zingiberales. It is unknown in Pandanaceae, rare in Palmae, and known from just four genera in Cyclanthaceae. Some authors (e.g., Harling, 1958) have assumed that distichy is primitive in seed plants. Beck et al. (1982), however, concluded that spiral phyllotaxy is more primitive—another cor- nerstone of the “Durian Theory" (Corner, 1954). Dahlgren & Clifford (1982) suspected that spiral phyllotaxy is basically primitive but that it may have been secondarily derived in some cases. The distribution of distichy in Araceae suggests that the latter may be the case here. That the first leaves of araceous shoots are nearly always distichous, even in species with ultimately spiral phyllotaxy (Engler, 1877), may be germane in this connection. Furthermore, Periasamy & Muruganathan (1986) reported that the phyllotaxy of leaf primordia in Arisaema is distichous. Tomlinson (1966) has sug- gested that distichy in Commelinaceae may have come about via neoteny, a hypothesis that ought to be considered for Araceae as well. Most Araceae with distichous leaves are climbers, so there is probably also a functional correlation. Modifications of the leaf sheath are rare in Ara- ceae. Їп a few taxa, such as Сайа, some Philo- dendron species, and several Schismatoglottidinae, the sheath is free at the tips and has been referred to as а "ligule-" or "stipulelike" structure (Dahl- gren & Clifford, 1982; French, in press). The most important petiole character in Araceae is the presence or absence of a geniculum. A ge- niculum (pulvinus) at the apex of the petiole occurs in most Pothoideae and throughout Monsteroideae (see Grayum, 1984), i.e., mostly i in vining genera. In Zamioculcadeae and som of Anthurium (e.g., A. oerstedianum) the geniculum is located in the middle of the petiole. Outside of these subfamilies, genicula are extremely rare (Anubias, some species of Homalomena, perhaps some species of Philodendron). All aroid leaf blades are homologous structures, and, though they may superficially resemble dicot leaves, develop in a manner typical of monocots (Kaplan, 1973). Aroid leaves typically undergo heteroblastic de- velopment, that is, there is a marked but gradual change in morphology from juvenile to adult forms. Only a few vining genera (Pothos, Rhaphido- phora, Monstera) show true heterophylly, defined as an abrupt switch from juvenile leaves to very different adult leaves. supervolute according to Cullen's (1978) survey of 16 genera representing every subfamily except Pistioideae. But involute ptyxis is now known to characterize Lagenandra (de Wit, 1978) and An- thurium sect. Pachyneurium (Croat & Sheffer, 1983). Supervolute ptyxis is the most frequent condition among monocots and is to be regarded as primitive for Araceae. Although lobed and compound laminae are com- mon in Araceae, leaf margins are otherwise vir- tually always entire; serrate margins are extremely rare, known only in a few species of Arisaema, and are certainly derived (see Hickey, 1978). Peltate leaves occur chiefly in Engler's subfam- ily Colocasioideae (Grayum, 1984). This condition is exceedingly rare in other subfamilies, known from a single species each of Anthurium, Philo- dendron, and Pinellia (P'ei, 1935), and from two species of Homalomena sect. curmeria (but see also Ariopsis). Lamina shape is highly variable in Araceae and has been discussed in depth by Grayum (1984) and Bogner (1987). Linear, lanceolate, elliptic, ovate, cordate, sagittate, hastate, pedately com- pound, and palmately (or radiately) compound leaves are widespread, and most or all of these shapes may occur even within single genera (Phil- odendron and especially Anthurium). Other types are less common. The narrowly linear, grasslike leaves of Gymnostachys are unique within the family. Truly pinnate leaves, in the dicot sense, i.e., with the leaflets abscising individually, are a only from Zamioculcas (once-pinnate) and Gonatopus (bi- or tripinnate). These genera prove inaccurate the assertion of Dahlgren & Clifford (1982) that such leaves do not exist among mon- ocots. Pinnately or even bipinnately lobed leaves are more common, occurring commonly in Mon- steroideae and sparingly in most other subfamilies. The development of leaf pinnation in various aroid subgroups is not homologous. In Monstero- ideae, pinnate leaves result from a necrotic process in which growth stops and tissue rots away (Schwarz, 1878; Ertl, 1932; Hotta, 1971; Madison, 1977). Apparently, the unusual decompound leaves of Ап- chomanes and presumably Pseudohydrosme also develop in this fashion (Riedl, 1978), as do leaf fenestrations in many Monsteroideae (Madison, 1977). Pinnation in Philodendron is achieved by means of differential growth rates. This **marginal type" process is also involved in the development of some palmately lobed aroid leaves, as well as in more complex situations, and is characteristic of several 640 Annals of the Missouri Botanical Garden Lasioideae, Colocasioideae, and Aroideae, including the tribe Spathicarpeae (Ertl, 1932; Hotta, 1971; Riedl, 1978; Periasamy & Muruganathan, 1986). The complex, deeply lobed leaves of many Las- ieae in which the lamina is basically trisect with the main divisions further divided (pedately and/ or pinnately) may represent a distinctive type. Here the early stages of adult leaves possess three dis- tinct leaflet primordia, which are subsequently fur- ther dissected, presumably via the marginal pro- cess, at least in Thomsonieae. Pedately compound or pedately lobed leaves are particularly common in Colocasioideae and Aro- ideae. Their development, at least in some cases, is known to involve marginal dissection (Ertl, 1932; Periasamy & Muruganathan, 1986). Curiously, pedate and tripartite leaves in Colocasioideae are restricted to New World genera (Chlorospatha, Caladium, Xanthosoma, Syngonium; but see also Protarum). They are more widespread in Aroideae (see Grayum, 1984). Truly palmately compound (or radiate) leaves are known with certainty only in Anthurium and Arisaema. Both of these genera also exhibit clearly pedate or pedatisect leaves, as well as tripartite leaves, which are not identifiable as pedate or palmate. It is highly doubtful that all of these occurrences of pedate and palmate leaves can be homologized. Madison (1978b) described two distinctive heter- oblastic sequences а, palmate Anthurium species and suggested that p may ha ve as many as seven ан in this genus sina: Нона (1971) has contended that in aroids, “Һе compound leaf is apparently an evolved type orig- inated by the dissection of a simple leaf”; this mirrors Engler’s (1884) opinion on the subject. though alternative interpretations are possible (see especially Corner, 1954; Hay, 1986; Barabe, 1987), I agree that elliptic, ovate or cordate simple leaves are the most primitive type in Araceae, and that other types, especially compound leaves, are ate leave derived. This view receives some support from the angiosperm fossil record (see Hickey, 1978). More- over, sagittate and especially lobed and compound leaf types often seem strongly correlated, in Arace- ae, with condensation of the axis and often with a tuberous, geophytic habit (see Grayum, 1984). In Biarum and Eminium simple leaves probably have been derived secondarily from compound leaves. Root morphology. For details of aroid root mor- phology, see the recent reviews of Grayum (1984) and French (in press). Hotta (1971) stated that “in general, the features of the root are useless for Although this will classification of the Araceae.” no doubt prove to be an exaggeration, external root morphology has yielded no taxonomically useful characters to date. Roots in all Araceae, as is typical of monocots in general, are adventitious beyond the seedling stage (Engler, 1905). Roots never become tuberous in Araceae (Hotta, 1971). In aroids with tuberous or cormose stems (many Aroideae, Lasioideae, and Calloideae), contractile roots occur and serve to pull the stem deeper into the soil (Hotta, 1971; French, in press). Climbing and epiphytic aroids of all subfamilies typically bear aerial roots. These are frequently of two types: stout anchoring roots and longer feeder roots, which may extend to the ground from great heights. These two types differ markedly anatom- ically and physiologically poh 1888; see French, in press, for an exhausti ion). Feeder roots arise only at the nodes, whereas anchor roots may arise along the internodes as well (French, in press). Epidermal anatomy. The epidermis of above- ground organs in Araceae is typically glabrous. Trichomes and other types of ornamentation are quite rare; their distribution, which is spotty and not of much phylogenetic significance, has been summarized by Grayum (1984) and French (in press). The araceous cuticle shows great variation in thickness and ornamentation (Solereder & Meyer, 1928; Crepet, 1978; Behnke & Barthlott, 1983; Grau, 1983; French, in press). As yet, however, this feature has not been adequately surveyed. Ground epidermal cells of aroid leaves are de- scribed by Grau (1983) and French (in press). Of much greater potential phylogenetic significance are stomata. One of the few generalizations that can be made concerning aroid stomata is that they exhibit the “perigeneous” type of development characteristic of monocots (Pant & Kidwai, 1966; Tomlinson, 1974), in which a sister cell of the guard cell initial becomes a neighboring cell. In all other respects, aroid stomata are quite variable. Most genera have strictly or essentially hyposto- matic leaves, or else amphistomatic leaves with more stomata on the lower leaf surface (Solereder & Meyer, 1928; Pant & Kidwai, 1966; see Table 2). In Pistia, most stomata occur on the upper leaf surface; Pant & Kidwai (1966) interpreted this as homologous with the epistomatic condition of Lemnaceae but it is more likely an example of convergent evolution (see Grayum, 1984). Only Orontium has truly or essentially epistomatic leaves (Solereder & Meyer, 1928; Grear, 1973) Among the traditional stomatal types, paracytic, Volume 77, Number 4 1990 Grayum Evolution and Phylogeny of Araceae tetracytic (the hexacytic variant), and anomocytic all occur in Araceae (Stebbins € Khush, 1961; Pant & Kidwai, 1966; see Table 2), paracytic being the most frequent (contrary to Dahlgren & Clifford, 1982). Anomocytic stomata are known only in Pistia, all genera of the tribe Orontieae (Grear, 1973), and perhaps in some species of Arisaema (according to Dahlgren & Clifford, 1982). According to Stebbins & Khush (1961), paracytic stomata are primitive and the other types derived. Tomlinson (1970, 1974), however, studied the de- velopment of the stomatal apparatus in monocots and showed that the traditional system of classifi- cation based on morphological descriptions of ma- genesis of ' stomata, for example, may involve as few as two or as many as eight cell divisions. Tomlinson (1974; see also Dahlgren & Clifford, 1982) proposed a new classification of stomata based on developmental sequences, but unfortu- tung stomata is largely artificial: the nately no aroids have yet been studied from this viewpoint. Hydathodes occur in many hygrophilous Arace- ae, especially Philodendroideae, Colocasioideae, and of leaves, either singly or in large groups (Muller, 1919; Engler, 1920b; Solereder & Meyer, 1928). Müller (1919) recognized three types in Araceae (see Ta- ble 2) but did not adequately document the distri- bution of these within the family. Her “Philoden- dron type" hydathodes, which differ only slightly from normal stomata, are the most widespread and probably should be regarded as the most primitive. Most aroid roots possess a simple (one-cell-layer- thick) epidermis; however, two epidermal layers are said to occur in roots of Amorphophallus, Schismatoglottis, Aglaonema, Dieffenbachia, and other genera, three in id нн four in Za- mioculcas, and up to six in Anthurium and Hom- alomena (Lierau, 1888; Solereder & Meyer, 1928). The multilayered epidermis of aerial roots in some epiphytic Anthurium species may become second- arily thickened, nonliving, and capable of absorbing water. This is the only clear case in Araceae where- in a true velamen occurs, of the type found in many Orchidaceae (Lierau, 1888; Solereder & Meyer, 1928; Dahlgren & Clifford, 1982; French, in press). The outermost (or single) epidermal layer forms the root hairs, which are usually simple but may be forked or branched, especially in certain aerial roots (Lierau, 1888; Solereder & Meyer, 1928). Certain taxa, such as Calloideae (Krause, 1908), appear to lack root hairs, probably accounting for Cronquist's (1981) incorrect generalization of Ara- Aroideae, at the attenuated “drip-tips” ceae as ““mycorrhizal, without root hairs." Two basic types of root hair development are known in Arace- ae (Leavitt, 1904; see also Dahlgren & Clifford, 1982; French, in press): Type I, observed in Agla- onema, Dieffenbachia, Zantedeschia, Caladium, and Arisaema, is characterized by the ability of any protodermal cell to form a root hair; in Type II (Monstera, Anthurium) only specialized cells can form root hairs. This intriguing distribution suggests that Type II root hairs, being associated with phenetically more primitive genera, are prim- itive in Araceae. A more extensive inventory is highly desirable. Vascular anatomy. Many modern general texts (e.g., Dahlgren & Clifford, 1982) perpetuate the notion, based on the very incomplete survey of Cheadle (1942), that Araceae lack vessels in the stem. Although this is true of most genera, vessels have now been reported from stems of several vining Araceae, including species of Pothos (So- lereder & Meyer, 1928; Hotta, 1971), Epiprem- num, Rhaphidophora, Scindapsus (Hotta, 1971), and Philodendron scandens (French, in press). Vessels of Araceae are primitive, with scalariform perforation plates, and apparently rather difficult to distinguish from the large, specialized form tracheids” (Hotta, 1971). According to Cheadle’s theory (see, e.g., Chea- dle & Tucker, 1961), the specialization of vessels was unidirectional, beginning in the roots. If true, this evidently implies that the vining growth habit in Araceae is derived, since vessels of the stem occur only in climbers (cf. Hotta, 1971). Vascular bundles are more or less scattered in aroid stems, as is typical for monocots (Cronquist, 1981), although the majority tend to be concen- trated centrally. The course of leaf traces in the cortex, a subject apparently without taxonomic im- portance, is discussed by French (in press). More significant is the occasional presence of an inde- **vessel- pendent, permanent cortical vascular system, which has been observed in numerous Araceae (see Table 2), including many Pothoideae and all genera of the tribe Monstereae (Solereder & Meyer, 1928; French & Tomlinson, 1980, 1981a-d, 1984). French & Tomlinson (1983) reported that per- manent cortical vascular systems are characteristic of New World Colocasioideae but are absent in the Old World genera (with the possible exception of Hapaline; see French, in press). Vascular bundles in Araceae are of four different types: simple collateral bundles, compound bun- dles, amphivasal bundles, and compound/amphi- vasal transitional forms. The characteristics and Annals of the Missouri Botanical Garden distribution of these types were summarized by French & Tomlinson (1986) and French (in press), who pointed out that the four categories intergrade. Collateral bundles occur mainly in scandent species and species with perfect flowers; they are probably the primitive type. Amphivasal bundles predomi- nate in Araceae with unisexual flowers. Compound bundles, which may occur with collateral bundles, are apparently restricted to Zamioculcas, Rho- dospatha, Stenospermation, Cercestis, Montri- chardia, Philodendron, and Dieffenbachia (al- though Solereder & Meyer, 1928, reported them from numerous other genera). The course of vascular bundles in aroid stems was studied cinematographically by French & Tomlinson (1980, 1981a-d, 1983, 1984), whose results were summarized by French (in press). Four “artificial” categories are recognized, one of which (Pattern 2) is very widespread and perhaps prim- itive; the other three are restricted to Pothoideae, Monsteroideae, and a few genera here included in Philodendroideae. The same authors simultaneous- ly studied bud trace organization in Araceae, and French (in press) has summarized these findings as well. Three major patterns have been distinguished, but their largely erratic distributions are not easily amenable to broad phylogenetic interpretation, though the information may be valuable in specific cases. Phloem sieve-tube plastids in Araceae are of the “P”-type, as in all known monocots (Behnke, 1981). Phloem (P-) proteins are general in Araceae, as in most monocots (Behnke, 1981). It is important to note, however, that only six aroid genera in three subfamilies have been investigated with regard to these phloem characters. Araceous stems are not known to exhibit sec- ondary growth. Vessels are not known to occur in aroid leaves. The most taxonomically emphasized feature of the vascular system of aroid leaves is the venation. The definitive study of aroid leaf venation is that of Ertl (1932). Hotta (1971) also studied leaf ve- nation and apparently endorsed most of Ertl’s con- clusions, as follows: although venation types in Araceae vary greatly, and the extremes may ap- pear very different, the pattern of primary venation is the same for all; there are no sharp boundaries between the more distinctive forms, and transitional forms are of regular occurrence Ertl (1932) concurred with Бале s (1920) earlier conclusion that leaf venation has по decisive systematic significance. Nevertheless, the following observations of Ertl are of interest: he divided the Araceae into two major groups on the basis of leaf venation. To Group I, with essentially parallel or curved venation, belong Gymnostachys, all of En- gler’s Philodendroideae, Calloideae, and Monster- oideae, plus Cryptocoryne and Ambrosina of the Aroideae; the Colocasioideae, with a distinctive ve- nation type of their own, are connected with Phil- odendroideae and Monsteroideae by transitional forms. Group II, with essentially reticulate, dicotyle- donlike venation, includes most of the Pothoideae, Lasioideae, Aroideae, and Pistia. Ariopsis of En- gler’s Colocasioideae seems most closely related to this group. Pothos, via subg. Pothos, appears to be transitional between the two major groups. Seed- lings of most species in Group II have parallel venation typical of adult leaves of Group I (Hotta, 1971), emphasizing the unity within the family. The more overtly parallel type of venation, basic to monocots, is probably primitive in Araceae. Ac- cording to Ertl (1932), the Araceae all belong among those plants with a basically parallel or pinnate-parallel arrangement of vascular bundles in the leaves. Although superficially similar, the “reticulate” -veined leaves of certain Araceae differ fundamentally from those of dicots in that the primary vascular bundles never break down into a network, but rather pass independently to the mar- gin or the leaf apex (Ertl, 1932); the reticulate pattern is determined not by the major veins, but by the minor ones. This notion was recently reaf- firmed by Krishnamurthi & Geetha (1986); on the other hand, Inamdar et al. (1983) asserted that leaves of several monocot species studied by them (including Colocasia esculenta, Epipremnum au- reum, and Pistia stratiotes) “show reticulate ve- nation typical of dicotyledons.” Leaf venation patterns are accorded great, sometimes even exclusive, emphasis by leading contemporary aroid phylogenists. The taxonomic significance of these patterns is so dubious, how- ever, and their interpretation so subjective, that it would probably be better to ignore this character altogether. When authorities cannot even agree on venation differences distinguishing the angiosperm subclasses, it seems pointless to attempt to apply this character at lower taxonomic levels. Roots of most aroids apparently have vessels, although Caladium and Pistia may be completely vesselless (see French, in press). Aroid vessels are always of the type with scalariform perforation plates (see Dahlgren & Clifford, 1982). The stele of aroid roots is of the typical polyarch monocot type in nearly all cases, with xylem and phloem Volume 77, Number 4 1990 Grayum Evolution and Phylogeny of Araceae alternating with one another (Lierau, 1888; Engler, 1905). See French (in press) for additional details. Among aerial roots, the central cylinder occu- pies a larger portion of the cross section in feeder than in anchor roots and possesses broader vessels and sieve-cells. Moreover, anchor roots contain more mechanical tissue (Engler & Krause, 1908; Solereder & Meyer, 1928). Branch roots in Arace- ae originate opposite the xylem, as in most mon- ocots (Dahlgren & Clifford, 1982; French, in press). here is never any secondary growth. Ground tissues: general organization. The or- ganization of the ground tissue in aroid stems, leaves, and roots has been discussed in detail by Grayum (1984) and French (in press). Only the most basic and phylogenetically significant features will be repeated here. Stems of Araceae only rarely contain pith, as in Gonatopus (French, in press). The sclerenchy- matous sheath of the outermost vascular bundles in aroid stems is sometimes fused into a ring (So- lereder & Meyer, 1928). In other species, a con- tinuous or interrupted ring of collenchyma is pres- ent in the outer cortex of the stem or petiole; this situation is known only in Homalomena, Schis- matoglottis, Philodendron, Aglaonema, Dieffen- bachia, and Zantedeschia of the Philodendro- ideae, plus Asterostigma and Spathantheum of Engler’s Aroideae (Engler, 1920b). Each vascular bundle may or may not be ac- companied by groups of fibers (see Grayum, 1984, for a listing); in the peduncle and petioles of Pel- tandra, Typhonodorum, many Colocasioideae, and most Old World Aroideae, the bundles are accom- panied by a collenchymatous sheath (Solereder & Meyer, 1928). A typical endodermis has been found in the aboveground axes or subterranean stems of most Araceae (Engler & Krause, 1908; Solereder & Meyer, 1928; French & Tomlinson, 1980, 1981a- d, 1983, 1984; see Table 2). A very unusual type of endodermis, in which the vascular bundles are surrounded individually, is reported in stems of Schismatoglottidinae, plus Peltandra and Typhon- odorum (French & Tomlinson, 19814; French, in press). A sclerotic or thin-walled hypodermis may oc- casionally be present, as in the stems of Cercestis (French € Tomlinson, 1981c) and the tubers of Colocasia fallax (French € Tomlinson, 1983). Typical aroid leaves have a bifacial mesophyll, with the spongy layer accounting for the majority of the thickness (Solereder & Meyer, 1928). Truly centric mesophyll structure is found only in Ty- phonodorum (Dalitzsch, 1886). The cells of the palisade mesophyll in Araceae are generally short and broad, and most often in a single layer (Dal- itzsch, 1886). The cells of the spongy mesophyll have a very characteristic, mostly four-rayed, stel- late form (Dalitzsch, 1886). See French (in press) for additional details. In leaves of some species of Anthurium, Phil- odendron, and Caladium there is a chlorophyll- free hypodermis of one or more cell layers, directly under the epidermis of one or both sides (Dalitzsch, 6 The pith of aroid roots is initially of thin-walled cells, but these may become variously thickened in many genera (Solereder & Meyer, 1928). French (in press) provided a lengthy account of root cortex organization in Araceae. In all investigated aroids, the innermost cells of the cortex form a typical endodermis with Casparian strips, at least in young roots (Solereder & Meyer, 1928). The outermost layer of the cortex may form a one-cell-layer-thick (at least in Monsteroideae), suberized exodermis, particularly in species with a velamen or multilay- ered epidermis. Though not universally present, an exodermis has been reported from a wide range о species (see Solereder & Meyer, 1928). A small- celled, sclerenchymatous hypodermis develops be- neath the exodermis in roots of Culcasia, Cerces- tis, Furtadoa, Homalomena, and Philodendron (French, 1987a); Shishkoff (1987) reported a ““di- morphic hypodermis” from a variety of aroid gen- ех, сат referring to the tissue here termed k formation is known in aerial roots of Anthurium, several Monstereae, and Phil- odendron, following the appearance of a cork cam- bium (phellogen) below the exo- or hypodermis (see rench, in press). Secondary wall thickening can occur anywhere in the cortex of aroid roots, least frequently in the outermost layers. Ground tissue: universal features. Engler (1920b) placed great importance on two features of aroid ground tissue: the presence or absence of tricho- sclereids (distinctively branched sclereid cells), and the presence or absence and structure of tannini- ferous cells and *'laticifers." He regarded these features as having the distinct advantage of being universal: they are, with few exceptions, consistent within a given species, regardless of the organ examined or the age of the plant (Engler, 1920b). Trichosclereids are characterized by their elon- gate shape and regularly branching development (Nicolson, 1959, 1960b). Among monocots, they Annals of the Missouri Botanical Garden TABLE 2. Vegetative anatomy of Araceae. Summary of important characters. Subfamily, genus I II Ш IV V VI VII ҮШ XI Pothoideae Gymnostachys othos Pedicellarum Pothoidium Zamioculcas Gonatopus Anadendrum Anthurium Spathiphyllum Holochlamys Heteropsis Rhaphidophora Monstera Amydrium Epipremnum Scindapsus Alloschemone Stenospermation Rhodospatha Lasioideae Orontium Lysichiton Symplocarpus Stylochaeton Cyrtosperma „asia Anaphyllum Podolasia Urospatha Dracontioides Dracontium Pycnospatha Philodendroideae' Calla Furtadoa Homalomena Aglaonema Aglaodorum Zantedeschia Anubias Schismatoglottis Piptospatha Bucephalandra Phymatarum Aridarum Heteroaridarum Hottarum Typhonodorum Peltandra Philodendron Dieffenbachia 1 пат Id Id II П H TI тш АҺН uy Hd > >> - =ч > > А e o yy о ч + + Volume 77, Number 4 1990 Grayum Evolution and Phylogeny of Araceae 645 TABLE 2. Continued. Subfamily, genus VII УШ XI Bognera Carlephyton Colletogyne Arophyton ngonia Asterostigma Synandrospadix Taccarum Gorgonidium "earum Spathantheum Spathicarpa Cercestis Culcasia Pseudohydrosme Montrichardia Callopsis Filarum Ulearum Zomicarpella Colocasioideae Zomicarpa Jasarum Scaphispatha Caladium Aphyllarum Xanthosoma Chlorospatha Syngonium Hapaline Steudnera Alocasia Protarum Aroideae Arum Dracunculus Helicodiceros Arisaema Ariopsis Pinellia Pseudodracontium Amorphophallus Ah Ah Ah Ah Ah Ah P(A?) PE + sz Bez Har Har o Hesxsee aad rr] To > ++ 646 Annals of the Missouri Botanical Garden TABLE 2. Continued. Subfamily, genus I П ПІ IV ү VI VII VIII IX X XI Arisarum Ah — F Ambrosina I — F Cryptocoryne I = te = C? Lagenandra == es — C? Pistia П Ае А + + —# C^ + KEY (V) | Compound vascular bundles: General: ( )—in some spec + —presen = informato from dif- — —absent erent so (VI) Cortical vascular system: + — present (I) Venation type: Gwe p — —absent ticulate) (УП) Stem endodermis: + — pres I— Group I P individual type parallel = T— transitional (VIII) Trichosclereids: + —present d—derive —absent (from group ашан 7: negative indicated) reports (rest based on (II) Distribution of stomates: A —amphistomatic: general statements) e—tendency to (IX) Secretion vessels:? epistomatic — — completely lacking h—tendency to C— single cells (idioblasts) hypostomatic F — secretion files H — hypostomatic Т — secretion tubes E — epistomatic КА: (III) Stomate type: + — present P — paracytic — -— absent T — tetracytic XD Prisme: A— unde (М) Prisms: (IV) Hydathode t T P— Philodendron- -type A— Alocasia-type C — Colocasia-type ! al аы if used sensu this paper, 1.е., y from French (1988). to include Calla (see Nicolson, 1984b). з For Lasioidene and a few other groups, French (1988) recorded that laticifers were lacking, but did not comment t. on whether or not secretion cells were presen are apparently unique to Araceae, although Gaudet (1900) described and illustrated what can only be called trichosclereids in the dicot genus Nym- phaea; as such, they must be considered a derived character, as Nicolson (1959) concluded. The distribution of trichosclereids within the Araceae is almost congruent with the subfamily Monsteroideae, in which all genera and species investigated possess trichosclereids in varying abundance, though not always in all organs. They are absent, for example, from the leaves of Epi- premnum (Nicolson, 1959) and from the roots of m the flowers in Amydrium (Carvell, 1989). In both genera of the tribe Spathiphylleae, trichosclereids are longer and narrower than in Monstereae and occur in much larger groups (Nicolson, 1959, 1960b; French, in press). Outside Engler’s Monsteroideae, trichosclereids are well documented only in a few species of Pothos (Nicolson, 1959, 1960b; Hotta, 1971) and in Po- thoidium (French & Tomlinson, 198 1b). Solereder & Meyer (1928) reported trichosclereids in both species of Montrichardia, but Nicolson (1959) was unable to confirm this. The complex subject of “laticifers”” and other secretory tissues in Araceae was carefully treated by Solereder & Meyer (1928). They pointed out that the contents of aroid “laticifers” are frequent- ly tanniniferous (i.e., not true latex) and are not well characterized for most species. They therefore employed the noncommital terms *'secretion cells," "secretion files," and “secretion tubes" (in as- Volume 77, Number 4 1990 Grayum 647 Evolution and Phylogeny of Araceae cending order of specialization) for the three grades of “secretion vessels” found in Araceae. Their terminology is adopted in this paper, the second two types being grouped, for convenience, under the heading “laticifer.” Individual secretion cells occur in the ground tissue of most organs, where they are distinctive mainly due to their mostly colorless contents (So- lereder & Meyer, 1928). Individual secretion cells are widespread in Araceae since species exhibiting the more specialized types of secretion vessels usu- ally also possess all of the more generalized types. The Pothoideae and Monsteroideae, however, pos- sess only secretion cells, or, depending on the ge- nus, appear to lack secretion vessels altogether (see Table 2). Among the remaining aroid subfamilies, only Symplocarpus, Lysichiton (Rosendahl, 1911), Stylochaeton (Solereder & Meyer, 1928), the tribe Lasieae, the subtribe Cryptocoryninae, and Pistia are known to possess solely individual secretion cells (French, 1988) All so-called *'laticifers" articulated type, as in all other monocots (Metcalfe, 1967) with the apparent exception of Cyclanthus (Wilder & Harris, 1982). Secretion files are linear series of secretion cells separated from one another by cell walls (Solereder & Meyer, 1928). Their contents may be colorless (Calla), brown (Zante- deschia), or milky (Dieffenbachia). They occur in association with vascular bundles, although in roots they may also occur in the cortex (some Philo- dendron; New World but not Old World Coloca- sioideae) (Lierau, 1888). The possession of laticifers is here interpreted as derived, since it is strongly correlated with the monoecious condition in Araceae. her- maphroditic aroids, only Orontium and Calla are laticiferous (French, 1988). Secretion files, or nonanastomosing laticifers, occur widely in Calloideae, monoecious Lasioideae, Philodendroideae, and Aroideae, as well as in Ariopsis of the Colocasioideae (Solereder & Mey- er, 1928; French, 1988). Secretion tubes, or anastomosing laticifers, are in Araceae are of the Among characterized by cellular fusion and extensive and sometimes multiple branching, resulting in the for- mation of complicated net- or latticeworks. Fusion may occur along the longitudinal cell walls of ad- jacent tubes, or along the walls of abutting branch ends (Solereder 8 Meyer, 1928). Secretion tubes are mostly associated with vascular bundles, though they may also occur throughout the ground tissue, even in the vicinity of the epidermis. Secretion tubes are found only in the Coloca- sioideae (with the exception of Ariopsis), including Scaphispatha, Protarum, and Zomicarpa, (French, 1988), all of Engler's Aroideae but here included in Colocasioideae. They appear to be lim- ited to the aboveground organs (Lierau, 1888). The contents of secretion tubes are generally milk y in New World genera but reddish in Old World tribes (pers. obs.; also noted by Trécul, 1866). Intercellular secretion vessels—secretion spaces or secretion ducts—occur in various organs of certain aroid genera. They are typically schizog- enous in origin and surrounded by an epithelium of one to several cell layers (Solereder & Meyer, 1928). * aroid roots, except for the genera Culcasia, Cer- “Resin”” canals are generally lacking in cestis, Furtadoa, Homalomena, and Philoden- dron (French, 1987b). In Philodendron, resinlike secretions from such ducts along the inner surface of the spathe appear to play an important role in pollination (von Martius, 1831; Warming, 1883; Schrottky, 1910; Pohl, 1931; Gottsberger & Amaral, 1984; Mayo, 1986). Schizogenous or lysigenous mucilage canals are known only from Pothoideae, Monsteroideae, Col- ocasioideae (Old and New World genera), and gen- era here included in the Philodendroideae (see French, in press Calcium oxalate crystals occur in every genus of Araceae yet investigated (Nicolson, 1959). Crys- tal morphology varies considerably, and the dis- tribution of various types within the family is of some taxonomic significance. For detailed reviews see Solereder & Meyer (1928), Nicolson (1959), Grayum (1984), and French (in press). Nicolson (1959) recognized four types of cal- cium oxalate crystals in Araceae. The most char- acteristic and well known of these, in Araceae as well as monocots in general, are raphides. These are elongate, needlelike crystals occurring in bun- dles. eo are found in the ground tissue of all ti d rarely even in the epidermis, рани іп ОНАКА cells of varying morphology. Raphides probably occur in every genus of Araceae (Nicolson, 1959). Certain features such as the length and arrangement of the crystals, the shape of the cells that contain them, and the nature of the cell wall vary intergenerically and probably have sys- tematic value (Solereder & Meyer, 1928); how- ever, too few data are presently available. A very specialized type of raphide cell of par- ticular systematic value is the *'biforine." In con- trast to typical raphide cells, with which they may occur, biforines are rather thick walled, except that at each end there is a thin-walled spot, which is generally narrower than the raphide-bundle. Bi- forine cells are either broadly rounded at the ends or are fusiform with acuminate or papillate ends (Solereder & Meyer, 1928). Several authors have 648 Annals of the Missouri Botanical Garden ascribed a “blowgun” (defensive) function to these specialized cells (Hegnauer, 1963; Tchiakpe, 1979; Middendorf, 1983; French, in press). According to Nicolson (1959), biforines are characteristic of subfamilies with unisexual flowers and do not occur in either Pothoideae or Mon- steroideae. Solereder & Meyer (1928), however, also repor ted biforines from Anthurium, Orontium, and Symplocarpus, none of which were investi- gated by Nicolson. More data on this cell type are desirable. So-called *'styloids" (Nicolson, 1959) represent the second main type of calcium oxalate crystal in Araceae. These are reported to be very restricted in Araceae, occurring only in the two genera of the tribe Zamioculcadeae. It is unclear, however, exactly how styloids differ from typical raphides (see detailed discussion in Grayum, 1984). The third of Nicolson’s main crystal types is the druse. These are “spherical, aggregated crystals of calcium oxalate with many sharp points, resem- bling the spiked head of the medieval mace" (Ni- colson, 1959). Druses occur in virtually all ara- ceous genera and in all organs, usually in ground tissue but sometimes also epidermally (Solereder & Meyer, 1928). Nicolson (1959) did not regard them to be of systematic value. The fourth main type, termed “prisms,” are mostly large, solitary crystals in the form of flat- tened rhombohedrons (Nicolson, 1959). Nicolson (1959) reported these only from Potheae and Mon- steroideae; they are occasional in the latter tribe and are characteristic of the former, where they are generally found in cells along vascular strands. Solereder & Meyer (1928) reported crystals of this description also from single species of Anthurium and Philodendron. “Crystal sand,” consisting of minute crystals held by Nicolson (1959) to be of unknown com- position, was reported from every aroid genus ex- amined and is hence considered of no systematic significance (Nicolson, 1959). Franceschi & Hor- ner (1980) regarded crystal sand as a form of calcium oxalate. FLORAL MORPHOLOGY AND ANATOMY General considerations. Aroid inflorescence and floral morphology have recently been exhaustively reviewed by Grayum (1984), and some pertinent details appear in Bogner (1987) and Grayum (in press b). The phylogenetically significant aspects of this field are discussed below, and some more salient features are summarized in Table : The aroid inflorescence is virtually without ex- ception a spadix (an unbranched, fleshy axis bear- ing crowded, sessile flowers) subtended by a single, mostly rather large bract, the spathe. It is always morphologically terminal (Ray, 1988), in contrast to Cyclanthaceae where axillary inflorescences are most common and considered primitive (Wilder, 1988). The general structure of the inflorescence is one of the most consistent and characteristic features of the Araceae. Deviations from the basic lan are rare, and such reports are sometimes rather dubious. The spathe. The ultimate bractlike or leaflike struc- ture below the spadix is ordinarily designated as the spathe. Only in Gymnostachys and Orontium is a spathe regularly lacking, according to Ray’s (1988) interpretations. In these genera the ultimate foliar organ is designated as a mesobracteole or sympodial leaf, respectively. In Pothoidium, the spathe appears to be lacking in the uppermost inflorescences, which are enclosed when young by the foliage leaves of the main stem (Engler, 1877). The absence of a spathe in genera such as Gym- nostachys and Orontium may represent the prim- itive condition in Araceae; these genera, however, exhibit many autapomorphic character states, and it is equally likely that the spathe has been sec- ondarily lost. The variation in gross morphology of the spathe within Araceae is extreme. The highly differen- tiated and specialized spathes of genera such as Gonatanthus or Cryptocoryne bear about the same relationship to the simple, bractlike spathes of Po- thos or many Anthurium species as does a pitcher leaf of Sarracenia or Nepenthes to an unspecial- ized dicot foliage leaf. Most modifications of the spathe are closely correlated with pollination bi- ology and the distribution of male and female flow- ers on the spadix. The classifications of spathe and spadix types that follow are more or less original, but compare also the systems of Engler (1884, 1920b) and Goebel (1931). The simplest types of spathes (Type I; see Table 3) appear to have only the function of enclosing and protecting the spadix in bud. Such spathes are often green or otherwise inconspicuously colored and probably play little if any role in pollination. Type I spathes are mostly found in genera with bisexual flowers, especially in Engler’s Pothoideae, but are also known from certain genera with uni- sexual flowers (some Spathicarpeae and Zomicar- peae). The next stage of specialization, Type II spathes, 649 Volume 77, Number 4 1990 Grayum Evolution and Phylogeny of Araceae L (Dv 494 1 1 IV ««[B.19A98,,"] = Н Ш DYIDASOUIA q] L 0% qy oy I S-Z у (Z1-6)9'PT 8-#+ H Ш WTITUOIDA L ү таа ‘qY (c)1 «D IV (6-1)1 сС-#+ Н Ш sapiouo2na(T == V эу +2 & IV 9 ‘FT 9 ‘p+ H Ill 0106017] = V 14d I I 1 9 `1 9 ‘P+ H I unsgoped = V 19d 1 1 IV al p+ H II wn Xydouj = ү 1deg 1 1 1 (9I (9+ H П DISD] L V 14 dd — (c- Doo I 1 (S-p)91 (S-p)9+ H П nui42ds01447) oy ‘p 1 + qv ‘y -G4 Id 00-Z-1 -I 0 T 9-v1 ic аши ITA AI и0]ә020]/415 = O dy I I 1 #1 b+ H Ш snd1w90 duás pee О эү ant ГА 1 VI pt H II uoqmpois T B чү 9 I I 1 91 9+ H I ummuoa) aPoprorseT L бшу tV 00 Z T +1 — H Il Dyivdsopoy y + У G«4d ЧУ tv 6-1 1 VI = H II uoiuti2dsouajg é шү g 1 1 1 (Эл — H II әиошәҹэѕоу ж wy ‘y 9 1 1 1 TT = H II snsdppuiog + V Idd Md — (9v-c I 1 #1 = Н П wnuwasdidgy EE V 6- TI ?d С 1 T VI — H II unapAup m V qv © © 1 га! — Н Il юлә$иорү + V €-Cald со 1 1 rT — Н Il psoydopiydny y = V ¿xd Ӯ 1 IV VI — H II sisdo12]2]] + V eg oo I 1 VI Jv c H III SÁUID]YIO]0 8-9 (4) + ү оү у ‘IV (b-1) (@Є(@) 1 (Av)T (8)9(b)+ H П unjadnj mds + (89)у эү (4 © 1 VI b+ H П ‘I штітушү E V g I I 1 VI A+ H I unipuappup + ү эү 1 © 0 dy AVI p+ ПИ AI sndojnuo») id suy эү 1 ГА 0 1 VI p+ аши ЛІ $D2]n201um7 pem V 14d I I 9-€ 1 9(£yT 9+ Н 1 umpioqioq E V ¿AV 1 € 1 9T (44)9+ H I AAA = у оу ‘qy 1 t T 9T 9+ H I soyiod + о dy 1 1 1 VI pt H I sAYIDISOUWA avaploylog IX X XI IIIA ПА IA AI III II snuas 'Aqrurejqng `5ләуэвтецә juej1odurm jo Атешшпс -eeeo?iy jo AZojoydiour [ео 4 '€ 318VL Annals of the 650 Missouri Botanical Garden ¿ sO qy I 8-9 8-S T L-1 ^GIN dI wnayjuny ind + ¿ qy 1 -E + dy (¿s)r(¿)dg ПИ III UWNIDIL) ¿ О qy 1 v 8-9 dy K8-9)*1 IW III umipiuog 102) ¿ ү эү 1 9-€ 9-# dy J8-v1 IW Ш um4920] + О qy 1 (с)є S-€ 1 A(9-S)+w1 IW HI xippdso4pupuAg + qy qy I S-€ (4)S-€ ү Jt-edg IN ӨШ юшЎ1]$олә1$р ¿ V эү © £-c £ dy Jc-€1 PITA III DIUOZUD y — О g I I A dy ‘Iy 1 As-zdg о “9 PITA JAI uoj4y dos = O g I I A 1 U7 ПИ ИП 20901907 — О g 1 1 Ab-€ т 49-с (11 “dg PITA ‘IW qu иода әр) ¿ ¿V g I I 0 IV dp-edg аши ИП D/9udog = Чу “y qv I €-Z S-p dy “ү JS-vdg аши JAI piyooquaffany (g) оу co-p (/%-/) + ЧУ ‘O ‘qy ‘IV -(D 9-HE-Z) (eae Аләл) dy 9-zdg CTITAMIITA Al иолриәроч L (Ч)О avd — MET I AS- dy Jc-vdg эи Al оірирәд — О 9-€4d Z-I I 9-€ IV 18-99 эши Al wnsopouoyd y ¿ (S0 g S1-Z1 1 0 dy ау Je-c*g о-ви — JAI III un10110]H * О ady oo I 0 qdy Кашан: эи AI UNIDPI4DO4919 4 + О q oo I 0 qdy je- t*g о-у IIA III UNIDPUY + О Я 6-1 1 0 ydy тея әу ЛІ UNIDIDUÁY Y ¿ О g oo 1 0 ydy тея JIN III pipunjpydaong sO ‘Ҷу £-Z 24 ‘Id со 1 0 dy 4с-189 2 ‘ШИ “TIA III vy10dsoidiq + OY t-z (did (Doo 1 (#-1) dy (De-31 (9 Y) INIA — AI III $1910) 90JDUSIYIS + (WW а ‘у àv со €-Z 0 1 Ae(z)dg GITA INW AI spiqnupy + V 8 ‘уЗу Ӯ S-I (£)0 dy e-zl¿ de ПИ Ill DIYISIPAUDZ ¿ ү 9d I "d £-I dy t-e*g ЧШИ Ш uniopov]dp — чу ‘y g 1 (Z){ (әле Аләл) dy aeg ПИ ӨШ puauon Iy (qv) + Чү ‘V T-C«dd 24 o-t* (Sr-c(D (£-2)0(0 L'dv'tiv*'1 (9-5)+-с9 — (ПИШИ АЛШ риәшојошон ¿ qy g oo I Gc dy 14g — аши Ш тортту + ү eg 6-9 1 1 31-011 H Il D]]D) ¡923pIO IPUIPO[IYJ IX X XI IIIA IIA IA А A Ш Il I ѕпиә8 “Ápmuejqng 'penunuo?) `Є 318v] 651 Volume 77, Number 4 1990 Grayum Evolution and Phylogeny of Araceae + 10) gdy (9>) мә 1 0 1 j-€1 TIN Al sAINIUNIDA(] + О па +9 1 0 dy t-e*g QUI AI uni 9?apro1y ¿ О g p 1 9-t ¿ Aedg ШИ AI UNIDIOL] + qy “y qd 01-9 1 0 dy Az1-) мәј 1 0 IV t-c*d = эШИ Al Sata IX X XI ША ПА IA А А1 III П 1 snus3 *Á[muejqns "panunguo) ‘g ATV], Volume 77, Number 4 1990 Grayum 653 Evolution and Phylogeny of Araceae oo [^] © со n =, — = © 5 5 oe k E E m 23 ES -£ 907, °щ 2 xi S255 265 & o res в E BOS ЗЕ [5 Beas оо чокар SSE раа 5 » 42111 оо а ү б ш = — © Ф d E: [- © а ga P с EG Som o o 6б фа „8 Gd 0 w 9 б-а O _, n | + — "а БЕ: ea se 3 ANN ш o uu EÉSvvvov Е в 657575735 3 Gir m Б 3 — Y g= 0000 e d o sis! || | | © "d Rd бж anoe yz a] >] e ec | | ps ae E з < < > o Sa — bb = = Lx бег PE € ETOS - осы д D .g мо 5 ё "E 23 D wp ‚Б ES © S w.E'g a NNN Ф E 2 £, 2 0 3 dadas = Б БББ тоз „22560 =... Ф 55 не!!!) eg o5 ornano Ba Sg ETS d. oe К< e — 5 < S 3 ©8 & ‚м, e = g е = Am — amphitropous a—derived from axile ' Properly Calloideae if used sensu this paper, i.e., to include Calla (see Nicolson, 1984b). are broadly expanded at maturity and/or brightly colored with some obvious role in attracting pol- linators. This stage is most characteristic of taxa with bisexual flowers (especially Anthurium and the Monsteroideae) but occurs rarely in monoecious genera (Nephthytis, Callopsis, Steudnera). e spathe їп most monoecious aroid genera, especially in Philodendroideae and Aroideae, somewhat funnelform to cucullate and/or tubular at maturity and wraps all around at least the lower portion of the spadix. Such a specialized type (Type Ш) can also be seen in a few genera with bisexual flowers, e.g., Symplocarpus and Dracontium. The most highly specialized spathes (Type IV) more or less completely surround the spadix at maturity and are constricted in one or two places, so that a lower and an upper “lamina” may be distinguished. This is the usual condition in Colocasioideae and occurs commonly in Philo- dendroideae and Aroideae; it is not known from genera with bisexual flowers. The constriction may be above the inflorescence, as in Stylochaeton and the Cryptocoryninae; between the sterile appendix and the male portion of the spadix, as in Zomi- carpa, Arisaema, at least some Alocasia species (pers. obs.) and most Arinae; or between the male and female portions of the inflorescence, as in most other genera. In Gonatanthus, there are two con- strictions: one between the male and female por- tions of the spadix and a more pronounced one above the entire inflorescence (Engler & Krause, 1920) Constrictions of the aroid spathe are known to play important roles in pollination biology (see, e.g., Gottsberger & A 84). he spathe is frequently decurrent on the pe- duncle and in some cases is fused to the fertile part of the spadix. In the closely related genera Spathicarpa and Spathantheum, the spathe and spadix are fused over the entire length of the latter (Subtype F, Table 3). Many other monoecious gen- era show an intermediate stage, with fusion only in the lower fertile regions, such that all of the female flowers and occasionally even some male flowers are borne unilaterally on the spadix (Sub- type f, Table 3). A tendency toward this condition, in which only the lowermost part of the female region is involved, can be seen in some species of Philodendron, Caladium, and Alocasia. e spathe may persist until the fruits are ma- ture, or it may gradually rot away. In Anaden- drum, Heteropsis, and the Monstereae, the entire spathe abscises cleanly following anthesis (Engler, 1905; Engler & Krause, 1908). In other taxa, such as Schismatoglottidinae (French & Tomlin- Annals Ani P M Garden son, 19814), Chlorospatha (pers. obs.), Syngon- ium (Croat, 1981), and Alocasia (Nicolson, 1968a; Burnett, 1984), only the lamina abscises and the tube persists. The spadix. А few species of Araceae do not tech- nically bear flowers on spadices but, rather, in spikes. The bisexual perigoniate flowers of Pedi- cellarum and some species of Pothos sect. Allo- pothos are rather distantly disposed on a slender, ligneous axis; male flowers of Filarum, Arisarum, and many caes of Arisaema may also be remote (Engler, 1920b; Nicolson, 1966). If the spadix is indeed derived from a spike and, ultimately, a raceme (see Stebbins, 1973), then these inflorescences must be regarded as primitive fram спе within Агасеае. In Pedicellarum, the flowers are supposed to be pedicellate (Hotta, 1976), which would make the inflorescence a raceme and hence potentially even more primitive. However, in the material of this genus that I have examined (Martin & Ismawi 536660 MO), the flowers are always sessile. Bracts supposedly never occur on the aroid spa- dix itself (Engler, 1884); however, Eyde et al. (1967) described ‘vascularized appendages” be- tween some flowers on spadices of two Pothos species, and bracts may occasionally occur on the lower flowers in some Anthurium species (Engler, 1905). According to Engler (1884) and Burger (1977), the bractless nature of the aroid spadix suggests a bracteal origin of the perianth members. The arrangement of individual flowers on the aroid spadix is usually spiral, although in some cases the flowers may be whorled, as in species of Asterostigma, Arum, Biarum, Lagenandra, and Cryptocoryne. The whorled arrangement appears to have been derived from a spiral organization (Engler, 1884; Benzing, 1969). The number of flowers on most aroid spadices is indefinite, but there is a tendency toward re- duction in some groups. The number of male flow- ers is reduced to 8-10 in Ambrosina, and to only one in Pistia; there are only a few female flowers in Aglaodorum, Ariopsis, Stylochaeton, Zomi- carpa, Filarum, and Cryptocoryne, and just one in Ambrosina and Pistia. Inflorescences of Ped- icellarum and some species of Pothos bear only a few bisexual flowers (five or fewer in the former genus). Two sexual systems predominate in Araceae: hermaphroditism, in which each inflorescence is more or less completely covered with bisexual flow- ers; and monoecy, in which each inflorescence bears separate male and female flowers. In most species of Arisaema some plants produce only male inflo- rescences and others only female inflorescences during a particular season, but may produce inflo- rescences of the opposite sex or even (depending on the species) monoecious inflorescences in a dif- ferent season (Hotta, 1971). The sexuality of a given plant is evidently correlated with such factors as age, size, and nutritional state. The situation in Arisaema has been referred to as “‘paradioecy” (Hara, 1971) or, better, as "sequential monoecy" (Bawa, 1980) or "sequential cosexuality" (Lloyd, 1979). The same condition may also occur rarely in Cryptocoryne, as in C. lucens (Jacobsen, 1976). No Araceae are known to be truly dioecious; Engler's (1905) tentative interpretation of dioecy in Pothoidium was in error (J. Bogner, pers. Unisexual flowers have clearly been derived from bisexual flowers in Araceae (Engler, 1884; Eyde et al., 1967; Hotta, 1971), as in monocots (How- arth, 1957) and angiosperms as a whole (Bawa & Beach, 1981). Thus, hermaphroditism is to be re- garded as primitive, monoecy derived, and “se- quential cosexuality" even more derived. One piece of evidence for this hypothesis is the distribution of male, female, and sometimes bisexual flowers on monoecious aroid inflorescences. In virtually all monoecious species the female flowers are situated on the lower part of the spadix with the male flowers above. Bisexual flowers, whenever present, occur in between. Based on the assumption that pollen would tend to fall or be carried downward by insects foraging from top to bottom on an inflorescence, Engler (1884) concluded that flowers in the lower part of a hermaphroditic spadix would on average receive more pollen and thus set more fruit than those in the upper part. Upper flowers would ul- timately become masculine and lower flowers fem- inine, he reasoned, in Lamarckian fashion, due to “disuse” of their pistils and stamens, respectively. It is perhaps significant in this regard that spadices of many nominally hermaphroditic aroids exhibit the predicted tendencies: Calla (Krause, 1908; Goebel, 1931) and Orontium (Schaffner, 1937; Hotta, 1971) frequently produce a few male flow- ers toward the tips of their spadices, and Heteropsis (Engler, 1905) and many Monstereae (Engler & Krause, 1908) often bear some basal female flow- ers. Furthermore, there is a tendency even in species with uniformly bisexual flowers (e.g., many An- thurium species) for only the lower flowers to set fruit (Engler, 1905; pers. obs.). The curious dorsiventrally segregated distribu- tion of male and female flowers in Spathicarpa (and to a lesser extent in the closely related genus Volume 77, Number 4 1990 Grayum 655 Evolution and Phylogeny of Araceae Spathantheum) appears strikingly at variance with i usual aroid pattern. This peculiar distribution ‚ however, been elegantly explained by Troll л, m a Uhlarz (1982) as a developmental con- sequence of the complete fusion of spathe and spadix, which likewise occurs only in these genera. The dorsiventral segregation of various flower types in some species of Hottarum (Bogner, 1983; Bog- ner & Hotta, 1983) is perhaps explicable in a similar manner. rmaphroditism is characteristic of Engler’s subfamilies Pothoideae (except as noted in Table 3), Monsteroideae, and Calloideae, and of the tribe Lasieae of subfamily Lasioideae. All other aroids are monoecious. The arrangement of flowers on monoecious aroid inflorescences is quite variable and may involve sterile male and/or female flowers in addition to fertile male, female, and occasionally bisexual flowers. Sterile flowers may be present in various positions, often occupying a considerable portion of the spadix and sometimes playing an important auxiliary role in pollination. Presumably the most primitive type of monoe- cious spadices in terms of floral distribution, if Engler’s scheme is correct, retain some bisexual flowers in the middle and have not yet evolved sterile flowers. This type of spadix (Type I, Table 3) is rather infrequent in Araceae but occurs in species of Carlephyton (Bogner, 1972) and in several Spathicarpeae (Engler, 1920a). A slightly more advanced stage is probably represented by spadices with the male and female regions abutting directly, without evidence of bisexual or sterile flowers. This condition (Type II) occurs in most major groups (see Table 3). Type III spadices lack bisexual flowers but bear sterile flowers, which may occur either above (Sub- type Ша) or below (Subtype IIIb) the fertile male portion of the spadix, or often in both locations (Subtype IIIc). In several genera of the Schisma- toglottidinae all three subtypes are known; Hotta (1982) suggested that Types Ша and IIIb have evolved independently from the Type II grade, and that Type IIIc might have been derived from either of the former. Sterile flowers in Araceae are almost invariably sterile male flowers, as indicated by their position (when apical on the spadix) and/or morphology. Only in Zamioculcas do centrally located sterile flowers clearly represent sterile female flowers (En- gler, 1905), though in several Arinae this is strong- ly suspected. In some other Arinae and a few species of Amorphophallus, central sterile flowers are so strongly modified that their sexual derivation is indeterminable. Sterile flowers may be only slightly modified or more or less strongly modified from their fertile counterparts. Sterile male flowers of some Astero- stigma species, for example, differ from the fertile synandria only by failing to produce pollen (Engler, 1884). Similarly, sterile male flowers in Typhono- dorum and most Schismatoglottidinae differ little from adjacent fertile male flowers. On the other hand, sterile flowers may be highly modified as fungoid (Pseudodracontium), hairlike (Amorpho- phallus sect. Dysamorphophallus and subg. Met- andrium, Arisaema sect. Arinae), or wartlike (Zomicarpa, some Amorpho- phallus species, various Arinae) structures. More often, apical sterile male flowers are strongly flat- tened and reduced upward, spreading apart and finally blending altogether into a smooth-surfaced to deeply rugose, bulbous, clavate, caudate or mor- chelloid appendage that may greatly overtop the spathe and play a major role in pollination, espe- cially as regards heat and odor production (Knoll, 1926; Vogel, 1962). Such is the case in most species of Amorphophallus, Arisaema, and Ari- nae, plus Zomicarpella and some Arophyteae (Bogner, 1972). Only one genus in Engler’s Phil- odendroideae (Peltandra) and two in Colocasioi- deae (Alocasia, Colocasia), possess sterile, naked apical appendages. Engler (1884) contended ad- amantly that such terminal appendages are elab- orated from the male flowers as described above, and do not, for example, represent the prolonged central axis of the inflorescence. The position of the central sterile flowers often oe to the site of the spathe constriction, such occurs, and a role in floral biology (involving the occlusion of the spathe) has occa- sionally been ascribed to these as well (see, e.g., Knoll, 1926). Centrally located sterile male flowers may be differentiated morphologically from apically located ones on t padix, as in Bucephalan- dra (Bogner, 1980) and many Arinae. In a few species of Theriophonum and Typhonium, distinct morphs of sterile flowers occur in the upper and lower parts of the central sterile region; these are believed to represent sterile male and female flow- ers, respectively (Engler, 1920a). For a more complete discussion of the mor- phology and spatial distribution of sterile flowers in araceous inflorescences, see Grayum ( . Fimbriata, various Compound inflorescences. Given that “first-or- der" inflorescences (i.e., spadices) in Araceae are often functionally analogous to single flowers, it is not surprising that they are frequently associated in “second-order, or compound, inflorescences. 656 Annals of the Missouri Botanical Garden Ray (1987b, 1988) analyzed and placed in ten- tative evolutionary and ecological perspective the various growth phenomena resulting in the pro- duction of compound inflorescences in Araceae. Ray (1988) pointed out that, with the exception of Orontium, all Araceae with bisexual flowers have either no bud or merely a vegetative bud on the peduncle base and never develop multiple inflo- rescences (he dismissed the complex and unique "inorescénce sympodium” of Gymnostachys as ‘condensed shoot system”). Based partly on this o he reasoned that the inability to pro- duce paired or compound inflorescences is the primitive condition in Araceae; species able to de- velop but a single inflorescence per article were deemed relatively inflexible reproductively, such that selection would tend to favor homeophyllous articles” in order that * inflorescences" might be produced “as Nuus as possible." But since this would tend to prevent seasonal flowering, such species "would need to mature all of these inflorescences" in order to maximize reproductive potential (Ray, 1988). Ал- thurium is cited as an exemplar of this continuous, aseasonal flowering strategy, necessitated by an inability to produce compound inflorescences. Reproductive flexibility is increased somewhat by the ability to produce two inflorescences per article, considered the next most advanced state by Ray (1988). In Orontium, Anubias, Peltandra, Alocasia, Caladium, Spathicarpa, and Callopsis (among those genera studied by Ray), a second inflorescence is produced from a bud at the base of the original peduncle (Ray, 1988). Compound inflorescences in these genera are determinate, however: no further proliferation is possible, since the second peduncle lacks a bud Intermittent homeophyllous growth is seen as a means for aroid species with a fixed one or two inflorescences per article of overcoming this limi- tation and achieving a modicum of reproductive flexibility by allowing for “the relatively rapid serial production of inflorescences" within a specified sea- son (Ray, 1988). It is, in fact, known only from species in this category. As defined in a previous section, volves an alternation of anisophyllous and homeo- intermittent homeophyllous growth in- phyllous phases. Various subtypes are recognized (Ray, 1987b), according to the number of leaves per homeophyllous article. Intermittent diphyllous growth is the most fa- miliar type, occurring in most species of Monstera and in Anadendrum (Ray, 1988). The diphyllous phase is associated with flowering and produces what is in effect a compound inflorescence. Al- though the spadices are solitary in the axils of the sympodial leaves, the compound effect is achieved by a condensation of the axis and by the fact that both leaves (the prophyll and sympodial leaf) of each diphyllous article are cataphylls. At the time of anthesis, the diphyllous “‘compound inflores- cences" of these species generally terminate the shoots. Intermittent triphyllous sympodial growth, so far observed only in Rhodospatha and Alocasia, in- terposes a mesophyll between the prophyll and sympodial cataphyll in the homeophyllous articles. In Alocasia, each article terminates in paired in- florescences, and the mesophylls are foliage leaves (Ray, 1988). Intermittent tetraphyllous growth is hypothetical (Ray, 1987b Monophyllous homeophyllous sympodial growth e production of indeterminate, con- each consisting results in th densed “inflorescence sympodia, of as many as 10 or more “first-order” inflores- cences, in the axils of individual sympodial leaves on the shoot (Ray, 1988). Each “first-order” in- florescence is terminal on an article bearing a single ээ leaf, the bracteole (always a cataphyll). Three distinct types of monophyllous growth occur in Araceae. (1) Axillary growth, in whic each succeeding inflorescence develops from a bud in the axil of the bracteole of the next lower order, is known only from Dieffenbachia and Philoden- dron, including species of subg. Pteromischum. (2) Gorgonoid growth, in which each succeeding inflorescence develops from a bud on the internode below and opposite to the bracteole, is known from Aglaonema, Syngonium, and Xanthosoma. (3) Mixed axillary gorgonoid growth is an alternation of the above two types and is known only from Homalomena rubescens. Presumably monophyl- lous inflorescence sympodia are known from many other genera than analyzed by Ray (see e.g., Gra- yum, 1984), and the application of his techniques to these is desirable. Production of inflorescence sympodia by axil- lary, gorgonoid, or mixed monophyllous homeo- phyllous growth is considered the most advanced condition in Araceae. Species in this category are able to take maximum advantage of the best season for reproduction by producing a theoretically un- limited number of inflorescences from the axil of a single sympodial leaf (Ray, 1988). Individual flowers: general considerations. All aroid flowers are actinomorphic and hypogynous, or fundamentally so. Various authors (Eyde et al., 1967; Hotta, 1971; Barabé, 1982) have described the ovary in Lysichiton and species of Pothos and Volume 77, Number 4 1990 Grayum 657 Evolution and Phylogeny of Araceae Cyrtosperma as “рагу inferior," and Barabé et al. (1987) insisted that that of Symplocarpus is truly inferior. According to Engler (1884) and Krause (1908), by contrast, the flowers of these genera are deeply embedded in the axis of the inflorescence. Either condition must be regarded as derived (Eyde, 1975). All bisexual aroid flowers that have been studied are protogynous (Engler, 1905; Engler & Krause, 1908; Krause, 1908; Engler, 1911), and most monoecious aroid inflorescences appear to exhibit second-order protogyny (see e.g., Gottsberger & Amaral, 1984). To date, the sole exception is Ar- isarum simorrhinum, in which stigma receptivity and anther dehiscence are said to coincide (Her- rera, 1988). Protogyny is quite uncommon among flowering plants (Bawa & Beach, 1981) and is frequently seen in association with beetle pollina- tion (Gottsberger, 1977; Bullock, 1981). The perianth. Araceous flowers may be charac- terized either by the presence or absence of a perianth (or perigonium). The distribution of the two conditions is correlated with floral sexuality: Araceae that regularly have unisexual flowers lack a perianth, with the exception of Stylochaeton and the Zamioculcadeae. The reverse correlation is slightly weaker: Calla, Pycnospatha, Heteropsis and all Monstereae have naked, bisexual flowers. There is little question that naked flowers have been derived from perigoniate ones in Araceae (Engler, 1884), as is the general trend in monocots (Howarth, 1957). Moreover, unisexual flowers and naked flowers have evolved repeatedly within the family (Grayum, 1984). The aroid perianth is always reduced, wholly sepaloid, and rarely any color other than greenish or brownish, and probably plays little if any role in pollination. There are always two alternating whorls of tepals (Engler, 1884), the total number generally being either four or six. The presumed transition from trimery to dimery appears to be relatively simple and to have occurred indepen- dently on many occasions; several genera contain species in each category, and the condition may vary on individuals, even within the same spadix (Engler, 1884). In Anadendrum, Spathiphyllum sect. Mas- sowia, Holochlamys, and Stylochaeton the tepals are completely fused into a rotate to cupulate or urceolate perianth. This is certainly a derived con- dition (Howarth, 1957). The perianth is dimorphic in Stylochaeton, being rotate in the male and ur- ceolate in the female flowers (Engler, 1920a). Hot- ta (1976) reported a connate perianth in Pedicel- larum, but in the material I examined (Martin & Ismawi S36660 MO) the six tepals are entirely distinct and in two whorls, just as in Pothos. Araceous tepals generally have a single vascular trace, but in Symplocarpus and Podolasia these are branched. Tepals in Zamioculcas each have three traces (Hotta, 1971). Valvate aestivation of the perianth is said to represent the primitive condition in Araceae (En- gler, 1920b); however, in Podolasia, Dracontium (Engler, 1911), and all three genera of the tribe Orontieae (Krause, 1908) aestivation has been re- ported as imbricate. These may represent cases in which basically valvate aestivation has been dis- torted, as suggested by Engler (1920b), but the distribution of this character state among such phe- netically primitive genera warrants further inves- tigation. The androecium. Stamen morphology varies great- ly in Araceae. Most Araceae have rather short (later elongating), broad, flat filaments topped by smallish, basifixed anthers. The stamens of Po- theae, Anthurium, and the Lasieae are especially similar in this respect (Engler, 1911). On the other hand, nearly all monoecious species have thick, short, often prismatic, flat-topped stamens, fre- quently with lateral anthers. Engler’s (1884) ex- planation is that stamens deprived of the protection of a perianth need to be sturdier. Long, slender filaments are uncommon in Arace- ae, occurring, for example, in Pseudodracontium, Stylochaeton, Arisarum, and in the synandria of Arisaema and most Spathicarpeae (Engler, 1920а). The stamens of perigoniate aroid flowers are generally the same number as the tepals and op- posite them, in two whorls (Eyde et al., 1967; Cronquist, 1981). Thus, rarely are there more than six stamens per flower. In some species of Dra- contium, however, there may be a third whorl, for a total of 9-12 stamens (Engler, 1911, 1920b). The stamens of all naked, unisexual flowers are considered to be reduced to a single whorl (Engler, 1884). The number of stamens per flower is thus usually reduced, frequently to three or two and, in a few cases, to just one (Colletogyne, Filarum, some species of Biarum, Arisarum, and other gen- era). At the other extreme, the synandria of male flowers of Typhonodorum may comprise up to eight stamens, those of some species of Alocasia as many as Stamens may be so crowded and irregularly arranged on the spadix in some monoecious aroids that it is difficult or sometimes impossible to be sure how many compose a single flower; this is the 658 Annals of the Missouri Botanical Garden case, for example, in Ambrosina (Engler, 1920a) and the subtribe Arinae (van Tieghem, 1907). The stamens in the Philodendron type of male flower, on the other hand, are more or less polygonal and fit together like pie slices in each male flower, so that the number of stamens per flower is obvious. This type of stamen occurs also in Culcasia, Pseu- dohydrosme, Nephthytideae, Montrichardia, and Homalomena. The number of stamens per flower in these groups may vary greatly on the same spadix, as Engler (1884) has documented for Ho- malomena rubescens. French (1986b) surveyed stamen vasculature in Araceae and reported that virtually all genera with bisexual flowers have 1-3 unbranched bundles per stamen, certainly the primitive condition for the family. Stamens of the demand genera have either forked or anastomosing bun Barabe & Labrecque (1983). uate that some of the stamens in the naked flowers of Calla may be tepalar in origin. The anthers of Araceae are basically tetraspo- rangiate and dithecal, as for angiosperms in gen- eral; cases of reduction in the number of sporangia are few (see Grayum, in press b). In Synandros- padix, the upper male flowers have disporangiate anthers, while the anthers of the central bisexual flowers are tetrasporangiate (Cocucci, 1966). The anthers of all Araceae are supposed to be impeltate (basifixed) (Dahlgren & Clifford, 1982) The mode of anther dehiscence in Araceae is mewhat correlated with sexuality. Dehiscence by longitudinal slits appears to be the primitive con- dition and is common only in taxa with bisexual Pothoideae, Monsteroideae, Calloi- deae, and Lasieae). Only a few monoecious genera have fully longitudinal anther dehiscence (see Table 3). The tendency is for these slits to become con- fined to the apical half of the anther, as in Het- eropsis and Cyrtosperma, and ultimately to be reduced to apical pores, as in most monoecious genera, in which the male flowers tend to be densely packe of aie and Cryp- САНЕ. the роге is located a conical tube. In lll Aridarum, and Phymatarum, each anther sac is surmounted by a horn; the pollen exits through a pore formed by the abscission of the apical part of the horn (Hotta, 1971) In a few taxa, such as Zomicarpa and some flowers (i.e., the apex of a species of Homalomena and Arisaema, the an- thers dehisce by an apical transverse slit. Anther dehiscence in Araceae is virtually always extrorse. Although Hotta (1976) reported introrse dehiscence in Pedicellarum, my examination of the specimen Martin & Ismawi 536660 (MO) showed anther dehiscence to be unequivocally ex- trorse. However, Engler (1905) reported introrse dehiscence in Zamioculcas, and Bogner (in litt.) has recently confirmed this. п many monoecious Araceae the stamens are connate in the male flowers, forming synandria (this detail was overlooked by Dahlgren & Clifford, 1982). In most cases the fusion involves only the nts; however, in a few species of Arisaema the anthers also are partially or fully (Sivadasan & Sathish Kumar, 1987) co most characteristic of Colocasioideae, where they occur exclusively in every genus. They are fairly frequent in Engler’s Philodendroideae and Aroi- deae, and also occur in Gonatopus and Pistia. In the simplest synandria the filaments are fused only toward the base. This condition occurs in some Schismatoglottidinae, Carlephyton, Gorgonidium, Dracunculus, and many species of Arisaema. The most specialized type is found throughout the Col- ocasioideae, in which the filaments, though com- pletely united, are basically short and thick with the anthers borne laterally or at the upper margin of the more or less flat-topped, polygonal structure. nnate. Synandria are This type of synandrium also occurs in Anubias, Bognera, Dieffenbachia, Typhonodorum, Peltan- dra, Arophyton, and Protarum. These synandria much resemble, and are no doubt derived from, the Philodendron type of male flower, discussed previously; in some species sutures may still be observed where the individual stamens must have used. Rarely synandria belonging to adjacent male flowers may fuse. In the apical part of the spadix of Gorgonidium mirabile, for example, numerous stamens may become united into branched, den- dritic ““super-synandria” in which the limits of in- dividual flowers are impossible to define. The sit- uation in Ariopsis is even more remarkable: adjacent synandria are connate laterally into a massive superstructure involving the entire male portion of the spadix, analogous to the syncarps of genera such as Syngonium and Cryptocoryne. Staminodia are usually the sole components of sterile male i present) generally have distinct staminodia. Like- wise, where fertile stamens are connate into syn- andria, staminodia (where present) are connate into synandrodia. Dieffenbachia, with connate fertile stamens but sterile flowers with distinct staminodia, is the sole exception to this rule (Engler, 1884). Staminodia associated with female flowers are encountered in Amorphophallus subg. Metan- Volume 77, Number 4 1990 Grayum 659 Evolution and Phylogeny of Araceae drium (Stapf, 1923), several genera of Philoden- droideae, and a few Aroideae. In Colocasioideae only Steudnera has staminodia in the female flow- ers. Contrary to the rule for sterile male flowers, staminodia in female flowers are generally distinct, even in genera with connate fertile stamens. This suggests that monoecy evolved prior to stamen connation in these groups. Female flowers with synandrodia are known only in Peltandra, Aster- ostigma sect. Rhopalostigma, and all Arophyteae (Bogner, 1972). The number of staminodia associated with each female flower, in cases where these staminodia are distinct, probably corresponds primitively to the number of stamens per fertile male flower. This correspondence is a frequent condition in Araceae. However, female flowers of Typhonodorum, Steudnera, and Spathicarpa typically have fewer staminodia than fertile stamens, and the number is typically reduced to just one in Homalomena (Engler, 1884), and occasionally in Schismato- glottis, Hottarum, and Aglaonema as well. On the other hand, there are more staminodia in female flowers of Protarum than fertile stamens in the male flowers (Engler, 1920b) The function of staminodia in female flowers is not usually known; Engler (1884) ascribed a pro- tective function to the tepaloid staminodia of Zan- tedeschia aethiopica as well as to the perianthlike synandrodia of Peltandra. Staminodia of Spathi- carpa (Troll, 1928) and of Aglaonema (Daumann, 1930) appear to secrete nectar. Additional aspects of androecium morphology in the Araceae are discussed in Grayum (in press b). The gynoecium. The gynoecium in all known Ara- ceae is unipistillate. The style is generally short and thick to conical; often it is entirely absent, and the frequently discoid stigma is thus sessile on the ovary. Long narrow styles are rare (as in species of Anthurium, Symplocarpus, Dracontium, Pyc- nospatha, some species of Amorphophallus and Philodendron, Stylochaeton, some Spathicarpeae, and Ambrosina), and the stigma is never elongate. According to Engler (1920a), most aroids have a densely papillate stigma, although Heslop-Harrison & Shivanna (1977) described the stigma of Alo- casia as smooth. Mayo’s (1989) detailed studies of style and stigma characters in Philodendron suggest that these organs deserve more intensive study throughout the fam In perigoniate or bisexual flowers, the number of locules in the ovary is generally equal to half the number of tepals (or stamens), so that species with six tepals per flower have fundamentally three- loculed ovaries, and species with four tepals fun- damentally two-loculed ovaries. Reduction to a single locule is frequent, however, throughout Araceae. A few taxa may have an unusually high number of locules, for example, Spathantheum (6-8) and Philodendron (to 12-14 in a few species, to as many as 47 in subg. Meconostigma; Mayo, 1989). Placentation is correlated with loculization to a certain extent. All Araceae with more than one locule in the ovary have some type of axile pla- centation (a circumstance overlooked by Dahlgren & Clifford, 1982). Three variations of axile pla- centation are known in Araceae: ovules may be borne along the entire length of the placentae (the rarest condition), or they may be reduced to one or a few borne either at the base or in the middle (both comparatively common; see Table 3). Aroid genera with unilocular ovaries may have parietal, basal, apical, or both basal and apical placentation. The first two conditions are most com- mon. In parietal placentation, the number of pla- (e.g., Arum) to many (e.g., Homalomena). Apical placentation is known only centae varies from one from Gymnostachys, Symplocarpus, and Lasia; apical and basal placentation in the same ovary occurs in Dracunculus, Helicodiceros, Therio- phonum, and Heteroaridarum, although in the last-mentioned genus the apical placenta is sterile (Hotta, 1976). Apical and basal placentation types have in some cases clearly been derived from parietal placen- tation, with the development of just one or a few ovules at a particular locus on the placenta. For example, in some species of Epipremnum with basal placentation, the largely naked placenta ex- tends the entire length of the ovary. Other species in the same genus have typical parietal placenta- tion. Hotta (1971), based on a study of ovary vascularization, concluded that basal placentation was derived from parietal in Alocasia as well. That basal placentation may also be directly derived from axile placentation, on the other hand, is sug- gested by the situation in Culcasia, where some species have two- or three-loculed ovaries with axile placentation (basal subtype), while others have uni- locular ovaries with basal placentation. Further- more, in Holochlamys (with unilocular, multiovu- late ovaries) there is evidence of a basal vascularized septum, indicating derivation from a plurilocular ovary with axile placentation, such as occurs in the closely related genus Spathiphyllum (Eyde et al., 1967). A similar situation seems to occur in Calla (Barabe & Labrecque, 1983). Spathicarpa, with unilocular, uniovulate ova- 660 Annals of the Missouri Botanical Garden ries and basal placentation, possesses an apical pendent lobe within the ovarian cavity that is in- terpreted as the vestige of a septum (Engler, 1884; see also Barabé & Chrétien, 1985). The closest relatives of Spathicarpa are all plurilocular with axile placentation. Thus it seems clear that basal (and presumably apical) placentation in Araceae may be derived directly from either axile or parietal placentation but must in either case be considered an advanced condition. Parietal placentation is here considered to have been derived in all cases from axile pla- centation via the intrusive-parietal intermediate stage, which is widespread in Araceae (see Table 3). This conclusion follows inevitably if derivation of unilocular ovaries from plurilocular ovaries is accepted as universal. This view receives support from Puri (1952), who concluded that parietal pla- centation is always derived from axile placentation in angiosperms as a whole. Since the derivation of basal and apical placentation from the parietal or axile conditions necessarily involves a reduction in ovule number, it also follows that multiovulate pla- centae are generally more primitive than few-ovuled placentae, as Engler (1884) concluded. This is a very strong trend in monocots in general according to Howarth (1957), and Eyde (1975) stated that the reverse trend is known in no group of angio- sperms. The pluricarpellate nature of unilocular ovaries in Araceae is generally easily established on the basis of such criteria as number of placentae (in species with parietal placentation), number of stig- ma lobes, multiples of ovules, and ovarian vascu- lature (see, e.g., Krause, 1908; Eyde et al., 1967; Hotta, 1971; Barabé, 1982; Barabe & Chrétien, 1985; Barabe & Labrecque, 1983, 1984, 1985; specific examples are discussed in Grayum, 1984). In other cases (e.g., Cyrtosperma, Lasia, Aglao- dorum, and Spathicarpa), putatively pseudomo- nomerous ovaries can be homologized with indubi- tably pluricarpellate ovaries of related genera (Engler, 1884). The diagnosis of pseudomonomery in these cases may be bolstered by occasional ob- servations of excentric sterile locules, sterile pla- centae (see Eyde et al., 1967), vestigial septa, extra vascular bundles, or variation in the orientation of placentae in flowers of the same spadix (Engler, 1884; Eckardt, 1937). Most Araceae, then, are either plainly syncar- pous or else have pseudomonomerous ovaries that can be homologized with clearly syncarpous ovaries of related genera. Pistia and most of Engler’s Aroideae, however, have unilocular ovaries with basal and/or apical placentation or with a single parietal (marginal?) placenta, and have no close relatives generally acknowledged to be syncarpous. Engler (1884) categorized these unilocular genera as unicarpellate and apparently believed them to e primitively so; Dahlgren $ Clifford (1982) adopted the same view. By contrast, I believe these to represent products of reduction from syncarpy, for the following reasons. (1) The flowers and in- florescences of these genera are in all other respects among the most advanced of any Araceae. (2) The Aroideae and Pistia are in my estimation (this will be discussed later) rather closely related to the basically syncarpous Thomsonieae. (3) In such a large family containing both syncarpous and uni- carpellate gynoecia, surely we ought to find a few species with pluripistillate, apocarpous gynoecia, but there is not even the slightest hint of such a condition within the family. (4) Arber (1925) and Eckardt (1937) reported the occasional presence of a second placenta in Arum. The hypothetical aroid ancestor must have had syncarpous, plurilocular ovaries with multiovulate, axile placentation. There is no strong evidence for apocarpy, unipistillate or pluripistillate, anywhere within the family. Pervasive throughout the Arace- ae are trends toward pseudomonomerous, uniloc- ular ovaries with pauciovulate, parietal to basal and/or apical placentation. But it is likely that high ovule number and high locule number have oc- casionally been secondarily derived. Furtadoa, in which the number of ovules in the unilocular ova- ries varies from 15 to 20 in the lowermost flowers to five, three, or even one in the uppermost female flowers of the same spadix (Hotta, 1981), may best exemplify a secondary increase in ovule number (see Grayum, 1984, for additional candidates). The high locule numbers noted previously for some species of Philodendron probably reflect a sec- ondary increase in carpel number, as Engler (1884) originally hypothesized, in spite of the assertion of Eyde et al. (1967) that “и would strain credulity” to postulate any such increase. Evidence for this secondary increase is provided in Grayum (1984). Mayo (1986) proposed that high locule numbers in Philodendron may have evolved in response to gall wasp parasitism. The primitive carpel number of Araceae is cer- tainly either two or three, though there appears to be little basis within the family for selecting between these alternatives. The conversion from dimery to trimery or vice versa seems rather trivial and to have occurred independently any number of times, at least among hermaphroditic taxa. Both floral Volume 77, Number 4 1990 Grayum 661 Evolution and Phylogeny of Araceae plans are frequent and occur together in several genera (Rhaphidophora, Rhodospatha, Spathi- phyllum, Orontium, Cyrtosperma, Lasia, Podo- lasia, Urospatha, and Dracontium), often in the same species or even on the same spadix. The same tendency can also be seen in the ovaries of some monoecious genera (see Grayum, 1984, for a list). Several clues, however, suggest that trimery is perhaps more primitive (see Grayum, 1984), and this is by far the most common condition in mono- cots as a whole, although the same duality occur- ring in Araceae is widespread in putatively related monocot families (Burger, 1977). As with male flowers, adjacent female flowers in Araceae may become fused to one another on the spadix. This is obviously a derived condition. The tendency is most pronounced in New World Col- ocasioideae. Pistils of Xanthosoma and Aphylla- rum have thick, spreading, discoid styles that are coherent with those of adjacent pistils (Madison, 1981). In Syngonium, the ovaries themselves are connate, resulting in a syncarpous multiple fruiting body. This also occurs in Cryptocoryne and Pip- tospatha sect. Gamogyne. The stylar regions of adjacent bisexual flowers in most Monstereae become fused following ab- scission of the stamens, thus forming a mantle that is shed as a unit, exposing the seeds upon ripening (Engler & Krause, 1908). Pistillodes in association with fertile male flowers are less common than the converse situation. Not surprisingly, pistillodes are seen to best advantage in those few genera with perigoniate, unisexual flowers: Zamioculcas, Gonatopus, and Stylo- chaeton. These are usually much-reduced struc- tures, but in Zamioculcas sterile ovules are still evident in the pistillodes of male flowers (Engler, 1884). In genera with naked flowers, evidence (if any) of a vestigial gynoecium is usually limited to a central, vacant spot, as in the lower male flowers of some species of Amorphophallus, Dieffenbach- ia, and Taccarum, or to the presence of a central, free-standing vascular trace, as in the synandria of some Colocasia and Arisaema species (Hotta, 1971). Actual pistillodes occur with regularity only in Furtadoa, where a conspicuous pistillode ac- companies each male flower nearly to the tip of the spadix (Hotta, 1981). No function is yet known for these. Ovule morphology and orientation in Araceae are fully discussed in Grayum (in press b); see Table 4 for polarities of important characters. French (1986c) reinvestigated aroid ovule orientation, and his results are included in Table 3. One of French’s more significant findings is that the traditional sys- tem of classification of ovule orientation types, i.e., *anatropous," etc., is appar- ently inadequate in Araceae; for example, ortho- tropous ovules in Philodendroideae and Coloca- sioideae differ significantly from those of Aroideae and are probably not homologous for this condition. French (1986c) also systematically surveyed ovary vasculature, characterizing three main pat- terns (simple bundles, multiple dorsal chalazal bun- dles, and radially symmetric chalazal bundles) high- ly correlated with ovule orientation and, somewhat more weakly, with ovule number and size. Simple bundles are the most widespread condition, and are exclusive to Araceae with bisexual flowers and/or a perianth; they are probably primitive. Fruits and seeds of Araceae have yet to be studied adequately, since these structures often fail to develop in greenhouse situations (Engler, 1884). As far as is known, aroid fruits are always berries. Since fleshy fruits are generally believed to have been derived from dehiscent fruits (Roth, 1977), this feature can be regarded as one of the few synapomorphies for the family Araceae. Araceous berries may be one- to many-seeded, but there is never a bony endocarp. Dahlgren & Clifford’s (1982) report of drupes in Pistia is in error (see, e.g., Crisci, 1971; Mercado-Noriel & Mercado, 1978; da Silva, 1981). The berries of Lagenandra are said to dehisce basally, releasing the seeds (Bogner, 1987); however, in most cases aroid fruits are probably bird (Peckover, 1985; Shaw et al., 1985) or bat (Bogner, 1987) dispersed. Seed mor- phology and endosperm retention in Araceae are discussed in detail in Grayum (in press b; see also Tables 3 and 4 in the present paper). Nectaries. Septal (gynoecial) nectaries, recorded from many monocot families including palms, do not occur in Araceae. Dahlgren & Clifford (1982) recorded only stigmatic secretions and an occa- sional instance of perigonal nectaries (e.g., An- thurium) from Araceae. However, staminodial nec- taries occur in Spathicarpa (Troll, 1928), and probably also in Aglaonema (Daumann, 1930); sterile bisexual flowers may function similarly in Monstera (Madison, 1977; Ramirez & Gomez, 1978; Ramirez, 1980, 1987 Extrafloral nectaries have only rarely been re- ported for Araceae, for example, at the apex of the petiole in Philodendron myrmecophilum (Madison, 1979b). J. C. French (pers. comm.) has found anatomically defined extrafloral nectaries only Ovule type anatropous Nucellus type crassinucellate Megaspore units tetrads Endosperm in seed present Cells of chalazal chamber -8 Base chromosome number = Pollen aperture type monosulcate Pollen shape boat-shape medium-sized 662 Annals of the Missouri Botanical Garden TABLE 4. Summary of taxonomically useful characters in the Araceae. Characters Primitive conditions Advanced conditions Growth habit rhizomatous or caulescent tuberous, epiphytic, vining Phyllotaxy distichous spiral iculum absent present Leaf form simple, cordate lanceolate; compound or lobed Leaf venation allel reticulate Stomata paracytic? anomocytic, tetracytic mpound vascular bundles absent present Coria vascular system absent present ermis absent present Trishoscloreide absent present Secretion vessels absent or single cells “laticifers” Biforines absent resent Prisms absent present Spathe ype Types II, III, & IV Spadix hermaphroditic monoecious Perigonium present absent or fused Locules of ovary 2 or 3 one or many Ovules per locule many few or one Placentation axile parietal, basal, apical Stamen leng elongate stout umber of stamens 4-6 fewer or more Anther dehiscence longitudinal apical, lateral, poricidal Staminodia in female flowers present absent or fused ropous tenuinucellate x — 15, 16, 17, 18, 19, и zonate, forate, inapertura lobose small, large, very large Pollen-unit tetrads Exine sculpturing foveolate-reticulate striate, psilate, spinose, etc. Pollen starch content absent present Pollen nuclear number III in Engler's sereni Lasioideae (8 of 11 genera examined) an ilodendroideae а а omalomena, ни and Anubia EMBRYOLOGY Systematically important aspects of aroid em- bryology have been recently and exhaustively re- viewed elsewhere (Grayum, 1985, 1986a, in press b; also see Tables 3 and 4). Based on the available evidence, the Araceae appear quite as variable embryologically as in most other regards. Intrafamilial variation in many char- acters, such as nucellar development, embryogeny, and endosperm development, is intriguing and probably of great systematic importance. However, with the exception of a few comparatively unim- portant characters (anther endothecium type, pol- len nuclear number, pollen starch content, ovule orientation, and endosperm retention), information in most embryological subfields is available for only a handful of aroid gener Embryological аг states that appear in- variable in Araceae include extrorse anther dehis- cence, periplasmodial anther tapetum with uni- nucleate cells, absence of perisperm, and helobial (haustorial) endosperm development. All other im- portant characters vary, but the hypothetical aroid ancestor is considered to have had four microspo- rangia per anther; endothecial thickenings present; isobilateral microspore tetrads; binucleate, starch- less pollen at anthesis; bitegmic, anatropous, cras- sinucellate ovules with a thick nucellar cap; uni- Volume 77, Number 4 1990 Grayum 663 Evolution and Phylogeny of Araceae cellular ovule archesporia; megaspores in tetrads, with the chalazal one functional; Polygonum-type embryo sacs; solanad or caryophyllad embryogeny; ab initio cellular division in the micropylar en- dosperm chamber; 2-8 cells in the chalazal en- dosperm chamber; and endosperm retained in the ripe seeds. Some important apomorphic variations on these themes are discussed under the individual taxonomic headings in the second half of this paper. POLLEN MORPHOLOGY Pollen morphology of Araceae has recently been reviewed in detail by Grayum (in press a). Refer- ence should be made to that paper, which includes scanning electron micrographs of pollen for most genera, and to Table 4 in the present paper; the definitive light-microscopy survey of aroid pollen is by Thanikaimoni (1969) External pollen morphology is exceptionally variable within the Araceae. Internal exine struc- ture is probably equally variable, although as yet little studied. The most primitive araceous pollen is regarded as monosulcate, boat-shaped, hetero- polar, bilaterally ion and small t dium-sized (22-34 um in maximum Ma: I is shed in monads, has foveolate to reticulate exine sculpturing, and is probably tectate-columellate (Grayum, in press a). The most advanced pollen is held to be inaperturate, globose, apolar, and ra- diosymmetric. Exine sculpturing is exceedingly variable; the evolutionary derivation of the various forms is complex, with similar types having evi- dently been derived via two or more different path- ways (see Grayum, in press a). This is the case with psilate and spinose exines, considered the most highly derived conditions; these two types are closely correlated with, and presumably specialized for, beetle and fly pollination modes, respectively (Gra- yum, 1986b). POLLINATION BIOLOGY Data on pollination о for the majority of Araceae are still very meager. This is especially true in the case of tropical species, which compose the bulk of the family. Pollination in the vast genus Philodendron, for example, is mainly known only from a few studies on two species of subg. Me- conostigma (see, e.g., Gottsberger & Amaral, 1984), and no careful field study of pollination in Anthurium has been published. The only tropical aroid genus that is somewhat well known in this regard is Amorphophallus, mainly from the work of van der Pijl (1937, 1953). Phylogenetically critical taxa (such as Potheae and Lasieae) with primitive floral and pollen characters are in par- ticular need of study. Even the temperate aroids, with the exception of the mainly European genera Arum and Dracunculus, are surprisingly poorly known in terms of pollination biology. Data on aroid pollinators were summarized by Grayum (1984, 1986b), including only field studies conducted in natural habitats. Even with this re- striction, the data must be evaluated carefully, since many are superficial or anecdotal. The broad- est generalization that can be made is that Araceae are, as far as is known, exclusively entomophilous. The overwhelming majority of species are polli- nated by beetles and/or flies. North American species of Arisaema have variously been said to be pollinated by thrips (Rust, 1980), by flies (Bier- zychudek, 1982), or by both (Rock, 1952). Bee pollination occurs rarely in Araceae. It is confirmed only in species of Anthurium and Spathiphyllum (Williams & Dressler, 1976), re- ported in Dracunculus canariensis (Boyce, 1986), and suspected in some Syngonium species (Mad- ison, 1979a; pers. obs.). Madison (1979a) de- scribed Montrichardia as pollinated by *'bees, pre- sumably”; however, dynastine scarab beetles have recently been collected from spathes of M. arbo- rescens in Costa Rica (M. Grayum & G. Schatz, unpublished data). Species of Monstera have also been reported as bee-pollinated (Madison, 1977; Ramirez & Gomez, 1978), but these reports are also dubious; at least two species of the closely related and florally similar genus Rhodospatha are now definitely known to be beetle-pollinated (G. Schatz & M. Grayum, unpublished data), and dy- nastine scarab beetles were recently discovered in inflorescences of Monstera oreophila in Chiriqui, Panama (M. Grayum, unpublished data). Symplo- carpus foetidus has been said to be partly polli- nated by hive bees (Trelease, 1879) in North Amer- ica, where Apis mellifera is not native. Beetle pollination in neotropical aroids virtually always involves large scarab beetles of subfamily па of the ao Scarabaeidae. Pollination documented or strongly indicated for species of Monstera, Rhodospatha, Homalomena, Philodendron, Dieffenbachia (Young, 1986), Montrichardia, Caladium, Xan- thosoma, and Syngonium and is perhaps general throughout most of these genera. Dynastine scar- abs have even been collected from spathes of a Homalomena species in Malaya (G. Schatz, pers. comm.). Beetle-pollinated paleotropical aroids more commonly attract beetles of other taxa, mainly Nitidulidae (see, e.g., Thompson & Rawlins, 1986), though other groups (nondynastine Scarabaeidae, c < 664 Annals of the Missouri Botanical Garden Scaphidiidae, Staphylinidae, Dermestidae, Scyd- maenidae, Silphidae) are occasionally implicated. The north temperate Lysichiton americanus is pollinated by staphylinids (Pellmyr & Patt, 1986). These beetles appear to have a less specialized relationship with Araceae than do the dynastine scarabs; several families are frequently reported together on a single inflorescence, often along with one or more families of flies (see, e.g., Bogner, 1976a, 1981a; Dakwale & Bhatnagar, 1985). Fly-pollinated aroids are mainly temperate or paleotropical; a few neotropical taxa, most notably species of Dracontium, are strongly suspected to be myiophilous. The flies involved are usually small, weak-flying types, such as fungus-gnats, fruit-flies, and various blood-sucking midges. Apparently they are deceived into entering aroid spathes by olfac- tory and perhaps other cues. Most frequently men- tioned are the families Mycetophilidae and Dro- sophilidae; others include Syrphidae, Chloropidae, Muscidae, Anthomyiidae, Centropogonidae, Sim- uliidae, Psychodidae, Sphaeroceridae, Calliphori- dae, Sciaridae, and Culicidae Fly and beetle pollination in Araceae are usually attended by complex and obviously highly evolved morphological, behavioral, and physiological spe- cializations on the part of the plant species, in- cluding enclosing, constricting spathes (sometimes secreting resin); unisexual, naked flowers segre- gated on the spadix; and often highly modified sterile flowers in one or more positions on the spadix, presumed or known to function, e.g., as food rewards for pollinators, or in nectar secretion, occlusion of the spathe orifice or (at least in the case of naked, terminal appendages) in the pro- duction of heat and/or odors Pollen exine sculpturing is strongly correlated with pollination mode in Araceae. This notion ap- parently first surfaced in the literature as the fol- lowing observation by Rock (1952: 73) concerning the North American species Arisaema atrorubens (= A. triphyllum): “The pollen grains appear to be especially designed for dispersal by insects; the tiny spherical grains are covered with numerous barbed spines.” Strong, family-wide correlations of spinose pollen with fly pollination and psilate pollen with beetle pollination have recently been docu- mented (Grayum, 1986b), even if functional rela- tionships such as that implied in Rock’s statement have yet to be elucidated. Thus beetle and fly pollination, which are basic to Araceae and associated with across-the-board specializations, are here regarded as primitive for Araceae. It is unclear which of these two modes might have characterized the hypothetical aroid ancestor. Out-group comparison suggests beetle pollination, which is much more common than fly pollination in palms, and is speculated to be more primitive in that family (Henderson, 1986). Cy- clanthaceae, so far as is known, are also basically beetle-pollinated (see Beach, 1982). Pollination by families of smaller beetles, particularly Nitidulidae, is much more common among flowering plants in general than pollination by Scarabaeidae (see Gra- yum, 1984) and is perhaps more primitive within Araceae, even though it appears to be attended by less extensive specializations. Pollination by Niti- dulidae and Curculionidae is especially widespread in palms (Henderson, 1986) Bee pollination in Araceae is known with cer- tainty only from genera with comparatively un- specialized floral and pollen morphology and is here considered a rather recent development within the amily Some evidence suggests that the very earliest aroids or an immediate ancestor might have been wind-pollinated. The pollenkitt of Araceae, for ex- ample, is not typical of entomophilous angiosperms. Whereas entomophily normally implies sticky pol- len, araceous pollenkitt is nearly always deficient, resulting, as for example in Arum, in dry pollen "equivalent to that of anemophilous angiosperms” (Hesse, 1980, 1981). Pohl (1931) reported that the pollenkitt of Philodendron bipinnatifidum is not resinous or oily as in most other angiosperms, but **plasma-like" and rapidly drying when exposed to air; it does not suffice to stick the pollen onto hard-bodied insects (this is apparently accom- plished in Philodendron by means of a resinlike secretion of the inner surface of the spathe; see von Martius, 1831; Warming, 1883; Schrottky, 1910; Pohl, 1931; Gottsberger & Amaral, 1984; Mayo, 1986). Camazine & Niklas (1984) showed that Symplocarpus foetidus, a phenetically prim- itive species, has “a facultative capacity" for wind pollination. The possibility that Araceae may be derived from anemophilous ancestors certainly de- serves further consideration. Heat production in the vicinity of the inflores- cence during anthesis is known from a wide range of aroids, including hermaphroditic genera, such as Monstera and Symplocarpus (Knutson, 1974) as well as monoecious genera in all major subfam- ilies. The exact site of heat production is the spadix, the energy being derived from respiration of starch (Knutson, 1974) or lipid (Seymour et al., 1984) contained in the rhizome or elsewhere. The physiological literature on heat production by aroid inflorescences is voluminous but mostly divorced from taxonomic, evolutionary, and eco- Volume 77, Number 4 1990 Grayum 665 Evolution and Phylogeny of Araceae logical considerations. The exact function of heat production has been much debated; apparently there is a major role in attracting pollinators, but whether this is effected by the heat directly, or indirectly via an increase in the volatility of amines, indoles, etc., is a bone of contention (see, e.g., Moodie, 1976). The only paper of systematic significance is apparently that of Leick (1915), who recognized four main types of heat production among those Araceae for which this feature was then known, based on the periodicity of heating episodes and their localization on the spadix. As modified an augmented by Engler (1920b), these are: (1) the Monstera type, with three episodes on successive days, the middle one being strongest, and no marked localization; (2) the Philodendron type, with epi- sodes on two successive days, the second stronger, localized in the male portion of the spadix; (3) the Alocasia type, with 3-5 maxima, the second being the main one, localized in the sterile apical appen- dage of the spadix; and (4) the Arum type (also known from Sauromatum, Dracunculus, and the Thomsonieae), with episodes on two successive days, the first being stronger and localized in the sterile appendage, the second (insignificant) episode con- fined to the male portion of the spadix. Meeuse (1978) has offered a limited evolution- ary interpretation of the various types of heat pro- duction in Araceae, suggesting that aroids with a more primitive inflorescence morphology, such as an open spathe, bisexual flowers, and other fea- tures, must have at least two heating episodes: one to attract pollinators, another to attract pollen vec- tors. More advanced aroids (with an enclosing or constricting spathe) can get away with a single pronounced heating episode at the time of stigma hie en a since in effect they have a audience” on hand for anther dehiscence. That such species yet retain weak, secondary heating episodes is taken as evidence that this type of heat "captive production evolved from one of the more primitive types in Araceae rather than directly from an ancestor not exhibiting heat production. KARYOLOGY Data on aroid chromosome numbers available up to 1984 are assembled, tabulated, and discussed at length in Grayum (1984). The raw data will not be reproduced here but are frequently alluded to under the individual taxon headings in the second half of the present paper. Karyological data for the entire family Araceae are collated, revised, augmented, and analyzed in an important recent contribution by Petersen (1989), to which the read- er is referred. Among the polytypic aroid genera, only Cryp- tocoryne (Arends et al., 1982; Reumer, 1984), Arisaema (Hotta, 1971), Amorphophallus (Chau- han € Brandham, 1985), Alocasia (Bhattachar- ya,1974), Arum (Bedalov, 1981), Biarum (Talav- era, 1976), and Anthurium (Sheffer & Croat, 1983) are reasonably well known chromosomally. The following relevant data can be distilled from the available counts in these and other genera. The lowest diploid number reported to be characteristic of any species in Araceae is 2n = 14, occurring oyi in Typhonium flagelliforme (Ayyangar, 1973; = 14 has also been reported once for Pistia кые ы but 2n = 28 is more typical of that species); the highest is 2n = 140, in Arisaema heterophyllum. One of the most frequently re- ported diploid numbers in Araceae is 2n = 28, and multiples of seven clearly prevail in the family. These considerations have led virtually all modern workers (Mookerjea, 1955; Larsen, 1969; Mar- chant, 1974; Raven, 1975; Ramachandran, 1978) to the conclusion that the primitive base number for the family is x = 7, which may also be the primitive base number for angiosperms as a whole (Raven, 1975). This is clearly the most reasonable viewpoint, considering the limitations of the data set, and is adopted here. It should be mentioned, however, that multiples of six appear to be dispro- portionately represented as base or co-base num- bers among phenetically primitive aroids (see Gra- yum, 1984), e.g., those groups with bisexual flowers (most Pothoideae, Monsteroideae, Calloideae, and Lasieae). The putative sister-family Acoraceae is based on x = 12. Thus it may prove ultimately that Araceae are based on x = 6, even though multiples of six are now uncommon within the amily. A fair number of aroid genera appear to have base chromosome numbers that are not multiples of seven. Particularly common are x = 1 18, 19, and 20. Genera in this category largely comprise the subfamily Philodendroideae, as in- terpreted in this paper Although neither intrageneric polyploidy nor aneuploidy are pronounced features of Araceae (see Goldblatt, 1980; Arends et al., 1982; Sheffer & Croat, 1983), paleopolyploidy and dp кеч have apparently played sizeable ro mo- some evolution (Jones, 1957; Goldblatt, 1980; ‘Shef. fer & Croat, 1983). Even allowing for primary base numbers as high as x = 13, 82% of all araceous species counted are polyploids (i.e., 2n = 28 or higher), as opposed to only 55% for monocots as a whole (Goldblatt, 1980). he basic scenario for chromosome evolution in 666 Annals of the Missouri Botanical Garden Araceae (Jones, 1957; Larsen, 1969; Marchant, 1974) envisions the evolution of a (paleo)aneuploid series based on x = 7, giving rise to the primary base numbers x = 6, 8, 9, and perhaps even 5, 10, and 11. Subsequent doubling of chromosomes must then have resulted in euploid series based on these numbers, this in view of the fact that (aside from a very few counts of 2n = 14, 16, 18, 20, and 22) no potential **paleodiploids" are known to exist. With a few minor reservations (see Grayum, 1984), this scheme is accepted here. Another important process in aroid chromosome evolution may have been amphiploidy, which might explain the rather frequent occurrences of 2n = 26, 30, and 34 on the basis of primordial hybrid- izations involving the primary base numbers 6 and 7, 7 and 8, and 8 and 9, respectively, to yield the secondary base numbers 13, 15, and 17 (Jones, 1957; Ramachandran, 1978). Amphiploidy may also explain some of the more unusual counts in Araceae, such as 2n = 22 (a putative allotetraploid of 5 and 6, known only in the taxonomically dis- parate genera Jasarum, Zomicarpa, and Ambro- sina) and 2n = 38 (a putative allotetraploid of 9 and 10, occurring only in Arophyton, whose clos- est relatives have chromosomes in multiples of nine). Hybridization of extant aroid species in the field has rarely been reported, though it occurs, for example, in Dieffenbachia (H. J. Young, pers. comm.). Of course, any of these putative amphiploids may equally well have arisen as a result of aneu- ploidy at the secondary level, followed by chro- mosome doubling. This, in fact, may be a more conservative and realistic explanation, and is es- pecially to be suspected in such genera as Ty- phonium, Eorum and e f ne ibe all with 982 apparent het Arends at al., In either case, chromosome numbers such as those discussed in the preceding paragraph must be re- garded as advanced. In summary, most chromosome counts in Агасе- ae appear to be explainable on the dual basis of ascending and descending aneuploidy at the sec- ondary (paleodiploid) level from a primitive base number of x = 7 (or perhaps 6), with subsequent (paleo)polyploidy in many cases. Ascendin ploidy at the primary (paleodiploid) level, though it no doubt has occurred up to at least x = 9 and probably even x = 11 in a few cases, does not ing aneu- seem to merit the emphasis it has received, given that so many genera are still clearly based on multiples of seven. The most primitive extant chro- mosome numbers in Araceae are probably, on the one hand, those ““paleodiploids”” of members of the original paleoaneuploid series, i.e., 2n = (12), 14, 16, 18, and perhaps 20 and 22; and, on the other hand, **paleotetraploids" of the primitive base num- ber, i.e., the widespread 2n — 28 (or perhaps 2n — 24, if x — 6 is accepted as primitive). Allotetra- ploids, secondary aneuploids, higher polyploids, and combinations of these would necessarily be more advanced. Allotetraploids based partly on x = 7 (or 6), 1.e., 2n = 26 and 2n = 30, should be relatively more primitive than those involving high- er base numbers, and so on. More chromosome data are needed, and older reports need to be thoroughly reevaluated and reas- sessed, as Sheffer & Croat (1983) did for An- thurium, before anything close to a clear picture for chromosome number evolution in Araceae emerges. Analyses of chromosome evolution and data on chromosome size for some individual aroid subtaxa are offered in Grayum (1984) and Petersen (1989). Aroid chromosomes range in length from about 1 um (Pistia stratiotes) to at least 13.6 um (Zam- ioculcas; Jones, 1957); most genera fall into the "medium" range (3-6 um) of Ramachandran (1978). The phylogenetic significance of these data is as yet unclear. PHYTOCHEMISTRY Phylogenetically relevant phytochemical data for Агасеае are found in Hegnauer (1963, 1986, 1987), Gibbs (1974), Dahlgren & Clifford (1982), and Grayum (1984). Data on useful characters are particularly sparse for this field, so no phytochem- ical characters were used in the character analysis undertaken during this study. Only the most basic and significant trends are recorded below. Inorganic compounds. Silica bodies are unknown in Агасеае (Dahlgren & Clifford, 1982). As dis- cussed previously, calcium oxalate crystals occur throughout the family (Nicolson, 1959). Carbohydrates. Polysaccharide mucilage is very abundant (and perhaps universal) in the Araceae, in monocots in general (de Wildeman, 1942; Gibbs, 1974). Czaja (1978) studied the structure and properties of starch grains in vascular plants and divided Araceae into two seemingly artificial groups on this basis. Aroid starch grains were also studied by Reichert (1913; see summary in French, in press), although too few genera were examined to permit any significant generalizations. Nitrogenous compounds. Amines are supposedly “common in the inflorescences of aroids” (Gibbs, Volume 77, Number 4 90 rayum Evolution and Phylogeny of Araceae 1974), although virtually all reports concern Amorphophallus and genera of the Arinae (Sau- romatum, Dracunculus, Arum). Free indoles have also been reported from the same genera. All of these compounds appear to be concerned with the production of heat and stench associated with pol- lination by flies and carrion beetles (Chen & Meeuse, 1971) and ought to be expected in any genus exhibiting this syndrome. Alkaloids are reported from a range of aroid genera (see Willaman & Li, 1970; Hegnauer, 1986) but have been identified in only a few cases. Terpenoidal compounds. Volatile terpenoids (ethe- real oils) are very uncommon in Araceae. Hegnauer (1963, 1986) reported that the dried rhizomes of the Old World species Homalomena rubescens and H. aromatica yield about 5% and 1.2% ethereal oil, respectively, apparently from the schizogenous (intercellular) ducts. New World species of Homa- lomena typically have highly aromatic foliage, rhi- zomes, and/or roots (pers. obs.). El-Din (1968) identified a bicyclic monoterpene as the principal component of the floral odor in Alocasia portei, suggesting a role in attracting insect pollinators. It is possible that volatile ter- penoids are more widespread in aroid inflores- cences TFYaterpesoidal saponins are rather frequent in Araceae, as in many other monocots (Hegnauer, 1963; Gibbs, 1974; Grayum, 1984). Although these compounds are absent from many aroid genera, their distribution does not appear to be of system- atic significance. The genus Anthurium contains both saponiferous and saponin-free species here are a few isolated reports of steroidal saponins in Araceae. Hegnauer (1963) regarded these as dubious. Many authors have referred loosely to 66 resin" r "latex" in regard to araceous secretions, but until recently it was doubtful as to whether true polyterpenoidal resin or latex actually occurred in the family. Gibbs (1974) provided one report of ““polyisoprenoids” in Araceae, from Arum. Most earlier authors (e.g., Trécul, 1866; Lierau, 1888; Krause, 1908; Solereder & Meyer, 1928; Heg- nauer, 1963) described the contents of aroid “‘la- ticifers" as phenolic (i.e., tannins) rather than ter- penoidal. However, French & Fox (manuscript) have now studied and characterized the apparently terpenoidal latex of various Araceae, and the con- tents of aroid resin canals are now known to be terpenoidal as well (French, 19874). French & Fox (manuscript) have investigated the ultrastructure of latex within the subfamily Colocasioideae and found it to be of systematic significance. They distinguished four types of par- ticles, which they called S bodies, rubber particles, bodies, and R bodies. S bodies and rubber par- ticles occur only in New World Colocasioideae and in the Old World genus Hapaline, whereas Y bodies are restricted to Old World genera except Hapaline. R bodies occur mainly in Old World genera. Fox & French (1988) found that sterol- esters are present in the latex of Xanthosoma, Syngonium, Colocasia, and Alocasia. However, free sterols occur only in latex of the former two genera, leading the authors to speculate that Old and New World Colocasioideae may be chemically distinguishable. Phenolic compounds. Syringin is apparently not present in Araceae (Gibbs, 1974) Flavonoids (including simple flavones, flavonols, anthocyanins, proanthocyanidins, Havans C-gly- cosides, and э flavonoids) are ANE the f. markers 101 у most valuable poses” (Harborne, 1982). Aroids are cana by a predominance of flavone C-glycosides, abun- dant flavonols, and a comparative dearth of simple flavones. The predominance of flavone C-glycosides and the frequent presence of proanthocyanidins are construed as primitive conditions (Williams et 1981) — al., Anthocyanin pigments are not generally re- garded as having great systematic value. In Arace- ae they occur in reddish fruits, some spathes and spadices, petioles, and in the epidermis of some leaf blades (Williams et al., 1981). Cyanidin 3-rutinoside is by far the most common form in Araceae, though several others are known. Ап unusual anthocyanin, the gentiobioside of pelar- gonidin and/or cyanidin, has been detected only in Anchomanes and Cercestis and is believed to indicate a close relationship between these genera (Williams et al., y The distribution of flavonols in Araceae is re- ported in Bate-Smith (1968), Williams et al. (1981), and Grayum (1984). Flavonols as a group occurred in only 27% of the species tested by Williams et al. (1981); they are frequent only in Engler’s Col- ocasioideae, Philodendroideae, and Calloideae, and are the predominant flavonoids in leaves of the tribe Orontieae. Flavonols were not detected in Lasioideae, Pistia, or Aroideae except Stylochae- ton, where they predominate in a manner **curi- ously similar" to the tribe Orontieae (Williams et al., Flavone C-glycosides are LI the most character- istic flavonoid constituents" of the Araceae and 668 Annals of the Missouri Botanical Garden were found in 82% of the species surveyed by Williams et al. (1981). The vast majority of species in all taxa except ne and Orontieae contained these compou imple flavones are rare in Araceae, occurring in only 6% of the species examined (Williams et al., 1981). Though rare, flavones do seem to have some systematic value in Araceae in that they are concentrated in the subtribe Arinae. Luteolin and chrysoeriol both occur in this taxon, the former exclusively. Araceae lack the distinctive flavone tricin, so characteristic of palms, grasses, and Cy- peraceae (Williams et al., 1981; Harborne, 1982). Proanthocyanidins are “extremely widely pres- ent in leaves of monocotyledons and their almost universal occurrence robs them of any systematic significance" (Harborne, 1982). Moreover, their presence may be masked by anthocyanins, at least in Araceae, making them difficult to detect (Wil- 1981). Proanthocyanidins are wide- spread in Araceae but appear to be absent from Monsteroideae, Orontieae, and Aroideae (except Stylochaeton and the Arophyteae) (Williams et al., 1981) Flavonoid sulphates appear to be rare in Arace- liams et al., ae, having been identified in only a few, mostly 1981) Phenolic (cinnamic) acids (caffeic, p-coumaric, sinapic, and ferulic) are widespread in Araceae according to Bate-Smith's (1968) survey. They seem particularly rich in Monsteroideae and poorly represented in Aroideae, but beyond that no gen- eralizations are possible. Harris & Hartley (1980) detected no phenolic acids bound to the cell walls unrelated genera (Williams et al., of Sauromatum or Pistia. Although leucoanthocyanidins have been alleged to be common in Araceae (Hegnauer, 1963), all records either pertain to weak or dubious reactions or are contradicted by negative or doubtful reports (see Grayum, 1984), with the exception of two mutually corroborating positive reports for Phil- odendron. Tannins are frequently attributed to Araceae in the early literature (e.g., Trécul, 1866; Solereder & Meyer, 1928); however, it appears that only Gibbs (1974) has rigorously and systematically tested for them. According to his data, tannins are present in some genera and absent in others. The systematic significance of their distribution, if any, is not yet clear. Anthurium appears to have both tanniniferous and tannin-free species. According to Hegnauer (1963), aroid tannins are of the type derived from leucoanthocyanidins, so the wide- spread occurrence of the latter compound in Arace- ae Is perhaps to be expected. Cyanogenic glycosides. The Araceae are gener- ally regarded as particularly rich in cyanogenic glycosides (Hegnauer, 1977; Dahlgren & Clifford, 1982). Only the tyrosine-derived glycoside triglo- chinin is known from Araceae (Nahrstedt, 1975). Ап enzyme is required to cleave off the cyanide from a cyanogenic glycoside (Hegnauer, 1977); the highly specific enzyme 8-glucosidase has been isolated from Alocasia macrorrhizon (Hósel & Nahrstedt, 1975). According to Nahrstedt (1975), cyanogenesis is known from all aroid subfamilies except Monster- oideae, Calloideae, and Pistioideae. The individual reports cited by Hegnauer (1963, 1986) and Gibbs (1974) tend to support this, although there are at least two positive reports for Calla and one for Lysichiton. Reports for many genera are contra- et rendering all of the unreplicated reports dubiou The а of cyanogenic апа попсуапо- genic genera in Агасеае, according to the available laboratory data (see Grayum, 1984), does not ap- pear phylogenetically significant. It is of interest, however, that crushed leaves of many Old World Colocasioideae yield a strong cyanide odor, whereas those of New World members of this subfamily do not. This test has been used in the field in Central * America to separate the indigenous Xanthosoma from some widely naturalized species of Alocasia (B. E. Hammel, pers. comm., substantiated by pers. obs.). Laboratory data are not sufficient to provide a sound basis for these observations. PHYLOGENY AND CLASSIFICATION CHARACTER ANALYSIS The possible phylogenetic (cladistic) relation- ships of the infrafamilial taxa in Araceae have been analyzed formally in the present study, with respect to the phenetic data assembled in the foregoing sections. Thirty-six characters were used in this analysis; these were selected for their apparent taxonomic value and also because they have been comparatively well surveyed in most cases. Char- acters not involved in the formal analysis have been considered and frequently brought into play in interpreting the often ambiguous implications of the cladistic analysis; the same is true for important characters that were discovered or surveyed after the analysis. The 36 principal characters used in the analysis are listed in Table Certain of Engler’s infrafamilial taxa of various rank were accepted a priori as natural assemblages. These generally comprise genera, subtribes, and Volume 77, Number 4 1990 Grayum 669 Evolution and Phylogeny of Araceae tribes; in only one case (Pistioideae) has an entire Englerian subfamily been considered as a unit. The taxonomic units employed in the analysis are the same as those used for the headings in the following discussion, except where indicated. Each of these units was compared with every other unit for all of the characters in Table 4, and all shared derived characters (i.e., potential synapomorphies) were noted for each comparison. Presumed autapomor- phies were not utilized; only the presumed most primitive condition of each character in a taxon was employed for analytical purposes. Similar au- tapomorphies of different taxa thought perhaps to reflect parallel tendencies are occasionally noted in the discussions. Similar character analyses to that described above, employing fewer characters, have already been performed on Araceae by Howarth (1957) using mostly floral and embryological characters. Her analyses dealt with larger and more hetero- geneous, sometimes artificial, taxa —Engler?s subfamilies and Hutchinson's tribes—and yielded via a rather simple calculation a so-called *'ad- vancement index” for each taxon. Such an ap- praisal, though perhaps providing a more or less adequate phenetic evaluation indicating the ap- proximate evolutionary grade of each taxon, says nothing about cladistic relationships. or the record, Howarth’s advancement indices for the eight Englerian subfamilies are as follows (higher numbers indicate greater phenetic ‘‘ad- vancement," i.e., a smaller proportion of primitive character states): Monsteroideae 36%, Lasioideae 36%, Pothoideae 38%, Calloideae 42%, Aroideae 54%, Philodendroideae 75%, Colocasioideae 79%, and Pistioideae 86%. Howarth’s advancement in- dices for Hutchinson’s tribes are included in Ap- > pendix 1. These figures are sometimes more in- formative, since the taxonomic units involved are smaller and consequently more likely to be mono- phyletic. The results of the character analysis undertaken in the present study are discussed under the fol- lowing headings, as they pertain to aroid subtaxa accepted as monophyletic. The linear arrangement reflects the cladogram proposed at the end of this section. Since the discussions that follow incorpo- rate much information presented in other parts of this paper, literature citations have frequently been omitted. Gymnostachys. The monotypic or ditypic Gymnostachys is the only aroid genus endemic to Australia. Because of its narrowly linear leaves, Gymnostachys bears a superficial resemblance to Acorus, with which it has been closely aligned virtually since its discovery. There is not, however, good evidence that Gymnostachys is related to Acorus, nor that it should be dissociated from Ara- ceae as the latter genus has been (see Grayum, 1987) Gymnostachys is isolated (French & Kessler, 1989) and phenetically rather primitive and shares few derived characters with other aroid taxa. It has been rather poorly investigated, especially em- bryologically. Although there are some striking sim- ilarities to the tribe Orontieae (foliar prophylls; lack of ovular trichomes; lack of a spathe; gynoecial features similar to Symplocarpus), I believe that Gymnostachys is most closely related to the Po- theae. The two taxa occur sympatrically and have milar patterns of anther endothecial thickening (French, 1985a); further, Gymnostachys is said to have 2n = chromosomes (J. Bogner, pers. comm.), an unusual number in Araceae but com- parable to 2n = 24, reported for species of Pothos. Thus Gymnostachys is here included as a tribe in subfamily Pothoideae. Judging from its occurrence in tropical rainfo- rests, where it grows alongside Pothos and Alo- casia, it seems likely that Gymnostachys entered Australia via New Guinea from Laurasia subse- quent to the collision of the Laurasian and Aus- tralian plates during the Miocene, about 10-12 myBP (Raven, 1979). Potheae. The Potheae, comprising the genera Pothos, Pothoidium, and Pedic netically rather primitive tribe, with the lowest ad- vancement index (23%) of any of Hutchinson’s 18 tribes (Howarth, 1957). This is slightly exagger- ated, however, due to Howarth’s almost complete cellarum, are a phe- reliance upon floral characters. Vegetatively, the tribe is more specialized, made up of mostly mono- podial climbers with geniculate petioles, distinctive (Pattern 3) stem vasculature, reticulate leaf ve- nation, trichosclereids (at least sometimes), prisms, vessels in the stem, a cortical vascular system, and a stem endodermis. The Potheae, which are still oorly known cytologically and are unknown em- bryologically, are here treated as the sister group to Anthurieae. All of the above character states are shared with at least some genera here considered to belong to the tribe Monstereae. Hence, the Potheae/ Anthu- rieae are here placed in a sister group relationship with the Monstereae and satellite taxa, thus ne- cessitating the union of Engler's subfamilies Po- thoideae and Monsteroideae. Nicolson (1960b) and Hotta (1970) have already raised this possibility, 670 Annals of the Missouri Botanical Garden and Carvell (1989) endorses such a merger on floral-anatomical grounds. Pedicellarum paiei is probably a species of Po- thos sect. Allopothos. All of the character states on which Pedicellarum was founded have been shown in this paper to be misinterpreted (introrse anther dehiscence) or variable within the species (pedicellate flowers, fused perianth members). Fused and distinct perianth members both occur in Pothos emotiflorus of sect. Allopothos (Grayum, 1984; Nicolson, 1984a). Although I once (Grayum, 1984) proposed a new combination in Pothos for Pedi- cellarum paiei, recent information (Nicolson, 1984a; Carvell, 1989) suggests that Pedicellarum is better retained for now. Zamioculcadeae. This isolated and bizarre Af- rican tribe of two genera (Zamioculcas, Gonato- pus) shows every indication of being a remnant of a very old African flora. Since the group is so highly idiosyncratic, its relationship to other aroid taxa is difficult to fathom. The greatest phenetic resemblances, according to the present analysis, are with the tribes Spathicarpeae and Monstereae. With the former taxon (in which Hutchinson ac- tually included Gonatopus), Zamioculcadeae share a tuberous growth habit, reticulate leaf venation, unisexual flowers, and the unusual base chromo- some number of x = 17. However, details of floral morphology strongly suggest that monoecy in both groups is of relatively recent and independent der- ivation. Zamioculcadeae are here treated as a sister group of Monstereae with which they share distichous, pinnate leaves, geniculate petioles, compound vas- cular bundles, and zonate, foveolate, starchy pol- len. Fully pinnate to bi- or tripinnate leaves are unique to Zamioculcadeae; however, Monstereae exhibit the greatest tendency toward pinnation of any other aroid subgroup; furthermore, zonate pol- len is restricted to these two tribes. There is little superficial resemblance between members of these taxa, although Zamioculcadeae must have been subjected to intensive selection pressure during the Neogene and Quaternary when the African flora was severely impoverished by widespread aridity (Raven & Axelrod, 1974) Bogner & Nicolson (in press) transferred Za- mioculcadeae to subfamily Lasioideae for unclear reasons. Floral anatomy, however, may provide some support for such an alliance (Carvell, 1989). The wide separation of Gonatopus from Zami- oculcas and the juxtaposition of the latter with Stylochaeton and the Arophyteae, as proposed by Hutchinson (1973), are altogether indefensible from the standpoint of character analysis. Anthurium. This huge neotropical genus is phenetically rather primitive in most respects and shares few derived characters with other aroid sub- taxa. The greatest resemblance seems to be with Potheae, near which most authors have placed Anthurium. These taxa share geniculate petioles, reticulate venation, cortical vascular systems, stems with an endodermis, and similar or perhaps iden- tical base chromosome numbers (apparently x = 12 in Pothos and in sect. Tetraspermium of An- thurium, and possibly the entire genus; see Sheffer & Croat, 1983). Although the forate pollen of Anthurium is unique in Araceae, the commonly spinulose-reticulate exine of this genus finds its counterpart only in Pothos sect. Allopothos (see rayum, in press a). Although Anthurium and Pothos differ in sev- eral important respects, including growth habit, phyllotaxy, and flavonoid pattern (Williams et al., 1981), they appear more similar to one another than either is to any other taxon, and they are here treated as sister groups. The di- or tritypic section Polyphyllium is unique within Anthurium in several important attributes (anisophyllous growth, internodal roots, inapertur- ate gemmate pollen, and other features), and may merit separate generic status. he unrivaled diversity of Anthurium in south- ern Central America and the Andean region is probably related to the Andean orogeny (Gentry, 1982). The practically invariable pollen morphol- ogy of such diverse sections as Pachyneurium (Grayum, in press a) suggests a relatively recent adaptive radiation. Spathiphylleae. The association of the oligo- typic New Guinean genus Holochlamys with the polytypic, largely neotropical Spathiphyllum in a single tribe is now widely accepted (see Grayum, 1984). This was a matter of controversy for some time, following Hutchinson’s ill-advised placement of Holochlamys in the tribe Lasiea The Spathiphylleae clearly ene to the Po- thoideae (where Spathiphyllum has been assigned by all members of the Schottian school) or Mon- steroideae (according to the Englerian school), as suggested by such characters as distichous leaves, geniculate petioles, possession of trichosclereids, absence of laticifers, and a thoroughly primitive floral plan (perhaps the most primitive in Araceae). The two subfamilies are combined in the present bue ess оаа ипдег Potheae). ca atively, either the Pothene or Мо" in having ап acaulescent or rhizomatous growth habit and in E 1 than Volume 77, Number 4 1990 Grayum 671 Evolution and Phylogeny of Araceae lacking prisms, vessels, a cortical vascular system in the stem, and a stem endodermis. Thus, Spathi- phylleae are here considered the most cladistically primitive tribe of Pothoideae. e most conspicuously derived feature of Spathiphylleae is its inaperturate, striate pollen, yet this is probably not homologous with similar pollen occurring in more phenetically advanced aroid taxa (see Grayum, in press a). The Spathiphylleae are of great biogeographic interest (see especially Nicolson, 1968b; Williams & Dressler, 1976; Grayum, 1984). The contem- porary occurrence of Spathiphyllum in Southeast Asia and in the Neotropics, and its absence from Africa, are probably best explained by a West Gondwanalandic origin followed by migration from Africa to Southeast Asia, and the subsequent ex- tirpation of the African elements due to increased aridity during the Neogene and Quaternary. This hypothesis and the only other plausible one (see Grayum, 1984) suggest that Spathiphyllum sect. Massowia (which occurs in Southeast Asia and the New World) is artificial, as Williams & Dressler (1976) intimated. Anadendrum. This distinctive Southeast Asian genus is largely unknown cytologically, embryo- logically, and anatomically. Its relationships are clearly with Pothoideae (Englerian school) or Mon- steroideae (Schottian school). Anadendrum ex- hibits a combination of characters that argues strongly for merging these two subfamilies; since they have been merged in the present treatment (see discussion under Potheae), the decision of where to place Anadendrum is obviated. Although it lacks trichosclereids, Anadendrum appears more closely related to Monstereae than to Potheae. For example, it exhibits intermittent diphyllous sympodial growth, which is otherwise known only from Monstera (Ray, 1988). Also, Anadendrum shows a tendency toward pinnatifid and even fenestrate leaves, uncommon in Araceae but characteristic of the former tribe. French & Tomlinson (1981b) reported that Anadendrum is much more similar in stem anatomy to certain Monstereae (especially Amydrium, the only mem- ber of the tribe without trichosclereids) than to Potheae. Five additional anatomical similarities of Anadendrum to Monstereae are provided in Nicol- son (1984b, appendix 2). As in all Monstereae, the spathe in Anadendrum abscises cleanly following anthesis. The peculiar inaperturate, spinulose-pi- late or spinulose-baculate pollen is out of place in either tribe. Anadendrum is here provisionally regarded as a sister taxon to Monstereae/ Zamioculcadeae. This Herera psis. small, чн не genus growth and absence of tricho- sclereids, typical of Engler’s subfamily Pothoideae, with inflorescence and floral features of Monster- oideae. The two subfamilies are combined in the present treatment (see under Potheae). he strong floral-morphological resemblances between Heteropsis and Monstereae (Type II spathes that abscise immediately after anthesis; naked, bisexual, four-merous flowers; long, flat sta- mens) are augmented by the medium-sized, foveo- late, starch-bearing, binucleate, zonate to diaper- turate pollen common to the two taxa (see Grayum, in press a). Heteropsis is also more similar to Monstereae in floral anatomy (Carvell, 1989). Thus I agree with all members of the Schottian school in aligning Heteropsis with Monstereae, in which it is here included as a subtribe. Stem anatomy of Heteropsis is said to be more similar to that of Potheae (French & Tomlinson, 1981c), but this may reflect the similar and probably convergent growth habits of these taxa. Monstereae. The eight (nine, if Heteropsis is included) genera of this pantropical tribe are strongly united vegetatively, florally, and palynologically. With a few exceptions (Heteropsis, Amydrium), Monstereae have been maintained intact by most prominent authors. The Monstereae are here tentatively considered most closely related to the highly derived African tribe Zamioculcadeae. They are also clearly related to elements of Engler’s subfamily Pothoideae, es- pecially Heteropsis (here included within Mon- stereae), Anadendrum, and Potheae. The Spathi- phylleae, treated by Engler as the sister group of Monstereae, are here considered more distantly related. The Englerian subfamilies Pothoideae and Mon- steroideae have been merged in the present paper because Engler’s characterizations of the two taxa (Pothoideae with reticulate leaf venation and lack- ing trichosclereids, Monsteroideae with parallel ve- nation and trichosclereids) do not circumscribe nat- ural groups (see discussions under Potheae, Anadendrum, Spathiphylleae, and Heteropsis). Relationships among genera within the Mon- stereae are not completely resolved. Rhaphido- phora, Epipremnum, and Monstera are so poorly differentiated that their integrity as genera has been questioned (see, e.g., Bakhuizen van den Brink, 958). Similarly, Scindapsus and Alloschemone have sometimes been united (Madison, 1976). Rhodospatha and Stenospermation share several unusual features and are probably sister genera (see Grayum, 1984) 672 Annals of the Missouri Botanical Garden Orontieae. Each of the three genera (Lysi- rican. Because of its reticulate leaf venation and chiton, Symplocarpus, Orontium) of this tribe (formerly known as Symplocarpeae) were also con- sidered individually in the present analysis. Phe- netically, the group is quite primitive and shares few derived characters with other taxa. Features appearing to unite the Orontieae are homeophyl- lous sympodial growth, uniovulate locules, absence of endosperm, anomocytic stomata (occurring else- where only in Pistia), a base chromosome number of x = 13 or 14, a north temperate distribution, and a characteristic and unusual flavonoid profile (Williams et al., 1981). Thus the group appears to be natural. Nonetheless, there appears to be a rather sub- stantial rift in the Orontieae, which has widened ominously as a result of Ray's (1988) recent stud- ies. Orontium differs markedly from the other two genera by having tetraphyllous growth, distichous phyllotaxy, laticifers, paired inflorescences, ab- sence of a spathe, six-merous flowers, basal pla- centation, and a base chromosome number of x = 13. It is here accorded separate tribal status. The occurrence of laticifers in Orontium, which (along with their absence in Lysichiton and Sym- plocarpus) was verified by Rosendahl (1911) and French (1988), is especially perplexing and may reflect an independent derivation of this character state. Orontium is unknown embryologically and deserves highest priority (along with Gymnosta- chys) in this respect. The relationship of the tribe Orontieae sensu lato to other taxa in Araceae is difficult to pin down, largely as a consequence of the dearth of advanced characters. The Lasieae are similar in most morphological d and likewise have a base E ARRA number of x = 13 or 14. Lindley (1847), Schott (1860), Hooker (1883), and Hutch- inson (1973) i classified Orontium and/or Sym- plocarpus among lasioid elements. Stylochaeton, though with many advanced floral character states, has a number of primitive features (absence of laticifers; perigoniate flowers; elongate stamens; a base chromosome number of x = 14; and retic- ulate, boat-shaped, monosulcoidate pollen) and has a flavonoid profile remarkably similar to that of the Orontieae. These taxa are treated here as con- stituting subfamily Lasioideae. Lindley (1847) placed the Af- rican genus Stylochaeton next to Cryptocoryne Stylochaeton. due to wholly superficial resemblances in spathe and spadix structure. Hutchinson included it with Zamioculcas and the Arophyteae for no conceiv- able reason other than that all three taxa are Af- unisexual flowers, Engler (1920a) treated Stylo- chaeton as a monotypic tribe of the Aroideae. All of these authors overlooked the fact that Stylochaeton is an isolated and phenetically prim- itive genus. Features primitive within Araceae in- clude a rhizomatous habit; cordate/sagittate leaves; absence of laticifers (French, 1988); perigoniate flowers; plurilocular ovaries with multiovular axile placentation; elongate stamens with longitudinal anther dehiscence; anatropous ovules; presence of endosperm; a base chromosome number of x = 14; and boat-shaped, reticulate, binucleate and (at least in sect. Spirogyne) monosulcoidate pollen. Most of the rather few advanced character states of Stylochaeton appear to represent autapomor- phies and hence are not useful for classification. The Type IV spathe is constricted above the spadix rather than lower down as is typical for Araceae; monoecy is attended by such unusual features as perigoniate flowers and elongate stamens and pis- tillodes (in the male flowers), and has likewise cer- tainly evolved independently. Starchy pollen is probably also independently derived in Stylochae- ton, since all other genera with reticulate, boat- shaped pollen (except Lysichiton) have starchless pollen. Styloc haeton cannot be included in any of the more “advanced” aroid subfamilies (Philodendroi- deae, Colocasioideae, or Aroideae), since it pos- sesses none of the important synapomorphies char- acterizing these groups. Here it is placed, x with Lasieae and Orontieae, in the subfamily sioideae, as a monotypic tribe in which monoecy is considered to have evolved independently from the rest of the family (just as Pycnospatha, of the Lasieae, has independently acquired naked flow- ers). Stylochaetonieae have been treated as the sister group of Symplocarpeae, which also lack laticifers and have a strikingly similar flavonoid profile (Williams et al., fer in their north temperate distribution, parallel leaf venation (according to Ertl, 1932), and seeds lacking endosperm. 1981). Symplocarpeae dif- Lasieae. The composition of Engler’s subfam- ily Lasioideae has emerged as one of the major shortcomings of his system. The two major tribes, the pantropical Lasieae (comprising eight to ten genera) and the paleotropical Thomsonieae, have apparently been grouped together solely on the basis of superficial vegetative similarities among some of their members. These tribes are actually separated by as profound a gulf as any two taxa in Araceae, as is well evidenced by Howarth’s (1957) Volume 77, Number 4 1990 Grayum 673 y Evolution and Phylogeny of Araceae advancement indices: 30% and 29% for Hutch- inson's tribes Lasieae and Dracontieae, respectively (the latter including Symplocarpus), and 84% for his tribe Pythonieae (including Zomicarpa, Zomi- carpella, and Xenophya). Apart from general resemblances in habitat, about the only features shared by Lasieae and Thomsonieae are a base chromosome number of х = 13 or 14 (here considered plesiomorphic) and seeds lacking endosperm. The phenetically primi- tive Lasieae are basically rhizomatous, lack latic- ifers (French, 1988), and have bisexual, perigoni- ate (except Pycnospatha) flowers and monosulcate, reticulate, generally small, starchless, binucleate pollen grains; the exclusively tuberous Thomsoni- eae are laticiferous, have unisexual, naked flowers and inaperturate, basically striate, medium-sized to large, pta da ишш pollen grains. All major aroid p er than those of the Englerian ¿hdi E raid 1847; Schott, 1860; Hooker, 1883; Hutchinson, 1973) have separated the two tribes widely, typically associating the Lasieae with homsonieae with subtribe the Orontieae and the Arinae or with tribe Zomicarpeae. These are sub- stantially the same affinities indicated by the pres- ent analysis. The neotropical genus Dracontium and the Southeast Asian Pycnospatha seem more ad- vanced than the rest of the Lasieae in their tuberous habit, deeply compound leaves (of the same general type seen in Amorphophallus and Anchomanes), more advanced (Type III) spathes, and rather large pollen; Pycnospatha has completely lost the per- igonium. These genera bear the closest superficial resemblance to the Thomsonieae, but clearly this is a case of convergent evolution, every bit as obvious and superficial as certain well known and long-accepted examples from the animal kingdom (e.g., New World and Old World porcupines, kan- garoo rats, and jerboas). Calla. Itis not clear why Engler grouped Cal- la and the Orontieae in the same subfamily. The only characters of any consequence that the two taxa have in common are unilocular ovaries and north temperate distributions. Apart from these, none of the character states appearing to unite the three genera of Orontieae sensu lato are known from Calla, which differs strikingly in having an- isophyllous growth, collateral bundles, naked flow- ers, multiovulate ovaries, numerous stamens per flower, endospermous seeds, a base chromosome number of x = 18, a markedly different flavonoid profile (Williams et al., 1981) and small, diaper- turate pollen. Lindley (1847), Schott (1860), Hooker (1883), and Hutchinson (1973) all grouped Calla with the Monstereae. The two taxa have a number of fea- tures in common: distichous leaves, parallel leaf venation, naked, bisexual flowers (an unusual com- bination in Araceae, otherwise known only in Рус- nospatha of the Lasieae), and diaperturate pollen (unique to these two taxa). However, Calla differs from Monstereae in lacking a geniculum and tricho- sclereids and, most problematically, in possessing laticifers (as verified by Rosendahl, 1911, and French, 1988). Calla shares the greatest number of derived features with /Vephthytis and Callopsis (here in- cluded in the Philodendroideae): laticifers, Type II spathes, unilocular ovules with basal placentation, (with Cal- lopsis, which was so named well before its chro- mosomes had been counted), and absence of a perigonium. The combination of laticifers and a base number of x = 18 is characteristic of sub- family Philodendroideae as here construed, to which Calla is thus provisionally assigned (as the cla- base chromosome number of x = 18 distically most primitive member in light of its distichous leaves, bisexual flowers, and aperturate pollen). Hotta (1970) also included Calla in Phil- odendroideae. An unfortunate nomenclatural consequence of including Calla in Philodendroideae (or any other subfamily except Aroideae) is the necessity of re- placing the latter name with the earlier Calloideae (see Nicolson, 1984b). Clearly the classification of taxa should not be guided or influenced by such nomenclatural considerations; however, I have continued to use the name Philodendroideae throughout the narrative portion of this paper and in the tables (except Table 5) for conceptual rea- sons. Culcasia. Engler placed this African genus in subfamily Pothoideae even though it is highly dis- cordant there due to its advanced inflorescence and floral morphology: Type III (enclosing) spathe, Type II (monoecious) spadix, naked flowers, a sin- gle ovule per locule, prismatic stamens, and anthers dehiscing by apical pores. Engler was swayed by certain vegetative features in common with Po- thoideae: distichous leaves, geniculate petioles and, especially, absence of laticifers (confirmed by French, 1988 All authors af the Schottian school, on the other hand, have placed Culcasia next to Cercestis of Engler’s subfamily Lasioideae (Schott, 1860; Hooker, 1883; Hutchinson, 1973). Like Culcasia, Cercestis is an African genus consisting largely of 674 Annals м A m Garden climbers; the genera also share a similar bud trace organization (French & Tomlinson, 1981b), inap- erturate starch-bearing pollen (Grayum, in press a), and the unusual base chromosome number of x = 21. The idea of a close relationship between Cul- casia and Cercestis now enjoys a wide consensus among aroid workers, even including those of the neo-Englerian school (e.g., Bogner & Nicolson, in press). But although the latter authors would retain both genera in the subfamily Lasioideae, there is now overwhelming evidence that Culcasia and Cer- cestis are intimately related with Philodendron and thus must be transferred to the Philodendroideae. This evidence, generated and assembled by J. C. French, is discussed under the next heading. Dis- tichous leaves and geniculate petioles are both al- ready known in Philodendroideae (from Heteroar- idarum and Anubias, respectively), however the lack of laticifers in Culcasia is highly bothersome. According to Carvell (1989) the floral anatomy of Culcasia is unique, even within the context of Philodendroideae. Homalomeninae, Schismatoglottidinae, Aglao- nemateae, Philodendron, Zantedeschia, Anubias, Dieffenbachia. of the Philodendroideae inasmuch as they have These taxa represent the “core” been aligned together in this subfamily by most important authors and are so treated here. Al- though each was considered separately for the pres- ent analysis, their interrelationships are difficult to resolve and it is most convenient to discuss them together. The seven taxa listed in the above heading are all fundamentally rhizomatous (except Zantedes- chia?) and have spiral phyllotaxy; petioles lacking venation; a ring of collenchyma in the outer cortex of the stem or petiole (except apparently Anubias and the Aglaonemateae); secretion files; biforines; advanced (Type III or IV) spathes; monoecious (Type II or III) spadices; naked flowers; plurilocular ovaries (except Schismatoglottidinae); numerous ovules per locule (except Dieffenbachia and Aglao- nemateae); axile placentation (basically parietal in Aglaonemateae and Schismatoglottidinae); gener- ally 2-4 stout stamens per male flower; anther dehiscence by apical pores (except Anubias and the Homalomeninae); anatropous ovules; endo- sperm (except Dieffenbachia and the Aglaone- mateae); staminodia on the female flowers (except Anubias and Philodendron); and psilate, starchy pollen (foveolate in some Dieffenbachia species, starchless in Anubias). Most of the exceptions in the above list can be considered autapomorphies within the group in question. Other presumed autapomorphies include homeophyllous growth (in many Philodendron species), compound vascular bundles (in Dieffen- bachia), inflorescence aympodia un Mielen hacha and many Philod Dief- fenbachia and Anubia s) aod trinucleate pollen (in Dieffenbachia, Zantedeschia, and the Aglaone- mateae). Philodendron, Homalomeninae, and Schisma- toglottidinae have in common the possession of intercellular spaces and ducts in the ground tissues. This feature probably influenced Engler's decision to group these taxa into a single tribe, the Philo- dendreae (here disbanded); however, it also occurs in Culcasia and the Nephthytideae. Another char- acter shared with the last-mentioned tribe is the possession of megaspore dyads, known only from Dieffenbachia, and Nephthytis (Grayum, in press b). Cytologically, the * quite variable but show a conspicuous absence of Homalomena, “core” Philodendroideae are multiples of seven, which are so common in the rest of Araceae and considered basic to the family. Schismatoglottidinae is based on х = 12 or 13; Philodendron, Dieffenbachia, and Zantedeschia on x = 16 or 17; Aglaonemateae and Homalo- meninae on x = 20; and Anubias on x = 24 (see Grayum, 1984). Although all of the **core" Philodendroideae are retained in the subfamily in the present treatment, the traditional Englerian subfamilial classification has been abandoned. Instead, the group has been broken into five informal alliances," each of which includes an admixture of taxa formerly assigned to other subfamilies. These have not been given formal taxonomic recognition since their “alliances” composition is mostly tentative and highly subjec- tive (the number of characters used in the analysis having been insufficient to resolve the group fully). I believe, however, that subfamily Philodendroideae as here circumscribed comes much closer to being a natural group than either of Engler’s subfamilies Lasioideae and Aroideae, from which most of the new taxa were transferred. The compositions of the five “alliances”” as here presented are considerably different from those originally conceived (see Grayum, 1984). Exten- sive changes have been necessitated by an out- pouring of new data from the laboratory of J. С. French, currently the most prolific researcher on Volume 77, Number 4 1990 Grayum 675 Evolution and Phylogeny of Araceae Araceae. In the case of Philodendroideae, these data (organized in French, 1987b, and Mayo, 1986) have now established the close relationship among Philodendron, Homalomeninae, and the African genera Culcasia and Cercestis (formerly of En- gler’s Pothoideae and Lasioideae, respectively). These taxa are unique in possessing resin canals in the roots (French, 1987b) and a sclerotic root hypodermis (French, 1987a). They share similar stem and stamen vasculature (French, 1986b) and extrafloral nectaries (J. С. French, pers. comm.), and all but Homalomeninae lack an anther endo- thecium (French, 1986a; all other Araceae have an endothecium). These four taxa here compose the “Philodendron alliance” (Homalomeninae being raised to tribal rank). Cercestis has been dissociated from Nephthytis, which lacks most of the above character states, and assigned to its own tribe, Cercestideae. Explanations for most of the other four “аШ- ances” e found under other headings: the monotypic “Calla alliance” bin under Сайа); the “Nephthytis alliance,”” comprising former mem- bers of Engler’s subfamilies Lasioideae (Vephthy- tis, Anchomanes and Pseudohydrosme, Montri- chardia) and Aroideae (Callopsis, most genera of Engler’s Zomicarpeae); and the “Peltandra alli- ” including Peltandra and Typhonodorum, can ance, Schismatoglottidinae (here raised to tribal status), and the tribe Arophyteae from Engler's Aroideae (see under those headings). The “Aglaonema alliance” includes the Aglao- nemateae, Anubias, Zantedeschia, Dieffenbach- ia, Bognera, and the tribe Spathicarpeae from Engler’s Aroideae. This is an especially tentative, difficultly defined group consisting mainly of those Philodendroideae that do not seem to be- the inclusion of “core” long in any other “alliance”; Spathicarpeae here is highly speculative. A more extensive character analysis will be Pj paei for Wiese taxa to be mds more confidently. ie y Philodendroideae is treated in this paper as a geler group to subfamily Po- thoideae (including Engler’s Monsteroideae). The combinations of character states found in such genera as Calla and Culcasia (see under those headings) strongly suggest that this is a proper alignment. Some important characters exclusive to members of Pothoideae and Philodendroideae are geniculate petioles, cork formation in the aerial roots, compound vascular bundles, collateral bun- dles, and prisms. (For more, see Grayum, 1984.) Peltandra, Typhonodorum. The striking sim- ilarities between these genera were pointed out by Buchet (1939) but have been largely downplayed by later authors. The two genera, which were con- sidered separately in the present analyses, are vir- tually identical in every important respect, sharing more derived characters than any other pair of taxa compared in the present study: sagittate leaves; stem endodermis (of the subtype individually sur- rounding the vascular bundles); secretion files; Type IV (constricting) spathe; monoecious spadix with sterile male flowers above and below the fertile male region; no perianth; unilocular, uniovulate or pauciovulate ovaries; parietal placentation; stout, anther dehiscence; ortho- tropous ovules; absence of endosperm; apparen connate stamens; apical base chromosome number of x = 50; staminodia in the female flowers; and starchy, trinucleate pol- len. Based on the above evidence, the monotypic, Madagascan genus Typhonodorum and the dityp- ic, North American Peltandra are here combined in the tribe Peltandreae. One might even argue that they be merged in the same genus; the only notable difference between the two genera is the fusion of the staminodia into synandrodia in Pel- tandra, and both fused and distinct staminodia are tolerated in the genus Asterostigma. Lindley (1847) and Schott (1860) placed Pel- tandra and Typhonodorum among colocasioid ele- ments on the basis of their similar leaf venation and connate stamens. Although the Peltandreae lack the secretion tubes characteristic of Coloca- sioideae, there are some similarities. The base chro- mosome number of Peltandreae (x = 56) is a multiple of seven, as in Colocasioideae (with x = 14) but unlike all “core” Philodendroideae. The colocasioid genus Hapaline especially resembles the Peltandreae in having sagittate leaves and in its gynoecial morphology, trinucleate pollen, and distribution of sterile male flowers. Yet all of these features may have been independently derived. The Peltandreae are here considered closest to tribe Arophyteae and subtribe Schismatoglottidi- nae. The latter group shares with Peltandreae the unusual characteristic of a stem endodermis sur- rounding each vascular bundle individually and tends to have a similar distribution of sterile male flowers. Schismatoglottidinae differ from Peltan- dreae in having distinct stamens, endospermous seeds, binucleate pollen, and a base chromosome number of x = 12 or 13 The unusual geographic disjunction of the two genera in Peltandreae may be explained according to the following scenario: the group must have 676 Annals of the Missouri Botanical Garden originated in tropical West Africa subsequent to its separation from South America, with Peltandra reaching North America via Eurasia and the North Atlantic route prior to the Eocene. Peltandreae were then presumably extirpated in their homeland as a result of increasing aridity, with Typhono- dorum (having viviparous germination) reaching Madagascar. A consensus has now been reached among aroid workers that this very poorly known, Bognera. monotypic South American genus, originally placed in the tribe Zomicarpeae (Madison, 1980), belongs among elements here included in the “Aglaonema alliance” of Philodendroideae (see Mayo, 1986; Bogner & Nicolson, in press). Character analysis m indicates а close phenetic resemblance with the Madagascan tribe Arophyteae (of the “Peltandra alliance"); however, this is unconvincing since very few character states are known for Bognera. The genus is likewise similar to the polytypic South American genus Dieffenbachia (sharing a fused spathe and spadix, sterile male flowers between the fertile male and female regions, connate stamens, apical anther dehiscence, and large, basically psi- late pollen). I feel that this is а much more likely relative, however Mayo (1986) implicates Anu- bias. This tribe consists of four mono- typic or oligotypic South American genera. Only Zomicarpeae. Zomicarpa is even moderately well known phe- netically. Ulearum and Filarum, which were analyzed as a pair apart from Zomicarpa, are unknown cytologically and anatomically. Zomi- carpella is virtually unknown in all respects, and was not included in the present analysis. The Zomicarpeae were treated by both Schott (1860) and Engler (1920b) and his followers as belonging with elements of Engler’s Aroideae—a standard repository in the latter system for taxa with unisexual flowers and reticulate leaf venation. In the present work they are included partly in the “Nephthytis alliance” of Philodendroideae, and partly (Zomicarpa) in the Colocasioid Filarum and Ulearum strongly piam the African genus Callopsis and, less so, Nephthytis. With Callopsis they share a rhizomatous habit, cordate leaves with reticulate venation, a simple spathe fused basally with the spadix, rather few female flowers, unilocular and uniovulate ovaries, basal placentation, anatropous ovules, 2—3 distinct stamens, and spinose pollen. It is, of course, nec- essary to hypothesize that the present ranges of these taxa represent the remnants of a once con- tinuous West Gondwanalandic distribution. Zomicarpa also shows significant similarities with Callopsis and Nephthytis, differing from them, as also from Filarum and Ulearum, in its subtuberous habit, compound leaves, Type IV (constricted) spathe, plurilocular ovaries, and transverse anther dehiscence. Whereas Filarum and Ulearum are obviously closely related to one another, Zomi- carpa stands apart. Їп addition to the above char- acters, it differs in its endospermous seeds and Atlantic coastal (Bahian) distribution (Filarum, Ulearum, and Zomicarpella occur in western Amazonia or in the Andean foothills). Most dis- turbing, however, is the recent discovery (French, 1988) that Zomicarpa has secretion tubes, here considered a unique synapomorphy for the Colo- casioideae. I have been prompted by this to transfer Zomicarpa into the latter subfamily as a monotypic tribe, even though another synapomorphy (syn- andria) is thereby lost. Secretion tubes have not been sought in the remaining three genera of En- gler's Zomicarpeae, but | predict they will be ab- sent. The latter genera are serted in the Callopsideae (Philodendroideae). here provisionally in- Arophyteae. This is a well-defined and wholly Madagascan tribe of three genera, none of which were known to Engler or earlier authors. Hutch- inson (1973) classed them along with Zamioculcas in his heterogeneous tribe “Stylochitonieae,”” which he effectively employed as a trash-basket category for miscellaneous African taxa of dubious affinity. Bogner (1972, 1975, 1978) has treated the Ar- ophyteae as a tribe within subfamily Aroideae. The Arophyteae are florally, cytologically, and palynologically one of the most highly advanced of all aroid taxa. They are a bit more primitive veg- etatively in having a rhizomatous habit and basi- cally cordate leaves. The strongest resemblances of this tribe to the Aroideae are its orthotropous ovules and inaperturate, globose, spinose, starchy, trinucleate pollen. Many other characters of Ar- ophyteae are discordant in this subfamily: seeds lacking endosperm, connate stamens, staminodia in the female flowers, fusion of the spathe and spadix, base chromosome number of x — 18, and tropical African distribution. With the exception of connate stamens, these features are also rare or unknown in Colocasioideae. Arophyteae share the greatest number of de- rived characters with Callopsis and Nephthytis, on one hand, and Peltandra and Typhonodorum, on the other. All of these are here placed in subfam- Volume 77, Number 4 1990 Grayum 677 Evolution and Phylogeny of Araceae ily Philodendroideae. With the East African Cal- lopsis, the Arophyteae share reticulate leaf ve- nation, basal fusion of the spathe and spadix, basal placentation, spinose pollen, and the base chro- mosome number x = 18. They differ from Cal- lopsis in having a Type III or IV spathe, connate stamens, staminodia in the female flowers, ortho- tropous ovules, seeds lacking endosperm, and trinu- cleate pollen. Except for staminodia, all of the latter may be regarded as advanced character states; whereas only lack of endosperm is shared with Nephthytis, all are shared with both Typhono- dorum and Peltandra. Furthermore, the stami- nodia in Peltandra are fused into synandrodia, just as in the Arophyteae, which is a very rare feature otherwise known only in Asterostigma sect. Rho- palostigma. Taken in conjunction with the Madagascan dis- tribution of Typhonodorum, the above evidence strongly suggests that the Arophyteae have a sister group relationship with the tribe Peltandreae. The only significant difference between the two groups is their base chromosome numbers (x = 18 and 56, respectively). Since Africa and Madagascar have apparently been separated since at least the mid-Cretaceous (Raven, 1979), the Arophyteae probably arrived at their present location via long-distance avian dispersal. Spathicarpeae. These eight South American genera appear to form a cohesive and natural as- semblage. Spathantheum and Spathicarpa stand somewhat apart in their total or partial fusion of spathe and spadix and dorsiventral segregation of male and female flowers; Spathicarpa differs fur- ther in having unilocular, uniovulate ovaries, basal placentation, and trinucleate pollen. Schott’s (1860) and Hutchinson's (1973) separation of these gen- era into their own tribe does not appear warranted. Spathantheum effectively bridges the gap between Spathicarpa and the remaining six genera, the latter of which have many of their own autapo- morphic features (Grayum, 1984 Lindley (1847) and all Schatten authors have tended to group the Spathicarpeae (formerly known as Asterostigmateae) with philodendroid genera, particularly near Dieffenbachia. Hotta (1970) also followed this course. Engler (1920a), on the other hand, included the Spathicarpeae in his subfamily Aroideae, presumably on the basis of their retic- ulate leaf venation. According to the character analysis performed here, the Spathicarpeae show a definite correspondence to members of the Phil- odendroideae, especially Dieffenbachia, and no compelling link to Aroideae. Mayo (1986) suggests a connection with the Arophyteae, also here in- cluded in Philodendroideae. Some of the more prominent putative synapo- morphies of the Spathicarpeae with members of the Philodendroideae are secretion files (French, 1988); a collenchyma ring in the outer cortex of the stem or petiole (exclusive to these two taxa); a base chromosome number of x = 17 (as in Dieffenbachia and Philodendron; cf. also Zami- oculcadeae); and inaperturate, boat-shaped, ver- rucate to psilate, starchy, binucleate pollen vir- tually indistinguishable from that of Philodendron With Dieffenbachia, the Spathicarpeae also share Alocasia type hydathodes (Múller, 1919), basal fusion of the spathe and spadix, connate stamens, and distinct staminodia and Homalomena. in the female flowers (the latter two characters in combination being highly unusual in Araceae). The compound leaves of some genera of Spathicarpeae are very similar to those of certain Philodendron species, e.g., in sect. Schizophyllum. The Spathicarpeae have here tentatively been included next to Dieffenbachia in the “Aglaonema alliance" of subfamily Philodendroideae. But al. though the tribe exhibits many specializations (tu- berous growth habit, spiral phyllotaxy, compound leaves with reticulate venation, monoecy, fusion of the spathe and spadix, and connate stamens), it is in many respects phenetically more primitive than the other members of this assemblage. The spathe is generally of a rather simple type; even in Spathi- carpa, it is flat, narrow and greenish. The spadix is basically of the most primitive monoecious type, i.e., with bisexual flowers still present between the male and female regions. The ovary is basically plurilocular, with axile placentation, and the sta- mens, though usually more or less connate, are mostly elongate and with longitudinally dehiscent anthers. Ovules are anatropous, and the seeds re- tain endosperm. The pollen of Gorgonidium var- gasii is subreticulate and appears monosulcoidate (see Grayum, in press a). It is thus likely that monoecy has evolved independently and relatively recently in Spathicarpeae, as in Zamioculcadeae and Stylochaeton. The distribution of the Spathicarpeae is also “primitive.” The eight genera are restricted to the fringe of Amazonia from the Bahia coast of Brazil through the subtropical and warm temperate re- gions of northern Argentina, Paraguay, Bolivia, and southern Brazil, and northward along the An- dean chain. This distribution more or less parallels 678 Annals of the Missouri Botanical Garden that of the tribes Lasieae and Zomicarpeae in South America and probably represents the vestige of a once more expansive range that has been reduced by vicariance events. Nephthytis and Cercestis. These two African genera, generally placed together in the tribe Nephthytideae, were considered separately for the present analysis. The Nephthytideae were treated among philodendroid genera by Schott (1860), Hooker (1883), and Hutchinson (1973); however, Engler (1911) included them in his subfamily La- sioideae, apparently because of their reticulate leaf venation. According to the character analysis per- formed for this study, the tribe clearly belongs in Philodendroideae (see under Homalomeninae, etc.). Character states of Nephthytideae considered synapomorphic with members of the Philodendroi- deae include monoecy, naked flowers, unilocular uniovulate ovaries with basal placentation, pris- matic stamens with poricidal anther dehiscence, seeds lacking endosperm, a base chromosome num- ber of x = 20 or 21, and inaperturate, boat-shaped, verrucate to psilate, starchy, binucleate pollen. These similarities are shared to the greatest extent with Montrichardia, Callopsis, Anchomanes and Pseudohydrosme, the Zomicarpeae, Culcasia, the Aglaonemateae, Peltandra, and Typhonodorum. The first five taxa in this list have been included with Nephthytis in the “Nephthytis alliance” of Philodendroideae (see under the appropriate head- ings for further details). Cercestis comprises mainly scandent species based on x = 21. Although species of Nephthytis are rhizomatous and based on x = 20, the original character analysis did not reveal any major dis- continuity between the two genera. Recently gen- erated anatomical data, however, strongly suggest that Cercestis is much more closely related to Culcasia (also consisting largely of scandent species based on x = 21) than to Nephthytis, and that the former two genera are more properly associated with Philodendron and the Homalomeninae (see under Homalomeninae, etc.). Thus Cercestis has hereby been removed from the Nephthytideae to its own tribe, Cercestideae, which, along with the monotypic Culcasieae, is transferred to the “Phil- The Nephthytideae have sub- sequently been enlarged to accommodate Ancho- manes and Pseudohydrosme (see next heading). Anchomanes and Pseudohydrosme. Schott (1860), Hooker (1883), and Engler (1911) agreed in placing these two African genera among lasioid elements in or near the tribe Thomsonieae. Hutch- odendron alliance.” inson (1973), however, classed them in his tribe Richardieae, which consisted largely of philoden- droid elements. Bogner & Nicolson (in press) es- sentially agreed with this alignment, including An- chomanes and Pseudohydrosme in the Nephthytideae (which, however, they retained in Lasioideae). Character analysis supports the latter position. The resemblances of these two genera to the Thomsonieae are striking but superficial: tuberous stems; few or solitary large, compound leaves; and reticulate venation. This type of growth habit is common in Araceae and has evolved independently in several more or less unrelated lines, e.g., Za- Spathicarpeae, Arinae, Lasieae, Zomicarpeae, and Colocasioideae (Chlorospatha). The compound leaves of Anchomanes and Pseu- dohydrosme differ in detail from those of Thom- sonieae and appear to develop via the necrotic mioculcadeae, process as in Monstereae rather than the marginal process typical of Thomsonieae (see Riedl, 1978). nchomanes and Pseudohydrosme share a few apomorphic floral characters with Thomsonieae (naked flowers, uniovulate locules, seeds lacking endosperm), but these are just as characteristic of, say, Aglaonemateae. The following features of these genera appear to place them squarely in the Phil- odendroideae: Type II spadix, absence of sterile flowers (typical of Montrichardia, Nephthytis, etc.; unknown in манын, a base chromosome number of x = 20 (common in cerca Thomsonieae are dal on x — 14); psilate, binucleate pollen (characteristic a Philodendroi. deae; pollen of Thomsonieae is basically striate and uniformly trinucleate); and a rare anthocyanin de- tected in Anchomanes and otherwise known only from Cercestis (Williams et al., 1981). Nicolson (in press) in placing Anchomanes and Pseudohydrosme di- rectly into the а (I had previously segregated them in their own tribe; Grayum, 198 I now follow Bogner $ hese taxa apparently have the same base shew mosome number, x = 20 (Pseudohydrosme is yet unknown cytologically), and differ principally in leaf architecture. However, sagittate and highly compound leaves occur together in several other tribes and even subtribes (Spathicarpeae, Dracon- tiinae, Arinae, and others). Anchomanes has acu- leate petioles (unknown in Thomsonieae), a con- dition which is also occasionally seen in Nephthytis afzelii (J. Bogner, pers. comm.). Although Pseudohydrosme has two species, only Р. gabunensis was included in the analysis. The second species, P. buettneri, is still unknown in Volume 77, Number 4 1990 Grayum 679 Evolution and Phylogeny of Araceae most respects. Florally it diverges markedly from P. gabunensis in having very different stamens and sterile male flowers at the apex of the spadix. It may be that this small genus is artificial. Montrichardia. The tritypic neotropical ge- nus Montrichardia shows strong similarities with genera here considered philodendroid and few sim- ilarities with the mainstream of Engler’s Lasioideae (Lasieae and Thomsonieae). The closest relatives of the genus appear to be the African genera Nephthytis and Anchomanes/ Pseudohydrosme, as recognized by most previous authors. Montri- chardia is placed along with those taxa in the “Nephthytis alliance” of Philodendroideae in the present work. The general resemblances among taxa of the “Nephthytis alliance” are listed under Nephthytis and Cercestis. The prickly trunk of Montrichardia recalls the aculeate petioles of Anchomanes, and its large, psilate, thin-walled, binucleate pollen is very similar to pollen of Anchomanes and Pseu- dohydrosme. The common ancestor of these taxa was presumably widespread across West Gon- dwanaland before that continent was bisected. Montrichardia is still unknown chromosomally, but I predict it to have Zn = 36, 40, or 42 chromosomes. Callopsis. Engler treated this monotypic East African genus as a separate tribe in his subfamily Aroideae; Hutchinson grouped it with his tribe Col- ocasieae. Character analysis suggests that Callop- sis is most closely related to taxa here composing the “Nephthytis alliance” of subfamily Philoden- droideae, particularly to Nephthytis (see under Nephthytis and Cercestis for shared characteris- tics of this group). Noteworthy points of agreement with the neotropical genera Filarum and Ulearum, also of this and here included in the Callopsideae, are discussed under Zomicarpeae. Callopsis also has a few features in common with the Madagascan tribe Arophyteae of the “Рейап- dra alliance.” “alliance” Colocasioideae. The following subtaxa of Col- ocasioideae were analyzed separately: Jasarum, Scaphispatha, Chlorospatha, Caladium/ Xan- thosoma/ Aphyllarum, Caladiinae, Syngonium, Steudnera, Remusatia/ Gonatanthus, Hapaline, Colocasia, and Alocasia. The subfamily is here discussed as a unit for purposes of convenience. The treatment of subfamily Colocasioideae looms as the most serious unresolved problem in the clas- sification presented in this paper. On the one hand, this subfamily (it is accepted here virtually as cir- cumscribed by Engler) is probably the most easily recognized of all aroid subfamilies. Moreover, the similarities among the various subtaxa are not su- perficial; the Colocasioideae are unified by an un- usual number of apomorphic character states: most have a tuberous growth habit, basically sagittate to cordate leaves with a peculiar type of venation, secretion tubes (unique to this subfamily), Type IV (constricted) spathes, sterile flowers below the fertile male portion, naked monoecious spadices with flowers, connate stamens (but see Zomicarpa) with poricidal anther dehiscence, a base chromosome number of x = 13 or 14, and inaperturate, basically starchy pollen. Peltate leaves are widespread in Colocasioideae and almost unique here (occurring otherwise only in Ariopsis, Homalomena, and a single species each of Anthurium, Pinellia, and Philodendron). п the other hand, there are some disturbing indications that the Colocasioideae may be diphy- letic. And diphyletic or not, it seems possible that Colocasioideae may be derived from within some other subfamily or subfamilies, which would thus render the latter group(s) paraphyletic and argue strongly that Colocasioideae be submerged. Subfamily Colocasioideae is provisionally kept together in this paper (except for Ariopsis). The following observations are pertinent to classification within the subfamily. Although not reflected in other classifications, there appears to be a deep and fundamental division in the subfamily roughly along geographic lines. The New World Coloca- sioideae differ in several important respects from the Old World members of the subfamily. Pedately compound leaves, for example, occur exclusively in New World Colocasioideae (but see also Pro- tarum), where they are widespread; a cortical vas- cular system is found in most New World Colo- casioideae but never in Old World members (French & Tomlinson, 1983); New World Colocasioideae differ markedly from all Old World genera except Hapaline in the color and detailed ultrastructure of their latex (Frenc “ox, manuscript), and preliminary data (Fox & French, 1988) suggest a chemical difference in the latex as well (although critical genera such as Hapaline, Protarum, and Scaphispatha were not studied); and there is some indication that cyanogenesis is limited to the Old World Colocasioideae. Pollen morphology is also very different in the two groups. Old World Colocasioideae most fre- quently have spinose pollen, but it is clear that the most primitive exine sculpturing in this group is 680 Annals of the Missouri Botanical Garden striate (as in Steudnera, Protarum, and some species of Colocasia; see Grayum, in press a). Striate pollen is also the primitive type in Thom- sonieae, and it occurs in other members of Aroi- deae. New World Colocasioideae most frequently have psilate pollen, consistent with pollination by dynastine scarab beetles, which is the rule among these taxa; only a few species of Syngonium (which may be bee pollinated) are known to have spinose pollen (see also Zomicarpa). The most primitive exine sculpturing in the New World taxa is retic- ulate (as in Jasarum and some species of Chlo- rospatha), which is also the case for Philodendro- ideae. Reticulate pollen is considered antecedent to striate pollen in Araceae (Grayum, in press a); thus, if the Colocasioideae are indeed monophy- letic, the striate pollen of Old World members must have evolved independently of striate pollen in oth- er Araceae. It is conceivable that all of the above differences reflect the long isolation of the two segments of the subfamily following the breakup of West Gon- dwanaland; however, the same phenomenon is not evident between New and Old World members of other araceous taxa with presumably the same biogeographic history. I entertain the hypothesis that the New World Colocasioideae are related to or derived from the Philodendroideae, and the Old World Colocasioideae are related to the Aroideae. Dieffenbachia shows parallels with New World Col- ocasioideae, particularly Chlorospatha and Syn- gonium, with which it shares unusual pollen fea- tures (Grayum, 1984, in press a). A separate cortical vascular system occurs in Philodendron, Dieffen- bachia, and New World Colocasioideae. On the other hand, starch grains of the Old World genus Colocasia are of the “Arum type” and do not resemble those of Peltandra or Dief- fenbachia (Reichert, 1913). The genus Protarum is in some respects intermediate between Aroideae (where Engler placed it) and Old World Coloca- sioideae. The present evidence for diphylesis does not seem to outweigh the imposing list of putative syn- apomorphies appearing to unite the Colocasioideae. Moreover, recent work involving restriction frag- ment analysis of chloroplast DNA, while supporting the division of Colocasioideae into two tribes along hemispheric lines and (apparently) the ouster of Ariopsis, suggests that the subfamily is indeed monophyletic (Kessler & French, 1989). Thus I have opted for the conservative approach of main- taining Colocasioideae essentially in the Englerian sense. The only changes in content involve the exclusion of Ariopsis (transferred to Aroideae) and the inclusion of Protarum and Zomicarpa (trans- ferred from Aroideae). However, character analysis strongly indicates sweeping changes in the internal classification of the subfamily. Engler & Krause (1920) divided the Coloca- sioideae into three tribes: the monotypic Syngoni- eae and Ariopsideae, plus an all-inclusive Coloca- sieae. This scheme is both unilluminating and highly artificial. In the present work the bulk of the subfamily is divided into two coordinate tribes, the Colocasieae (consisting entirely of Old World gen- era) and the Caladieae (predominantly of New World genera), to reflect the differences discussed above. Within the Old World tribe Colocasieae, En- gler’s subtribe Steudnerinae is clearly artificial on the basis of floral and pollen characters. Steudnera differs greatly from the other two genera in its small, boat-shaped, striate, binucleate pollen. In this sense it resembles Protarum and some species of Colocasia (Grayum, in press a); however, Steud- nera is so distinctive in its Type II spathe and lack of sterile male flowers that it clearly deserves sep- arate subtribal status. Remusatia and Gonatanthus, on the other hand, are so similar morphologically and palynologically that they probably ought to be combined. These genera lack a sterile apical appendage on the spadix and have trinucleate pollen. I include them in a separate subtribe (Remusatiinae) from Colocasia and Alocasia (Colocasiinae), which have a naked apical appendage and binucleate pollen (with the exception of Colocasia esculenta). The Colocasieae include the highly aberrant ge- nus Protarum. In the New World tribe Caladieae, Jasarum, Scaphispatha, and Syngonium stand out so much vegetatively, florally, cytologically, and/or paly- nologically that they all deserve subtribal status: Jasarum, by virtue of its Guayana Shield distri- bution, rhizomatous, aquatic habit, linear-lanceo- late leaves, absence of a cortical vascular system, unilocular biovulate ovaries, basal placentation, seeds lacking endosperm (Bogner, 1985a), base chromosome number of x = 11, and fully reticulate pollen; Scaphispatha, in its primitive spadix (lack- ing sterile flowers), unilocular ovaries with basal placentation and unique exine sculpturing; and Syngonium in its vining habit, uniovulate locules, connate gynoecia, and absence of endosperm. Scaphispatha is a still poorly known genus that Engler originally included in Aroideae. It is the only genus of Colocasioideae other than the Asian Steudnera to lack sterile flowers on the spadix. Volume 77, Number 4 1990 Grayum Evolution and Phylogeny of Araceae The pollen, though unique in Araceae, is most nearly comparable to certain striate-derived types of Old World species (Colocasia gigantea, Amor- phophallus konjac; see Grayum, in press a). Thus it is conceivable that Scaphispatha will prove to be a member of Colocasieae. Zomicarpa, here considered a colocasioid due to its secretion tubes (French, 1988), has been kept in a separate tribe to emphasize its uniqueness in having distinct stamens. This New World genus shows some interesting similarities to Pinellia and Arisaema, of the Aroideae (Grayum, 1984). Its classification must still be considered highly spec- ulative. The remaining New World Colocasioideae, Ca- ladium, Chlorospatha, Xanthosoma, and Aphyl- larum, are probably best treated as a single sub- tribe, Caladiinae. Syngonium is much closer to the Caladiinae than to either Jasarum or Scaphispa- tha. Mayo & Bogner (1988) presented compelling evidence that Aphyllarum should be submerged in Caladium, but stopped short of formally ef- fecting the reduction. Remaining to be discussed is the enigmatic Southeast Asian genus Hapaline. This is, in many ways, the most phenetically advanced Old World colocasioid genus. It is distinctive in its nonpeltate leaves, fusion of spathe and spadix, uniovulate ova- ries, base chromosome number of x = 13 (rather than x = 14), and trinucleate pollen (also in Re- musatia). All of these features occur in the genus Pinellia (of Aroideae), to which Hapaline shows the greatest phenetic similarity. Hapaline is also unique among the Colocasioideae in having sterile male flowers above and below the fertile male re- gion; this is characteristic of Peltandra and Ty- phonodorum, which Hapaline resembles in other respects. But since Hapaline has now been shown definitely to have secretion tubes (French & Fox, manuscript), its membership in Colocasioideae is virtually assured. The placement of Hapaline with- in the subfamily remains problematical. Hapaline, unlike all other Old World Colocasioideae, secretes a milky, white latex from its laticifers. Moreover, microscopic examination of this latex reveals that it is in every detail (presence of S bodies and rubber particles, and absence of Y and R bodies) exactly like that of New World genera (French & Fox, manuscript). And although Hapaline does not ex- hibit a well-defined cortical vascular system (typical of New World Colocasioideae), there is a “retic- ulum of laticifers" in the stem cortex with which "a more detailed study might reveal a reduced vascular system” in association (French, in press). The pollen of Hapaline is of a derived type (globose, spinose, trinucleate) occurring in Colo- casioideae of both hemispheres. Nonpeltate leaves are more characteristic of New World Colocasioi- deae, and fusion of the spathe and spadix is oth- erwise known only in the neotropical genus Chlo- rospatha (Grayum, 1984). I here treat Hapaline as a subtribe in the oth- erwise New World tribe Caladieae, entirely on the basis of the above latex characters, which consti- tute the best evidence currently available. This decision entails the implication that the tribes Col- ocasieae and Caladieae must have diverged prior to the breakup of West Gondwanaland (still assum- ing that Colocasioideae are monophyletic) and clears the path for Scaphispatha (and possibly Zomi- carpa) to be transferred to Colocasieae should that prove necessary. The cladistic alignment of subfamily Coloca- sioideae, when considered as a monophyletic unit, is difficult to interpret. I have rather unconfidently derived the group from Philodendroideae. The Ca- ladieae are more primitive than the Colocasieae in having basically reticulate pollen and, as previously discussed, they show certain similarities with mem- bers of the Philodendroideae. Furthermore, all of the apomorphic characters uniting the Colocasioi- deae also occur in one genus or another of Phil- odendroideae (Grayum, 1984). Typhonodorum and Peltandra have strikingly colocasioid venation and have even been included in Colocasioideae, though they lack secretion tubes (probably the most im- portant synapomorphy for the subfamily). The dis- tinctive gorgonoid type of inflorescence sympodium occurs only in Philodendroideae and Colocasioideae (Ray, 1988) and, according to French (1986c), orthotropous ovules occurring in Colocasioideae are more similar in detail to those of Philodendroideae than of Aroideae. Derivation of Colocasioideae from within Phil- odendroideae would render the latter group para- phyletic, thereby suggesting that Colocasioideae ought to be submerged. Although there is a prec- edent for this (Colocasioideae were included in their entirety in Philodendroideae by Hotta, 1970), I prefer to maintain this distinctive and well-known subfamily for as long as the cladistic evidence for submerging and/or dividing it remains equivocal. Protarum. Engler (1920a) treated the mono- typic genus Protarum as a separate tribe of subfamily Aroideae. Hutchinson (1973) treated it as an integral part of the tribe Areae. Protarum 682 Annals of the Missouri Botanical Garden agrees with Aroideae (including Thomsonieae) in its tuberous habit, pedately compound leaves, ad- vanced (constricted) spathe, naked apical spadix appendage, unilocular ovaries, orthotropous ovules, base chromosome number of x = 14 and, striate pollen. Character analysis, however, indicates that Bog- ner & Nicolson (in press) were correct in trans- ferring Protarum to the Colocasioideae. In the first place, all of the above character states also occur in and are in most cases basic to the Old World Colocasioideae, except for pedate leaves, which are widespread in New World Colocasioideae. Prota- rum has fully connate stamens, as do all Coloca- sioideae (but extremely few Aroideae). Further- more, the striate pollen of Protarum is, in terms of its small size and peculiar pattern of striation, much more similar to that of Steudnera than to the pollen of any Aroideae (Grayum, in press a). And French (1988) has recently confirmed the presence of secretion tubes in Protarum. Protarum is the only aroid genus endemic to, or even occurring in, the Seychelles Islands. Since the genus is phenetically distinctive and taxonom- | to these ically isolated, recent long-dist islands can be ruled out. The Seychelles are be- lieved to represent continental fragments of the Indian Plate, from which they are thought to have separated about 70 myBP—at about the same time India commenced its northward movement (Raven, 979). Protarum probably reached the Seychelles via overseas dispersal from Africa or Madagascar, at some point after the separation of those islands from the Indian Plate (see Grayum, 1984). The Colocasioideae were then presumably extirpated or driven out of mainland Africa due to increased aridity (only the widespread and highly vagile Re- musatia vivipara, which is certainly a very recent invader, occurs there today). Arinae. glerian subtribe Arinae represent the “core” of the subfamily Aroideae. The group has been classed together by all authors at least since the time of Lindley (1847), although frequently with an ad- mixture of other more or less distantly related taxa. The Arinae are held together by having a tu- berous habit; basically sagittate leaves that often The eight genera composing the En- become compound; reticulate venation; secretion advanced, frequently constricted spathes; monoecious spadices with sterile flowers and/or naked regions above and below the fertile male portion; unilocular multiovulate ovaries with basi- files; cally parietal placentation; mostly stout, distinct stamens; orthotropous ovules; seeds with endo- sperm; a base chromosome number of x = 14 (or perhaps x = 7); simple flavones (particularly lu- teolin, unique to this taxon); onagrad embryogeny; inaperturate, globose, spinose or spinose-reticulate pollen that is mostly starchy (except Arum) and trinucleate (with the possible exception of Ther- iophonum); and perigonia absent. Few subtaxa of Araceae are defined by more synapomorphies than this one; yet the relationship of Arinae to the rest of the family is less clear-cut. The closest overall phenetic resemblances are shown by Ariopsis and Arisaema, the latter of which is here regarded as the sister taxon to Arinae. These genera agree in their tuberous habit, reticulate Venetum: secretion files, naked flowers, unilocular ‚ orthotropous ovules, seeds with ee base chromosome number of x = Я and inaperturate, globose, spinose pollen. Arisae- ma further agrees in having compound leaves, a naked spadix appendage, and starchy pollen; Ariopsis agrees in its parietal placentation and trinucleate pollen. Lindley (1847) and Schott (1860) placed the Arinae in close juxtaposition with tribe Thomso- nieae of Engler’s subfamily Lasioideae. There is no question that Thomsonieae are completely out of place in Lasioideae (see discussion under Lasieae); they indeed agree well with Arinae and are here included with that group in subfamily Aroideae. Thomsonieae and Arinae share the following fea- tures: tuberous growth habits; compound leaves (in most Arinae); paracytic stomata; Philodendron- type hydathodes; reticulate venation; secretion files; advanced (Type or IV) spathes; monoecious spadices with a sterile and usually naked terminal appendage; basically distinct stamens; a base chro- mosome number of x = 13 or 14; inaperturate, starchy, trinucleate pollen; similar starch grain structure (Reichert, 1913); and production of amines › and free indoles in the inflorescence at anthes The шыш differ from Arinae in having an altogether more primitive gynoecial structure (plurilocular ovaries with axile placentation, and anatropous ovules), seeds with endosperm, and ba- sically boat-shaped, striate or striate-reticulate pol- len (which, however, occurs elsewhere in Aroideae). But the number of apomorphies shared by Thom- sonieae with Arinae far surpasses those few shared with Lasieae. The geographic distribution of the Arinae is Tethyan, from Great Britain (Arum) and the Ca- nary Islands (Dracunculus) through the Mediter- ranean region eastward to the Himalayan region and Central Asia (Eminium) and on to the Far Volume 77, Number 4 1990 Grayum 683 Evolution and Phylogeny of Araceae East, New Guinea, and Australia (Typhonium). It is probable that the group originated in the ho- mogeneous flora that stretched along the northern margin of the Tethys Sea from western Europe to Indomalaya during the Eocene (Wolfe, 1975) and that it has only recently extended into tropical Africa (Sauromatum venosum) and Australasia. Engler’s tribe Areae is a sprawling, uninfor- mative, and partly artificial assemblage comparable to his tribe Colocasieae, and the former tribe is here reduced to coincide with his subtribe Arinae. Most of the other former subtribes of Areae have been raised to tribal rank to emphasize their dis- tinctiveness and isolation. Arisaema. as discussed under the previous heading, is prob- ably the Arinae. Arisaema has been placed fairly near Arinae by all important authors and does not The closest outgroup to Arisaema, differ from Arinae in many important respects: lack of sterile flowers below the fertile male region, basal rather than parietal placentation, usually more or less connate stamens, absence of luteolin, and bi- nucleate pollen. Arisaema also shows a resem- blance to Ariopsis (see under Arinae), especially in gynoecial and pollen characters, but differs markedly in leaf morphology and spadix structure. It shows some similarities to the neotropical genus Zomicarpa as well (tuberous habit, compound leaves, similar gynoecial structure, constriction of the spathe between the fertile male and sterile apical portions). Arisaema has perhaps the widest distribution of any aroid genus: from central Africa and the Ara- bian Peninsula through south and east Asia and into North America as far south as Mexico. All 10—11 sections of the genus and 61 of the 150 or so species occur in the Himalayan region, ac- cording to Li (1980), who believes the genus orig- inated there. This is probable since none of the outlying portions of the range (Africa, Southeast Asia, or North America) has a single endemic sec- tion. Furthermore, the most phenetically primitive section (the monotypic Dochafa, with male and female flowers on the same spadix) ranges from the Himalayan region west to Yemen. The adaptive radiation in Arisaema must have been relatively recent, however, in view of the vague separation among sections and the almost invariable pollen morphology throughout the genus, and likely owes to the new habitats made available during the uplift of the Himalayas, beginning with the collision of the Indian and Laurasian plates during the Eocene (about 40-45 myBP). See Grayum (1984) and the first half of this paper for further discussion. Ariopsis. This monotypic Asian genus has been placed among colocasioid genera by all major au- thors of the Schottian and Englerian schools. Fea- tures of the genus are peltate leaves, y thodes (otherwise known only from Steudnera and most h an association giant Cae -type hyda- Colocasia), parietal placentation, and connate sta- mens. The last-mentioned feature is unique in Ara- ceae, however, in that stamens of adjacent flowers are connate, in addition to those within each flower; thus, the stamens of Ariopsis are quite likely not homologous in this respect with those typical of Colocasioideae. Character analysis suggests that Ariopsis is bet- ter transferred to subfamily Aroideae. It shares more advanced characters with the Englerian sub- tribe Arinae than with any other group: tuberous habit; reticulate venation (Ertl, 1932); secretion files; unilocular multiovulate ovaries with parietal placentation; orthotropous ovules; inaperturate, globose, spinose, trinucleate pollen; a base chro- mosome number of x = ; and a unicellular chalazal endosperm chamber. That Ariopsis lacks the characteristic colocasioid venation is signifi- cant; that it lacks secretion tubes (French, 1988), the most important synapomorphy for the subfam- ily, is even more so. Moreover, ovular vasculature and morphology of Ariopsis are atypical for Col- ocasioideae (French, 1986c). Ariopsis also has numerous features in common with Arisaema and Pistia. Ariopsis and Pistia share reticulate venation, basal fusion of spathe spadix, a stipitate male portion of the spadix nown in Colocasioideae), multiovulate ovaries with parietal placentation, connate stamens with ume = dehiscence by lateral pores (unique in Araceae to these two genera), orthotropous ovules, endosper- mous seeds, a ase chromosome number of x = 14, and trinucleate pollen. The similarities in spadix structure between Ariopsis and Pistia/ Ambrosina are so striking that Ariopsis is here treated as a sister taxon to the latter two genera. Pinellia. In some respects the East Asian ge- nus Pinellia is a typical member of subfamily Aro- ideae: it has a tuberous habit, ovate to compound leaves, reticulate venation, secretion files, a base chromosome number of x = 13, and inaperturate, globose, spinulose, starchy, trinucleate pollen. It is unusual, however, in sometimes having peltate leaves (P’ei, 1935), full fusion of the spathe with the female portion of the spadix, and unilocular uniovulate ovaries with basal placentation, hem- ianatropous ovules, and transverse anther dehis- cence 684 Annals of the Missouri Botanical Garden Largely because of the last-mentioned features, Pinellia shows greater overall tic resem- blance to members of the Philodendroideae (Ту- phonodorum, Peltandra, Callopsis, Zomicarpeae) and Colocasioideae (Hapaline) than to any Aro- ideae. It agrees best with the latter genus, with which Pinellia shares a tuberous habit; basal fusion of the spathe and spadix; sterile male flowers (or naked regions) above and below the fertile male portion of the spadix; unilocular, uniovulate ova- ries; more or less anatropous ovules; endospermous seeds; an identical base chromosome number (x = 13); and inaperturate, globose, spinose, starchy, trinucleate pollen. In spite of the above evidence, Pinellia has here been conservatively maintained as a separate tribe in Aroideae. The genus lacks several important synapomorphies characterizing the Colocasioideae (including Hapaline): colocasioid venation, con- nate stamens, and secretion tubes. Those advanced characteristics shared with genera of other subfam- ilies are here interpreted as homoplasies. The geo- graphic distribution of Pinellia tends to support its retention in Aroideae: it is the only aroid genus endemic to East Asia, a stronghold of this subfamily (Typhonium, of the Arinae, ranges into East Asia, Arisaema is highly diverse there, and Amorpho- phallus has an endemic section). Thomsonieae. This pair of Old World genera was associated with the tribe Lasieae by Engler (1911), and with the Aroideae by most other au- thors. Character analysis strongly supports the lat- ter view, which has been adopted in this paper (see detailed discussions under Lasieae and Arinae). The essence of the argument is that there is but a single important apomorphy shared by Thom- sonieae with Lasieae that is not also shared with the Arinae, i.e., seeds lacking endosperm. The em- phasis of this single character over the extensive list of putative synapomorphies linking the Arinae and Thomsonieae is indefensible. The Thomsonieae also share certain features with Ariopsis, Arisarum, Ambrosina, and Pistia, though they do not appear closely related to any one of these. However, the occurrence in the last- named three genera of boat-shaped, striate pollen (uncommon in Araceae but basic to Thomsonieae) is viewed as significant, and the Thomsonieae have here been treated as the sister taxon to all four of these genera. The distribution of Thomsonieae appears to be basically Laurasian. Pseudodracontium is exclu- sively South Asian. The vast and protean genus Amorphophallus, though extending throughout Africa (including Madagascar) and eastward to northeastern China, southern Japan, and Australia, likewise appears to be basically South Asian. The most primitive pollen type in the genus (striate) is almost restricted to the Asian species of the genus. The monotypic subgenus Metandrium (Stapf, 1923) and eight of the 11 sections of subg. Amorpho- phallus are endemic to or centered in continental South Asia (there are, however, two endemic Af- rican sections and one section restricted to East Asia, suggesting phallus in these regions is of greater antiquity than that of, say, Arisaema, Sauromatum, or Typhoni- um). Subfamily Aroideae is here interpreted to include Engler’s subtribe Arinae (elevated to tribal rank) and tribe Thomsonieae, plus Ariopsis, Arisaema, that the presence of Amorpho- Pinellia, Arisarum, Ambrosina, the Cryptocory- ninae (here elevated to tribal rank), and Pistia. The sister group relationship of the subfamily as a whole is difficult to interpret. Although phenetically rather advanced overall, the Aroideae retain some primitive characteristics (e.g., microreticulate, per- haps starchless pollen; plurilocular ovaries with ax- ile placentation; a base chromosome number of x = 14 some respects, it would be most parsimonious to or even x = 7; and distinct stamens). In pair the Aroideae with another “advanced” ara- Philodendroideae or Colo- casioideae; one might thereby minimize the number of separate derivations of laticifers, naked flowers, ceous subfamily, i.e., and monoecy. However, I have tentatively decided to align Aroideae with subfamily Lasioideae, mainly because of the prevalence of base chromosome numbers of x = 13 or 14 in both groups and the pronouncedly Laurasian distribution of the tribe Orontieae and all Aroideae. Although the Lasioideae are in general much more primitive than Aroideae, laticifers, monoecy, and naked flowers are all known to occur in the former subfamily (laticifers in Orontium and throughout the Lasieae; monoecy in Stylochaeton; naked flowers in Pycnospatha of the Lasieae). Cryptocoryninae. Southeast Asian genera Cryptocoryne and Lage- nandra have been placed by all important authors of both the Englerian and Schottian schools among such genera as Arisarum, Ambrosina, Pinellia (all of Engler’s Aroideae), and Pistia. This alignment has been based mainly on shared gynoecial char- acters: unilocular, pluriovulate ovaries with parietal The two obviously related or basal placentation; orthotropous ovules; and en- The character analysis per- formed in the present pis showed a fair resem- dospermous seeds. Volume 77, Number 4 1990 Grayum 685 Evolution and Phylogeny of Araceae blance of Cryptocoryninae to these genera (mainly due to the gynoecial characters listed above), but even greater agreement with certain members of the Philodendroideae: the Aglaonemateae, Peltan- dra and Typhonodorum, but most strikingly with the Schismatoglottidinae Subtribe Schismatoglottidinae conforms in all respects to the gynoecial characteristics of the Cryptocoryninae. The two subtribes, which I pre- viously (Grayum, 1984) included together in the same tribe, are both basically rhizomatous and show strong aquatic tendencies. Both at least occasion- ally have a stem endodermis, and they tend strongly toward conical, horned or tubular anther apices (unique in Araceae). The two have inaperturate, basically psilate, boat-shaped, starchy pollen, a type that is virtually unknown outside of Philodendroi- deae. Syncarpy, which is very rare in Araceae, occurs in both taxa (Cryptocoryne and Piptospa- tha sect. Gamogyne), and both have a Southeast Asian distribution. e Cryptocoryninae are extremely well known cytologically but are difficult to comprehend. La- genandra is clearly based on x = 18—a number typical of Philodendroideae but highly unusual in Aroideae. Most counts in Cryptocoryne are also typically philodendroid, i.e., x = 17, 18, and 21; however, many species are based on x = 14. Cryp- tocoryne is the only aroid genus to combine these fundamentally different base numbers French’s (1988) startling revelation that Cryp- iocoryninae lack laticifers would appear to provide additional support for the original notion that the subtribe is most closely related to Pistia. It is, however, conceivable that laticifers have been con- vergently lost in these taxa, which share an aquatic lifestyle. But this argument cannot be made in the case of ovary vasculature, in which respect the Cryptocoryninae closely resemble Pistia and differ from all Philodendroideae (French, 1986c). Thus, the balance has recently tipped back in favor of an Aroideae alignment for the Cryptocoryninae. I herein elevate Cryptocoryninae to tribal rank, and return it to the subfamily Aroideae, near Pistia. These three genera, which were analyzed separately, have been associated more or less closely with one another Arisarum, Ambrosina, Pistia. by most authors on aroid classification. Arisarum and Ambrosina have usually been included among elements here placed in subfamily Aroideae (e.g., by Schott, 1860; Hooker, 1883; Engler, 1920a; and Hutchinson, 1973). Pistia has variously been tribe Arinae” (Hooker, 1883) or as a separate, monotypic tribe (Hutch- treated as a subtribe of the “* inson, 1973) or subfamily (Engler, 1920a). Several authors, including Schott (1860), have excluded the genus from Araceae altogether. Lindley (1847) and Buscalioni & Lanza (1935) included Ambro- sina together with Pistia in a separate family (Pis- tiaceae), to which Lindley also appended the genera composing the modern family Lemnaceae. The notion that the Lemnaceae are related to or even derived from Pistia is by now quite firmly entrenched in the literature of botany (see, e.g., Takhtajan, 1980; Cronquist, 1981). Pistia is, how- ever, a perfectly good (if superficially aberrant) aroid in all important details, whose closest relatives include the genera being considered along with it in this section. It appears to have nothing what- soever to do with the Lemnaceae; that is, the Lem- naceae were not derived from Pistia nor, most likely, from anywhere near Pistia. The Lemnaceae may have a sister group relationship with the Ara- ceae, which is about the closest conceivable con- nection they might have to Pistia, but even this is debatable. The two taxa have merely converged upon the floating, aquatic mode of existence and its attendant morphological adaptations. Full doc- umentation for these statements is presented in Grayum (1984). Character analysis indicates that Arisarum, Am- brosina, and Pistia are closely related to one another and somewhat more distantly to the rest of subfamily Aroideae. Basic characters of the group typical for Aroideae include a tuberous habit (ex- cept Pistia), reticulate venation (supposedly par- allel in Ambrosina; Ertl, 1932), secretion files (ex- cept Pistia), advanced spathe and spadix types, unilocular, multiovulate ovaries with orthotropous ovules, distinct stamens (connate in Pistia), en- dospermous seeds, a base chromosome number of x = 14 (x = 11 in Ambrosina), a Mediterranean (Tethyan) distribution (except the cosmopolitan Pistia), же Ru pollen. They share with the Ат f leaf morphology and gynoecial "es ue AME although in their basically boat-shaped, striate-reticulate pollen they agree more with Thomsonieae. The troublesome absence of laticifers in Pistia (French, 1988) may be related to neoteny or reduction in this genus. Pistia and Ambrosina appear more closely re- lated to each other than either is to Arisarum. Both genera exhibit basal fusion of the spathe and spadix, a stipitate male portion of the spadix, re- duction of the female portion to a single flower, apparently two stamens per male flower, and strik- ingly similar pollen exine sculpturing (see Grayum, in press a). They differ, however, in their growth habit, mode of anther dehiscence, base chromo- 686 Annals of the Missouri Botanical Garden some numbers (x = 14 in Pistia, x = 11 in Am- brosina), and pollen nuclear numbers (III in Pistia, II in Ambrosina). Thus, though these genera are here considered closely related, they are main- tained in separate tribes. Pistia and Ambrosina show more similarities to Ariopsis than to Arisarum. In fact, Pistia shares more putative apomorphies with Ariopsis than with Ambrosina. These mostly pertain to a strong over- all resemblance in inflorescence structure, gynoe- cial morphology, stamens, ovules, and base chro- mosome number. Ariopsis differs from Pistia, Ambrosina, and the Cryptocoryninae in having globose, spinose, starchless pollen, and hence has been treated here as a sister group to both of those taxa rather than to Pistia alone. Arisarum differs from Pistia and Ambrosina in several important respects: spathe free from the spadix, a nonstipitate male portion of the spadix, possession of a sterile apical appendage, a single stamen per male flower, several female flowers per inflorescence, and transverse anther dehiscence. The genus shows certain similarities to most of the other taxa in Aroideae (especially Thomsonieae, Arinae, and Ariopsis), but no compelling relation- ship to any one taxon. It is apparently quite isolated and is here maintained in its own tribe It should be clear from the foregoing discussion that, despite its peculiar aspect, Pistia is an integral part of the subfamily Aroideae. It clearly cannot be accorded separate subfamilial status since to do so would render the Aroideae paraphyletic. AROID CLADISTICS: A BEGINNING A preliminary, quasi-cladistic model of the pu- tative phylogenetic relationships of the important infrafamilial taxa in Araceae appears in Figure 1. This diagram was constructed in a piecemeal fash- ion, in a manner thought to best reflect the impli- cations of the available data; it is not the product of a rigorous, computer-mediated algorithm. This subjective interpretation was necessary due to the small amount of characters employed in the anal- ysis compared with the number of taxa under con- sideration. Computer analysis of aroid cladistics can be expected to provide valuable new insights, and work toward that goal is currently underway in at least two different laboratories. Meanwhile, highest priority should be given to the search for new kinds of systematically useful characters in Araceae, and to the expansion of the data base (for these and already established characters) to cover all genera of the family. The aan dichotomy in Araceae (Fig. 1), as visualized as a result of the present investiga- tions, pairs the Pothoideae and Philodendroideae, on one hand, against the Lasioideae and Aroideae, on the other. The former group consists of fun- damentally rhizomatous or scandent plants with basically distichous leaves, a base chromosome number possibly other than x = 7 or 14 (x = 127), and a largely Gondwanalandic distribution. The latter group consists of fundamentally rhizomatous or tuberous plants with basically spiral phyllotaxy, a base chromosome number o , and a basically Laurasian distribution. The position of Colocasioideae relative to these two groups remains unclear. Explanations for the two secondary di- chotomies, Pothoideae/Philodendroideae and La- sioideae/ Aroideae, are found under the headings *Homalomeninae, etc. *"Thomsonieae," re- spectively. Tertiary and ied dichotomies are discussed under the appropriate headings. I consider the following aspects of Figure 1 to be particularly tentative: the position of Coloca- sioideae (which may even be diphyletic); the der- ivation of Zamioculcadeae, an isolated group of obscure affinity; the position of Spathicarpeae with- in Philodendroideae (this tribe may be more cla- distically primitive than indicated); and the com- position of the subfamily Lasioideae, the members of which are associated mainly on the basis of symplesiomorphies. The cladistic relationship among the various *'alliances" of subfamily Philodendroi- deae is highly speculative, as is the composition of some of these groups (particularly the “Aglaonema alliance") AROID CLASSIFICATION: NEW AND OLD PERSPECTIVES Inasmuch as aroid cladistics are still in the em- bryonic stage, the infrafamilial classification of Ara- ceae is bound to remain in a state of flux. Thus, the classification at the end of this section (Table 5) is necessarily tentative, as are all other existing aroid classifications. Although there is a natural impatience on the part of workers in the field for a rigid and finalized system, our present compre- hension of aroid phenetics falls far short of what will be needed for the construction of the finely resolved cladogram upon which such a classifica- tion must be based. The linear order of the classification presented here reflects the constraints of the cladogram (Fig. 1); no additional significance should be imputed to the sequencing of taxa within larger groups. With the exception of Philodendroideae, no paraphyletic (according to the cladogram) groups have been accepted; otherwise, no attempt has been made to Volume 77, Number 4 1990 Grayum 687 Evolution and Phylogeny of Araceae o + o o o o Ф s o o o o о со 4 2 o Б $ c c “© с ос о 3 М > 7 Г“ Ф 2 2 = Ez с 5 os Ф Ф 4 4 o o о - = х= 07 S=> <= а в & E = 5 2 < Za Et P« o< о s = 2 2 9 2 s E 2 а $ 3 8 o = = © 5 = a 2 = = 9, 2 a c o c o © E o £ © o < [7] « a < = о > < a a o о Pothoideae Philodendroideae FIGURE 1. rigidly apply the criteria of cladistic classification. Since this classification is recognized as tentative, an effort was made to maintain nomenclatural sta- bility by accepting Englerian ranks. In a few cases (particularly his tribes Colocasieae and Areae), however, this was deemed inadvisable. No system- atic effort was made to ensure that various taxa at the same rank were equivalent in terms of their "level of peculiarity,” and new suprageneric taxa have been described only where absolutely nec- essary. The Araceae have here been grouped into five subfamilies in light of the information organized in the foregoing section. The system resulting from this arrangement most closely resembles that o Hotta (1970) in that Acorus has been removed from Pothoideae; Pothoideae and Monsteroideae Orontieae to the Lasioideae, in both systems). The present system differs from Hotta’s mainly in that Acorus has now been removed from the Zomicarpeae —" -— " . Colocasioideae Pinellieae Arisaemateae Areae Arisareae Pistieae Cryptocoryneae Ambrosineae Ariopsideae Thomsonieae Lasieae Orontieae Symplocarpeae Aroideae Lasioideae Proposed cladogram for major aroid subtaxa. Araceae altogether (Grayum, 1987), Pistia has been inserted directly into the Aroideae, and the Colocasioideae are here retained as a subfamily (rather than submerged in Philodendroideae, as Hotta has done). Hotta is probably at least partly correct on the latter issue, however: it is perhaps inevitable that at least part of the Colocasioideae will ultimately have to be subsumed in Philoden- droideae. I tentatively maintain the former subfam- ily, owing to a good deal of uncertainty over the exact manner in which the two taxa are related, and particularly to some lingering suspicion that the Colocasioideae may in fact be diphyletic (see under Colocasioideae). Significant departures from the Englerian sys- tem below the subfamily level are as follows (the rationales for all of these changes have been pre- sented in previous sections): Heteropsis is moved from the Potheae into the Monstereae; the Mon- stereae are here envisioned as being more closely related to the Potheae than to the Spathiphylleae. Culcasia is transferred out of Pothoideae into 688 Annals of the Missouri Botanical Garden TABLE 5. A preliminary new classification of the Araceae (with nomenclature based on Nicolson, 1984b). Family Araceae Juss. 1. Subfamily Pothoideae Engl. . Tribe das Nakai (Gymnostachys) 2. Tribe Spa арча Engl. (Spathiphyllum, Holochlamys) Tribe Anthurieae Engl. (Anthurium) Tribe Potheae Engl (Pothos, Pedicellarum, Pothoidium) Tribe Anadendreae Bogner & J. French (Anadendrum) Tribe Monstereae Engl. жада a. Subtribe Heteropsidinae Engl. (Heteropsis) b. Subtribe Monsterinae Schott (Rhaphidophora, Monstera, Amydrium, Epipremnum, Scindapsus, Alloschemone, Stenospermation, Rhodospatha) Tribe шш ы Engl. (Zamioculcas, Gonatopus) II. im Calloideae Schott Calla Alliance 8. Tribe Calleae Schott (Сайа) В. к АШапс das Engl. (Vephthytis, Anchomanes, Pseudohydrosme) in Tribe Callopsideae Engl. (Callopsis, Ulearum, Filarum, Zomicarpella) . Tribe Montrichardieae Engl. (Montrichardia) C. а АШа iM Engl. (Anubias) 13. Tribe Zantedeschieae Engl. (Zantedeschia) Y Tribe Aglaonemateae Engl. (Aglaonema, Aglaodorum) . Tribe re dan Schott (Mangonia, Asterostigma, Synandrospadix, Taccarum, Gorgonidium, Gearum, Spathantheum, Spathicarpa e Tribe Dielfenbachieae Engl. (Dieffenbachia) . Tribe el Mayo & соон (Bognera) D. Peon Alliane . Tribe Peltandreae Engl. (Peltandra, Typhonodorum) B Tribe Arop »hyteae Bogner (Arophyton, Carlephyton, Colletogyn 20. Tribe Schismatoglottideae Nakai (Schismatoglottis, tas Bucephalandra, Phymatarum, Ari- dar “He teroaridarum, Hottarum E. Philodendron Allia 2]. Tribe ЕВА Engl (Culcasia) 22. Tribe Cercestideae Grayum' (Cercestis) A ipn Homalomeneae (Schott) M. Hotta (Furtadoa, Homalomena Philodendreae Schott (Philodendron) Ш. sani Cease Engl — 5. Tribe Zomicarpeae Schott (Zomicarpa) a Tribe in Engl. a. Subtribe Protarinae (Engl.) Grayum' (Protarum) b. Subtribe Steudnerinae Engl. & К. Krause (Steudnera) . Subtribe Remusatiinae Grayum' (Remusatia, Gonatanthus) d. Subtribe Clos Schott (Colocasia, Alocasia) 27. Tribe ые be же — (Jasarum) b. О. dessen Grayum' (Scaphispatha) с. Subtribe Caladiinae Engl. € К. Krause (Caladium, Xanthosoma, Chlorospatha, Aphyllarum) d. Subtribe Syngoniinae к. (Syngonium) e. Subtribe Hapalininae Engl K. Krause (Hapaline) . Subfamily Lasioideae Eng ribe се Engl. (Symplocarpus, Lysichiton) 29, Tribe Orontieae К. Br. ex Dumort. (Orontium) 30. Tribe Lasieae Engl. a. Subtribe Dracontiinae Schott (Cyrtosperma, Lasia, Anaphyllum, Podolasia, Urospatha, Dracon- tioides, Dracontiu b. Subtribe Рус oie Bogner (Pycnospatha) 31. Tribe Stylochaetoneae Schott (Stylochaeton) Volume 77, Number 4 1990 rayum 689 Evolution and Phylogeny of Araceae TABLE 5. Continued. V. Subfamily Aroideae Engl. 3 e Thomsonieae Blume (Pseudodracontium, Amorphophallus) ri 33. Tribe Arisareae Dumort. (Arisarum) ribe Pinellieae Nakai (Pinellia) 35. Tribe Pi isti 36. Tribe Cryptocoryn 37. Tribe Ambrosineae Schott (Ambrosin 38. Tribe Ariopsideae Engl. (Ariopsis) N 39. Tribe Arisaemateae Nakai (Arisaema) w P 4 neae Blume (C Ty ptocoryne, Lagenandra) a) 40. Tribe Areae Engl. (Arum, Dracunculus, Helicodiceros, Theriophonum, Typhonium, Sauromatum, Eminium, Biarum ' Validation of new taxa and new combination: Cercestideae Grayum, trib. nov. Philodendroidearu P lantae africanae praecipue scandentes c analibus ЕЯ folia spiraliter disposita sine geniculo; е sine endothecio pollen amylaceum continentes; radices sine hypodermate sclerotico canalibus resiniferis; x Protarinae, stat. nov.; Protareae Engl., in Engler & Prantl, Nat. Pflanzenfam. Nachtr. 3: 29, 34 бу Remusatiinae Grayum, subtrib. nov. Colocasiearum. Plantae orbis antiqui pone globoso spinoso trinucleato; spadix sine appendice sterili apicali. Jasarinae G іеагит. atosae у enezuelensium caulibus sine systemate vasculari corticali foliis linearo- lan i aud ovarium uniloculare Шеше basali ovulis 2; semina sine endospermio; exinium pollinis omnino = 11. reticulatum Se 'aphispat a Grayum, subtrib. nov. Caladie um. Plantae brasiliensium sine floribus sterilibus ovariis unilocularibus placentatione basali; exinium pollinis verrucis applanatis multangulis. the Philodendroideae in the neighborhood of Phil- odendron; Cercestis is removed from the Nephthy- tideae to its own tribe, Cercestideae, while the former tribe is enlarged to accommodate Ancho- manes and Pseudohydrosme (formerly of the Thomsonieae); Cercestideae, Nephthytideae, and Montrichardieae are all transferred from Lasioi- deae to Philodendroideae; Peltandra and Typho- nodorum are joined in a single tribe, the Peltan- dreae; the Callopsideae, Spathicarpeae, and Arophyteae are also moved from Aroideae to Philo- dendroideae. The Zomicarpeae are broken up, Zomicarpa being assigned to the Colocasioideae (as a separate tribe) and the remaining three genera to the Callopsideae (Philodendroideae). The last- mentioned subfamily, here much и has been divided into five informal “allianc The internal classification of eee Coloca- sioideae is changed to reflect a basic dichotomy, largely between New and Old World taxa. Pro- tarum is transferred from the Aroideae into the Old World tribe Colocasieae as a separate subtribe; Steudnera is dissociated from Remusatia and natanthus and is included in its own subtribe; the latter two genera (which probably ought to be com- bined) are placed in a new subtribe, Remusatiinae; Jasarum and Scaphispatha are each accorded separate subtribal status in the New World tribe Caladieae; Syngonium is included in this tribe as e typic subtribe Syngoni orld genus Hapaline agrees phenetically with the Ca- ladieae and has been included there in its own subtribe. The Lasioideae are expanded to include Stylo- chaeton (transferred from Aroideae) and the tribe Orontieae (sensu lato); the Thomsonieae, however, are moved out of Lasioideae and into Aroideae; Ariopsis is moved from the Colocasioideae into the Aroideae. The following departures from Engler’s usage (below the subfamily level) may be noted in the present system: nine tribes or subtribes (Potheae, Philodendreae, Nephthytideae, Colocasieae, Steud- nerinae, Caladiinae, Orontieae, Thomsonieae, Zomicarpeae, and Areae) were judged to be more or less artificial, and variously disbanded; ten tribes or subtribes underwent a change of rank (Heter- opsideae to Heteropsidinae; Sch шоона to Schismatoglottideae; Homalomeninae to Homalo- meneae; Protareae to Protarinae; Syngonieae to Syngoniinae; Arisarinae to Arisareae; Atherurinae to Pinellieae; Arisaematinae to Arisaemateae; Am- brosininae to Ambrosineae; and Cryptocoryninae to Cryptocoryneae); subtribe Colocasiinae and tribe Callopsideae were enlarged; five tribes or subtribes (Acoreae, Typhonodoreae, Philodendrinae, Alo- 690 Annals of the Missouri Botanical Garden casiinae, and Arinae) were eliminated; and 10 (Gymnostachydeae, Anadendreae, Monsterinae, Беген, di adi Remusatiinae, Jasarinae, , and Pistieae) were у added о оп. Many of the changes in the accompanying clas- sification (Table 5) represent a return to pre-En- glerian concepts. Although Engler’s contributions revolutionized our understanding of aroid phenet- ics, I believe he made a fundamental and rather grave error in unduly emphasizing often superficial vegetative features (e.g., leaf venation and mor- phology) in constructing his classification. Engler’s treatment of the tribes Lasieae and Thomsonieae provides a particularly glaring example of super- ficially convergent taxa juxtaposed on this basis; his inclusion of the genera Anchomanes and Pseu- dohydrosme within the Thomsonieae constitutes a somewhat less obvious case. There are very good reasons why reproductive characteristics of higher plants should be accorded more taxonomic “weight” than vegetative features (see, e.g., Stebbins, 1970) combination event significantly affecting the re- . Any mutation or re- productive organs (e.g., a change in style length, petal color, timing of anther dehiscence or stigma receptivity, pollen exine ornamentation, chromo- some features, etc.) would be much more likely to result in reproductive isolation, followed by spe- ciation and a consequent cladistic dichotomy than would an event of similar magnitude affecting only vegetative organs. Therefore, within a given clade, one should expect to encounter considerably more variation in vegetative than in reproductive struc- tures—as has long been intuitively appreciated by plant taxonomists. With regard to the Araceae, this would suggest that the strikingly similar leaf morphology of, say, Dracontium and Amorpho- phallus ought not to be emphasized so much as their radically different floral and pollen morphol- ogies. LITERATURE CITED ARBER, А. 1925 act аа x A Study. eS Univ. Press ARENDS, J. C., J. D. BASTMEUER & N. (m 1982. Ceanainn numbers and taxonomy in Cryptocory- ne (Araceae). II. Nordic J. Bot. 2: 453-463. AYYANGAR, K. R. 1973. Karyological affinities of Ty- paa flagelliforme Bl. and Typhonium trilo- tum Schott. Proc. Indian Sci. Congr. Assoc. 60(III): )3 BAKHUIZEN VAN DEN BRINK, R. С. 1958. Are Epiprem- num Schott, Rhaphidophora Hassk., and Monstera Adans. congeneric? Blumea suppl. 4: 91-92. BanaBÉ, D. Vascularisation de la fleur de ps ploc dien foe didus Araceae). Canad. J. Bot 936-1544. — 987. La néoténie s les Araceae pd sper rmae). uu . Séances Acad. 305, sér. III: 609-612 & L. ШЙНЕН. 85. Anatomie florale de iip OPE ad (Araceae). Beitr. Biol. 03- Pflanze & 1 . LABRECQUE. 1983. Vascularisation de la fleur de е ae (Araceae). Canad. J. Bot 61: 1718-17 Vascularisation de la fleur de Lysichitum camtschatcense (Araceae). Canad. J. Bot. 62: 1971- 1985 е de Іа fleu d'Orontium РА L. (Aracées). Bull. Soc cn France 132: ‚ 5. FORGET а T a 1987. Organo- génèse de la fleur de е foetidus (Arace- ae). Can . Bot. 6 46-4 plants and their taxonomic Eu . Mono- : i 5 56 Evolution of dioecy in flowering plants. Annual Rev. it Syst. 11: 15-39. — — — Ф J. Н. BEACH volution of sexual systems in и de Ann. Missouri Bot. Gard. 74. 68: 25 ВЕАСН, J. H. 1982. Beetle pollination of Cyclanthus bipartitus (Cyclanthaceae). Amer. J. Bot. 69: 1074- 1081. Beck, C. R. 5снмїр & G. W. ROTHWELL. 1982. iin morphology and the primary vascular system eed plants. Bot. Rev. (Lancaster) 48: 691-815. NOM M. 1981. Cytotaxonomy of the genus Arum (Araceae) in the Balkans and the Aegean area. Bot Jahrb. Syst. 102: 183-200. BEHNKE, H.- Siebelement-Plastiden, Phloem- Protein und Evolution der p ure kotyledonen. € Deutsch. Bot. Ges. 94: 641- 662. H 9 w evidenc ce from Schall sce Ои der Tribus _ Areae (Aro de Araceae) auf v E | ge (editor) Advancing Frontiers in Cytogenetics. Hin- o., Delhi 1982. usta малени, Р. The demography of jack-in- e-pulpit, a forest perennial that changes sex. Eco Ms 52: 335-351. BLanc, P. 197 Contribution à а des Aracées. I arques sur la croiss sympodiale . Rem chez PAnthurium scandens Engl., Е “Philodendron fenz- lii Engl. et le Philodendron speciosum Schott. Rev Gén. Bot. 84: 319-331 980. б ирин sur les flagelles des Arace- ae. Adansonia, sér. 2, 20: 325-338. BOGNER, J. 1972. Revision bod Arophyteae (Araceae). Bot. po Syst. 92: ; AP Ru A In: J.-F. Leroy (editor) Fore de Madagascar et des em Mu- séum National d'Histoire Naturelle, Par 1 76a. t aces N. E. Вг. Palmengarten 40: 83-8 1976b. Die systematische йш. уоп Aco- ра Conwentz, einer fossilen Bern- Volume 77, Number 4 1990 Grayum 691 Evolution and Phylogeny of Araceae stein. Mitt. Bayer. Staatssamml. Paláontol. Hist. Geol. 16: 95-98. . 1978. A critical list of the aroid genera. Aroi- deana 1: 63-73. The genus Bucephalandra Schott. Атала 3: 134-143. 981a. ае gabunensis Engl. colinas 4: 31-3 1981b. A new RPM species from Borneo. Aroideana 4: 57-63. ———. 19 А new Hottarum е species from poo Plant Syst. Evol. 142: 49-5 . Two ede MM from Habd Plant E 145: 159- fede p Bunting, an ut. aroid from Guyana Highland. Aroideana 8: 55-63. Syst. 1985b. One new name and five new combi- nations in Araceae. Aroideana 8: 73-79. . 1987. Мага. variation in aroids. Aroi- deana 10: 4-16. М. Hotta. 1983. A new Hottarum (Arace- ae) from Sarawak. Adansonia, sér. 4, 5: 27-31. & NICOLSON. A revised classification of the Araceae with dichotomous keys. Willdenonia (in press). , S. J. Mayo & M. SivADASAN. 1985. w species and changing concepts in Amorphophallus. 4-25. Boyce, P 1 med on aroids. Bull. Alpine Gard. Soc. Gr. Brit. 5-42. Bovp, L. 1932. Pusat ds seedlings. е зон studies in к post-seminal . & Proc. Bot. evelopment o Soc. Edinburgh 31: " 2 BUCHET, s. 1939. Nouveau genre malgache d' ipi à pistils enveloppés. Bull. Soc. Bot. France 86: BULLOCK, S. H. 1981. Notes on the phenology of in- PERE and pollination of тош rain forest palms osta Rica. Principes 1 “a 105 жонине. С. S. 1979. Sinopsis de las Araceae de Ver zuela. Revista Fac. Agron. Mees 10: 139- 290. 1 Piperales and the monocots. 84. The heed Alocasia. үе BuscALIONI, L. & D. Lanza. 1935. Le basi morfolo- giche, anatomiche e teratologiche della nuova fa- miglia delle Pistiaceae (Buscalioni e Lanza) rappre- sentate dai due generi Pistia ed Ambrosinia. Malpighia 34: 103-180. CAMAZINE, S. & K. J. NikLas. 1984. Aerobiology of : sag Soe between the spathe and n Amer : 843-850. CARVELL, W. N. 1989. Floral i of the Pothoi- deae and oo (Araceae). Amer. J. Bot. 76(suppl.): 2 CHAUHAN, К. “ E & P. E. BRANDHAM. 1985. Chro- mosome DNA ow E Amorphophallus Гессе. Kew Bull. 40: 758. CHEADLE, V. I. ss occurrence and types of vessels in us various organs of the plant in the Monocotyledonae. Amer. J. Bot. 29: 441-450. J. M. TuckER. 1961. Vessels and phylogeny of Monocotyledoneae. Recent Advances Bot. 1: 161- 165. CHEN, J. & B. J. D. MEEUSE. 1971. Production p free indole by some aroids. Acta Bot. Neerl. 20: 627- n. 63 Cocucci, A. E. 1966. Карой d de а pd vermitoxicus. Kurtziana 3: 15 The T morphology "d the reticulate-veined Liliiflorae. Telopea 2: 401-4 CORNER, E. J. Н The 2 Theory eR II. Phytomorphology 4: 152 = W. 7 "iis ie ui of angiosperms m the "iud of North America: an aroid inflo- rescence. Rev. Palaeobot. Palynol. 25: 241-252. Crisci, J. V. 1971. Flora баш Araceae. Revista Mus. La Plata, Secc. Bot. 193-284. Croat, Т.В. 1979. The er ant of Araceae. Pp. 291-308 in K. Larsen & T. B. Holm ‘Nielsen (edi- tors), К э Botany. Academic Press, London. 1981. A revision of QU ge (Araceae). Ann. Missouri Bot. Gard. 68: 565- & R. BAKER. 19 Studies in VETE II: m section Polyphyllium. Selbyana 2: 230- R. D. SHEFFER. 1983. The sectional group- ings of Anthurium yp d Aroideana 6: 85-123. CRONQUIST, А. 1. An Integrated System of Classi- fication of uie Plants. Columbia Univ. Press, ew or CULLEN, J. A preliminary bi of ptyxis (ver- nation) in the angiosperms. Notes Roy. Bot. Gard. Edinburgh 37: 161-214. ZAJA, А. Т. 1978. Starke und Stárkespeicherung bei Gefasspflanzen und einigen Vorfahren im Hinblick auf Evolution: Versuch einer Amylotaxonomie. Fi- scher, Stuttgart. DAGHLIAN, C. P. . A review of the fossil record of monocotyledons. Bot. Rev. (Lancaster) 47: 517- 555. DAHLGREN, R. M. T. & H. T. CLIFFORD. 1982. The Monocotyledons: A Comparative Study. Academic Press, London. DAKWALE, S. & S. BHATNAGAR. 1985. Insect-trapping behaviour and diel periodicity in Sauromatum gut- tatum Schott (Araceae). Curr. Sci. 54: 699-702. DaLrrzscH, M. 1886. Beiträge zur Kenntnis der Blatt- anatomie der Araceen. Bot. Centralbl. 25: 153-156, 184-187, 217-219, 249-253, 280-285, 312-318, -349. DAUMANN, E. 1930. Nektarabscheidung in der Blü- tenregion einiger Araceen. Planta 12: 38-48 DILCHER, D. L. & C. P. DAGHLIAN. 1977. йозе аа иы of angiosperms from the Eocene: Philodendron leaf remains. Amer. J. Bot. 64: 526-534 DoYLE, J. A Fossil evidence on early evolution of the monocotyledons. Quart. Rev. Biol. 48: 399- ECKARDT, T. 1937. Untersuchungen über Morphologie, Entwicklungsgeschichte und systematische Bedeu tung des pseudomonomeren Gynoeceums. Nova Acta Leop. 5: 1-112. EL-DiN, S. M. 1968. Warmeperiode und Duftstoff des Blütenkolbens der Aracee Schizocasia portei. Na- turwissenschaften 55: 658-659. A 76. Vergleichende р e is über die morphologischen Verhaltnisse der Araceae. Natürliches System der Araceae. Nova Pe Acad. Caes. Leop.-Carol. German. Nat. Cur. 39: 133-155. 1877. Vergleichende Untersu change uber die morphologischen Verhältnisse der Araceae. II. 692 Annals of the Missouri Botanical Garden Ueber Blattstellung und оа der Ara- ceae. s. Leop.-Carol. German. Beitráge zur Kenntnis der Araceae V 12. "Über den Entwicklungsgang i in der Familie der Araceen und über die Saar y derselben. 336. Bot. Jahrb. Syst. 5: 14 Я 1905. Potho tim ae. pus 1-330 in A. Engler (editor), Das Pflanzenreich 21: (IV.23). asioideae 1-130 in A. Engler (editor), Das нер" 48: (IV.23). omalomeninae und Sc Ан dinae Pp. 1- 134 in A. Engler (editor), Das Pflan senec fs: (IV.23). Oa. Aroideae und Pistioideae. Pp. 1- in A. Tee (editor), Das Pflanzenreich 73: (IV. e EL 920b. е qus generalis et index famili- ae pan e Pp. 1 n А. Engler (editor), Das Pflanzenreich 74: e De A). . KRAUSE. 1908. Monsteroideae. Pp. 4- 138 in А. Engler (editor), Das Pflanzenreich 37: (IV.23). & 1920. Colocasioideae. Pp. 3-13 in А. Engler (editor), Das Pflanzenreich 71: (IV.23. “+ EnTL, Р. О. 1932. Vergleichende Untersuchungen über die eres der Blattnervatur der Araceen. Flora 126: EypE, R. H. 1975. The bases of angiosperm phylogeny: floral anatomy. Ann. Missouri Bot. Gard. 62: 521 537. NICOLSON & P. SHERWIN. 1967. survey of floral anatomy in Araceae. Amer. J. Bot. 54: 478-497. Fox, M. С. € J. C. FRENCH. 1988. Systematic occur- rence of sterols in latex of Araceae: subfamily Col- r. J. Bot. 75: 132-137. . HORNER, JR. 80. Calcium oxalate үйөз in plants. Bot. Rev. (Lancaster) 46: 27. FRENCH, J.C. 1985a. Patterns of endothecial wall thick- enings in Araceae: petición Pothoideae and Mon alga nd Amer. J. Bot. 72: 472-486. 85b. o of endothecial wall thick- a in а ае: subfamilies Calloideae, Lasioideae and Philodendroideae. Bot. Gaz. (Crawfordsville) 146: 21-533 Patterns of endothecial wall thick- enings in Araceae: subfamilies Colocasioideae, Aroi- deae and Pistioideae. Bot. Gaz. (Crawfordsville) 147: 79 1 Patterns of stamen vasculature in the Araceae. ell J. Bot. 73: 434-449, vule нна in Araceae. Bot. Сая (Colore) 147: 478-495 87a. Systematic occurrence of a р hypodermis in roots of Araceae. Amer. J. Bot. а Systematic survey of resin canals іп roots of Araceae. Bot. Gaz. (Crawfordsville) 148: 360-371. Systematic occurrence of anastomo sing lies in Xr cae Bot. Gaz. (Crawfordsville) 149: 71-81. Vegetative anatomy of the Araceae. In —— J. Ar- ditti & L. Nyman (editors), The Biology of Araceae. Cornell Univ. Press, Ithaca, New York (in press). G. Fox. Latex ultrastructure in the Arace- ae: subfamily neo (manuscript). & C. T. KessLER. 1989. Molecular systematics of Araceae: are d and ги aroids? Amer. J. p 76(suppl.): 2 — — — & P. B. TOMLINSON. 1980. pcd ob- servations on е vascular ms of certain 6 in C. D. Bickn ell D. F. Cutler & M. Gregory (editors), Petaloid Monocotyledons. Academic Press, & —, 81a. Vascular patterns in stems of Araceae: subfamilies Calloideae and Lasioideae. Bot. Gaz. (Crawfordsville) 142: 366-381. 1981 ascular patterns in к of Araceae: subfamily Pothoideas. Amer. J. Bot 113-729. —— ——. 1981 Vascular te of ipd ө: aubtamily Monsteroideae. Amer. J. Во. 68: 1115-1129 Vascular e: BEER ni Phiodenirodeae Bot. Gaz P riara 142: s E cular patterns in stems of Araceae: siblumilies iu M Aroideae and Pistioideae. Amer. A Bot. 70: 756-771. 984. Patterns of in vascula- 1: 1432- ture in Ру од Amer. J. Bot. 7 1443. 986. Compound vascular bundles uction and signif- in Е i gon constr 4. nce. Kew Bull. 41: 56 GALIL, J. 1978. Morphologi: ages on Arisarum vulgare Targ.-Tozz. Israel J. Bot. 27: 77-89. GAUDET, J. 1960. Ontogeny of [olia sclereids in Nym- phaea pages Amer. J. Bot. 47: 525-532. GENTRY, A. 982. Neotropical floristic diversity: sages connections between Central and So GIBBS, R. D. 1974 i of Flowering aen Volume ПІ. e ill-Queen's Univ. Press, Montreal. GOEBEL, К. 1931. Blütenbildung und a (Anthokladien und Infloreszenzen). Gustav Fischer, ena. GOLDBLATT, P. 1980. ná in angiosperms: mono cotyledons. Pp. 219-2 nw. Lewis (editor), Polyploidy: Biological со Plenum Press, New York. GOTTSBERGER, G. 1977. Some aspects of beetle polli- nation in the evolution of flowering plants. Plant Syst. Evol., Suppl. 1: 211-226. MARAL, JR. 198 in Brazilian о species. Ber. Deuts G : 391- 4. Pollination strategies ch. Bot. ed йа foliares de las Aráceas cultiva das en la ciudad de Tucumán y sus alrede- dores. Lilloa 36: 15-41. GRAYUM, M. H. 1984. Te and phylogeny da the Araceae. Ph.D. Dissertation. Univ. Massac setts, а Massachusett 985. Evolutionary a ecological signifi- cance of starch storage in pollen of the Araceae. Amer. J. Bot. 72: 1565-1577. 1986a. Phylogenetic implications of pollen Volume 77, Number 4 1990 Grayum 693 Evolution and Phylogeny of Araceae nuclear number in the Araceae. Plant Syst. Evol. 151: 145-161. 1986b. Correlations between г bi- ology and pollen morpholo gy in the Araceae, with some implications for angiosperm evolution. Pp. 313- 327 in I. K. Ferguson & S. Blackmore (editors), Pollen and Spores: Form and Function. Elsevier Press, London. . 1987. A summary of evidence and arguments supporting 2 я of Асогиѕ from the Агасеае. Taxon 36: З ©. external pollen ultrastructure of the Araceae, and some putatively related taxa. Mono- Systematic embryology of the Araceae. Bot. Rev. rai (in press „ JR. 1973. Observations on the stomatal о E Be (Araceae). Bot. Gaz. (Crawfordsville) 134: 151-153 GREGOR, H.-J. & J. BOGNER. 1984. Fossile Araceen Mitteleuropas d ihre rezenten Vergleichsformen. Wi eris Naturae 19: 1-12. Hara, Н. 1971. A revision = the Eastern Himalayan B of ke enus Arisaema (Araceae). Pp. 321- 54 in H. Нага (editor), The Flora of Eastern Him- ее Second Report. Univ. Tokyo Press, Tokyo. HARBORNE, J. B. 1982. Flavonoid compounds. Pp. 264- 274 in R. M. T. Dahlgren & H. T. Clifford, The Monocotyledons: A Comparative Study. Academic Press, London. HanLiNG, С. 1958. Kogia, of the Cyclanthaceae. cta Horti pur 18: Harris, P. J. & R. D. unc 1980. Phenolic con- stituents of the cell walls of monocotyledons. Bio- 3-160 . 1986. Cyrtosperma Griffith and the origin of the aroids. Ph.D. Dissertation, Hasse College, xfor 1988. Cyrtosperma (Araceae) and its Old World allies. Blumea 33: 427-4 ——. = llecting Alocasia in New Guinea. Aroideana: (in press). Hoces, R. 1963. Chemotaxonomie der Pflanzen. Band 2. Monocotyledoneae. Birkháuser Verlag, Ba- sel, 1977. Cyanogenic compounds as systematic markers in Tracheophyta. Plant Syst. Evol., Suppl. 1: =2 986. Chemotaxonomie der Pflanzen. Band 7. Birkhiuser Verlag, Basel. Phytochemistry and chemotaxonomy of the e Aroideana 10: 17-19. EC A. 1986. A review of ebat studies n the Palmae. Bot. Rev. (Lancaster) 52: 221-259. Eee, J. cite Reproducción sexual y As ción vegetativa en Arisarum M rrhinum Durieu (Araceae). Lagasealia 15: 25-4 HESLoP-HARRISO & K. R. E 1977. The receptive surface of the angiosperm stigma. Ann. Bot. (London) 41: 1233-1258. Hesse, М. 1980. Entwicklungsgeschichte und Ultra- struktur von Pollenkitt und Exine bei nahe ver- wandten entomophilen und ты Angio- spermsippen der Alismataceae, Liliaceae, Juncaceae, рек Роасеае und Araceae. Plant Syst. Evol. 134: 207. 1981. The fine structure of the exine in re- lation to the stickiness of angiosperm pollen. Rev. dard е 35: 81-92. 9 Origin of the major features of Cour. Forsch.-Inst. Senckenberg 3 - HorTTUM, R. E. 1955. wth-habits of mono e dons—variations on a theme. ee S 399-413. g > D. 1883. a Pp. 955-1000 in G m & Í D. Hooker, Genera Plantarum, Vol- — ndon Hosen, W. & A. NAHRSTEDT. 1975. Spezifische Glu- Physiol. Chem. 356: 1265-1275. A system of the family Araceae in Japan and adjacent areas. Mem. Fac. Sci. Kyoto 1971. Study of family Araceae. General re- —! о J. Bot. 20: 269-310. 976. Notes on Bornean oa III. Pedi- a and Heteroaridarum, two new genera of the aroids. Acta Phytotax. Geobot. 27: 61-65. 19 A new genus of the family — from West ‘Sumatra. Acta Phytotax. Geobot. 32: 142-146. . 1982. On the differentiation of subtribes Hom- alomeninae and Schismatoglottidinae (Araceae) in Malesia. Acta Phytotax. Geobot. 33: 127-139. 1 Macano) [сой Sumatra with a genus Furtadoa. Gard. Bull. Straits Settlem. 38: 43- 54 HOWARTH, J. 1957. The phylogeny of monocotyledons. Ph.D. Dissertation. Cambridge Univ., Cambridge HUTCHINSON, J. 1973. The Families of Flowering Plants. I Press, London INAMDAR, J. . М. SHENoY & N. V. Rao. 1983. Leaf с. of — monocotyledons with re- ticulate venation. Ann. Bot. (London) 52: 725-735. ЈАСОВЅЕМ, N. 1976. Notes on Cryptocoryne of Sri Lanka ape Bot. Not. 129: 179-190. Jones, С. E. Chromosome numbers and phylo- genetic as in the Araceae. Ph.D. Disser- tation. Univ. Virginia, Charlottesville, Virginia. Kaptan, D. R. 1973. The problem of leaf morphology and evolution in the monocotyledons. Quart. Rev. Biol. 48: 437-457 KESSLER, C. T. ENCH. 1989. Molecular systematics a Colocasioideae (Araceae). Amer. J. Bot. (suppl.) 76: 251. KNOLL, F. 1926. м Arum-Blitenstande und ihre Be- sucher. Abh. К. К. Zool.-Bot. Ges. Wien 12: 381- 481. Knutson, R. M. 1974. Heat production and tem ature apps in eastern skunk cabbage. сеснен 186: ; KRAUSE, E Calloideae. Pp. 140-155 in A. Engler Жн Das Pfanzenreich 37: (IV.23). KRISHNAMURTHI, M. S. & N. GEETHA. 1986. Minor veins an e vein endings in Araceae. J. Swamy Bot. Club 3: 145-148. LARSEN, K. pn Cytology of vascular plants. III. A 694 Annals of the Missouri Botanical Garden study of Thai aroids. Dansk Bot. Ark. 27(1): 39- 59 Leavitt, В. С. 1904. Trichomes of the root in vascular cryptogams and oo Proc. Boston Nat. Hist. Soc. 31: 273-3 LEICK, E. 1915. und ihre blütenbiologische Deutung Bot. Ges. 33: 518-525. Li H. imalayas — Hengduan Mountains — the centre of distribullon and differentiation of the genus Arisaema; to discuss the problems about the origin and migration of this genus. Acta Bot. Yunnanica 2: 6. Dio Erwarmungstypen der Araceen . Ber. Deutsch. LINDLEY, J. The Vegetable Kingdom or, The Structure, Clausibcation: and Uses of Plants, Illus- trated upon the Natural System. Bradbury & Evans, London LLovp, D. С. 1979. Parental strategies of angiosperms. New Zealand J. Bot. 17: 595-606 Mapison, M. 1 llose pibe and Scindapsus (Асен), Selbyana 1: 325-327. gue of Monstera (Araceae). Contr. -101. Gray Herb. 207 a. The genera of Araceae in the northern Andes, Aroideana 1: 31-53. 9 he species of Anthurium with pal- mately divided leaves. Selbyana 2: 239-281. 979a. Protection of developing seeds in neo- vial Araceae. Aroideana 79b. Notes on some aroids along the Rio ER. Brazil Aroideana 2: 67-7 new species of Ulearum from Brazil. Árcideana 3: 101-1 Notes on € cg (Araceae) and its alins, Selbyana 5: 34 ——— & B. TI IFFNEY. к? e seeds of the Monstereae: their morphology and fossil record. J. 2 Chromosome variation in Ára- . Acoreae to Lasieae. Kew Bull. 28: 199- MARTICORENA, С. & M. QUEZADA. 1985. Catálogo Pi la flor asc de Chile. Gayana, Bot. 42: 1-15 Martius, K. F. P. von. 1831. Ueber die Art der e а bei einigen Aroideen und úber die Charac- teristik mehrere Gattung dieser Familie. Flora 14(2): -460. Systematics of Philodendron Schott rescence Р бозлы of gynoecial structure in Philodendron (Araceae). J. Linn. Soc., Bot. 100: 39-172. M. Barroso. 1979. A new pedate-leaved species of Philodendron from Bahia, Brazil. Aroidea- 2-94. & J. BOGNER. 1988. A new species of Caladi- um (Araceae) with notes on generic delimitation in the TRA Caladieae. Willdenowia 18: 231- EN h - D. 1978. The physiology of some sap- yophilous flowers. Pp. 97-104 in A. J. Richards (editor, The Pollination of Flowers by Insects. Ac- ademic Press, London. MERCADO-NORIEL, L. К. & B. T. MERCADO. 1978. Floral anatomy and seed morphology of water lettuce (Pis- tia stratiotes). Philipp. Agric. 61: =2 METCALFE, С. R. 1967. ша of latex in the plant kingdom. ‚ Bot. 21 5-127 1983. The remarkable shooting idio- blasts. Aroideana 6: 9- Моор, G. E. E. 1976. Heat production and pollination in Araceae. Canad. J. Bot. 54: 545-546. MOOKERJEA, A. 1955. Cytology of different species of aroids with a view to trace the basis of their evolution. к. & N. W. Un. 982. Major Pan of evolution in palms. Bot. Rev. (Lancaster) 48: 69. MULLER, J. 1981. Fossil pollen records of extant an- giosperms. Bot. Rev. (Lancaster) 47: 1-142. MÜLLER, L. 1919. Uber Hydathoden bei Araceen. Akad Wiss. Wien Sitzungsber., Math.-Naturwiss. Kl., Abt. I, 128: NAHRSTEDT, А. 1975. mnn in Araceen. Phy- tochemistry 14: 2627-262 NicoLsoN, D. H. 1959. The occurrence of trichoscler- eids and crystalline deposits in the Monsteroideae ae). dide ace M.S. Thesis. Cornell Univ., Ithaca, New ork. 1960a. A brief review of classifications in the mend Baileya 8: 62-67. 1960 e occurrence of trichosclereids in the Monstecoidene (Araceae). Amer. J. Bot. 47: 598- 1966. Filarum, a new genus of Peruvian йн. Brittonia 18: 347-349. 968a. The diet Xenophya Schott (Arace- ae). “Blanes 16: 115-11 968b. The genus ошай іп en east Malesian and west Pacific islands. Blumea 16: 119- 121. . 1968c. A revision of Amydrium (Araceae). Blumea 16: 123-127. 1969. A revision of the genus onen TOM Smithsonian Contr. Bot. 1 9 1984a. A second collection of Pedicellarum (Araceae). Aroideana 7: 56-57 . 1984b. Suprageneric names attributable to Araceae. Taxon 33: 680-690. 1987. History of aroid systematics. Aroideana 10: 23-30. РАМТ, D. D. & P. Е. Кіру. 1966. Structure of leaves Seed dispersal of Amorpho- phallus paconiifolius by birds of paradise in Papua New Guinea. Aroideana 8: 70- Р, C. 1935. Notes on Pinellia ol China. Contrib. Biol. Lab. Chin. Assoc. Advancem. Sci., Sect. Bot 1- 10: PELLMYR, O. & J.M. Patr. 1986. tion of olfact and visual = in pollination 2 res hiton Mise d canum (Araceae) by a staphylinid beetle. Madrono 3: 47 54. PERIASAMY, К. & E. A. id dae d Meis bn togeny of аа oa compound lea 3. о de ЕЕ oc. Indian re Sd. (Plant Sci. 96: Prose ү А arca a е of Ara- eae. Nordic J. Bot. 9: 119 Volume 77, Number 4 1990 Grayum 695 Evolution and Phylogeny of Araceae PIJL, L. VAN DER. 1937. Biological and ae observations on the inflorescence of Am l- lus. кү Trav. Bot. Néerl. 34: 157- 167. А On the flower biology of some plants a va. Ann. Bogor. 1: 77- 1931. Ana tomische und ökologische Unter- u en am Blütenstand von Philodendron Sel- loum Schott, mit ade erer Berücksichtigung der Harzkanále und der Beschaffenheit der Pollenkitt- stoffe. Planta 15: 506-529. V. 1952. Placentation in angiosperms. Bot. Rev. (Lancaster) 18: 603-651. N, K. 1978. M rs vu on south ian Araceae. Cytologia 43: 289-3 кушш: B., W. Informe ao bon 18: 367-368. 1987. Informe cientifico. Brenesia 28: 117- . D. Gómez P. 1978. Production of nectar and gums by flowers of Monstera deliciosa (Araceae) and of some species of Clusia (Guttiferae) Ге 6 by New World Trigona bees. Bened а 14/15:4 REN P. H. 1975. The bases of angiosperm phylog eny: cytology. Ann. Missouri Bot. Gard. 62: 124. 76 e 1979. Plate d fes southern hemi- sphere biogeography. n K. Larsen & L. B. Holm-Nielsen (editors), Tropical Botany. Academ- ic Press, London. & D. I. AXELROD. 1974. Angiosperm bio- geography and iri о movements. Апп. Missouri Bot. Gard. 6 3. Ray, T. S. 1987a. Let ype in the Araceae. Amer. J. d 74: 1359- 87b. Diver a shoot organization in the a Amer. J. Bot. 74: 1373-1387. 1 . Survey ut phe organization in the Araceae. Amer. J. Bot. 75: 56-84. 1989. Diversification of growth habits in the Araceae. Amer. J. Bot. 76(suppl.): 267. жый к. E. Т. 1913. The differentiation and speci- ficity of starches in relation to genera, cee etc. pne Institute of Washington Publ. 17 REUMER, J. W А boire di d evolution in Cryptocoryge (Araceae). Genetica 65: 149-158. йй. Н. 1978. Не inrich "Wilhelm Schott as a devel. opmental morphologist. Aroideana 1: 77-85. RITTERBUSCH, А. 19 Morphologische Untersuchun- gen zur Wuchsform von Philodendron. Bot. Jahrb. The ecological life history of Ari- saema atrorubens (Ait.) ce M.S. Thesis. Purdue , W. Lafayette, Indian Rosina С.О. 19 6e aee on morphology of underground stems of V ce e and Wa m Minnesota Bot. Stud. IV, Bot. Ser. . Pollen movement and reproduction in Arisaema triphyllum. Bull. Torrey Bot. Club 107: 539-542. — J.H. 1937. The те e - golden-club. Bot. (Binghamton) 43 SCHLITTLER, 1. 1955. Уос Miel bare die re ichti tig- sten Grundgestalten bei den le aac eine mit spezieller Berücksichtigung der Liliaceen und einiger Bemerkungen zu deren оар Viertaljahrsschr, Naturf. Ges. Zürich 100: 193. Rm ematis Aroidearum. Kafer TOM ). Z. Wiss. Insektenbiol. 6: 67- emarginata Mann.). 68. pns se F. 1878. Uber die Entstehung der Locher S., М.С. Respiratory gas exchange during thermo- genesis in е selloum C. Koch. Planta 161: 229-2 SHa w, D. E., A. aaa & К. A. HILLER. ocasia е and birds іп D Eo idea r SHEFFER, R. EA & T. B. CROAT. 1983. Chromosome = (A X IT Amer J. Bot. 70: 858-871. SHISHKOFF, N. 1987. Distribution of the dimorphic hy- Bot podermis A roots in angiosperm families. Ann. -15. Observações sobre a biologia reprodutiva de Pistia stratiotes L. (Araceae). Acta Amazonica 11: 487-504. SIVADASAN, М. & C. SATHISH KUMAR. 1987. A new species of Arisaema (Araceae) from India with a note on ras and evolution of staminate flowers. Aroi- deana 10: 1. SOLEREDER, Н. & F. J. MEYER. 1928. Systematische Anatomie E Mon okotyledonen. Heft 3. Principes- Synanthae-Spathiflorae. Gebrüder Borntraeger, Ber- lin STAPF, O. 1923. Amorphophallus cirrifer. Bot. Mag. 149: tab. 9000. STEBBINS, С. L. . Adaptive radiation in angio- sperms, I: pollination mechanisms. Annual Rev. Ecol. Syst. 1: 307-326. 1973. Evolutionary trends in the inflorescence of angiosperms. Flora 162: 501-528 & С KHusH. 1961. ans | in the or ganization of the stomatal complex in the leaf epi- dermis of monocotyledons and its bearing on their phylogeny. Amer. J. Bot. 48: 51-59. е A. 1980. Outline of the classification of wering pints No аш Bot. Rev. (Lan- aster) 46: 225 TEN S. 19 ‘Revisión de las especies Bo i oy el genero Biarum Sc ieee ius 6: 296. ТСНІАКРЁ, L. 1979. Contribution d'une рар africaine: анн difirmis Engl. Ph.D. Dis- 969. “Es ha ie ha ху « Bist: m Pondichéry, Trav. Sect А Tech. 5 ): l- THOMPSON, 5. А. "n E RAWwLINS. 1986. Floral ecology of Anchomanes difformis (Araceae). Amer. J. Bot. 73: 791. TIEGHEM, Р. VAN. 1907. Remarques sur унн orale еї la structure de Povule des Aracées. Ann. Sci. Nat. Bot., sér. 9, 5: 312-320. TiLLiCH, H.-J. 1985. altares und verwandtschaft- 696 Annals of the Missouri Botanical Garden liche Beziehungen der Araceae. Gleditschia 13: 63- TOMLINSON, Р. В. 1962. Phylogeny of the Scita neae — morphological and anatomical dioe реч Evolution 16: 192-213 1966. Anatomical data in the о of is J. Linn. Soc., Bot. . 197 Md E an under- standing of their morphology and anatomy. Pp. 92 in R. D. Preston (editor), Advances in Botanical Research. Academic Press, London. 1974. Development of the stomatal complex s a taxonomic character in the monocotyledons. 128. Des vaisseux propres dans les Aroi- dées (seconde partie). Compt. Rend. Hebd. Séances Acad. Sci. 62: 29-33. TRELEASE, W. 1879. On the fertilisation of Symplo- carpus foetidus. Amer. Hees 13: 580-581 TroLL, W. 1928. Über Spathicarpa sagittifolia; mit einem Anhang über die Stellung der Spathicarpeen m System rie Araceae. 23: 286-316. Una H. (Araceae); Wuchsform und Infloreszenz. Beitr. Biol. Pflanzen 57: 389-429. VocEL, S. 1962. Duftdrüsen im Dienste der Bestáu- ung; über Bau und Funktion der Osmophoren. Akad. Wiss. Abh. Mainz Math.-Naturwiss. Kl. 10: 601- 76 WALKER, J. W. Classification and phylogeny of the mono- cotyledons а гір ———— & A. С. WALKER. 1984. Ultrastructure of Lower (ша. angiosperm pollen and the origin and early evolution of flowering plants. Ann. Missouri Bot. Gard. 71: 464-5 WARMING, E. 1 Tropische Fragmente. I. Die Be- staubung von diodos ipit Schott. Bot Jahrb. Syst. 4 WILDEMAN, É. DE. : cf ee e a ш Acad. Roy. Sci. Belgique, Cl. Sci., sér. 5, 28: 169. WILDER, G. J. 1988. Inflorescence position as a taxo- nomic character in the Cyclanthaceae. Bot. Gaz. (Crawfordsville) & . HARRIS. 5 bipartitus Poit. (Crawfordsville) 143: 84 WILLAMAN, J. J. € 1982. Laticifers in Cyclan- (Cyclanthaceae). Bot. Gaz -93. 1970. Alkaloid-bearing plants and their contained alkaloids, 1957-1968. Lloydia 33(suppl.): 1-286. WILLIAMS, С. A., ]. В. е ат 5. J. Mayo. 1981. Anthocyanin pigments and leaf flavonoids i in the fam- ily Araceae. Phytochemistry 20: 217-234. WiLLIAMS, N. Н. & R. DRESSLER. Pres pollination of Spathiphyllum (Araceae). Selbyan 49-356. WiLLis, J. В. 1949. The Birth and Spread of Plants. Chronica TE Co., Waltham, Massachusetts. Wrr, H. C. D. 978. Revisie van het genus La- genandra Dale pu Meded. Landbouwho- geschool 78-1 WoLFE, J. A. pur aspects of plant кыш? of the Northern Непизрһеге during the Late tacai and Tertiary. Ann. Missouri Bot. Gard. 62. 79, £s A. DovrE & V. M. Pace. 1975. The bases of angiosperm Phylogeny: oe Ann. Mis- souri Bot. Gard. 62: Younc, Н. J. 1986. Beetle кы of ЕИН longispatha (Araceae). Amer. J. Bot. 73: 931- APPENDIX 1. (1973 The aroid classification of Hutchinson Tribe Acoreae Acorus L. ee R. Br. Tribe Orontie Orontium L Lysichiton Schott e Spathiphylleae athiphyllum Schott "E Anthurieae Anthurium Schott Tribe Dracontieae 29%! 33% 25% 33% 29% он Engl. ymploc > 7 y» =й UT RS CU er V o Spo Куб ў WW ee a S A q ‹ Qs ^. TS VEZ LAO а J / f 2. (am « y* = EE Ç e oye EUN 7 E >=, x t - A ES g {Б A МШЕ: "v 2 a - ыў. > zc ы Oy Babe - 2 7 et e E UN С CES pe ISE m У) "A KO) N ja DEA AE Ь FAMILY MUS SA y JA % YEr ох! EE DNI iet PARADISI, ACA ANT TAIN VARIE TlES мА) D е / A ы SA , (8 = = 7j с i YP. | 2 be R35 х; й D -— —! CHN > C : ы TEAS, d AAA "Rhizogram" of the families of the е Rhizomatous patterns of growth аге used to represent FIGURE 2. evolutionary branching relationships (from Reynolds, (Heliconiaceae, Musaceae, and Strelitziaceae) with raphide sacs, symmetrical guard cells, and anom- alous root structure (the last only in Musaceae and Strelitziaceae); Group 2 (Costaceae, Marantaceae, and Zingiberaceae) with asymmetrical guard cells, lack of raphide sacs, and normal root structure; Group 3 (Cannaceae) with symmetrical guard cells (variable), no raphide sacs, and normal root struc- ture; and Group 4 (Lowiaceae) with raphide sacs, 1927). asymmetrical guard cells, and normal root struc- ture. Tomlinson then went on to discuss evolutionary trends in floral and vegetative characters as a basis for understanding the historical relationships of the four groups in the order. He concluded that the *... Strelitziaceae was the most primitive family within the Scitamineae, in the sense that its mem- bers possess the greatest number of primitive floral Volume 77, Number 4 Kress 707 199 Phylogeny and Classification of Zingiberales Heliconiaceae RAPHIDE -SACS Musaceae yo. N ANE 7 Cannaceae ~ — __Proto-Scitamineae yn | Qo ае á ^ / ( ` . РЕ. st / | P» ; V D / \ Marantaceae Cooking . \ АЅҮММЕТЕІС a ` H D GUARD -CELLS Zingiberaceae Р = — = uad 2—7 Tomlinson's two-dimensional as of the evolutionary relationships of the eight families of the RE Zingiberales (from Tomlinson, 1962, his fig. features. Thus existing scitaminean flowers [and though somewhat ambiguous in the correspondence other vegetative and anatomical characters] can between the classification and its diagrammatic rep- be regarded as derivative and advanced in varying resentation, was the first phylogenetic analysis of degrees, compared with the strelitziaceous flower." the eight families of the order that was based on Based on the assumption that the Strelitziaceae a methodical examination of character distribution, were the оні toa prote: -scitaminean” ancestor, His use of a priori evolutionary “trends” in floral Tomlinson lly represented the evolu- traits (e.g., the designation of the strelitziaceous tionary relationships of the families (Fig. 3). How- flower as primitive) to develop concepts of primitive ever, the relationships depicted in his figure do not and advanced states of other features may be crit- correspond to his four groups. In Group l, the — icized as a questionable method of character po- Heliconiaceae, Musaceae, and Strelitziaceae are larization. In addition, the relationships of the Mu- positioned together in the figure, but each is оп а saceae, Heliconiaceae, and Strelitziaceae were never separate line directly from the proto-scitaminean fully resolved in his work. The analysis, however, ancestor. Group 4, the Lowiaceae, is set in between provided a concise rationale for linking the Mar- Group 1 and the other families on its own distinctive — antaceae to the Cannaceae and the Zingiberaceae line from the ancestor. Group 2 is divided into two to the Costaceae, as well as for the distinct position separate lines. Two families of the group, Costaceae іп the order of the Lowiacea and Zingiberaceae, share a common branched line. Bisson et al. (1968), as "i result of a lengthy The third family of Group 2, the Marantaceae, — caryological analysis of the Zingiberales, supported shares а common line with the only member of — Tomlinson's (1962) inferences on the evolutionary Group 3, the Cannaceae. The affinity of these last relationships of the eight families. They informally four families (plus the Lowiaceae) is depicted on proposed three suborders: (1) Strelitziaceae, Mu- the figure by a broken circle drawn around them, ѕасеае, and Heliconiaceae; (2) Zingiberaceae and indicating the possession of asymmetrical guard Cosaceae; and (3) Marantaceae and Cannaceae; cells (apparently rer including the Can- excluding the Lowiaceae due to the lack of ade- naceae). Likewise, the a y of the families of quate cytological data. Group 1l (plus pi is {е by a circle representing the presence of raphide sacs. Dahlgren & Rasmussen’s analysis. The most re- Tomlinson's publication on the Scitamineae, al- cent analysis of the phylogenetic relationships of 66 708 Annals of th Missouri „ш Garden Maranthaceae Zingiberaceae : Cannaceae Costaceae Heliconiaceae Musaceae : Strelitziaceae vo © Ф © © 3 o — FIGURE 4. fig. 9) the Zingiberales was by Dahlgren & Rasmussen (1983) in their publication on the evolution of the monocotyledons. They were not interested in the relationships of the order per se, but rather chose the group to demonstrate their method of cladistic analysis, which they advocated for an evaluation of the relationships of the monocotyledons as a whole. They selected the Zingiberales, accepting the eight families proposed by Nakai (1941), as an example “because of its anchalenges status as a monophyletic group . Dahlgren & Ron Коней the basic pro- cedures of Hennigian phylogenetic systematics by (1) selecting an outgroup, (2) polarizing characters of the ingroup based on the states present in the outgroup, and (3) grouping taxa according to shared derived character states (synapomorphies). They recognized that no unequivocal outgroup could be selected for the Zingiberales, but chose their Com- meliniflorae (discussed earlier) as the best possible outgroup. The morphological and anatomical char- acters described by Tomlinson (1962, 1969) pro- vided the basis for the 40 two-state character set for the analysis. Their cladogram was constructed by “hand” according to a method described in the “The best-defined section of the dax inae gram (Fig. 4) was the monophyletic “ginger group" (Zingiberaceae, Costaceae, Marantaceae, and Can- naceae) united by the lack of raphide sacs, single fertile stamen, and abundant perisperm. Two pairs of sister groups within the ginger group were also each defined by several apomorphies: the Costa- Cladogram of the Zingiberales resulting from the analysis of Dahlgren & Rasmussen (1983, their ceae + Zingiberaceae by the fused staminodes; the Cannaceae + Marantaceae by the asymmetric flowers, bisporangiate anther, and specialized petal- oid staminodes. Within the remaining four families of the order (the “‘banana group"), the Musaceae and Heliconiaceae were shown to be sister taxa based on the unique presence of the perianth tube, and the Strelitziaceae and Lowiaceae were united by their distichous phyllotaxy. However, due to the conics "presenten i the distribution of two char- stegmata), some doubt was expressed as to Miete. the banana group formed a separate clade distinct from the ginger group. The thick-walled silica cells (steg- mata) shared by the four banana families suggests that they are a monophyletic group. The presence of a terminal inflorescence in the Heliconiaceae, Musaceae, and the members of the ginger group unite those taxa. Dahlgren & Rasmussen consi ered the stegmata as the * apomorphy" and hence supported the monophyly of the banana group. Dahlgren & Rasmussen's phylogeny of the Zin- giberales does not conflict significantly with the scheme proposed by Tomlinson (1962). Both anal- yses accepted the two sister-group relationships within the ginger group (Marantaceae + Canna- ceae; Costaceae + Zingiberaceae), and both con- sidered the relationships of the four families of the banana group as somewhat unresolved. The mon- ophyly of the four families of the ginger group is strongly supported by Dahlgren & Rasmussen. They also provided evidence for uniting as sister groups *more significant syn- Volume 77, Number 4 1990 Kre 709 Phylogeny and Classification of Zingiberales STR HEL MUS LOW ZIN COS MAR CAN STR HEL MUS LOW ZIN COS MAR CAN e Ur STR LOW HEL MUS ZIN COS MAR CAN LOW STR HEL MUS ZIN COS MAR CAN Tu du STR LOW HEL MUS ZIN COS MAR CAN MUS HEL STR LOW ZIN COS MAR CAN C D STR LOW HEL MUS ZIN COS MAR CAN G E 5. Seven equally parsimonious trees from Analysis One in which the character set of Dahlgren & FIGUR Rasmussen (1983) was used. the Heliconiaceae and Musaceae as well as the Lowiaceae and Strelitziaceae, but questioned the strength of this evidence. Their analysis is more explicit than Tomlinson’s in its statements on the cladistic relationships of the families and the char- acters that define the monophyletic groups. Neither of the investigations suggested any new formal ranks. Because of the formalized, explicit construction of the cladogram of Dahlgren & Rasmussen, it is possible to follow clearly their methods and provide a critique of the phylogeny. A close look at their data matrix reveals several problems, inconsisten- construct the cladogram were coded incorrectly (e.g., characters 4 and 5: “uniseriate” is confused with *unicellular" trichomes; character 23: free median petal is found in the Musaceae and in the Strelitziaceae; character 30: all staminodes fused is an autapomorphy for the Costaceae, not the Zingiberaceae). Sixteen characters are coded (in some cases incorrectly) as being present in one family only, i.e., as autapomorphies for the terminal taxa, and therefore are of no value in determining relationships among the taxa (character 2: pseu- dobulbs; character 7: oil cells; character 9: artic- ulated lacticifers; character 10: mucilage canals; character 11: axillary inflorescence; character 15: two-flowered inflorescences; character 17: resu- pinate flowers; character 21: corolla tube; char- acter 23: free median petal; character 27: median stamen of outer whorl missing; character 30: all staminodes fused; character 32: hypanthiumlike neck on ovary; character 33: petaloid style; char- acter 34: less than three fertile locules per ovary; character 35: basal placentation; character 40: schizocarp). Of course autapomorphies are of crit- ical importance in defining the monophylesis of the terminal taxa, but they provide no information on cladistic relationships. Another problem with the character analysis is the coding of characters that are variable or polymorphic within taxa (e.g., char- acter 8: stegmata; characters 11, 12: inflorescence type; characters 24, 27: stamen number; character 34: number of locules per ovary). Polymorphic characters can be legitimately used in cladistic analysis if evidence is provided for designating which of the states is plesiomorphic in the taxon. Dahlgren & Rasmussen did not provide this information. Finally, misinterpretation of some character state homologies (especially in perianth features; see be- low) led to inaccurate coding in several taxa. The most significant deficiency of the phylogeny is the lack of any demonstration that the arrange- ment of taxa and characters in the cladogram is the most logically acceptable or parsimonious one. No justification was provided as to why this clado- gram is the best representation of the relationships of the taxa based on the distribution of the available characters. To test the hypothesis that the clado- gram presented by Dahlgren & Rasmussen is the most parsimonious, the same 24 characters (the 16 autapomorphies listed above were omitted) with 710 Annals of the Missouri Botanical Garden no recording or changes in polarization were rean- alyzed using a computer-assisted maximum parsi- mony method (D. Swofford’s PAUP program; see below). This method for inferring phylogenies, which places no restrictions on character state changes and estimates minimum length trees, is based on the same principles as the method used by Dahlgren & Rasmussen. Seven equally parsimonious clado- grams (each with the same number of character state changes) were found (Fig. 5). The topology of each of the seven cladograms, which included the arrangement(s) presented by Dahlgren & Ras- mussen (Fig. 5C, G), differed significantly in the relationships of the eight families. The sister-group relationships of the Heliconiaceae + Musaceae and the Marantaceae + Cannaceae were the only two consistencies in all seven of the cladograms. These results indicate that the analysis by Dahlgren & Rasmussen was not only incomplete, but that the character set they used is not sufficient (with the included inaccuracies) to determine unambiguously the phylogenetic relationships of the families of the Zingiberales. THE PHYLOGENY OF THE ZINGIBERALES The phylogenetic analysis performed by Dahl- gren & Rasmussen (1983) was the first attempt to infer historical relationships of the eight families of the Zingiberales using the methods of phylo- genetic systematics. However, as pointed out above, their results are suspect due to certain flaws in the analysis. Nevertheless, their cladistic methods were sound, and sufficient documentation was provided to repeat the analysis and test their hypothesis of the phylogeny of the Zingiberales. An independent, original analysis of the Zingi- berales using cladistic methods was initiated in the present investigation to clarify the conflicts, incon- sistencies, and omissions of previous investigations. The aim of the study was to devise an unambiguous character set that would allow the estimation of a fully resolved cladogram to serve as the basis for a new hierarchical classification reflecting phylo- genetic relationships. MATERIALS AND METHODS The eight families of the Zingiberales proposed by Nakai (1941) and accepted by Tomlinson (1962, 1969), Cronquist (1978, 1981), Dahlgren & Ras- mussen (1983), and Dahlgren et al. (1985) were used as the operational taxonomic units in the present analysis. The autapomorphies listed earlier in the individual family descriptions are accepted as evidence for the monophyly of each of the families. The characters used for inferring relationships were taken from a number of sources, including my own published and unpublished investigations. Lane (1955), Tomlinson (1956, 1959, 1961, 1962, 1969), and Dahlgren et al. (1985) were consulted for basic morphological and anatomical characters. Conflicts in the distribution of character states among these references were resolved by original observations where possible or by further reference to the works of other workers on specific families or characters, e.g., Humphrey (1896), Gatin (1908), Cheesman (1947), Holttum (1950, 1970), Simmonds (1962), Larsen (1966), Bisson et al. (1968), Mahanty (1970), Burtt (1972), Maas (1972, 1977), Wagner (1977), Williams & Har- borne (1977), Goldblatt (1980), Olatunji (1980), Andersson (1981), Barthlott € Frólich (1983), Kirchoff (1983a, b, 1986), Kress (1984), and Rog- ers (1984) In the Zingiberales, as is true in most plant groups, many characters are variable within the family. Unless the primitive state of a polymorphic character is known a priori, that character cannot be unequivocally coded and used in the analysis (Mickevich & Mitter, 1981). For this reason some variable characters were omitted from the analysis, e.g., cytology, stomatal type, septal nectaries, en- dosperm type, embryo shape, fruit type. Additional characters that may provide evidence on relation- ships, such as leaf flavonoids, are not known for all families (and are often variable in those taxa in which they are known) and were not included in the analysis in most cases. After attempting to assign states to more than 50 characters of the eight families, 32 characters (primarily floral mor- phology and vegetative anatomy) initially appeared unambiguous or “solid” enough to incorporate into the analysis (Table 2). All characters were defined to have two states to avoid the problems of devising a transformation series for multistate characters. Autapomorphies of the families, listed earlier in the family descriptions, were not included in the anal- is. Pollen characters were specifically omitted from the analysis even though a considerable amount is now known about the pollen of various families (e.g., Muller-Stoll, 1956; Saad & Ibrahim, 1965; Erdtman, 1966; Punt, 1968; Skvarla & Rowley, 1970; Kress et al., 1978; Stone et al., 1979, 1981; Kress & Stone, 1982, 1983; Hesse & Waha, 1983; Kress, 1986; Stone & Kress, unpublished). This information was not included because the cladogram resulting from the present analysis will Volume 77, Number 4 Kress 711 1990 Phylogeny and Classification of Zingiberales TABLE 2. Characters used in Phylogenetic Analyses TABLE 2. Continued. Two and Three. Character Character states Character Character states 1. Anticlinal walls of leaf epidermal cells 2. Leaf guard cells 3. Leaf adaxial hypo- dermis + D ‚ Leaf longitudinal veins +5. Leaf veins transverse +6. Transverse vein sheathing cells 7. Air canals in leaf 8. Root stele 9. Raphide sacs 10. Internal silica cell bodies — hat- shaped 11. Internal silica cell bodies — trough- shaped 12. Internal silica cell i use shaped +13. oo cells silica bodies 14. Vessels 15. Phyllotaxy 16. Flower shape +17. Sepals and petals fused into tube ++18. All sepals ++19. All petals +++20. Median petal 21. Fertile stamen number 22. Inner whorl medi- an stamen 23. Outer whorl medi- an stamen sinuous (0)! not sinuous (1) symmetrical (О)! asymmetrical (1) >1 cell layer (0) 1 cell layer (1) independent of leaf surface (0)! attached to leaf surface (1) sheathed (0)! not sheathed (1) thick-walled (0)! thin-walled (1) 1 arc (0)! 2 arcs (1) polyarc (0)! with medullary vessels & phloem (1) present (0)! nt (1) absent (0)! present (1) absent (0)! present (1) absent (0)! present (1) present (0)' absent (1) roots & stems (О)! roots only (1) spirally arranged (0)! distic hous (1) zygomorphic (0)! asymmetric (1) no (0)! yes (1) not fused (0)! fused at least at base (1) not fused (0)! fused at least at base (1) free (0) fused to lateral petals (1) =5 (0)! 1 (1) present/fertile (0)! fertile (О)! not fertile (1) 24. Outer whorl medi- present (0)! an stamen absent (1) 25. Outer whorl lateral fertile (О)! stamens staminodia (1) 26. Inner whorl lateral fertile (О)! stamens staminodia (1) 27. Staminodes not fused (0)! variously fused for most of length (1) tetr ipie (0) bispora 28. Anther(s) 29. Style т p modified (1) 30. Ovule placentation — axile (0)' basal (1) 3l. Aril absent (0)! present (1) +++32. Well-devel- absent (О)! oped peri- present (1) sperm helobial (0)! nuclear (1) absent (0)! present (1) 33. Endosperm 34. Chalazosperm ' Character state present in —€— M and coded as primitive in the Zingibera + +++ Added to Аан Pies. be used to infer the evolution of pollen in struc- ture in the Z ] subsequent i (Kress & Sune: unpublished; see below). The Bromeliiflorae of Dahlgren et al. (1985) was selected as the outgroup of the Zingiberales and the characters were polarized accordingly. The state common to the two groups was coded as the primitive state. For characters variable in the Bro- meliiflorae (15, 17, 21), the state present in the Bromeliales (Bromeliaceae) was chosen as the prim- itive state in that superorder. In several cases, char- acter states were bur adi: or nonexistent for some taxa and were coded as “missing.” The computer program PAUP (Phylogenetic Analysis Using Parsimony; Swofford, 1985 used to infer the most parsimonious phylogeny. The “branch and bound” option (BANDB) of PAUP, which guarantees that the shortest trees will be found (if fewer than ten taxa), was used in con- junction with the ancestor rooting option (ROOT 712 Annals of the Missouri Botanical Garden n by character matrix for the eight families of the Zingiberales and 34 Thr TABLE 3. Tax "imu Abs Two and characters used in Character Family 2 3 4 5' 6 7 8 9 10 11 12 12 13: 14 14° 15 16 17 Lowiaceae 1 1 1 0 0 о 1 0 0 1 O 0 0 1 1 1 1] 0 0 Musaceae 1.0 о 1 l 9.0 1 0 0 1 0 l l 1 0 0 1 Heliconiaceae 0 0 1 0 0 1 1 0 0 0 1 0 0 1 1 1 1 0 1 Strelitziaceae 9 0 0 l 1 9 | 1 о 0 0 1 1 0 о 0 1 0 0 Zingiberaceae 1 1 1 1 0 1 0 0 1 о 0 9 0 0 ] 0 1 0 0 Созїасеае 1 1 0 0 0 0 0 0 1 0 0 1 1 1 1 1 0 0 0 Саппасеае 1 0 1 1 0 1 0 0 1 0 0 1 1 1 1 1 1 1 0 Marantaceae oO |] 1 1 0 1 0 0 1 1 0 0. о 1 0 о 1 1.0 O = Primitive state. 1 = Advanced see 9 = Missing char ! = Omitted in jupes Three. = Recoded in Analysis Three. * — Added to Analysis Three. — ANCESTOR) and the Farris method of HTU character state optimization (OPT — FARRIS). Three separate analyses were run. The first was a test of the parsimony of the phylogeny derived by Dahlgren & Rasmussen (1983) and used the same characters (24, excluding the 16 autapo- morphies) and polarity coding. The second and third analyses incorporated 34 independently de- rived characters (autapomorphies also excluded; Tables 2 and 3) and used the Bromeliiflorae as the outgroup. After construction of the cladogram from the first independent character set (Analysis Two), character evolution was traced on the cladogram. As a result of the patterns revealed by the clado- m, each character was then reevaluated for defendable hypotheses of homology and accurate gra coding. Five characters (4: longitudinal veins; 5: leaf transverse veins; 6: transverse vein sheathing; 13: superficial cells with silica bodies; 17: sepal and petal fusion) could not be unambiguously coded because of variability in the outgroup, variability within families of the ingroup, or initial inaccurate orphological examination, and were omitted. Five characters (12: druse-shaped silica bodies; 14: ves- sels; 18: fusion of the sepals; 19: fusion of the petals; 23: outer whorl medium stamen) required recoding because of original mistakes in coding due to faulty literature reports. Two new informative characters (20: fusion of median petal; 33: peri- sperm) that were not included in the first original analysis were found. As a result of this character reevaluation, a second analysis (Analysis Three) was performed using the same PAUP options and a revised data set with characters omitted, recoded, or added as just described (29 total characters; Tables 2 and 3). RESULTS Analysis One: reevaluation of Dahlgren & Ras- mussen study. The reanalysis of the character set of Dahlgren & Rasmussen, as discussed above, yielded seven equally parsimonious trees (Fig. 5), each with a total of 41 evolutionary steps (including 17 homoplasies; Consistency Index — 0.585). The seven trees included the two cladograms (Fig. 5C, G) of their analysis. Analysis Two. The first of the independent anal- yses using the original set of 32 characters polar- ized with the Bromeliiflorae produced three equally parsimonious trees (Fig. 6A-C), each with 63 char- acter state changes, including 31 homoplasies (Consistency Index = 0.508; F-value = 6.164- 6.480). Each of these cladograms is fully resolved with all branching points dichotomous. The largest clade consistent in all three trees includes the Low- iaceae, Costaceae, Zingiberaceae, Cannaceae, and Marantaceae with the same sister group relation- ships in each tree. Autapomorphies of the Mar- antaceae-Cannaceae clade are asymmetric flowers (character 16), unfused staminodes (character 27), bisporangiate anthers (character 28), and a mod- ified style (character 29). This clade is united with the Zingiberaceae by the absence of the median stamen in the outer staminal whorl (character 24). Synapomorphies of the clade formed by the Cos- taceae and Zingiberaceae-Cannaceae- Maranta- ceae are the absence of raphide sacs (character Volume 77, Number 4 Kress 713 1990 Phylogeny and Classification of Zingiberales TABLE 3. Continued. Character 18 18: 19 192 203 21 22 23 23 24 25 26 27 28 29 30 31 32 33 34 1 то 0 0 0 1 0 0 0 0 0 9 0 0 0 1 0 9 0 0 1 то 0 0 10. 0 0 0 0 9 0 0 о 0 0 1 0 0 0 1 1 1 0 oO 1 1 0 0 0 9 0 0 1 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 9 0 0 0 1 о 1 O0 1 1 0 1 1 1 0 1 9 1 1 1 1 0 0 о 1 1 0 0 l 1 0 1 1 1 O 1 1 0 1 1 1 0 0 о 1 1 0 1 0 0 1 1 1 1 o 1 9 1 1 1 0 1 1 0 O0 1 1 1 0 0 1 1 1 1 o 1 9 1 1 1 0 1 1 1 1 1 1 1 9), presence of a single fertile stamen (character 21), and staminodia in the lateral positions of the inner and outer staminal whorls (characters 25, 26). The Lowiaceae are united to this four-family clade by the asymmetric guard cells (character 2; lost in the Cannaceae) and the fused sepals (char- acter 18; not shared by the Cannaceae—Maran- taceae clade; this character was recorded for Anal- ysis Three). The variable lineages of the three equally par- simonious trees in Analysis Two are due to the unstable positions of the Musaceae, Heliconiaceae, and Strelitziaceae. In the first tree (Fig. 6A), the nig Ре бын й а о г by the trough-shaped internal ѕШса cell bodies (character 11) and the fused sepals and petals (character 17; this misinterpreted character 15 omitted in Analysis Three), as well as several other homoplasious characters (character Ó: transverse vein sheathing cells; character 19: petals; char- acter 31: no aril). In the other two trees (Fig. 6B, C) the Musaceae and the Strelitziaceae form a monophyletic lineage based on the shared un- sheathed transverse leaf veins (character 5; vari- able and omitted in Analysis Three), the root stele with medullary vessels and phloem (character 8), and three homoplasious characters (character 3: multilayered leaf adaxial hypodermis; character 4: attached leaf longitudinal veins; character 22: in- ner whorl median stamen absent). No uniquely derived characters unite any of these lineages to the Lowiaceae—Costaceae—Zingiberaceae—Canna- ceae—Marantaceae group in the three trees. In all three of the trees, four characters change states between the outgroup (the Bromeliiflorae) and the common ancestor of the Zingiberales, and hence are additional synapomorphies of the order: two arcs of air canals in the leaf axis (character 7), distichous phyllotaxy (character 15), staminodia fused (character 27; not applicable to four of the eight families that possess one or no staminodium), and nuclear endosperm (character 33). Among the three trees there are 11 uniquely derived character states, mostly floral traits, nine of which are common to all trees (character 9: raphide sacs absent; character 16: asymmetric flowers; character 21: one fertile stamen; character 24: outer whorl median stamen absent; character 25: outer whorl lateral stamens sterile; character 26: inner whorl lateral stamens sterile; character 27: staminodes not fused; character 28: bispor- angiate anthers; character 29: modified style). Analysis Three. In the reanalysis of the original 32 characters in which characters were omitte (characters 4, 5, 6, 13, 17), recoded (characters 12, 14, 18, 19, 23), or added (characters 20, 32), a single shortest tree (Fig. 7) was found with a total length of 53 steps, including 24 homoplasies (Con- sistency Index = 0.547; F-value = 4.939). Fifteen characters are uniquely derived (character 8: poly- arc root stele; character 9: raphide sacs absent; character 16: asymmetric flowers; character 19: petals fused at base; character 20: median petal fused to lateral petal; character 21: one fertile stamen; character 22: inner whorl median stamen present; character 23: outer whorl median stamen not fertile; character 25: outer whorl lateral sta- mens sterile; character 26: inner whorl lateral sta- mens sterile; character 27: staminodes fused; char- acter 28: anther bisporangiate; character 29: style modified; character 32: well-developed perisperm; character 33: helobial endosperm). The cladogram is fully resolved with all branching points dichot- omous. The Cannaceae and Marantaceae form a terminal monophyletic lineage defined by asym- metric flowers (character 16), bisporangiate an- thers (character 28), and a modified style (char- acter 29). The Zingiberaceae and Costaceae, a second terminal group, are united by the uniquely Annals of the 714 Missouri Botanical Garden e d ES xs & «X y S « « С « « « С^ Y SS С & ss С СЯ & > SEAS FP PE a a um FF FPP KM Т Ar e? К E SS e ruf d Ку КЧ X SU SIS SY SUV E & X QUAS QN ar uS WL’ S S quar 9 ^ Oe су SVN © 3>0 d 2>0 Uu 1051 L 1, 1 320 13>0 2>0 1 A 1 7>0 A 1 J u 1»0 10>1 Т qd 1 15 >0 31>0 4» р 521 1620 3450 — 31x 1021 2221 $5! п> 3>0 15>0 3420 3>0 10>1 1121. 1350 N> B>1 11>1 0 »0 12>0 1550 14>0 7 5>1 22>1 А = 3151 3351 FiGURE 6. the left с ot the arrows (> apomorphic character states at that no ers listed in Ta derived fused staminodes (character 27) and he- lobial endosperm (character 33) as well as the homoplasious fused sepals (character 18; also in the Lowiaceae). Synapomorphies of the clade made up of these four families are the loss of raphide sacs (character 9), reduction in fertile stamens to one (character 21), staminodia in the lateral po- sitions of the inner and outer whorls of stamens (characters 25, 26), and the well-developed peri- sperm (character 32). The single arc of air canals in the leaf axis (character 7; also in Musaceae), absence of the median stamen in the outer staminal whorl (character 24; regained in Costaceae), the presence of chalazosperm (character 34; lost in Zingiberaceae) are homoplasies that also unite these families. The Heliconiaceae are ihe sister group of the four-family "ginger group," sharing the basally fused petals (character 19), the median petal fused to the lateral petals (character 20), the presence of the median stamen of the inner staminal whorl (character 22), and the infertile stamen (i.e., stam- inode) in the median position of the outer staminal whorl (character 23), all of which are uniquely bles 2 and 3; numbers to the right of the arrows are Three equally Jeep usd bep e of the Zingiberales resulting from Analysis Two. Numbers i derived states on the tree. The Lowiaceae are joined to this five-family clade on the basis of the shared polyarc root stele (character 8), asymmetrical guard cells (character 2; symmetrical in the Cannaceae), and the one cell-layered leaf hypodermous (char- acter 3; multiple layers in the Costaceae). The sister taxon of the lineage formed by the Lowiaceae-Heliconiaceae-Costaceae- Zingibera- ceae-Cannaceae-Marantaceae is the Strelitzia- ceae. Three homoplasious character states unite them: two arcs of air canals in the leaf axis (char- er 7; reversed to one arc in the ginger group), distichous phyllotaxy (character 15; reversed in the Costaceae), and the arillate seeds (character 31; lost in the Heliconiaceae and Cannaceae). These seven families share a common ancestor with the Musaceae, the eighth family of the order. Five character states derived in the common ancestor are homoplasious in the order: nonsinuous anticlinal walls of the leaf epidermal cells (character 1; independently becoming sinuous in Heliconi- aceae and Marantaceae), root stele with medullary vessels and phloem (character 8; polyarc in the Lowiaceae-Heliconiaceae-ginger group), vessels Volume 77, Number 4 1990 restricted to the roots (character 14; vessels in- dependently reappearing also in stems in the Stre- litziaceae, Zingiberaceae, and Marantaceae), ab- sence of the median stamen of the inner staminal whorl (character 22; regained in the Heliconi- aceae-ginger group), and nuclear endosperm (character 33; helobial in Zingiberaceae—Costa- ceae). DISCUSSION Character analysis. Continual reassessment and reexamination of character state identification, dis- tribution, and homology is the basis of any phy- logenetic analysis. In the present investigation of the Zingiberales, the problem posed by the seven conflicting but equally parsimonious trees (Fig. 5) derived from the Dahlgren & Rasmussen character set was partly resolved by the reevaluation of their ata. The three cladograms subsequently con- structed in Analysis Two (Fig. 6) forced a further evaluation and reanalysis of the original characters leading to the single shortest tree produced in Anal- ysis Three (Fig. 7). Floral characters have presented much poni in assigning probable homologous states. For ample, Dahlgren & Rasmussen (1983) е interpreted all three perianth characters that they used to unite the Musaceae and Heliconiaceae (their character 19: all sepals fused with petals into a tube; character 20: some sepals not part of perianth tube; character 22: petals part of a perianth tube). Although there is a ““perianth tube” in both fam- ilies, in the Musaceae it is made up of three fused sepals and two fused petals (the median petal free), while in the Heliconiaceae it is formed by two fused sepals and three fused petals (the median sepal is free). These **tubes" are certainly not homologous structures, and hence cannot unite the two families. In none of the families of the Zingiberales are “all sepals fused with [all] petals into a perianth tube" as they have defined their character number 19. The “corolla tube” (their character 21), assigned by Dahlgren & Rasmussen only to the Maranta- ceae, is also present in the Zingiberaceae, Costa- ceae, Cannaceae, and Heliconiaceae and is a syn- apomorphy of those five families according to Analysis Three. The position in the two staminal whorls of the modified or missing stamens is also an important character that must be coded carefully. In the independent analyses conducted here, each posi- tion was designated as a separate character in assigning homologies. The sterile or lost stamen in the median position of the outer staminal whorl (characters 23, 24) and the fertile stamen in the Kress 715 Phylogeny and Classification of Zingiberales & < < v v v © € С C C & © L q OSO X SRA ко ҮС voa (QV С AU NS A # Ws о? c gU С ү? SOC ^ C v o > Eo! dp g!d d gp od d м 2>0 14>0 3>0 1 10>1 12>1 11> FIGURE 7. Most parsimonious cladogram of the Zin- giberales resulting from Analysis Three. Numbers to the left of the arrows (>) refer to characters listed in Tables 2 and 3; numbers to the right of the arrows are apo- morphic character states at that node. median position of the inner staminal whorl (char- acter 22) are thus interpreted as synapomorphies of the Heliconiaceae, Zingiberaceae, Costaceae, Cannaceae, and Marantaceae. The progressive evolutionary modification of the stamens into stam- inodes can be interpreted as beginning with the alteration of the outer median stamen in the Hel- iconiaceae followed by the outer and inner lateral stamens in the common ancestor of the other four families (the ginger group). The loss of the inner median stamen in the Musaceae, Strelitziaceae, and Lowiaceae is thus either an apomorphic loss in the common ancestor of the order that was regained in the common ancestor of the Heliconiaceae + ginger group, or three independent losses in those three families. The fact that at least one member 716 Annals of the Missouri Botanical Garden TABLE 4. Linnaean classification of the Zingiberales polyarc root stele (character 8) is interpreted as a based on the cladistic relationships expressed in Figure 7 using the sequencing convention of Nelson (1972, 1974) and Wiley (1981). Superorder Zingiberiflorae Dahlgren, Clifford € Yeo Order Zingiberales Nakai Suborder Musineae Se subord. nov. mily Musace ‚ L. Juss Suborder Strelitzineae "eh albanil nov. Family Strelitziaceae Hutchinson Suborder Lowineae Kress, subord. Family айан» Ridley Suborder Heliconineae Kress, subord. nov. Family Heliconiaceae Nakai Suborder Zingiberineae Kress, subord. nov Superfamily Zingiberariae Кор, paro, nov. Family ES Lin ley Family Costac a эши Cannariae Kress, ка поу. mily Cannaceae A. L. Juss Family Marantaceae Been of the Strelitziaceae (Ravenala) and possibly the Musaceae (Ensete; Dahlgren et al., 1985) possess this sixth stamen supports the second hypothesis and suggests the necessity of recoding character 21 (number of fertile stamens) in these two families. In the Costaceae and the Zingiberaceae the staminodes are variously fused to each other (char- acter 27), which was interpreted here as a syn- apomorphy of the two families. However, on closer inspection this fusion may not be homologous: all five staminodes are fused to form the labellum in the Costaceae, while only the two inner staminodes are united in the Zingiberaceae. Furthermore, sev- eral members of the Cannaceae show partial fusion of the staminodes, which in some cases can b fused basally to the corolla tube. If this character c is homologous in these three families, then it is plesiomorphic in the ginger group (subsequently lost in the Marantaceae) and not a synapomorphy of the Costaceae- Zingiberaceae. n the Marantaceae and the Cannaceae, the number of staminodes (three or four) varies among taxa. The presence of four staminodes was chosen as the state found in the common ancestor of these families (Fig. 1), which implies an independent loss of a staminode in each lineage. Additional inves- tigations of the distribution and ontogeny of the various staminodes in the families of the ginger group are needed (e.g., Kirchoff, 1983a, b, 1986). Only two of the 14 vegetative characters used The in the third analysis are uniquely derived. synapomorphy of the Lowiaceae-Heliconiaceae- ginger group lineage, even though a similar feature is found in the outgroup. The absence of raphide sacs (character 9) evolved in the common ancestor of the ginger group and unites those four families. These two characters also furnished critical evi- dence for Tomlinson's (1962) recognition of nat- ural groups in the Zingiberales. His third character, guard cells (character 2), is not uniquely derived in the present interpretation of the relationships. Monophyletic groups. The results of the present analyses differ from previous classifications pri- marily in the recognition of the paraphyly of the group of “banana families" (Musaceae, Strelitzia- ceae, Heliconiaceae, and Lowiaceae). Earlier in- vestigators (e.g., Bentham & Hooker, 1883; Schu- mann in Engler, 1900; Winkler in Engler & Prantl, 1930b; Lane, 1955; Tomlinson, 1962) united the genera of the banana group on the basis of the shared symmetrical guard cells, raphide sacs, and/ or stamen number (five or six), all of which have been interpreted here as primitive (plesiomorphic) characters present in the common ancestor of the order. The inclusion of the members of these four families (or excluding Lowiaceae: Lane, 1955; Tomlinson, 1962) into the single family Musaceae s.l. has no cladistic basis. The position of the Lowiaceae has always been controversial. Many unique floral and vegetative characters isolate this family from the other fam- ilies of the order. In this analysis only the polyarc root stele provides evidence for its cladistic rela- tionship to the Heliconiaceae and the ginger group. The distichous phyllotaxy and the arillate seeds are the main features that unite the Strelitziaceae with the other six families of the order and exclude the Musaceae as the most this analysis the arborescent nature of the Stre- litziaceae is a derived character and not a primitive feature as Tomlinson (1962) suggested. The position of the Heliconiaceae as sister group to the four ginger families has not been acknowl- "primitive" taxon. In edged by any previous worker. The four floral synapomorphies shared by the Heliconiaceae and the ginger group provide evidence for uniting these families. The bananalike vegetative characters of Heliconia are plesiomorphies shared with the Mu- saceae and Strelitziaceae that have until now tend- ed to obscure placement of this genus in the order. The close relationship of the four families of the ginger group (Zingiberaceae, Costaceae, Canna- ceae, and Marantaceae) has always been recog- nized (e.g., Tomlinson, 1962; Dahlgren et al., nie and the monophyly of this family group рр Volume 77, Number 4 Kress 717 1990 Phylogeny and Classification of Zingiberales TABLE 5. Phylogenetic key to the order Zingiberales based on Figure 7. — £e Lacticifers present; flowers unisexual, plants monoecious. (Cellulae lactiferae praesentes; flores unisexuales, antae monociae. Suborder Musineae (Musaceae) dr es absent; flowers bisexual 2 a ot stele with medullary vessels and phloem; stem woody; two lateral petals fused to enclose anthers. (Radicis stela vasibus а и et phloemate instructa; caulis lignosus; petala duo lateralia connata, antheras includentia.) Suborder Strelitzineae (Strelitziaceae) Root stele polyarc; stem not Mus two lateral petals not so fused 3 a. Sepals fused to form a solid tube; petals not fused; median petal free, forming labellum; median stamen of outer whorl fertile; median stamen of inner whorl absent. (Sepala connata, tubum solidum formantia; petala non connata; petalum medianum liberum, labellum formans; stamen medianum verticilli exterioris fertile, interioris absens.) o... Suborder Lowineae (Lowiaceae) Sepals not forming a solid tube; petals fused at least at base; median petal not а labellu median stamen of outer whorl modified, sterile; median stamen of inner whorl present 000 4a. Two arcs of air canals in leaf axis; raphide sacs present; fertile stamens 5; lateral stamens of inner and outer whorls fertile; perisperm absent (Arcus duo canalium aeriorum in axe folii siti; saci raphidibus instructi praesentes; stamina quinque fertilia, lateralia verticil- lorum interiorum exteriorumque fertilia; perispermium absens. F N c w = ш; order дез еы (Heliconiaceae) 4b. One arc of air canals in leaf axis; raphid sacs absent; iul stamen 1; lateral stamens of inner and outer whorls sterile; perisperm present. (Агсиз unus canalium aeriorum in axe folii situs; saci raphidibus instructi absentes; stamen singulu um iini stamina lateralia verticillorum interiorum exteriorumque sterilia; perispermium praes . Suborder Zingiberineae da. Flowers zygomorphic; sepals fused at base; style unmodified, situated between anther sacs; anther tetrasporangiate; endosperm helobial. (Flores zygomorphi; sepala in base connata; stylus immutatus, E sacos antherarum situs; anthera tetrasporangifera endospermium instar Helobiar Super rfamily агаа ба. Phyllotaxy distichous; aromatic oils present in vegetative body; inner lateral stam inodes fused into a labellum Zi obian 6b. Phyllotaxy — а, aromatic oils not present in vegetative body; all stam- inodes fused into a labellun Costaceae 5b. Flowers ¡es and бе style Pens separated from anther; anther bispor- angiate; endosperm nuclear. (Flores asymmetrici; sepala libra; stylus mutatus, ex anthera discretus; anthera bisporangifera; ыы поп cre um. . Superfamily Cannariae Та. Mucilage cells absent from stems; pulvinus present; sigmoid E wei with evenl aced cross veins in leaves; terminal pairs of enantiomorphic flowers; inner stamens modified into cucullate and callous staminodes; style not petaloid; single ovule per locule arantaceae 7b. Mucilage cells in stems; pulvinus absent; lateral veins oblique in leaves; petaloid style; flower pairs not enantiomorphic; inner staminodes and style petaloid; multiple ovules per locule Cannaceae by a large suite of synapomorphies. The contro- the four families of the ginger group and the Can- versy over the recognition of the Costaceae as a naceae- Marantaceae. separate family from the Zingiberaceae is not re- solved except to show that these two families form — A new classification of the Zingiberales. One of a distinct monophyletic lineage not cladistically in- the goals of this investigation was to derive a new consistent with the recognition of a single family classification of the order that reflected cladistic Zingiberacea relationships. A strict cladistic hierarchical classi- e сайак presented here and the work fication based on Figure 7 would be very complex. y Dahlgren € Rasmussen (1983) provide the only Because of the paraphyletic nature of the banana explicit, fully resolved representations of the phy- families, five ranks would be required between the logenetic relationships of the Zingiberales. The con- order and family levels, four of which would include sistencies and inconsistencies among these trees only a single family. For this reason a modified are an indication of the problems and complexities classification based on the cladistic relationshige of character analysis and the resultant interpre- — but following the *sequencing convention" of Nel- tation of sister-group relationships among the fam- son (1972, 1974) po Wiley (1981) is suggested ilies. The only consistent lineages in all trees are (Table 4). The sequencing convention allows mono- 718 Annals of the Missouri Botanical Garden FIGURE 8. phyletic groups to be placed at the same rank (here suborders) and listed in the order of the branching sequence, thus reflecting the cladistic relationships without providing a separate categorical rank at each branching point. In the case of the Zingiber- ales this convention allows the retention of the eight terminal taxa at the rank of family and still provides a classification that exactly reflects the cladistic relationships. The classification depicted in Table 4 lists five new suborders, two new superfamilies, and eight families. An alternative to erecting new taxa below the ordinal level is simply to list the eight families in order of their branching on the cladogram. Such a classification would cause confusion in recon- structing the relationships of the the families of the ginger group in which the Cannaceae- Maranta- ceae share a common ancestor not shared by the Zingiberaceae—Costaceae. The problem could be “Rhizogram” of the Zingiberales based on Analysis Three and Figure 7. resolved by accepting the Zingiberaceae—Costa- ceae as a single family, as done in the past, which would then allow the legitimate listing of the seven families in order of their branching sequence. How- ever, one might counterargue (with less conviction) that the Cannaceae and Marantaceae should also be combined into a single family, something most taxonomists would be reluctant to do. CONCLUSIONS The history of the classification of the Zingi- berales shows that as information on new char- acters becomes available, new hypotheses on re- lationships among the families and taxonomic rank are proposed. In the present investigation, reeval- uation of character state distributions and homol- ogies coupled with the methods of phylogenetic systematics has provided a new classification based Volume 77, Number 4 1990 Kress 719 Phylogeny and Classification of Zingiberales on cladistic hypotheses. However, many phyloge- netically useful characters in the Zingiberales re- main to be studied carefully, appraised, and veri- fied. Investigations of basic floral and inflorescence anatomy, morphology, and ontogeny, such as cur- rently being carried out by Kirchoff (e.g., 1983a, b, 1986) and Kunze (e.g., 1985, 1986), will pro- vide fundamental data to be incorporated in future analyses. Chemical data are still lacking for most of the families of the Zingiberales. No phylogenetic analysis utilizing molecular characters, such as chloroplast DNA restriction site variation, has yet been attempted on any taxa in the order. In practice, most taxonomists can ignore the subordinal and superfamilial ranks in the classifi- cation proposed here if their goal is identification and placement of taxa only (see Table 5, Fig. However, those biologists wishing to understand the evolution of taxa or characters within the order are dependent upon the phylogenetic information provided by the cladistic classification (Tables 4, 5). For example, as discussed earlier, pollen char- acters were not used in constructing the cladogram so that their evolution within the order could be inferred from the tree. The pollen wall of most of the families of the Zingiberales is characterized by a highly reduced exine and much-elaborated intine (Erdtman, 1966; Kress et al., 1978; Stone et al., 1979; Kress & Stone, 1982; Kress, 1986). As an exception the Costaceae and Zingiberaceae contain taxa with a well-developed exinous layer in the pollen wall (Punt, 1968; Stone et al., 1981; Kress, 1986). If the distribution of pollen wall features currently known for these families is superimposed on the cladogram of Figure 7, the simplest expla- nation of wall evolution is obvious. Reduction of the exine layer in the common ancestor of the order accounts for the presence of exineless pollen in six of the eight families. The much-thickened exine layer found in the pollen of the Costaceae and Zingiberaceae is therefore secondarily derived in the common ancestor of these two families, a hypothesis earlier proposed by Stone et al. (1981). The evolutionary explanation may be complicated than this, especially because the thick- ened exine is not found in all genera of the Zin- giberaceae (Kress & Stone, unpublished). The im- portant conclusion is that the presence of a thick exine in these families is not indicative of their position in the order, but rather the much more "primitive" independent evolution of the trait. The diversifi- cation of aperture type and the elaboration of the intine layer in the Zingiberales can be investigated in the same manner (Kress & Stone, unpublished). It is hoped that the present phylogeny of the Zin- giberales will serve not only as a basis for the classification of taxa in the order, but also as a model for understanding character evolution in the monocotyledons. LITERATURE CITED kie x 1981. The neotropical genera of Mar- antaceae. Circumscription and relationships. Nord. J. Bot E 24 BanrHLOTT, W. & FROHLICH. 1983. Mikromor- phologie und Orien ntierungsmuster epicuticularer Wachs-Kristalloide: ein neues sytematisches Merk bei Monokotylen. Pl. Syst. Evol. 142: 171-185. BENTHAM, G. & KER. 1 enera Planta- rum, Volume 3. 1. Reeve & Co.; Williams & Nor- gate, n. Berry, E. W. 1921. Tertiary fossil plants from Ven- ezuela. Proc. U.S. Nat. Mus. 59: 553-579 Bisson, S., S. GUILLEMET & J. L. HAMEL. 1968. Con- кы a Pétude caryo-taxinomique des эч es Mem. Mus. Natl. Hist. Nat., Ser. B, Bot. 59- Burtt, ^ L. 1972. General introduction to papers Zingiberaceae. Notes Roy. Bot. Gard. Edinburgh 3l: 155-165 R. M. SMITH. 1972. Tentative keys to the subfamilies, tribes and genera of em Notes Roy. Bot. Ga n Edinburgh 31: 6. id aud of the bananas. 1981. of Flowering Plants. Col ia Univ. Press, New DAGHLIAN, С. l. A review of the fossil record of monocotyledons. Bot. Rev. (Lancaster) 47: 517- 555. 1982. Zingiberalean leaves from the Lower Eocene of Texas. Bot. Soc. Amer. Misc. Publ. No DAHLGREN, R. & F. М. RASMUSSEN. 1983. Monocot — Evolutionary Biology 16: 255- Clifford, & Р. F. Yeo. n The Кеш, A the Monocotyledons. Springer-Verlag, erlin. Bees a o ея Schumann). Das Pflan- eic a 2 Marantaceae (K. Schumann). Das dcc IV. 1904. Zngberacee (K. Schumann). Das Pfanzenreich IV. & K. зы 1889. Musaceae, Zingibera e, Cannaceae, Marantaceae (O. G. Petersen). Di Natürlichen Pflanzenfamilien, lst edition. 2(6): 43 930a. Zingiberaceae, Maranta- ceae (T. H. ракет Die Natúrlichen Pflanzenfami- lien, 2nd edition. _ 541-640, 654 & 930b. Musaceae, Сапа пасеае (Н. Winkler). Die ыы, Pflanzenfamilien, 2nd edi- tion. 15a: 505-541, 640-654. 720 Annals of the Missouri Botanical Garden e E 1966. Pollen Morphology and Plant Tax- . Hafner Publ. Co. Pu. W. L. & B. E. Косн. Comparison of fruits and seeds of fossil Spirematospermum m (Zin- p with those of living Cenolophon. Bull. Geol. Soc. Denmark 20: 192-195. & ——. 1972. Der Arillus der tertiaren CITReEacHíe Spirematospermum wetzleri. Lethaia : 47-6 GATIN, С. L. Re cherches anatomiques sur lem- es Cannacées et des Mu- sacées. Ann. Sci. Nat. Bot. 8: 113-14 GOLDBLATT, Р. 1980. Polyploidy i in angiosperms: m cotyledons. Pp. 219-239 in W. H. Lewis (editor), Polyploidy: Biological Relevance. Plenum Publ. Corp., York. HESSE, . WAHA. the pollen wall i in Strelitzia reginae (Musaceae). Pl. Syst. Evol. 141: 285-298. Hickey, L. J. & R. К. PETERSON. 1978. л a fossil genus d the ginger family from taceous to Early Eocene sediments of Western In. terior North America. Canad. J. Bot. 56: 1136- 1152 HorrrUM, К. E. 1950. The Zingiberaceae “ Malay Peninsula. Gard. Bull. Straits Settlem. 13: 1-249. 19 The genus Orc hidantha ну Саг 4. Bull. Straits Settlem. 25: 239-247. m J. E. The development of the seed 1 the Scitamineae. Ann. Bot. (London) 10: 1-40 ы J. . The Families of Flowering Plants, Volume 2. Monocotyledons. Macmillan & Co., Lon aon. 959. The Families of Flowering Plants, Vol- ume ‚2. Monocotyledons, 2nd edition. Clarendon Press, Oxfor à 1973. The Families of Flowering Plants, 3rd edition. Clarendon Press, Oxford. Jain, R. K. 1963. е dies of Musaceae. І. Musa car- diosperma sp. n banana fruit from the Deccan bove series, India. Paleobotanist 12: йени. : L. 1789. Ordo I. Musae, Ordo II. Cannae. p. 61-64 in Genera Plantarum. Paris йш a В. К. 1983a. Floral ие іп буе С i ке Маг gerer and in a (Canna- ceae). Amer. J. Bot. ——— 198: B3b. тене, growth of the aa a in the five genera of the Marantaceae and in Canna (Cannaceae). Bot. Gaz. (Crawfordsville) 144: 110- 1 1986. Inflorescence structure and develop- ment in the ии Ин aer geniculata (Mar- antaceae). Canad. J. Bot Koch, B. E. & W - De URL Samen von Spirematospermum er miozáne Fasterholt- Pa ora Dànemark. iia 1971. da und KRESS, W. de д рие of Central American Heliconia о th pendent inflores- a J. Arnold Arbor. 65: 429-532 —————,. 6. Exineless ah structure and polli- nation os of tropical е ота (Heliconiaceae). -345 їп S. Blackmo . K. Fergusor (editors), Pollen and Spores: Linnaean Society, ie =) ›. 32 "Form and Function. 1983. The fine structure of 1990. The taxonomy of 2 World Heliconia (Heliconiaceae). Allertonia 6: 1- & D. E. STONE. 1982. йз a the sporoderm in оа with special reference to the pol- len grains of Canna and Heliconia. Grana 21: 129- 148. & ———. 1983. Morphology and phyloge- netic significance of exine- dans ш? п of Heliconia (Heliconiaceae). S. 4. : & S. C. dn Ecos Ultrastruc- ture of exine-less pollen: е а (Heliconiaceae). Amer. J. Bot. 65: 1064- KUNZE, Н. АД Die a der Marantaceen und ihr Zusammenhang mit dem Typus der Zingi- le а Beitr. Biol. Pflanzen 60: 93- 140. Infloreszenz — und туа GM von Оаа піва maxillarioides (Ridl.) К. Schum. (Lowiaceae). Beitr. Biol. Pflanzen 61: 221-234. Lane, I. E. Genera and generic relationships in Musaceae. Mitt. Bot. Staatssamml. Miinchen 13: 4-131 n K. Chromosome cytology and ipod ip of the Lowiaceae. Nat. Hist. Bull. Siam Soc. 2 21- 24 ee 1. 1835. P. 69 in Key Bot. Lon МАА J. М. 1972. Costoideae (Zingiberaceae). Monograph No. 8 of Flora Neotropica. —— 1977 А ‚ Renealmia v ane Zingiber- oideae) C ). Mono- graph No. 18 d Flor Neotropica МАНАМТҮ, Н. К. 1970. A cytological study of the Zingiberales wi to their taxonomy. Cytologia Зо: 13-4 49. MickEvicH, M. К. & C. МїттЕК. 1981. Treating poly- morphic characters in systematics: a phylogenetic treatment of i perder data. Pp. 45-58 in V. A. Funk & ooks бейи}, Advances in Cla- distics: Proceedings of the First Meeting of the Willi Hennig Society. The New York Botanic Garden, New ork. MULLER-STOLL, W. 1956. Uber das Verhalten des Exine nonaperturater — -Pollen bei Quellung und Keimung. Grana Palyno NAKAI, T. 41. Notulae i Planta Asiae Orientalis (XVI). Jap. J. Bot. 17: 189-2 NELSON, G. J. 1972. Miis p and clas- sification. Syst. Zool. 21: 227-23 1974. Cisco as an expression of phy- берне етут Syst. Zool. 22: 344-359 OLATUNJI, O. A. 1980. e structure and development . Notes Roy. Bot Pollen morphology о of the American species of the subfamily Costoideae (Zingiberaceae). Rev. Paleobotan. Palynol. 7: 31-43. REYNOLDS, P. K. The Banana. Houghton-Mifflin Co., Riverside Press, Cambridge RIDLEY, н. DE 9 Lowiaceae. FI. Malay Penins. 4: 29] -293. ROGERS, с К. 1984. The Zingiberales (Cannaceae, Marantaceae, and ен іп the Southeastern United States. J. Arnold Arbor. 65: 5-55. Saab, S. I. & R. К. IBRAHIM. 1965. Palynological and biochemical studies of Scitamineae. J. Palynology 1: 62-66. Volume 77, Number 4 1990 Kress 721 Phylogeny and Classification of Zingiberales iss чё W.&P.J.M.Maas. 1971. The genus Canna orthern South America. Acta Bot. Neerl. 20: 663- 680. SIMMONDS, N. W. 1962. The Evolution of the Bananas. Longman, Green & Co., London. SKVARLA, J. & J. R. RowLEY. 1970. The pollen wall of Canna and its iine to the germinal apertures of other pollen. Amer. J. Bot. 57: 519-529. STEBBINS, С. L. 197 г Flowering Plants: Evolution Above the Species Level. Belknap Harvard Univ. Cambridge, Massachusetts STONE, D. E., S. C. SEL LERS & W. J. Kress. 1979. Ontogeny of ee po ollen in Heliconia, a — relative. Ann. Missouri Bot. Gard. 66: 701- Я . 1981. ae ae evolutionary implications of a neotenous exine in po (Zingiberales: Costaceae) pollen. Amer. Press, San D. L. 1985. PAUP: Phylogenetic Analysis sing Parsimony Ver. 2. I Illinois Natural History Survey, Champaign, Illino limas, A. L. 1980. Outline of the classification of flowering plants A Bot. Rev. (Lan В 2 = i phylogenetic classification of e Angiospermae. Evol. Biol. 9: 35-106. абан, Р.В. 1956. Studies in the systematic anat- omy of the Zingiberaceae. J. Linn. Soc. (Bot.) 55: 547-592. 1959. An anatomical approach to the clas- nd sification of the Musaceae. J. Linn. Soc. (Bot.) 55: 79-809 961. Morphological and anatomical char- вазе: of the Marantaceae. J. Linn. Soc. (Bot.) 55-78. 1962. Phylogeny of the Scitamineae—m phological and anatomical considerations. "amat 3. 1969. Classification of the Zingiberales (Scita- mineae) with special reference to anatomical evi- dence. Pp. 295-302 in C. R. Metcalfe (editor), Anat- omy of the Monocotyledons, Volume 3. Clarendon Press, Oxfor TRIVEDI, B. S. & C. 1971 of L. VERMA. The structure . from the of Madhya Pradesh, India. Palaeontographica 132B: "od Vessel types of the monocotyledons: . Bot. Not. 130: 383-402. Wiley, E. O. 1981. Phylogenetics: The Theory and Practice of Phylogenetic Systematics. John Wiley & ork. Sons, New Made: C. A. & . В. HARBORNE. 1977. The leaf flavonoids of the Zingiberales. Biochem. Syst. Ecol. 5: 221-229. WoDEHOUSE, R. P. the living representatives of the Bull. Torrey Bot. Club 59: 313- 340. 1932. Tertiary id I. Pollen of n River flora. PHYLOGENY AND CLASSIFICATION OF THE HAEMODORACEAE! Michael G. Simpson? ABSTRACT phylogenetic analysis of the monocot family Haemodoraceae is presented to assess the classification and A interrelationships of tri monophylesis of the family (as ake d order to assign character polen, two families, Philydraceae and Pon with the Pe including: (1) unifacial leaves (Philydra- tectate-cohime ted) and A some family genera. In eriaceae, were hypothesized as piss ). A detailed ceae) and (2) verrucate pollen wall ‘sculpturing aei non ectate-columellate analysis of the selection, definition, and coding of characters and character states is presented. Computer parsimony algorithms were used to co nstruct most parsimonious trees. Utilizing all characters, including several for which polarity could not be determined, two equally most parsimonious cladograms and Xiphidium could not be a priori established as monophyletic, and the genera Anigozanthos and Conos are ith paraphyletic. Evolutionary events, as portrayed in the cladograms, are reviewed 1 evolution of t anatomy, ovary position, ovule morphology, seed morphology, and chromosome number. Possible biogeographic scenarios support a Gondwanan origin for the Haemodoraceae with one major vicariance event occurring by continental separation of present Antarctica from South America- Africa. With regard to interfamilial relationships, the Haemodoraceae are hypothesized as the sister group of the family P with both f. milies more distantly related to the Philydraceae. Relationships to the Typhales, ж аш and Zingiberales are still ambiguous, but the possibility of a close relationship of the Haemodoraceae-Pontederiaceae to the Zingiberales is considered. The Haemodoraceae R. Br. are a monocot fam- ily of 14 genera and approximately 80 species with distributions in southern Africa, northern South America, Central America, Mexico, eastern North America, Australia, and New Guinea (Fig. 1). Mem- bers of the family are characterized as perennial, rhizomatous and stoloniferous or (more rarely) cor- mose to bulbous herbs with mostly basal to sub- basal, equitant leaves and a terminal, generally cymose inflorescence (Geerinck, 19 969a; Hutchinson, 1973; Robertson, 1976; present study). The leaves are ““ensiform” (unifacial), re- sembling those of /ris. The flowers, typical of mono- cotyledons, are bisexual, with 6 tepals, 1-3-6 sta- mens, and a tricarpellate gynoecium developing into a capsular fruit. Flower symmetry is actino- morphic or zygomorphic; ovary position, ovule type, ovule number, and placentation are variable. Tri- chomes characteristically cover pedicels, hypan- thia (if present), and outer perianth surfaces, often forming a dense tomentum. Several genera of the family possess a red sap in the roots and rhizome, accounting for the common name Bloodwort Fam- ily. The Haemodoraceae have had some interesting economic uses. Several Australian species were used as a “nutritious food" by t roasted and consumed the “roots” the rhizomes; Millspaugh, 1887). Narcotic effects have been attributed to the eastern North American Lachnanthes caroliniana (Lam.) Dandy (red root), the “roots” (again, likely rootstocks) of which were "esteemed as an invigorating tonic by the aborig- ines, especially the Seminoles, in whom it is said to cause brilliancy and fearless expression of the eye and countenance, a boldness and fluency of speech, and other symptoms of heroic bearing, with, of course, the natural opposite after-effects” (Millspaugh, 1887). Millspaugh also described a recipe for a red root tonic, with numerous medicinal ' Support from National Science Foundation grants DEB-8109909 and BSR-8400157 is o This study on the initial stages of this study. Robert Ornduff prov t Duke University, n A. White for help vided a plate S for Figures 2 and 3. U.S.A. ? Department of "Biology. San Diego State University, San Diego, California 92182, ANN. MISSOURI BoT. GARD. 77: 722-784. 1990. Volume 77, Number 4 1990 Simps 723 lil and Classification of Haemodoraceae Lachnanthes = "E оаа 5 MP v Xiphidium x. Ж келми, de Н y ~ WA E X , Schiekla Aur aS X Dx) Pyrrorhiza үү A | Я Sa | a а a iphidium c : al Uae dd А : Jj» ; NEL " x | | | | av. , | pt " | | | A +] tris A №. Barberetta! | 4 | y | ip | a ы Е Anigozanthos) \ © yy : g i Macropidia Phlebocarya Mer Tribonanthes | | 1 | [EET | згапсоа f | Ы crn 4 | | | | | | no n P^ PNE: A MERCATOR PROJECTION abuses NETS if ffi... -5 og] | a 150 EJ so eo 30 mu m FIGURE 1. Xiphidium c. = Xiphidium coerule benefits, including remedies for “‘rheumatic stiff- ness of the neck and shoulders," “ roid fevers, pneumonia, various severe forms of brain disease, rheumatic wry-neck, and laryngeal cough." Charles Darwin (1872) in The Origin of Species (citing an example of selection) described the consumption of Lachnanthes by feral pigs in the southern United States; Virginia farmers had recorded that pigs with light-colored hair were poi- soned by eating red root whereas dark-haired pigs were unaffected. Cooke & Edwards (1981) stated that this correlation between hair color and selec- tive poisoning is presumed to be a photodynamic phenomenon, evidence being the induction of pho- totoxicity in microorganisms by extracts of Lach- nanthes (Kornfeld & Edwards, 1972). The Aus- tralian Haemodorum corymbosum Vahl produces a red pigment (termed haemocorin), which has antitumor activity (Schwenk, 1962) and antibac- terial activity (Narasimhachari et al., 1968). Sev- eral Australian members of the Haemodoraceae, including Blancoa canescens Lindl. (red bugles), Haemodorum corymbosum (blood root lily), Mac- ropidia fuliginosa (Hook.) Druce (black kangaroo paw), Tribonanthes Endl. spp., and numerous species and forms of Anigozanthos Labill. (kan- typhus and thy- Geographic lp ini of the genera of the Haemodoraceae. Xiphidium x. = Xiphidium xanthorhiza; leu garoo paws, cats paws) and Conostylis R. Br., are horticulturally grown for their showy flowers (Dixon & Hopper, 1979). Lachnanthes caroliniana is listed as an agricultural pest, being a “rather ag- gressive weed in commercial cranberry (Vaccinium macrocarpon) bogs" (Robertson, 1976). HISTORY OF CLASSIFICATION As Geerinck (1968) and Robertson (1976) not- ed, the classification of the Haemodoraceae has been variable and uncertain, authors having pro- posed several different combinations of tribes and genera. Robert Brown (1810) first recognized the Haemodoraceae as a formal taxonomic unit com- posed of three southern African genera, Dilatris, Lanaria, and Wachendorfia, and four Australian genera, Anigozanthos, Conostylis, Haemodorum, and Phlebocarya. Diagnostic characteristics of the family were the six-parted, generally superior peri- anth (and thus an inferior ovary), capsular fruit, equitant leaves, and three or six stamens, if three then opposite the inner perianth lobes. Brown spe- cifically distinguished the Haemodoraceae from the Iridaceae, members of which possess flowers with three stamens opposite the outer whorl of tepals. 724 Annals of the Missouri Botanical Garden TABLE 1. Classification of the Haemodoraceae sensu Bentham & Hooker (1883) etnia perianth persistent; biseriate, the inner +/— enclosed by the outer; tube above the ovary n or shortly developed. Stamens 3 or rarely 6 1. Haemodorum 6. Lachnanthes 2. Wachendorfia 7. Barbere 3. Schiekia 8. е ни 4. Hagenbachia 9. Lanaria 5. Dilatris 10. Phlebocarya ncs C Каву ває perianth persistent; lobes Piden se of uniseriate species subvalvate. Stamens 6. Ova weh with numerous ovules. Tribonanthes 15. Macropidia D Conost ylis 16. Lophiola 13. Blancoa l7. Aletris 14. Anigozanthos Tribe Ophiopogoneae: perianth marcescent, persistent be- neath the fruit; segments subequal, similar, flat. Ovary locules 2-ovulate. Pericarp after anthesis ruptured, not enlarged. Seeds berry- жй, subglobose, extruded. Raceme unbranched. Flowers small. 18. Peliosanthes 20. Liriope 21. Sansevieria 19. нан Tribe Conanthereae: perianth at length around and above the ovary circumscissilly deciduous, segments subequal, similar or the exterior small and dissimilar, flat. Stamens or staminodes 6, equal or 1 or 3 of the other dissimilar; ocules of anthers frequently terminally pored or rarely short dehiscent. Ovary locules (except Odont ostomum) s superior, loculicidally with numerous ovules. Capsule dehiscent. Flowers in loose panicles or rarely racemose or solitar 22. Conanthera 23. Cyanella 24. Zephyra 25. Tecophilaea 26. Odontostomum Subsequent to Brown’s treatment, additional gen- era were included in the Haemodoraceae. Lindley (1830) placed the American genera Lachnanthes, Lophiola, and Xiphidium in the family. Endlicher (1836-1840) added the genera that Lindley con- tributed plus Aletris, Androstemma, Blancoa, Hagenbachia, Tribonanthes, Vellozia, and Bar- bacenia (the latter two genera of the tribe Vellozieae sensu Brown, 1810). Bentham & Hooker (1883) provided the first critical treatment of the Haemodoraceae, consid- ering the family to be intermediate between the Bromeliaceae and Iridaceae. Four tribes were des- Euhaemodoreae" (= Haemodoreae), “Conostyleae” (= Conostylideae), Ophiopogoneae, and Conanthereae (Table 1). Bentham & Hooker distinguished the tribes based on stamen number (3 vs. 6), perianth duration (persistent or decidu- ous), anther dehiscence (poricidal in the Conan- thereae), ovary position (inferior vs. superior), and inflorescence type (Table 1). Vellozia and Bar- bacenia were excluded from the family as consti- tuted by Endlicher, and placed in the Amarylli- daceae by Bentham ooker. The genus Androstemma was considered a generic section of Conostylis. Although subsequent treatments of the aemodoraceae have varied considerably in the position and/or rank of certain genera and tribes, the four tribes proposed by Bentham & Hooker have remained essentially intact. Pax (1888) in Die naturlichen Pflanzenfami- lien, limited the Haemodoraceae to Bentham's tribe (Eu)Haemodoreae with the deletion of Lanaria and Phlebocarya and the addition of Pauridia (Table 2). According to this treatment, the Conostylideae (= Bentham’s Conostyleae minus Aletris and plus Lanaria and Phlebocarya) and Conanthereae (mi- nus Odontostomum of Bentham’s classification) were transferred to the Amaryllidaceae, subfamily Hypoxidoideae, with the tribes Alstroemerieae and Hypoxideae. The tribe Ophiopogoneae was placed in the Liliaceae. Thus, of the original four tribes of Bentham, Pax considered only two, Conanthere- ae and Conostylideae, to be closely related. Ballion (1894) transferred the genera of the Haemodor- aceae, sensu Pax, to the Amaryllidaceae. He ar- gued that the Haemodoraceae are an artificial tax- on essentially indistinguishable from members of the Liliaceae and Amaryllidaceae. Pax (1930) and Pax & Hoffmann (1930) did not support Ballion’s view and made no changes in the group's classi- fication relative to the previous edition. Hutchinson (1934, 1959) advanced an original classification of the family. He united the Hae- modoreae and **Conostyleae" (= Conostylideae) as two tribes of the Haemodoraceae, considering the family (as defined) to be a natural assemblage; and he treated the tribe Conanthereae of Bentham & Hooker (1883) as a distinct family, the Tecophi- laeaceae (Table 3). Hutchinson classified the Hae- modoraceae with five other families in the order Haemodorales, considering the group to be inter- mediate to the Amaryllidaceae and Orchidaceae. he Tecophilaeaceae, however, were placed in the Liliales and thought to be rather distantly related to the Haemodoraceae. In his third edition, Hutch- inson (1973) added the newly discovered South American Pyrrorhiza (Maguire & Wurdack, 1957) to the tribe Haemodoreae. Subsequent treatments of the Haemodoraceae have continued to vary with regard to tribal inter- Melchior relationships and generic placement. Volume 77, Number 4 1990 Simpson 725 Phylogeny and Classification of Haemodoraceae TABLE 2. Classification of the Haemodoraceae and Amaryllidaceae, subfamily Hypoxidoideae sensu Pax (1888). Haemodoraceae 1. Haemodorum Sm. 2. Barberetta Harv. 3. Hagenbachia Nees 4. Dilatris Berg. 5. Lachnanthes Elliott Amaryllidaceae subfamily Hypoxidoideae Tribe Alstroemerieae Alstroemeria Blh. : Bomarea Mirb Tribe Hypoxideae 1. Curculigo Gartn. Tribe Conanthereae 1. Conanthera Ruiz & Pav. 2. Cyanella L. Tribe Conostylideae 1. Lanaria Ait. . Phlebocarya R. Br. Macropidia Drummond . Tribonanthes Endl. > о Do Wachendor fia L. Schiekia Meissn. Xiphidium Aubl. Pauridia Harv. Po > Qu . Leontochir Philippi 2. Hypoxis L. 3. Zephyra D. Don Tecophilaea Bert. » Lophiola Ker Blancoa Lindl. Conostylis R. Br. . Anigozanthos Labill. PM M (1964) grouped the tribes Haemodoreae, Conosty- lideae, and Conanthereae as the Haemodoraceae, thus mirroring (with the exception of tribe Ophio- pogoneae) Bentham's classification. In the most recent classification of the family, Geerinck (19692) essentially concurred with Hutchinson in recogniz- ing two tribes: Haemodoreae (10 genera) and Cono- stylideae (3 genera) (Table 4). Geerinck's system differs from that of Hutchinson in removing Lanar- ia and Hagenbachia from the family (to status "incertae sedis"), transferring Lophiola from the Conostylideae to the Haemodoreae, treating Blan- coa as a section of the genus Conostylis and treat- ing Macropidia as section of Anigozanthos. As a result of a multivariate morphometric analysis of Macropidia fuliginosa and 12 species of Anigo- zanthos, however, Hopper & Campbell (1977) argued for the reinstatement of Macropidia as a distinct genu The Dto classification of the Haemo- doraceae has also been variable. Hutchinson (1973) classified the Haemodoraceae with the Apostasiace- ae, Hypoxidaceae, Philydraceae, Taccaceae, and Velloziaceae in his order Haemodorales. Cronquist (1981) placed the Haemodoraceae in the order Liliales of the subclass Liliidae, ““near”” the families Pontederiaceae, Cyanastraceae, Philydraceae, and Liliaceae. In contrast, Takhtajan (1980) grouped the Haemodoraceae, Hypoxidaceae, and Vellozia- ceae in the suborder Haemodorineae of the Liliales. The Haemodoraceae were grouped with the Phi- lydraceae and Pontederiaceae by Dahlgren (1980) and by Dahlgren & Clifford (1982). More recently, Dahlgren & Rasmussen (1983) grouped the Hae- modoraceae, Pontederiaceae, and Typhales (Ty- phaceae and Sparganiaceae) as a tritomy (sharing a presumably derived amoeboid tapetum) in their superorder Bromeliiflorae. The Philydraceae were treated as a more basal clade, united with the above in having distichous leaves. Dahlgren & Rasmussen also included the Bromeliaceae and Velloziaceae as basal clades of the Bromeliiflorae (see Interfamilial Classification). OBJECTIVES The primary objective of the present study is to assess the phylogenetic relationships of the Hae- modoraceae. A detailed analysis of the characters possessed by taxa is included and the rationale for character coding is discussed. A phylogenetic anal- ysis, using these data, is presented in an attempt to answer the following: (1) Are the Haemodora- ceae monophyletic? (2) Are the genera in the fam- ily monophyletic? (3) What is the basis for the traditionally recognized tribes Conostylideae and aemodoreae? (4) What monophyletic subgroups of genera are evident and what are the character changes evident from the cladistic analysis? (5) Can inferences be made as to biogeographic history 726 Annals of the Missouri Botanical Garden TABLE 3. Classification of the Haemodoraceae and E 4. Classification of the Haemodoraceae sensu Tecophilaeaceae sensu Hutchinson, 1934, 1959, 1973. Pyrrhorhiza added in 1973. Haemodoraceae Tribe Haemodoreae: “ very short or absent; stamens 3 or rarely 6” perianth-segments 2-seriate; tube 1. Barberetta 7. Phlebocarya 2. Dilatris 8. Pyrrorhiza 3. Haemodorum 9. Schiekia 4. Hagenbachia 10. Wachendorfia 5 . Lachnanthes 11. . Lanar Xiphidium an Tribe Conostyleae: valvate; tube often fairly long and curved; stamens 6; flowers always tomentose or woo "perianth-segments 1-seriate, sub- 1. Anigozanthos 4. Lophiola 2. Blanca (sic) 5. Macropidia 3. Conostylis 6. Tribonanthes Tecophilaeaceae 1. Conanthera 5. Tecophilaea 2. Cyanastrum 6. Walleria 3. Cyanella 7. Zephyra 4. Odontostomum of the Haemodoraceae? (6) What families are most closely related to the Haemodoraceae and what is the evidence for this relationship? (7) What is the position of the Haemodoraceae within the Bro- meliiflorae (sensu Dahlgren € Rasmussen, 1983)? MONOPHYLESIS OF THE HAEMODORACEAE It is essential in a phylogenetic study to dem- onstrate that the group to be analyzed is monophy- letic (in the sense of Hennig, 1966; equivalent to “holophyletic” of other authors), i.e., that it in- cludes all and only all descendants of a common ancestor as evidenced by one or more synapo- morphies. Despite past discrepancies in classifica- tion, the Ha oraceae are recognized in the present study to comprise a natural, monophyletic group made up of 14 genera: Anigozanthos, Bar- beretta, Blancoa, Conostylis, Dilatris, Haemo- dorum, Lachnanthes, Macropidia, Phlebocarya, Pyrrorhiza, Schiekia, Tribonanthes, Wachen- dorfia, and Xiphidium. The primary evidence for the monophylesis of the family is chemical com- position. The Haemodoraceae are chemically unique in being the only family of vascular plants to possess arylphenalenones," de- -1 H-phenalen-1-one; Edwards, 1981). These compounds are responsible for the floral pigmentation and/or red coloration A Geerinck (1969a Haemodoraceae Tribe Haemodoreae: “flowers glabrous or with simple or glandular trichomes; tepals distinct or rarely ba- sally connate; functional stamens 6 (in 2 whorls of 3) or 3 (the outer whorl absent or replaced by 2 staminodes); anthers nonappendicular at apex; ovary superior, half-inferior, or inferior" 1. Barberetta 6. Phlebocarya 2. Dilatris 7. Pyrrorhiza 3. Haemodorum 8. Schiekia 4. Lachnanthes 9. Wachendorfia 5. Lophiola 10. Xiphidium Tribe Conostylideae: “flowers covered with simple or branched trichomes (rarely both); perianth tube pres- ent; stamens 6, in 2 whorls of 3; anthers sometimes ed ae apically; ovary half-inferior or inferi- or . Anigozanthos (including Macropidia) 2. Conostylis (including Blancoa) 3. Tribonanthes Genera of uncertain affinities: Hagenbachia & Lanaria prominent in the roots and rootstocks of family members. The occurrence of phenalenones was first reported in Haemodorum corymbosum Vahl by Cooke & Segal (1955), who named the isolated phenalenone glycoside **haemocorin." Subsequent- ly, the following ten species in eight genera of the family have been found to possess phenalenones or derivatives thereof: Haemodorum corymbosum, Н. distichophyllum, Lachnanthes caroliniana, Phlebocarya ciliata, Wachendorfia paniculata, W. thyrsiflora, Xiphidium coeruleum, Anigozan- thos rufus, Conostylis setosa, Macropidia fuligi- nosa (Cooke & Edwards, 1981, and references therein). The eight genera not investigated to date for the presence of phenalenones are, in the au- thor's view, very closely related to those that have been, as determined by morphological and paly- nological similarity; it is hypothesized that, when chemically analyzed, they will be found to have arylphenalenones as well. It should be emphasized otherwise biologically known only in four genera of the Hyphomycetes (Fungi Imperfecti) and in one genus of the Discomycetes (Ascomycotina); these, however, are synthesized by a different bio- chemical pathway (Cooke & Edwards, 1981) and are obviously not homologous with those of the Volume 77, Number 4 1990 Simpson 727 Phylogeny and Classification of Haemodoraceae TABLE 5. Embryological characters of the Haemodoraceae and relatives. Tapetal Microspore Taxon type divisio Nucellus type Documentation Haemodoraceae Anigozanthos Amoeboid ^ Successive Crassinucellate Stenar (1927) Dilatris Amoeboid ^ Successive Crassinucellate De Vos (1956) Lachnanthes Amoeboid Successive Crassinucellate Simpson (1981, 1988) Wachendorfia Amoeboid Successive Crassinucellate Dellert (1933); De Vos (1956) Xiphidium oeboi Successive Crassinucellate Stenar (1938) Bromeliaceae Glandular Successive Crassinucellate Dahlgren et al. (1985) Cyanastraceae Cyanastrum Glandular Simultaneous — Crassinucellate Fries (1919); Nietsch (1941) Hypoxidaceae Hypoxis Glandular Successive Tenuinucellate De Vos (1948) Pauridia Glandular Successive Tenuinucellate De Vos (1949) Philydraceae Helmholtzia Glandular Successive Crassinucellate Hamann (1966) Orthothylax Glandular Successive Crassinucellate Hamann (1966) Philydrella Glandular Successive Crassinucellate Hamann (1966) Philydrum Glandular Successive Crassinucellate Hamann (1966) Pontederiaceae Eichhornia Amoeboid Successive Crassinucellate Banerji & Gangulee (1937); Schurhoff (1922 Monochoria Amoeboid Successive Crassinucellate Banerji & Haldar (1942) Sparganiaceae Sparganium Amoeboid Successive Crassinucellate Dahlgren & Clifford (1982) Taccaceae Schizocapsa Glandular Simultaneous — Crassinucellate Hakansson (1921) Tecophilaeaceae Cyanella Glandular Simultaneous — Crassinucellate De Vos (1950) Odontostomum Glandular Simultaneous — Crassinucellate Cave (1952) Typhaceae Typha Amoeboid ^ Successive Crassinucellate Dahlgren & Clifford (1982) Velloziaceae Vellozia Glandular Successive Schnarf (1931); Stenar (1925) Tenuinucellate (Р. seudocrassinucellate) Haemodoraceae. Because of the uniqueness of these compounds and their restriction to the Haemodora- ceae, their presence is hypothesized as a synapo- morphy, uniting the family as a monophyletic group. Other major similarities that the 14 genera have in common are: (1) occurrence of a fibrous layer (““mechanischen Cylinder") in the stem (Schulze, 1893); (2) presence of unifacial leaves with par- acytic stomata (Schulze, 1893; Stenar, 1927, 1938; Green, 1959; Simpson & Dickison, 1981; Simp- son, unpublished); (3) common embryological de- velopment, including occurrence of an amoeboid b ncn successive microsporogenesis, and cras- t (Table 5, and references therein); En (4) a similar and intergrading non-tectate-col- dence that any of these features are synapomorphic for the Haemodoraceae; they may, however, be synapomorphic for two or more families within the complex (see Interfamilial Relationships). The family Tecophilaeaceae have often been classified as the tribe Conanthereae of the Hae- modoraceae, but they definitely do not belong in 728 Annals of the Missouri Botanical Garden the latter. All investigated members of the Tecophi- laeaceae (delimited as the genera Conanthera, Cy- anella, Odontostomum, Tecophilaea, Walleria, and Zephyra; Simpson, in press) differ from the Haemodoraceae in having: (1) bifacial leaves and anomocytic stomata (Schulze, 1893; Simpson, un- published); (2) a glandular tapetum and simulta- neous microsporogenesis (Table 5); (3) phytome- laniferous seeds (except Walleria; Huber, 1969); and (4) pollen grains with a foveolate to reticulate sculpturing, an apertural operculum (except Cy- anella orchidiformis), and a tectate-columellate exine architecture with an inner endexinous layer (Simpson, 1985b). In addition, members of the Tecophilaeaceae lack fluorescent cell wall-bound compounds found in the Haemodoraceae (Harris & Hartley, 1980; see Outgroup Taxa). All the evidence suggests that the Tecophilaeaceae are comparatively distantly related to the Haemodora- ceae. The taxonomic placements of Hagenbachia, Lanaria, Lophiola, and Pauridia have been vari- able in past treatments. Each has been included in the Haemodoraceae by various authors (e.g., Pax, 1930; Hutchinson, 1973; Melchior, 1964; Gee- rinck, 1969a; see History of Classification). How- ever, a major conclusion reached from the present study is that no synapomorphies are known that unite any of these genera with the Haemodoraceae. The South American Hagenbachia brasiliensis (monotypic in its genus) is undoubtedly a case of taxonomic misplacement. It clearly belongs as a species of the genus Chlorophytum of the Lili- aceae, as Ravenna (1977) determined. Corrobo- rating this is the fact that “Hagenbachia” differs from all Haemodoraceae in having a tectate-col- umellate exine wall (Simpson, unpublished) and lacks UV-fluorescent cell-wall-bound compounds (Simp- son, unpublished). The monotypic, eastern North American Lo- phiola (which resembles some Haemodoraceae in having unifacial leaves, a corymb of helicoid cymes, and tomentose flowers and inflorescence) differs from the Haemodoraceae in many respects, in- cluding: (1) absence of a stem fibrous layer (““mech- anischen Cylinder"), absence of subsidiary cells, and differing trichome anatomy (Schulze, 1893; Simpson & Dickison, 1981); (2) reticulate pollen with a tectate-columellate architecture (Simpson, 1983; Zavada, 1983a); (3) glandular tapetal de- velopment (Simpson, 1981); (4) absence of the diagnostic arylphenalenones (Edwards et al., 1970); (5) absence of UV-fluorescent cell-wall-bound compounds (Simpson, unpublished). Ambrose (1980, 1985) presented convincing evidence for the classification of Lophiola in the Liliaceae, Me- lanthioideae (7 Melanthiaceae of Dahlgren & Clif- ford, 1982); there is no doubt that it belongs with at least some members of that group. The monotypic South African genus Lanaria resembles members of the Haemodoraceae in hav- ing a corymb of helicoid cymes and in having multiseriate, dendritic trichomes remarkably sim- ilar to those of some Haemodoraceae (see Char- acter Analysis). However, Lanaria shows many differences from the Haemodoraceae, including: (1) bifacial leaves without stomatal subsidiary cells (Schulze, 1893; Simpson, unpublished); (2) glan- dular tapetal development and simultaneous mi- crosporogenesis (Table 5); (3) reticulate pollen grains with a tectate-columellate exine wall structure (Simpson, 1983); (4) phytomelaniferous seedcoat (Huber, 1969); and (5) absence of UV-fluorescent cell-wall-bound compounds (Simpson, unpub- lished). (Lanaria has not been investigated chem- ically for the presence of arylphenalenones.) Sim- ilarities nod Lanaria and the Tecophilaeaceae have prompted some (e.g., De Vos, 1961, 1963 Dahlgren & ein 1982) to include the genus in that family. However, Dahlgren (pers. comm.) argued for the recognition of a segregate family, Lanariaceae, with close affinities to the Tecophi- laeaceae. Finally, the monotypic southern African genus Pauridia (usually placed in the Hypoxidaceae but sometimes classified in the Haemodoraceae) differs from the Haemodoraceae in having: (1) bifacial leaves; (2) tenuinucellate ovules and a glandular tapetum (Table 5); (3) disulculate pollen grains with a tectate-columellate exine having an endexinous basal layer (Simpson, 1983); and (4) absence of UV-fluorescent cell-wall-bound compounds (Simp- son, unpublished). Pauridia has not been investi- gated for the presence of arylphenalenones. No characters evidently unite Pauridia to the Hae- modoraceae; the genus is here retained in the Hy- poxidaceae. MONOPHYLESIS OF FAMILY E In a cladistic analysis all defined operational taxonomic units (OTUs) should either be mono- phyletic taxa, be split up into monophyletic groups, or have exemplar species assigned for them. erwise, it is possible that one or more species of previously circumscribed genus “A” may be more closely related to species of genus “B” than to other species of genus **A." In the Haemodoraceae, the monophyly of six genera— Barberetta, Blan- Volume 77, Number 4 1990 Simpson 729 Phylogeny and Classification of Haemodoraceae coa, Lachnanthes, Macropidia, Pyrrorhiza, and Schiekia—is accepted by virtue of their being monotypic. (See Platnick, 1976, for an alternate view.) A review of the monophyly of the remaining eight genera is essential before a valid cladistic analysis can be undertaken. The genus Dilatris (five species) is commonly distinguished from other family members by having an inferior ovary and one ovule per carpel. Each of these features are possessed by other members of the family and thus cannot be recognized as synapomorphies (being unique only in combina- tion). Dilatris has a trichome type not found in other genera of the family (see Character Analysis), yet if the proposed evolutionary gradation of tri- chome types (Fig. 28) is valid, then it is possible that the trichomes of Dilatris may not be uniquely derived for the genus as a whole. A feature that may show synapomorphy for the genus is the pres- ence of dotted glands in the distal region of tepals (see Fig. 73). These glands were observed in D. pilansii and D. corymbosa but were not found in species of any other genus in the family. It is hypothesized that these tepal glands are likely syn- apomorphic for the genus as a whole. Haemodorum (20 species) has a semibulbous underground rootstock, which is almost certainly derived from a primitive rhizomatous rootstock, present in almost all other family members (of both tribes) and in outgroups. This type of rootstock may not be unique to Haemodorum, as three other genera have a cormose rootstock (see Character Analysis, Stem type). Most species of Haemo- dorum are glabrous, which itself may be synapo- morphic for the genus. The trichomes observed in H. spicatum (see Character Analysis, Trichome anatomy; Fig. 13) may be homologous with the pilate e found in several genera of the tribe; thus, vestiture may not be a reliable indicator of synapomorphy (and therefore monophylesis) for the species of Haemodorum. In view of these dif- ficulties and because relatively few of the 20 species of Haemodorum were observed in this study, monophylesis cannot be established for the genus. No evident synapomorphies occur for investi- gated species of the genus Wachendorfia (five species). Possible derived characters in Wachen- dorfia, relative to the family as a whole, include zygomorphy and one ovule per carpel. However, these features also occur in other genera of the family and cannot be recognized as synapomor- phies for this genus. The distinctive perianth ap- ertures in Wachendorfia are also found in the genus Schiekia (see Character Analysis, Perianth apertures; Fig. 51). Monophylesis of Wachendorf- ia, therefore, cannot be affirmed in the present tudy. Xiphidium consists of X. coeruleum and АХ. xanthorhiza, which differ only in minor morpho- logical features and are likely more closely related to one another than to any other genus. However, щл because no definitive synapomorphy is evident for Xiphidium, its monophyly cannot be affirmed. Tribonanthes has a unique, ““stem tuberous” rootstock present in all five species (Pate & Dixon, 1981). This rootstock type may be derived for the genus; however, a cormose stem type is present in other genera and could be indicative of a common evolutionary origin for them (see Character Anal- ysis, Stem structural type). Species of Tribonan- thes do, however, have one feature that is very likely unique for the genus: the presence of dis- tinctive appendages arising from the connective of the anther (see Character Analysis, Stamen con- nective appendages; Fig. 59). This feature is ac- cepted as an autapomorphy, and Tribonanthes is hypothesized to be monophyletic. e three species of Phlebocarya are quite sim- ilar to one another, differing primarily in leaf shape and vestiture. The similarities among Phlebocarya species in inflorescence and floral morphology pro- vide good evidence of their very close relationship. In addition, flowers of Phlebocarya have a uni- locular ovary and epitropous ovules; these are unique within the Haemodoraceae and may be synapomorphies for the genus. Therefore, Phle- bocarya is accepted as being monophyletic. Conostylis is the largest genus in the family, with ca. 25 species. No feature appears to be synapomorphic for the genus. The species of Cono- stylis show considerable variability in vegetative and floral morphology. In fact, С. androstemma and C. bealiana have an elongate perianth tube very similar to and possibly homologous with that of the monotypic Blancoa (see Character Analysis, Perianth tube). Although much more detailed stud- ies of this genus are needed to resolve its inter- generic relationships, it is very likely paraphyletic; there is no evidence that one or more species of Conostylis might not be more closely related to Blancoa or even to species of Anigozanthos. Anigozanthos is the second largest genus in the family, with ca. 10 species. Species of Anigozan- thos, together with the monotypic Macropidia, almost certainly constitute a monophyletic group. Both have zygomorphic perianth tubes derived via a unique mechanism (see Character Analysis, Peri- anth splitting; Figs. 53, 54). Anigozanthos differs from Macropidia in trichome color and in having either two or numerous ovules per carpel (as op- 730 Annals of the Missouri Botanical Garden posed to one per carpel in Macropidia). However, trichome color is quite variable among species of Anigozanthos, and ovule number is likely not a shared derived feature for the genus. Because the hiracters distiomiabios Aio y Мас. ropidia are variable or likely to be plesiomorphic, it is uncertain that Anigozanthos is monotypic; one or more species of Anigozanthos may be more closely related to Macropidia than to other species of Anigozanthos. Thus, five of the eight nonmonotypic genera of the Haemodoraceae cannot be reasonably shown, by evidence of synapomorphy, to be monophyletic. In this cladistic analysis of the Haemodoraceae, the following exemplar species are designated for these five genera: Haemodorum spicatum, Wach- endorfia thyrsiflora, Xiphidium coeruleum, Co- nostylis androstemma, Conostylis aurea, Conosty- lis bealiana, Anigozanthos flavidus, and Anigozanthos rufus. Wherever Haemodorum, achendorfia, and Xiphidium are used in the analysis, it should be assumed that only the species indicated above applies. For Conostylis and Ani- gozanthos, it should be kept in mind that very few of the species in the genera will be considered and that these are exemplars. Future studies consid- ering all species of these genera will be needed to assess fully their phylogenetic relationships. OUTGROUP TAXA In the following cladistic analysis character state polarity of ingroup taxa was determined using out- group comparison. This basically entails perform- ing a cladistic analysis on the ingroup plus one or more closely related taxa (outgroups), which serve to root the cladogram. The plesiomorphic state for the ingroup (at the outgroup node; see Maddison et al., 1984) is that which yields maximum par- simony among ingroups and outgroups. The major difficulty in applying outgroup comparison, how- ever, is in determining which taxa indeed share most recent common ancestry with the Haemo- doraceae. As previously discussed (see History of Classification) the interfamilial classification of the aemodoraceae has been quite variable, a number of families having been proposed as close relatives. Therefore, in the present study, every monocot family that has ever been classified with or con- sidered closely related to the Haemodoraceae was assessed as a possible outgroup. These families are: Apostasiaceae, Bromeliaceae, Cyanastraceae, Hy- poxidaceae, Philydraceae, Pontederiaceae, Spar- ganiaceae, Taccaceae, Tecophilaeaceae, Typha- ceae, and Velloziaceae. A comprehensive analysis of the phylogenetic relationships of these families to one another is beyond the scope of the present investigation and will be pursued in the future. Selective features from the literature and from ongoing studies by the author were assessed in order to determine the most likely closest out- groups. It was hoped that at least the two closest outgroups could be identified, as a minimum of two outgroups is required for unequivocally assessing character state polarity (see Maddison et al., 1984). vidence for the sister group relationship of the Haemodoraceae comes mainly from studies of pol- len wall ultrastructure (Simpson, 1983, 1987). All investigated members of the Haemodoraceae have a 1-3-layered, non-tectate-columellate exine struc- ture, which is almost certainly derived among the monocotyledons as a whole (Simpson, 1983; see Character Analysis). In contrast, all investigated mem the Apostasiaceae, Bromeliaceae, Cyanastraceae, Hypoxidaceae, Philydraceae, Sparganiaceae, Taccaceae, Tecophilaeaceae, Ty- phaceae, and Velloziaceae have a typical tectate- columellate exine structure (Ayensu & Skvarla, 1974; Nilsson et al., 1977; Brighigna et al., 1981; Simpson, 1983, 1985a, b; Zavada, 1983b). A tectate-columellate exine structure is presumed to be a plesiomorphic condition among the monocots (Zavada, 1983b) and cannot be utilized to define monophyletic groups. Among the taxa previously proposed to be closely related to the Haemodora- ceae, only the Pontederiaceae are similar in pollen ultrastructure. Several members of the Pontede- riaceae possess an exine sculpturing and structure identical to that of members of the Haemodoraceae (Simpson, 1987; see Character Analysis, Pollen sculpturing, Exine wall structure). The palynolog- ical similarities between the Haemodoraceae an Pontederiaceae constitute excellent evidence for ers о the close relationship of the two families and are a here as synapomorphies linking the modoraceae and Pontederiaceae as sister taxa e dentification of the next most closely related outgroup of the Haemodoraceae-Pontederiaceae complex is rather uncertain, however. Dahlgren & Rasmussen (1983) proposed that the presence of an amoeboid tapetum in the Haemodoraceae, Pon- tederiaceae, Typhaceae, and Sparganiaceae con- stitutes a synapomorphy for these four families, uniting them as a monophyletic group within their Bromeliiflorae. However, because an amoeboid ta- petum occurs in numerous other monocot taxa, its use as a synapomorphy for these families seems less than certain, particularly with respect to dif- fering opinions as to the classification of the Ty- Volume 77, Number 4 1990 Simpson 731 Phylogeny and Classification of Haemodoraceae phales (Typhaceae and Sparganiaceae). Details of leaf morphology may provide less ambiguous evi- dence. Of the possible outgroups cup. the Haemodoraceae are similar only to ceae, Pontederiaceae, аз апа Турһа- сеае in possessing distichous leaves, a feature that Dahlgren & Rasmussen (1983) considered syn- apomorphic for these families. Of these families, only the Haemodoraceae and Philydraceae have unifacial (= ensiform) leaves. The presence of uni- facial leaves is generally considered to be apo- ilydra- morphic among the monocotyledons. Occurrence of such leaves in the Philydraceae and Haemo- doraceae is tentatively hypothesized as synapo- morphic for the two families and evidence for their recent common ancestry, especially in light of nu- merous other similarities of the two families (see below). The puru directionality of this fea- ture may n further consideration, as Walker (1989) arce unifacial leaves to be plesio- morphic for the monocots as a whole. Several other features link the Haemodoraceae to the Philydraceae and/or Pontederiaceae (and in many cases to other families), but the relative ancestry of these characters is uncertain or non- conclusive in outgroup selection. For example, among all outgroup candidates, the Bromeliaceae, Philydraceae, Pontederiaceae, Sparganiaceae, and Typhaceae are similar to the Haemodoraceae in having fluorescent, lignin-precursor acids (ferulic, diferulic, and p-hydroxybenzoic) bound to unlig- nified cell walls (Harris & Hartley, 1980). In con- trast, these bound acids are absent in investigated members of all other considered outgroup families (i.e., Hypoxidaceae, Taccaceae, Tecophilaeaceae, and Velloziaceae). The data base for this character is quite small. Only one to a few genera or species have been investigated for many monocot families, and numerous families have yet to be investigated at all. Dahlgren & Rasmussen (1983) hypothesized that the presence of these fluorescent cell-wall- bound acids is a derived feature within the mono- cots, since these compounds are lacking in pre- sumably closely related dicotyledons. Although many more taxa need investigation with regard to this feature, and although its biochemical signifi- cance needs elucidation, the presence of these flu- orescent cell-wall-bound acids seems to constitute good evidence for the close relationship of the above families (see Interfamilial Relationships). n addition, among the possible outgroups, only investigated members of the Haemodoraceae, Phil- ydraceae, Pontederiaceae, and Sparganiaceae pos- sess a common anatomical feature: presence of distinctive placental sclereid idioblasts (work in progress; see Character Analysis, Placental scler- eids). These compounds are present in members of the Zingiberaceae as well (see Interfamilial Re- lationships). Although few taxa in families other aemodoraceae have been investigated for this feature, it appears to provide yet another piece of evidence linking the Haemodoraceae, Phil- ydraceae, Pontederiaceae, and Typhales. In summary, the Pontederiaceae are chosen as the hypothesized sister taxon to the Haemodora- ceae because of a similar and hypothetically de- rived pollen wall structure. The identification of the next most closely related outgroup is less cer- tain. The Philydraceae are tentatively selected as this next most closely related outgroup because of the occurrence of similar, presumably derived, uni- Philydraceae and Haemodor- outgroup families show similarity to facial leaves in the aceae. Both the Haemodoraceae in anatomy (placental scler- eids) апа chemistry (fluorescent cell-wall-bound compounds), further supporting a close relation- ship. Certainly, additional studies are needed to assess interfamilial relationships in the complex (see Interfamilial Classification). However, rather than treat these outgroup families as unresolved or po- lytomous (Maddison et al., 1984), the evidence seems strong enough to utilize the Philydraceae and Pontederiaceae as most closely related out- groups to the Haemodoraceae in ascertaining the directionality of character state transformations. CHARACTER ANALYSIS The following is a list and discussion of those characters and character states thought by the author to be important in resolving intrafamilial relationships. It should be stresse that the initial genetic or not. discontinuities between the states are included in the analysis. Included in the character analysis are: (1) selection of characters; (2) selection and defi- nition of character states; (3) assessment of ho- mology of characters and character states; and (4) assessment of polarity of character states based on comparison with the designated outgroups (Phily- draceae and Pontederiaceae) or other criteria. oth outgroups were treated as operational taxo- nomic units (OTUs) in the data matrix. Characters with common states in all members of the Hae- modoraceae (characters 52-55) are included in the analysis only to establish relationships of the two outgroups to the ingroup. А given character was coded as missing data (**?”” in the data matrix) 732 Annals of the Missouri Botanical Garden if the taxon is polymorphic for the character, if data are unknown (e.g., chromosome numbers of Schiekia and Pyrrorhiza), or if X-coding is used (see below). Multistate characters were initially coded as two or more binary characters for ease of discussion. Where a multistate morphocline is illustrated in the character analysis, a character number in brackets portrays a coding such that taxa possessing the character to the left of the arrow are coded as state “0” and those to the right are coded as state “1.” For certain characters, the “X character” meth- od (Doyle & Donoghue, 1986) was utilized, which codes the character state for certain taxa as ^X" 93 (equivalent to missing data, **?," in the computer algorithm), allowing for either of two alternative state changes. This technique is valuable in that the number of characters assigned to a given trans- formation series (morphocline) may often be re- duced, thus minimizing unintentional weighting and unintentional bias, e.g., with regard to uncertain patterns of evolutionary direction. MATERIALS AND METHODS For studies of floral trichome anatomy and peri- anth cell types, small pieces of tepals and pedicels were removed and mounted in 50% glycerin. Prep- arations were left unstained or were occasionally stained with 0.01% aqueous Toluidine blue. For studies of perianth aestivation, immature buds were embedded in paraffin, and serial cross sections were prepared according to standard anatomical tech- nique (Johansen, 1940). For observations of pla- cental cell types, ovaries of mature flowers were paraffin-embedded and longitudinal sections were prepared as above. All sections were stained with safranin, iron hematoxylin, and fast green. Line drawings were made using a Wild Heerbrugg brightfield microscope with camera lucida attach- ment. Photographs were taken with a Leitz Wetzlar or Nikon Microphot-FX photomicroscope using Panatomic X film (ASA 32). Plant material was fixed in either formalin/ace- tic acid/alcohol (noted “FAA” below) or 4% glu- teraldehyde (“СОТ below). Some material (“DRIED” sheets and reexpanded in Aerosol OT for 2-5 days, below) was obtained from herbarium followed by several water rinses and then fixation in FAA. Materials and methods for the ultrastruc- tural observations of pollen grains are discussed in Simpson (1983, 1985a, b, 1987). Documentation for the taxa studied in the present work is as follows (parentheses indicate herbaria that house vouch- ers): HAEMORDORACEAE Anigozanthos flavidus DC. “FAA”-—M. G. Simpson 241X81J (DUKE) Anigozanthos rufus Labill. *FAA"—M. G. Simpson 271X81F (SDSU) Barberetta aurea Harv. “FAA”—R. Ornduff 7661 (UC) Blancoa canescens Lindl. “GLUT” — Simpson 181Х81АА (DUKE) Conostylis androstemma F. Muell. “DRIED” — S. R. Preif 1409 (K) C. aurea Lindl. “FAA”—M. G. Simpson 131X815 (SDSU) C. bealiana Е. Muell. *FAA"— Arboretum, U.C. Santa Cruz, 3XI80 Dilatris corymbosa Berg. “ЕАА” Р. Gold- blatt 3242 (MO) D. pilansii Barker "FAA"—P. V. D. Meriwe 30X81-2 (STEU) Haemodorum simplex Lindl. *GLUT"— Simpson 201X81A (DUKE) Н. spicatum R. Br. “FAA”—M. G. Simpson 161X81C (DUKE) Lachnanthes caroliniana (Lam.) Dandy "FAA"—M. G. Simpson 14VI80A (DUKE) Lanaria lanata (L.) Dur. & Schinz “DRIED”— R. D. A. Bayliss 4369 (US) Lophiola aurea Ker-Gawler “FAA”-—M. С. Simpson 14VI80B (DUKE) Macropidia fuliginosa (Hook.) Druce “FAA” — М. G. Simpson 181X81 DD (DUKE) Pauridia minuta (L.f.) Dur. & Schinz “DRIED” —P. MacOwan & H. Bolus 291 M. G. M. G. (US) Phlebocarya ciliata R. Br. “GLUT?” —M. G. Simpson 161X814 (DUKE) P. pilosissima F. Muell. “FAA” — son 161Х81К (DUKE) Pyrrorhiza neblinae Maguire & Wurdack “DRIED”—B. Maguire, J. J. Wurdack & G. S. Bunting 37222 (US) Schiekia orinocensis (Kunth) Meisn. “FAA” — B. Maguire 41569 (NY) Tribonanthes australis Endl. “DRIED”—-A. J. A. T. Hotchkiss, 23VIII1953 (US) T. variabilis Lindl. *FAA"—M. С. 8IX81A (DUKE) Wachendorfia paniculata L. “FAA” —P. V. D. Merwe 30X81-1 (STEU) W. thyrsiflora L. "FAA" —R. Ornduff 7691 (UC) M. G. Simp- Fames Simpson "FAA"—J. M. Xiphidium coeruleum Aubl. MacDougal 1043 (DUKE) Volume 77, Number 4 1990 Simpson 733 Phylogeny and Classification of Haemodoraceae HYPOXIDACEAE Curculigo capitulata (Lour.) Kuntze “FAA” — G. Simpson 29VI80 (FTG) Hypoxis micrantha Pollard “FAA”— Simpson 5V82A (DUKE) М. С. LILIACEAE TRIBE OPHIOPOGONEAE Liriope muscari (Decne.) L. Н. Bailey “FAA” — М. G. Simpson 7VII81A (DUKE) PHILYDRACEAE Helmholtzia acorifolia F. V. Mueller “FAA” — M. G. Simpson 81-16A (DUKE) novo-guineensis (Krause) Skottsberg “DRIED”—_L. J. Brass 12859 (A) Orthorthylax glaberrimus (Hooker fil.) Skotts- berg “FAA” —U. Hamann 1183 (Herb., U. Hamann, Berlin) Philydrum lanuginosum Gaertner *FAA"—E. F. Constable; U. Hamann ra (NSW) Fphiiydrelia pygmaea (R. Brown) Caru “FAA” —M. G. Simpson 281X814 OR. PONTEDERIACEAE Heteranthera reniformis R. & P. “GLUT”— M. G. Simpson 4VIII82A (DUKE) Pontederia cordata L. “GLUT”—-M. С. Simp- son 4VII182B (DUKE) Reussia rotundifolia (L.f.) Castell “DRIED” — G. T. Prance 23284 & J. F. Ramos (US) SPARGANIACEAE Sparganium eurycarpum Engelm. “FAA/ GLUT”—M. G. Simpson 21VI86A (SDSU) STRELITZIACEAE Strelitzia reginae Ait. “FAA” — M. G. Simpson 11XI86A (SDSU) TACCACEAE Tacca integrifolia Ker.-Gawl. '*FAA"— М. С. Simpson 23V82 (Duke Univ. greenhouses 81-0379) TECOPHILAEACEAE Conanthera bifolia R. & P. *DRIED"—E. P. Killip & E. Pisano 39690 (US) C. trimaculata Don. *DRIED"—C. Grandjot (MO 1126476) Cyanastrum cordifolium Oliv. O. Daramola 41029 (MO) "DRIED'— B. Cyanella alba L.f. *FAA"—R. Ornduff 7463 (UC C. hyacinthoides L. *FAA"—R. Ornduff 7501 (UC C. lutea L.f. var. lutea "FAA"—R. Ornduff 7565 (UC) Odontostomum hartwegii Torr. *FAA"—UCBG 53.845 Tecophilaea violiflora Bert. ex Colla “DRIED”—-O. Buchtien 10VIII1895 (US) Walleria mackenzii Kirk. *DRIED"—J. Bu- chanan 1891 (US) W. muricata N. E. B. “DRIED?” —N. C. Chase 5182 (MO) Zephyra elegans D. Don. “DRIED”—E. Wer- dermann 776 (US VELLOZIACEAE Barbecenia seubertiana Goeth. & Henr. “FAA” —Hatschbach 30095 (Duke Univ. greenhouses) CHARACTER CODING Character #1. and stem ation. Dilatris, Haemodorum, Lachnanthes, Pyrrorhiza, Wachendorfia, and Xiphidium have a red, red-orange, or maroon coloration of the roots and underground stems (Thiselton-Dyer, 1896- 1897; Adamson & Salter, 1950; Maguire & Wur- dack, 1957; Geerinck, 1969a; Simpson, pers. obs.), accounting for the family name Haemodoraceae (Gr. haima, blood), the Bloodwort Family. This reddish coloration results from the presence of one or more forms of the distinctive class of. chemical compounds, phenalenones. Although all investi- gated family members contain phenalenones (Cooke & Segal, 1955; Cooke et al., 1958; Cooke & Edwards, 1981; see Introduction), only the above six genera show red pigmentation in the roots and rootstocks. The possible adaptive significance of this coloration is unknown; it may simply be cor- related with a high concentration of one or more forms of this class of compounds. The fact that these pigments are toxic to certain livestock, at least in Lachnanthes (see Characterization and Economic Importance), may be significant in this Root color- regard. Character #2. Stem structural type. Most members of the Haemodoraceae have an elongate to congested, sympodially branched rhizome, com- monly bearing proliferative stolons, although four genera deviate from the rhizomatous habit. Pyr- rorhiza, Tribonanthes, and Wachendorfia possess 734 Annals of th Missouri a Garden an underground corm (illustrated for Wachendorf- ia in Fig. 2). In Tribonanthes the corm (termed a “root tuber," sensu Pate € Dixon, 1981) de- velops from a downward-directed axillary bud which penetrates the outer scale leaves of the old corm (initially resembling a root); further growth results in the formation of a globose mass of tissue. Pyr- rorhiza and Wachendorfia (Fig. 2) have a basal cluster of globose corms; it is not known whether they develop similarly to those of Tribonanthes. Haemodorum has a somewhat bulbous corm, not like a typical bulb, but consisting of an aggregate of the swollen, fleshy bases of primarily nonpho- tosynthetic leaves (Thiselton-Dyer, 1896-1897; Pate & Dixon, 1981; Simpson, pers. obs.). The bulbous corm of Haemodorum is tentatively coded as homologous to that of the above three genera. (Barberetta is somewhat intermediate between the rhizomatous and the cormose taxa, having short, horizontal, rather fleshy proliferative shoots; how- ever, these are probably a slight specialization of the rhizomatous/stoloniferous habit and are not coded as evolutionarily intermediate to the cormose stem type.) The hypothesized morphocline for stem struc- tural types in the family is: [#2] RHIZOMATOUS + CORM OR BULBOUS CORM. However, such a hypothesis seems to be very tentative. Deviations from a strictly rhizomatous stem habit likely have occurred secondarily more than once (e.g., via strong selective pressure for dormancy) and thus may not indicate homology. Of the outgroups, a rhizomatous/stoloniferous stem is present in all members of the Pontederia- ceae. In the Philydraceae three of the five species have a rhizomatous stem type, but Philydrum lanuginosum has a basal caudex (Dahlgren et al., 1985), and Philydrella pygmaea has corms sim- ilar to those of the cormose Haemodoraceae (Pate & Dixon, 1981; pers. obs.). Thus, the Philydraceae are coded as polymorphic for stem structural type. Interestingly, Philydrella is found with Tribonan- thes (in southwest Australia); they occur together in similar habitats (low, winter-wet flats). It is prob- able, however, that the common stem habit of these two taxa is not by homology, but is the result of separate, secondary adaptations to a winter-wet, summer-drought environmental regime. Character #3. Plicate leaves. Barberetta and Wachendorfia have longitudinally plicate leaves (Figs. 3-5), a condition found in no other member of the Haemodoraceae nor in any of the outgroups. (Helmholtzia of the Philydraceae has unifacial leaves with a pseudo-costa but no evidence of pli- cation.) Plication in these two genera arises by the occurrence of longitudinal folding and development of tissue ridges opposite the major vascular bundles (Fig. 5). The presence of similar plicate leaves in Barberetta and Wachendorfia constitutes strong evidence for their common evolutionary origin within the Haemodoraceae and is coded as derived (from an ancestral leaf with smooth posture). Characters #4-6. Inflorescence type. In- florescences in the Haemodoraceae are quite vari- able, but most have a common theme, consisting of a network of helicoid cymes (e.g., Fig. 6) ar- ranged as a panicle, raceme, corymb, or capitulum. The inflorescence of Haemodorum differs in being a raceme, panicle, or corymb of either flower pairs or cymules containing paired flowers. In addition, corymblike aggregates of bifurcate or trifurcate helicoid cymes occur in Dilatris and Lachnanthes, and bifurcate helicoid cyme aggregates (in the form of a raceme, panicle, corymb, or capitulum) occur in Anigozanthos, Blancoa, Conostylis, Macro- pidia, Phlebocarya, and Tribonanthes. Barber- etta is unique in the family in having a simple raceme. The possible evolutionary intergradation be- tween these varied inflorescence types is uncertain. A coding of inflorescence types that may likely represent homologies in the family is related to the cyme unit itself rather than to the type of aggre- gation of these cyme units. The morphocline used in the present study is: CYME ABSENT -[#4]-> CYME SIMPLE -[45]^ CYME BIFURCATE -[#6]> CYME BIFURCATE OR TRIFUR- CATE. In this morphocline only the simple raceme of Barberetta would be coded as lacking a cyme unit. The flower pairs or cymules in Haemodorum are interpreted as being a modification of the bi- furcate cyme. A tendency for trifurcate cyme units is found only in Dilatris and Lachnanthes. Among the outgroups all members of the Phily- draceae lack cyme units; the inflorescence is either a simple spike or a spike of spikes. The inflores- cence type in the Pontederiaceae is generally a spike or raceme of simple cyme units. À priori, it seems most probable that the simple cyme inflo- rescence unit, which is common in the monocot- yledons as a whole, may be most ancestral for the Haemodoraceae; this hypothesis will be tested by the cladistic analysis. Characters #7-13. Trichome anatomy. Trichomes are present on the inflorescence axes, bracts, outer perianths, and/or ovary surfaces of all genera of the Haemodoraceae and on the leaves Volume 77, Number 4 1990 Simpso 735 Phylogeny and Classification of Haemodoraceae FIGURES 2-5. "d aru paniculata plant, showing plicate leaves; х 0.35.— 0 of some family members. Distinctive pilate tri- chomes (Figs. 7, 9), consisting of a basal rosette of generally 3-5 cells (having characteristic trans- verse ridges; see Fig. 11), a uniseriate column of (1-)2-5(-7) cells, and a terminal, ovoid glandlike cell, are found in Barberetta, Dilatris, Lach- nanthes, Pyrrorhiza, Schiekia, Wachendorfia, and Xiphidium. As their anatomical similarity reveals, the trichomes in these taxa are undoubtedly ho- Vegetative characters of Wachendorfia. — 2. eire rootstock of W. paniculata; x0.3 4. Leaf of W. thyrsiflora. Note plication of blade; — 9. Leaf cross section of W. thyrsiflora. Note plication and ridges of tissue at major vascular bundles; x 7.4. mologous. Within these seven genera various com- binations with other trichome types may occur. Barberetta (Fig. 9) and Xiphidium (Fig. 20) pos- sess only the pilate trichome type. Pyrrorhiza (Fig. 17), Schiekia (Fig. 18), and Wachendorfia (Fig. 19) have both pilate trichomes and sharply taper- ing, unicellular trichomes, both trichome types with a basal rosette of cells. Dilatris possesses the typ- ical pilate trichome type (Figs. 10, 11) plus long, 736 Annals of the Missouri Botanical Garden Ficu JRES 6-8. is iie rx trichome morphology i А the е — 6. Xiphidium coeruleum, helicoid cyme inflorescence unit; x 7.0. — 7. Pilate trichomes of retta aurea; —8. Multiseriate, dendritic trichome of аа pov Note en: lateral der as (arrow); pt Volume 77, Number 4 1990 Simpson 737 Phylogeny and Classification of Haemodoraceae many -celled, uniseriate, tapering trichomes with a basal rosette of cells (Fig. 12). Lachnanthes has mostly long, uniseriate, tapering trichomes (Fig. 14) with no distinctive basal rosette (Fig. 15). How- ever, unicellular trichomes with a glandlike ter- minal cell and a basal rosette of epidermal cells (resembling those of the pilate trichome type) oc- casionally occur on the leaf margins in Lach- nanthes (Fig. 1 Species of а are usually glabrous throughout. However, at least H. spicatum (Fig. 13) has uniseriate, generally three-celled trichomes without specialized basal epidermal cells. These trichomes have a rounded, somewhat elongate ter- minal cell with densely brown-colored cytoplasmic contents, resembling (and possibly homologous with; see below) the terminal cell of the pilate trichome type Anigozanthos, Blancoa, Conostylis, and Mac- ropidia have identical short to very elongate, mul- tiseriate, highly branched, “dendritic” trichomes (Figs. 8, 21, 22). The bases of these dendritic trichomes consist of a few small, thick-walled cu- boidal cells (as in Fig. 22); cells that form the branches are decurrent along the trichome axis (see Fig. 8). Tribonanthes has long, many-celled, generally uniseriate trichomes (Fig. 27) that are characteristically branched at the base and have 2-4 rounded to short-cylindrical basal cells (Fig. 26). One species of Phlebocarya, Р. pilosissima, has branched dendritic (Fig. 24) to stellate (Fig. 25) trichomes; these are similar to but less highly branched than the dendritic trichomes of the above four genera. A second species of Phlebocarya, Р. ciliata, is, like Haemodorum, usually glabrous but has occasional, slightly elongate, unicellular tri- chomes (Fig. 23) resembling those of Haemodorum spicatum. Thus the trichomes of Phlebocarya might be interpreted as a morphological and evo- lutionary intermediate to those of Haemodorum and the taxa with dendritic trichomes. A hypothesized intergradation series for tri- chome anatomy in the Haemodoraceae is seen in Figure 28. Note that the homology of the trichomes of Haemodorum with the pilate trichome type of Barberetta and Xiphidium and with the unicellular type of Phlebocarya is questionable. The seven genera with pilate trichomes (having a basal rosette of epidermal cells) are arranged in a linear series depending on the presence and length of an ad- ditional trichome type, whether: (1) absent (Bar- beretta, Xiphidium); (2) unicellular (Pyrrorhiza, Schiekia, and Wachendorfia); (3) long-uniseriate (Dilatris); or (4) long-uniseriate lacking a basal rosette (Lachnanthes). The basally branched, uni- seriate trichomes of Tribonanthes and the highly branched, dendritic trichomes of Anigozanthos, Blancoa, Conostylis, and Macropidia are depict- ed as intergrading with the sparsely branched dendritic trichomes of Phlebocarya. The long-uni- seriate trichomes of Lachnanthes and the long- uniseriate, basally branched trichomes of Tribo- nanthes are possibly homologous; both trichome types lack a basal rosette of epidermal cells. Because of the uncertainty of some of the in- tergrading states of Figure 28, and because the groupings of this morphocline would tend to bias the cladistic analysis if (as often happens) any con- flicts in character evolution are evident, the char- acter “trichome anatomy” was subdivided into the following discrete two-state characters (see Table 7 and Fig. 165A): Character #7: Presence/absence of pilate tri- chomes. Haemodorum, whose trichomes are quite different than those of other taxa, is coded as possessing pilate trichomes. Lachnanthes, which has unicellular trichomes with a basal rosette of cells (similar to that of other taxa), is coded as “X?” because of the uncertainty of homology with the pilate type. The genus Phlebocarya, which has unicellular trichomes in one species (P. ciliata), is also coded as “X”” for this character so as not to bias the possibility of homology between its tri- an type and that of Haemodorum (see Figs. 28, A). Character #8: Presence/absence of trichomes which, if pilate, have a basal rosette of epidermal ells. The trichomes of Haemodorum, although pilate, lack this distinctive basal rosette. The uni- cellular trichomes of Lachnanthes, while coded as questionably pilate, do possess a basal rosette; thus, Taxa lacking pilate trichomes are coded as (including Phlebocarya). Character #9: Presence/absence of trichomes with a sharply tapering apex. The basally branched tri- chomes of Tribonanthes are interpreted as being sharply tapering, as are the uniseriate trichomes of Dilatris and Lachnanthes and the unicellular trichomes of Pyrrorhiza, Schiekia, and Wach- endorfia. Taxa with multiseriate, dendritic tri- chomes were also coded as possessing tapering trichomes because of the presumed homology of the sharply tapering trichome branches with the sharply tapering apices of the uniseriate trichomes. Character #10: Trichomes which, if tapering, are either unicellular (Pyrrorhiza, Schiekia, and Wachendorfia) or multicellular. Taxa lacking tri- 738 Annals of the Missouri Botanical Garden FIGURES 9-27. Тгісһоте anatomy in the Haemodoraceae. — 9. Barberetta aurea, pilate trichome. Note terminal granular cell and basal rosette of cells; x120. 10-12. Dilatris pilansii. — ‚ Pilate trichome; х120.— 11. Basal rosette cells with apical transverse ridges; x 73.— 12. Long, uniseriate taperin g trichome with basal rosette cells; x54.—13. Haemodorum spicatum, short, uniseriate (generally 3-celled) trichome. Note oblong, oo terminal — 14. 5 cell; х133. 14-16. Lachnanthes carolinian ong, uniseriate tapering trichome; х55. . Close-up of trichome base. Note absence of rosette Lis x #120 — 16. Unicellular trichome of leaf margins. Note basal rosette x146.—17. Pyrrorhiza neblinae, pilate (left) and unicellular, sharply tapering (right) basal rosette cells; x 51 (left), x 40 (right). — 18. Schiekia orinocensis, pilate (left) and unicellular, sharply tapering (right) trichomes; x73.—19. Wachendorfia thyrsiflora, pilate (left) and са med bg (right) trichomes; — 20. Xiphidium ее: pilate trichome with basal rosette cells; . Ani- gozanthos flavidus. —21. Long, dendritic ry with decurrent lateral branches; x40. d Short Fd ice trichome with small cuboidal basal cells; x 69. — 23. Phlebocarya ciliata, unicellular trichomes with granular contents; x67. 24, 25. Phlebocarya pilosissima. — 24. Dendritic trichome, with decurrent lateral branches and cuboidal basal cells; x 73. — 25. Stellate trichome; x91. 26, 27. Tribonanthes variabilis. — 26. Trichome base, showing cuboidal basal cells and basal, lateral branches; x120.— 27. Uniseriate, tapering, basally brached оше x5 Volume 77, Number 4 Simpson 739 1990 Phylogeny and Classification of Haemodoraceae e -—- == pua BARBERETTA 2 XIPHIDIUM SCHIEKIA : WACHENDORFIA DILATRIS H PYRRORHIZA A4» LACHNANTHES HAEMODORUM PHLEBOCARYA PIL. FIGURE 28. WE (== = | ANIGOZANTHOS BLANCOA CONOSTYLIS MACROPIDIA 06 E TRIBONANTHES Hypothesized intergradation series of trichome types in the Haemodoraceae. Note the ла (indicated by **?") of homology of the trichomes of Haemodorum. (See text for character state coding chomes with sharply tapering apices are coded as “Хх +? Character #11: Trichomes which, if tapering, are either uniseriate or multiseriate, the latter also being dendritic with decurrent branches and with mul- tiseriate cuboidal basal cells. Taxa lacking tapering trichomes are coded as “X.” Character #12: Trichomes which, if tapering, are unbranched vs. branched. Branched trichomes in- clude both the multiseriate, dendritic trichomes and the basally branched trichomes of Tribonanthes. Taxa lacking tapering trichomes are coded as “Х.” Character #13: Presence/absence of trichomes which, if tapering, possess a basal rosette of epi- dermal cells. Lachnanthes, Tribonanthes, and the multiseriate taxa are coded as lacking the basal rosette (in the tapering trichomes) possessed by Dilatris, Pyrrorhiza, Schiekia, and Wachendorf- ia. Taxa lacking tapering trichomes are coded as 6 3 For character #9 it might be argued that taxa with dendritic trichomes should not be coded as “tapering”; i.e., the sharply tapering branches of these multiseriate trichomes may not be homolo- gous with the sharply tapering apices of the uni- seriate trichomes. However, if taxa with multiseri- ate trichomes (Anigozanthos, Blancoa, Conostylis, Macropidia, and Phlebocarya) are coded as either “0” or as “X” for this character, the topology of the most parsimonious cladogram(s) is unaffected (see Cladistic Analysis). Trichomes of the outgroup families show some resemblances to those of the Haemodoraceae. For 740 Annals of the Missouri Botanical Garden FIGURES 29-43. basal cells; x 52. — 32. Close-up of trichome illustrated in Figure 31, showing oblique | /7 42 43 0 Trichome anatomy in the Philydraceae and Pontederiaceae. 29-32. Helmholtzia acorifolia. — 29. Three-celled, uniseriate trichome, middle cell cuboidal with granular contents; x117.—30 trichomes, the terminal cell with orange-brown contents; x 146.— . Four-celled, pilate ong, uniseriate, tapering trichome; note short cross-walls at junction of cells (above) and trichome base with short, basal cells (below); ese 33-34. Philydrum lanuginosum. — 33. Trichome — 34. Long, uniseriate, к ачы trichome; х40. 35- filament; x 40. — 36. Short, uniseriate pilate trichome of style; note granular contents of terminal cell; . x 146. — 37. Pilate itii of outer tepal surface. Note granular contents of terminal cell; x 106. 38-40. Pontederia cordata . —38. Four-celled uniseriate trichome. Note granular contents of terminal cell; Х51. —89. а ог Pos celled trichomes. Note small terminal cell with alc granular contents and (in middle t sub brown ergastic substance; х 73. — 40. Five-celled erus trichome, havi ving granular contents in terminal cell and dni -brown е substance іп middle cell; x51. -43. Reussia rotundifolia. —41. Five-celled, uniseriate trichome; x 39.— 2. Six-celled, uni- seriate trichome, vis granular contents in supra-basal cell; x 39. —43. Eight-celled, ше а E x example, Helmholtzia (Figs. 29, 31, 32) and Phil- ydrum (Figs. 33, 34) of the Philydraceae have elongate, uniseriate, tapering trichomes with two or three isodiametric basal cells and, in the more distal regions, steeply inclined, overlapping end walls (see Figs. 32, 33). These trichomes most resemble Tribonanthes, which, however, are ba- sally branched and have transverse, not inclined, end walls. Helmholtzia possesses, in addition to the above trichome type, occasional three- to four- celled, pilate trichomes (Fig. 30) with an ovoid to slightly elongate terminal cell containing a clear orange-brown ergastic substance similar to that found in perianth idioblasts in this and other genera (see Perianth tannin cells/idioblasts). These tri- chomes show some resemblance to the pilate tri- chomes of the Haemodoraceae, differing primarily in the contents and appearance of the terminal cell and in lacking the distinctive basal rosette of epi- dermal cells. Within the investigated members of the Pontederiaceae, the genus Heteranthera is largely glabrous but has some floral trichomes; these include: (1) multicellular, uniseriate staminal filament trichomes (Fig. 35); (2) short, pilate stylar trichomes with a terminal cell containing granular contents (Fig. 36); and (3) larger pilate trichomes, located on the outer tepal surfaces, with a globose terminal cell containing granular contents (Fig. 37). The so-called pilate trichomes of Heteranthera re- semble somewhat the pilate trichomes in the Hae- modoraceae, but (as in Helmholtzia) they lack the distinctive basal rosette cells. The one investigated species of Pontederia has several different, inter- grading trichome types, ranging from: (1) linear, Volume 77, Number 4 1990 Simpson 741 Phylogeny and Classification of Haemodoraceae uniseriate, 3-6-celled, with a densely granular ter- minal cell (Fig. 38); (2) short, 2-3-celled and cap- itate with a spherical to ellipsoid terminal cell (Fig. 39); and (3) uniseriate, of variable length, with one or more cells enlarged and containing a clear, or- ange ergastic substance similar to that in Helm- holtzia (Fig. 40). The pilate trichomes of Ponte- deria are like those in Helmholtzia of the Philydraceae and show some resemblance to the pilate trichomes in the Haemodoraceae. Reussia of the Pontederiaceae has several types of unise- riate perianth trichomes (Figs. 41-43), some of which have a terminal cell containing densely gran- ular contents. Many similarities are seen between the Hae- modoraceae and the outgroups with respect to tri- chome anatomy. In the present analysis if one or more outgroup(s) possessed a trichome type similar to that designated in the character analysis, then that feature was coded as being homologous with the condition found in the Haemodoraceae. It is thus hypothesized that the similarity of the pilate trichomes of Heteranthera (Pontederiaceae) or the uniseriate trichomes of Helmholtzia (Philydrace- ae), as examples, with members of the Haemodor- aceae reflects common ancestry. A possible diffi- culty with this, however, is that both outgroups are polymorphic with regard to trichome type. How- ever, when the two were coded as polymorphic (**?””) for all trichome characters (characters #7- 13) in a separate cladistic analysis, the topology of the resultant most parsimonious cladograms re- mains unchanged (see Cladistic Analysis). Character #14. Perianth apertures. Schiekia (Fig. 47) and Wachendorfia (Fig. 51) are similar in that the outer posterior tepal is basally fused to the two outer latero-anterior tepals and to the two inner latero-posterior tepals (Figs. 50, 52). At the basal junctions between the two outer latero- anterior tepals and the outer posterior tepal are distinctive slitlike pouches (termed ““apertures”” for Wachendorfia by Ornduff & Dulberger, 1978). Ornduff & Dulberger reported that in Wachen- dorfia paniculata nectar is produced from these perianth apertures. However, in view of the pres- ence of septal nectaries in Wachendorfia species (see Septal nectaries), it is probable that these apertures function solely as a collection site for nectar secreted from the ovary. The significance of these sites is unknown, as no insect or other visitors have been described for Wachendorfia. The perianth apertures in Schiekia probably func- tion similarly to those in Wachendorfia, although no observations have been published. All other genera of the Haemodoraceae with basically dis- tinct tepals have no basal fusion of the tepals and no perianth apertures. Many genera of the Hae- modoraceae have a syntepalous perianth, but fu- sion in these taxa is in the form of a complete basal perianth tube (see Perianth tube) and is treated as nonhomologous with the rather unique tepallary fusion in Schiekia and Wachendorfia. Among the outgroup taxa, all members of the Philydraceae have a four-parted perianth, the up- per component consisting of the fusion product of the inner posterior and outer latero-posterior tepals (Hamann, 1966). Members of the Pontederiaceae possess six imbricate tepals variously fused into a basal tube; in some taxa (e.g., Pontederia) lateral and anterior open slits are present at the base of the perianth tube. However, neither outgroup has the distinctive perianth apertures seen in Schiekia and Wachendorfia and are coded as lacking this feature. Characters #15, 16. Perianth tube. А basal perianth tube is possessed by five genera in the Haemodoraceae: Anigozanthos (Figs. 53, 54), Blancoa (Fig. 55), Conostylis (Fig. 56), Macro- pidia (Fig. 57), and Tribonanthes (Fig. 58). (Cono- stylis breviscapa lacks a perianth tube and has distinct tepals. In view of the numerous (ca. 24) species of Conostylis that possess a perianth tube, the distinct tepals of C. breviscapa are tentatively hypothesized to have evolved secondarily from an ancestral perianth tube; this matter needs further investigation.) All other members of the family, including Phlebocarya (Fig. 60), lack a complete perianth tube. As discussed under Perianth ap- ertures, the unique tepallary fusion in Schiekia and Wachendorfia is coded as nonhomologous with the tubular perianth of the above genera and is treated as a separate character. Of the taxa with perianth tubes, Anigozanthos, Blancoa, Conosty- lis androstemma, C. bealiana, and Macropidia have a very elongate perianth tube (Figs. 53-55, 57). The evolution of an elongate perianth tube in at least Anigozanthos, Blancoa, and Macropidia is almost certainly correlated with selective pres- sure for bird pollination (Hopper, 1977; Hopper & Burbidge, 1978; Keighery, 1981). Therefore, presence of an elongate perianth tube is designated as derived from an ancestor with a short perianth tube: PERIANTH TUBE ABSENT -[+15]> SHORT PERIANTH TUBE =[#16]— LONG PERIANTH TUBE. As discussed above (see Perianth apertures), the outgroup taxa have some variation of perianth fusion. All members of the Philydraceae have a 742 Annals of the Missouri Botanical Garden Go $ P274, 02 P > Т pA le A = » y e < #7 f А Ao Я ПОРТ AM Ee Vu yr 2 —— _ Es Haus 44-52. ia removed, show two sta xS. 7.— 46. Floral no: tT -50. Schiekia of two Кын i (s; X3.6. small (caducous) anther with basal constriction. — 50. minodes (s) a orinocensis — 48. Staminode, adaxial view; х8. Tue Floral diagram. 51, 52. Wachendorfia paniculata. — 51. "Ww le dis orp in the Haemodoraceae. 44-46. iA ad Vin — 44. Whole flower, of t an d s 1 single stamen; x2.8 Adaxial view of staminode; hole flower. Note ide erianth aperture (pa) ui one 49. One of two latero-posterior stamens. Note ho flower. Note perianth aperture (ра); x 2.8. — 52. Floral diagram. rather specialized four-parted perianth, the upper component consisting of the fusion product of the inner posterior and outer latero-posterior tepals. Because tepallary fusion in the Philydraceae is incomplete and rather specialized, it is coded non- homologous with the short or long perianth tube of the Haemodoraceae, which results from fusion of all six tepals. Members of the Pontederiaceae possess six imbricate tepals which vary from es- sentially distinct to being connate and forming a short or long perianth tube; thus, the Pontederi- aceae are coded as polymorphic for both charac- ters. Character #17. Perianth symmetry. Pyr- rorhiza (Fig. 44), Schiekia (Fig. 47), and Wach- endorfia (Fig. 51) have zygomorphic perianths. All other family members have basically actinomorphic perianths, with the exception of Anigozanthos and Macropidia, in which zygomorphy is thought to have been derived independently (see Perianth splitting). Volume 77, Number 4 Simpson 743 1990 Phylogeny and Classification of Haemodoraceae i sree что Е Do LE. иы. KES at Seat A VAT pg MER FIGURES 53-60. 54. A. си x1.35. ancoa canescens; X ginosa; X 2.4, 58-59. Шш variabilis. — 58. Whole flower; х 2.4. — 59. Шз E \ ЖД т [ү P Wie te, i oo Floral d in the opi uir 53, 54. Anigozanthos. —53. A. d m 0.—56. Conostylis aurea; x3. sn fa uli- Stamen, adaxial (left) and abaxial (right) views. Note connective appendages; х 2.0.— 60. Phlebocarya ciliata; x3.7 Among the outgroups, all species of the Phily- draceae have zygomorphic perianths. However, be- cause this family possesses a rather specialized perianth consisting of fusion of the posterior tepals and reduction of the latero-anterior tepals, perianth symmetry in the Philydraceae is coded as having uncertain homology (**?””) with that in the Hae- modoraceae. In the Pontederiaceae perianth sym- metry is either actinomorphic or more rarely zy- gomorphic and is coded as polymorphic (**?”). Among angiosperms as a whole, zygomorphy is generally considered to be a derived feature, usu- ally correlated with specialized pollination systems (Faegri & van der Pijl, 1966; Sporne, 1975). The relative ancestry of this feature and its significance in pollination mechanisms in the Haemodoraceae will be discussed with reference to the cladistic analysis. Character #18. Perianth splitting. Two oth- er family genera, Anigozanthos (Figs. 53, 54) and Macropidia (Fig. 57), also have zygomorphic peri- anths. However, the perianths of these taxa are syntepalous and tubular, not basically apotepalous as in the zygomorphic Pyrrorhiza, Schiekia, and Wachendorfia. More importantly, zygomorphy in Anigozanthos and Macropidia arises primarily by the longitudinal "splitting" of the tube along an anterior line (Fig. 54). Zygomorphy in these two genera almost certainly has evolved independently from (and is not homologous to) that in Pyrrorhiza, Schiekia, and Wachendorfia and is treated as a separate character. Among the outgroups and monocots as a whole, such perianth splitting is absent. It is extremely likely that zygomorphy in Anigozanthos and Mac- ropidia was derived from an ancestral tubular, actinomorphic condition, as occurs in Blancoa. This hypothesis is supported by the fact that all three of these genera have identical valvate peri- anths during the early bud stage (see below). Zygo- morphy in Anigozanthos and Macropidia prob- 744 Annals of the Missouri Botanical Garden ably evolved due to strong selective pressure for specialized bird pollination (Hopper & Campbell, 1977; Hopper & Burbidge, 1978) Character #19. Perianth aestivation. Anigozanthos, Blancoa, Conostylis, Macropidia, and Tribonanthes possess a valvate perianth at anthesis, in which the perianth lobes of the mature flower show no evidence of overlap (Figs. 53-58, 62). In Anigozanthos, Blancoa, and Macropidia the perianth lobes are valvate even during the earliest bud developmental stage (Fig. 61). How- ever, Conostylis species (Fig. 63) and Tribon- anthes (Fig. 64) clearly have an imbricate perianth aestivation in the bud stage; only when the flowers open are the tepals valvate. In all other genera of the Haemodoraceae, tepals are imbricate through- out floral development. An imbricate perianth, or evidence of such in cases of fusion, occurs in all species among the outgroups. It seems very likely then that the val- vate perianth of the above five genera is a derived feature. The a perianth of Anigozanthos, pidia may represent a further specialization, one probably correlated with the long perianth tube in these taxa (see Perianth tube). Because other species of Conostylis having elongate perianth tubes have not been studied for this feature, possession of a valvate perianth throughout floral development is not coded separately from a valvate perianth arent developmentally valvate lancoa, and Macro- present only at flower anthesis. Character #20. Perianth tannin cells. Distinctive perianth idioblast cells are present in ocarya, and Tribonanthes -67). These perianth idioblasts vary in length (1.5-6 times longer than broad), are ori- ented parallel to the tepal axis in the subepidermal layer, and are completely filled (presumably the vacuoles) with an orange to red-brown ergastic substance. The cells are scattered in the tepal among the more predominant clear parenchymatous cells. The ergastic substance looks like oil; yet staining reaction for fats and oils with Sudan IV (Johansen, 1940) was negative. Safranin red positively stains these contents, indicating the possible presence of tannins. The idioblasts are located throughout the perianth and occasionally in the ovary wall and placentae. Among all investigated members of the Haemodoraceae, perianth idioblasts are not present in the following genera: Anigozanthos, Barber- etta, Blancoa, Conostylis, Dilatris (two species), Lachnanthes, Macropidia, Pyrrorhiza, Schiekia, Wachendorfia (two species), and Xiphidium. All of the former taxa possess only clear, generally rectangular parenchyma cells comprising the non- vascularized tissue of the perianth (illustrated for Wachendorfia in Fig. 70). (Incidentally, raphide sacs are present in the tepals of all investigated members of the Haemodoraceae and both out- groups. a m the outgroup families, perianth tannin cells nb identical to those in Haemodorum, ebocarya, and Tribonanthes are present in all genera of the Philydraceae (Fig. 68) and in all investigated genera of the Pontederiaceae (Fig. 69), except for Heteranthera. Because of the great anatomical similarity between the perianth idio- blasts of the outgroups and those found in the Haemodoraceae, it seems highly probable that they are homologous structures and are so coded. The absence of these perianth idioblasts in Heteranthe- ra is tentatively hypothesized to be a secondary derivation within the Pontederiaceae. Characters #21, 22. Stamen number. The number of fertile stamens per flower in taxa of the Haemodoraceae is either six, three, or one. Schiek- ia has three fertile stamens, the anterior one of which is considerably larger than the other two, plus two staminodes (Fig. 48) positioned latero- anteriorly in a whorl outer to the fertile stamens (Fig. 50). Because the two staminodes of Schiekia may represent vestiges (or evolutionary precursors) of an outer stamen whorl, the androecium of Schiekia is coded as evolutionarily intermediate between six stamens per flower and three or one stamen(s) per flower. Pyrrorhiza has only one fer- tile stamen, plus two staminodes (Fig. 45) mor- phologically similar to those in Schiekia but dif- fering by being in the same whorl as the single FIGURES 61 - 70. Note valvate arrangement of tepals (0); х x30.—63. Conostylis priesii. Mature bud cross section overlapping inner tepal whorl (it); x 14.— 64. Tribonanthes variabilis. outer tepal (ot) overlapping inner Meh (и); x 26. 5 0. picatum; х 20 с eboca a AO acorifolia; X194. — p tei he cordata; х1 tanniniferous Шош х 164. — 61-64. Perianth aestivation. 61, 62. Anigozanthos flavidus. — 61. Immature bud cross section. 6.— 62. Mature bud cross section. Te als (t) remain valvately arranged; Note imbricate aestivation, with outer tepal whorl (ot) Bud cross section, intermediate stage. Note Perianth tannin idioblast cells (id). —65. Haemodorum .— 61. Tribonanthes variabilis, x 200. —68. Helmholtzia 4.— 70. Perianth cells of Wachendorfia thyrsiflora, which lack Volume 77, Number 4 impson 745 1990 Phylogeny and Classification of Haemodoraceae 746 Annals of th Missouri Eon Garden FIGURES 71-76. . Haemodorum spicatum; X3.4.— fertile stamen, i.e., opposite the inner whorl of tepals, not the outer as in Schiekia (Fig. 50). The staminodia in Pyrrorhiza are thus hypothesized to be homologous with stamens of an inner whorl. Based on this interpretation, the assigned morpho- cline for stamen number in the oo is: 6 STAMENS +-[#21]— 3 STAMENS + 2 LATERO-ANTERIOR STAMINODES эе, 3 ОК 1 STAMEN(S) One difficulty with the above morphocline con- cerns Schiekia and Pyrrorhiza. Although Schiek- ta may be intermediate between a six-staminate and three-staminate condition (because of the pres- ence of two latero-posterior staminodia), it is more likely intermediate between a three-staminate con- dition and the one-staminate morphology of Pyr- rorhiza. This latter interpretation is based on the presence in Schiekia of two reduced latero-pos- terior stamens with caducous anthers that greatly resemble and are likely homologous with the two staminodia of Pyrrorhiza. However, so as not to bias the present study, the latter hypothesis is treat- ed as an independent character (see Stamen di- morphism). In general, fewer than six stamens in mono- cotyledons is a condition thought to have arisen Floral morphology in the Haemodoraceae. — 71 idis — 72. Whole flower longitudinal T Note inferior ovary; x 3.0.— x 3.4. 72-73. Dilatris . Barberetta aurea; х3.4. . Outer tepal. Note apical glands; . Xiphidium coeruleum; x3.4.— 76. Lachnanthes caroliniana; by reduction from the ancestral condition of six stamens in two whorls (Dahlgren & Clifford, 1982); however, such generalized trends must be viewed with caution. With regard to outgroup comparison, all members of the Philydraceae have one stamen per flower, which interestingly is median anterior in position (similar to that in, e.g., Pyrrorhiza). The Pontederiaceae can have either six, three, or one MN per flower and are coded as poly- morphic (“2”). Characters #23-25. Stamen dimorphism. In the members of the Haemodoraceae with six stamens, all six anthers are of equal size. However, of the taxa with three anther-bearing stamens, in Dilatris (Fig. 72), Haemodorum (Fig. 74), Schiek- а (Figs. 47-50), and Xiphidium (Fig. 75), the anther of the adaxial stamen (relative to cyme axis) is significantly larger than those of the abaxial stamens. (Schiekia also has two staminodia; see characters #21, 22, Stamen number.) All three stamens and anthers are equal in the other genera with three stamens: Barberetta (Fig. 71), Lach- nanthes (Fig. 76), and Wachendorfia (Figs. 51, 22). The filament of the odd stamen may be either longer (Haemodorum, Schiekia, and Xiphidium) or shorter (Dilatris) than the filaments of the two Volume 77, Number 4 1990 Simpson 747 Phylogeny and Classification of Haemodoraceae equal stamens. Because of the positional similarity of the odd anther, however, these are all viewed as homologous features. In Schiekia the median anterior stamen is considerably enlarged relative to the two latero-posterior stamens (Fig. 47). The anthers of the two latero-posterior stamens of Schiekia are somewhat caducous (Figs. 47, 49), and the filaments of these stamens greatly resemble the staminodia of Pyrrorhiza (Figs. 44, 45). There- fore, it is hypothesized here that the two latero- posterior stamens of Schiekia are homologous (and evolutionarily intermediate) to the two latero-pos- terior staminodia of Pyrrorhiza. Furthermore, the occurrence of two latero-posterior staminodia in Pyrrorhiza may be interpreted as an extreme end- point in anther dimorphism. Thus, a hypothesized morphocline for anther dimorphism in the family is: ANTHERS OF EQUAL SIZE -[#23]- 1 LARGE ANTERIOR + 2 SMALL POSTERIOR ANTHERS -[#24]> 1 LARGE ANTERIOR + 2 CADUCOUS LATERO-POSTERIOR AN- THERS -[+425]> 1 ANTERIOR ANTHER + 2 LATERO-POSTERIOR STAMINODIA. Among the outgroup families stamen dimor- phism (or apparent stamen reduction) is common. All species of the Philydraceae have a single stamen in posterior position, as is the odd stamen in the dimorphic members of the Haemodoraceae. How- ever, because the homology of stamen dimorphism in the Philydraceae is uncertain, and because the presence of a single stamen in this family was taken into account previously (characters #21, 22, Sta- men number), the Philydraceae were coded as uncertain (**?”) for characters #23-25. Anther dimorphism varies considerably in the Pontederi- aceae. Some species exhibit no anther dimorphism. Dimorphic stamens are present in Heteranthera, Monochoria, Pontederia, and Scholleropsis; in Heteranthera there is usually one large anterior stamen and two smaller latero-anterior stamens. (Note the positional difference to that in the Hae- modoraceae.) In some Heteranthera species and in Hydrothrix of the Pontederiaceae, only one stamen is present. The Pontederiaceae are coded as polymorphic (**?””) for characters #23-25 Character #26. Stamen connective append- ages. Distinctive lobed appendages are present on the upper abaxial stamen connective in all species of Tribonanthes (Fig. 59). Such stamen append- ages are not found in the family or outgroups, although Anigozanthos may have mucronate an- thers. Characters #27, 28. Pollen aperture. Eight genera of the Haemodoraceae have a monosulcate pollen aperture type, as in Haemodorum (Fig. 77), whereas six genera have porate apertures (Figs. 78-80, 82; see Simpson, 1983). Of the latter, five genera have 2-3 apertures (Figs. 78-80). Tribo- nanthes differs in having 5-7 porate apertures (i.e., oligoforaminate; Fig. 82). Thus, this character is coded as: MONOSULCATE -[#27]> 2-3- PORATE -[+428]> OLIGOFORAMINATE. Of the six genera with porate pollen apertures, four (Anigozanthos, Blancoa, Conostylis, and Macropidia) are similar in having protruding, hemispheric aperture walls essentially devoid of exine (Figs. 78, 80); the other two genera, Phlebo- carya (Fig. 79) and Tribonanthes (Fig. 82), have rather flattened aperture walls with scattered exi- nous elements (Simpson, 1983). Although this ad- ditional feature is not taken into account in the character coding, it will be discussed after the cladistic analysis. monosulcate aperture type is considered to be ancestral for both the monocotyledons and the angiosperms (Zavada, 1983b; Walker & Doyle, 1975) and is likely ancestral for the Haemodora- ceae. Among the outgroup taxa, all members of the Philydraceae have a monosulcate aperture (il- lustrated by Helmholtzia in Fig. 81; see Simpson, 1985a). All members of the Pontederiaceae have disulculate apertures (illustrated by Pontederia in Figs. 83, ; see Simpson, 1987). Because the disulculate aperture type in the Pontederiaceae is probably derived from a monosulcate condition, and is almost certainly not homologous with the diporate aperture type in the Haemodoraceae, the Pontederiaceae are coded as having the equivalent of a monosulcate aperture type. Characters #29, 30. Pollen sculptur- ing. Seven genera of the Haemodoraceae (all with monosulcate apertures) possess verrucate ex- ine wall sculpturing, consisting of appressed wart- like projections of exine (illustrated by Haemo- by Dilatris in Fig. 86). Schiekia, also ате чазы differs from the above dorum in in having foveolate sculpturing with minute outer pores (Fig. 87). Six genera of the family, all of which have porate apertures, have distinctive ru- gulate (brainlike) exine sculpturing (Fig. 88; see Simpson, 1983). All three types of pollen wall sculpturing seen in the Haemodoraceae are found in the outgroups. Within the Philydraceae three of four genera have foveolate (to reticulate) sculpturing (Fig. 89), which somewhat resembles Schiekia; Philydrella of the Philydraceae possesses what is described as a ru- gulate sculpturing, but which does not greatly re- Annals of the Missouri Botanical Garden 748 саи? л ad ES Ee eat FIGURES 77-84. spicatum (monosulcate); х 3,3 79. Ph ус) ciliata (д ойша acorifolia баш боа x2,570.— 84. Pade cordata (disulculate);— 83; х [сысы х Pollen shape and aperture Top o ds in the Haemodoraceae and outgroups. — 77. Haemodorum 20.—78 hos flavidus (diporate with hemispheric apertures); x1,110.— 2,160. "30. Const beliana (triporate with hemispheric apertures); x1,390. — ribonanthes variabilis (oligoforaminate); х 1,760. 83- 70. semble the rugulate sculpturing found in six genera of the Haemodoraceae (see Simpson, 1985a). cause the Philydraceae have two of the sculpturing types of the Haemodoraceae, they are coded as polymorphic (**?") for both characters. In fact, it seems quite likely that the resemblance in sculp- turing between the Philydraceae and the Haemo- doraceae is homoplasious anyway, as all Philydra- ceae have a quite different exine structure (see below, Exine wall structure). In the Pontederia- ceae all investigated genera except Pontederia have verrucate sculpturing (illustrated for Heteranthera Volume 77, Number 4 1990 Simpson Phylogeny and Classification of Haemodoraceae 749 ao. ^V ` k rsen — FIGURES 85-91. in Fig. 90 and Zosterella in Fig. 91), which is virtually identical to that found in seven genera of the Haemodoraceae. In view of the similarity of exine structure between the Pontederiaceae and the verrucate Haemodoraceae (see Exine wall structure), it is very probable (and is coded as such) that the two families are homologous in terms of exine sculpturing as well. The most likely morphocline for pollen sculp- turing in the Haemodoraceae seems to be the fol- lowing: FOVEOLATE -[429]^ VERRUCATE Pollen wall sculpturing in the Haemodoraceae and outgroups. — 85. (verrucate); X 6,610.— 86. Dilatris pilansüi (verrucate); x 4,400. — 87. Schiekia orinocensis (foveolate); x 8,090. 88. Anigozanthos flavidus (rugulate); x 2,420.—89. Helmholtzia acorifolia (foveolate-reticulate); x 4,740.— 90. Heteranthera reniformis (verrucate); x 2,480.— 91. Zosterella dubia (verrucate); х 5,970. Haemodorum spicatum —-[430]^ RUGULATE. Although the intergra- dation between the sculpturing types is not clear in itself, it is quite probable that the verrucate sculpturing type is ancestral for the Haemodora- ceae as a whole, evidence for this being the identical sculpturing type (and exine structure; see below in the Pontederiaceae. Thus, the rugulate and fove- olate sculpturing types in the Haemodoraceae are hypothesized to have evolved independently from an ancestral verrucate type (to be tested by the cladistic analysis). 750 Annals of the Missouri Botanical Garden dap 92-99. n grain, о above 93. Close-up of aperture border of enlarged proxima 1al ( ); x5,070. 95- e psilate, micropore-pitted aperture (b) and proximal verrucae (upper left); x 2,710.— a. Palynological features in the Haemodoraceae. 92-94. Wachendorfia thyrsiflora. —92. Whole No border (b) and enlarged proximal verrucae 4. Close-up 5. Whole pollen grain, aperture facing. Note psilate, micropore-pitted arid border (b); x 2,430.— 96. pipe surface, showing aperture border (b) and enlarged proximal verrucae (arrow); X of proximal surface. Note absence of enlarged verrucae; Character #31. Smooth pollen grain apertural borders, consisting of a band of deni material with minute perfo- reagent are present in Wachendorfia (Figs. 92, 93), Barberetta (Figs. 95, 96), and Dilatris (Figs. 97, 98). differ slightly from those in the other two genera Pollen apertural border. The pollen apertural borders in Dilatris in having larger perforations (being “foveolate””) in contrast to the micropores present in Barberetta 090. 97-99. Dilatris corymbos showing aperture border (b); x 925. — n Close-up of psilate, foveolate nas border (b); x 2,460. 2,570. hole pollen grain, aperture facing, — 99. Close-up and Wachendorfia. Because of their overall sim- ilarity, all three genera are coded as possessing this feature. Similar pollen grain apertural borders are absent in other investigated members of the Hae- modoraceae, although a tendency for such a border may be seen in Xiphidium (Simpson, 1983). An apertural border is absent also in all investigated members of the Philydraceae (Simpson, 1985a and Pontederiaceae (Simpson, 1987). Volume 77, Number 4 1990 Simpson 751 Phylogeny and Classification of Haemodoraceae Character #32. Pollen with large proximal verrucae. Large verrucate exine elements are present on the proximal pollen grain surface of Barberetta (Fig. 96) and Wachendorfia (Figs. 92- 94). Similar elements do not occur elsewhere in the family or among any investigated outgroups. In particular, Dilatris, which resembles Barberetta and Wachendorfia in having a pollen apertural border, lacks any indication of enlarged verrucate elements on the proximal pollen grain surface (see Fig. 99). Characters #33-37. Exine wall struc- ture. Five basic types of pollen grain exine wall structure can be identified in the Haemodoraceae (see Simpson, 1983). Lachnanthes (Figs. 190, 101) and Haemodorum (Fig. 102) have a “one-lay- ered” exine structure, consisting of baculate (rod- shaped) elements that are closely appressed and generally basally fused. Ten other genera of the family have a two-layered exine wall, with inner and outer layers delimited by a distinctive “сот- missural line” (Figs. 103, 104); these inner and outer exine layers are ektexinous based on cyto- chemical tests and have similar TEM staining prop- erties (Simpson, 1983). Of these ten genera, six differ in having an inner exine layer composed of papillate exinous elements (Fig. 104), while those of Phlebocarya are restricted to the apertural re- gion. Pyrrorhiza (Fig. 105) and Schiekia (Fig. 106) possess, respectively, two- and three-layered exine walls. These two genera resemble one another (and differ from other family members) in that the subexterior exine wall is granular and discontinuous in composition. In Haemodorum and Lachnanthes, the single- layered exine wall resembles and is probably ho- mologous with the outer exine layer of those taxa that have a two-layered structure. Їп fact, the pollen exine wall of Haemodorum has an occasional scanty inner exine layer, perhaps indicative of an ancestral inner layer (Simpson, 1983). (Develop- mental studies to test this hypothesis are in progress by the author.) Similarly, in Pyrrorhiza and Schiekia the outermost layer of exine is probably structurally homologous to the single exine layer of Haemodorum and Lachnanthes. This can be seen near the apertural region, where all but the outermost exine layer disappears (Simpson, 1983). As seen in Figure 112 (after Simpson, 1983), a gradation among observed exine wall structural types of genera in the Haemodoraceae can be identified. However, because of the uncertainty of some of the intergrading states of this morphocline, and because its length could possibly bias the study, the character “ехіпе wall structure" was divided into the following two-state (absence or presence) characters: Characters #33, 34: Number of exine wall layers, coded as the following linear morphocline: 1-LAY- ERED -[4$33]^ 2-LAYERED <[+434]- 3-LAYERED. Character #35: Exine wall, if two-layered, with papillate inner exine elements. This includes Phleb- ocarya, which possesses papillate inner exine ele- ments only near the apertural region. Taxa without a two-layered exine (Наатофогиль, Lachnanthes, and Schiekia) are coded as “X.” Character #36: Exine wall, if two-layered, with only papillate elements making up the inner wall. This condition is present in Anigozanthos, Blan- coa, Conostylis, Macropidia, and Tribonanthes. Again, the three taxa which lack a two-layered exine are coded as “X.” Character #37: Subexterior exine wall discontin- uous. This feature links the common exine mor- phology of Pyrrorhiza and Schiekia. Figure 165B illustrates the character coding for exine wall struc- ture (characters #33-37) Among the outgroups, all Philydraceae have a typical homogeneous tectate-columellate ехіпе structure (Fig. 107) with characteristic lamellar deposits inner to the foot-layer (see Simpson, 1985a). The directionality of the tectate-columel- late exine structure with reference to the specific types occurring in the Haemodoraceae is uncer- tain; thus, the Philydraceae are coded as having uncertain homology (**?") for characters 433-37. In contrast, several members of the Pontederiaceae have an exine structure very similar to that in the Haemodoraceae (see Simpson, 1987). A “‘one-lay- ered" exine structure, identical to that of Hae- modorum and Lachnanthes of the Haemodora- ceae, is seen in two genera of the Pontederiaceae: Eichhornia and Hydrothrix (Fig. 110). Heter- anthera, Reussia, and Zosterella (Fig. 111) of the Pontederiaceae have a two-layered exine with a "commissural line" delimiting inner and outer lay- ers, resembling that found in ten genera of the Haemodoraceae. Four genera of the Pontederia- ceae, Heteranthera, Monochoria (Fig. 108), Pon- tederia (Fig. 109), and Scholleropsis, have what is described as a tectate-columellate exine struc- ture. However, in these taxa the foot-layer is thin or discontinuous, the columellae are short and ill- defined, and the tectum is thick and composed of rod-shaped elements (resembling the rod-shaped elements of the verrucate members of the Hae- 752 Annals of the Missouri Botanical Garden FIGURES 100-111. Ехіпе wall ultrastructure in the Haemodoraceae and outgroups. — 100, ‚ Lachnanthes caroliniana (one-layered exine composed of laterally appressed and basally fused rod-shaped duse x11,4 (1 (101).—102. Haemodorum simplex (one-layered, compose ed of laterally appressed rod-shaped elements); x 29,900.— 103. Xiphidium coeruleum (two-laye e exine, inner layer papillate; note commissural line at arrows); X 20,700.— 104. Tribonanthes variabilis. feid layered exine, inner layer papillate; note commissural , Volume 77, Number 4 1990 Simpson 753 Phylogeny and Classification of Haemodoraceae ala Ged WACHENDORFIA Кр GA BARBERETTA ARO =» GA A SCHIEKIA PYRRORHIZA FIGURE 112. LACHNANTHES HAEMODORUM an / \ кое PHLEBOCARYA | Ss Й ургоо TRIBONANTHES ANIGOZANTHOS BLANCOA CONOSTYLIS MACROPIDIA XIPHIDIUM DILATRIS Hypothesized intergradation series of exine ultrastructure morphology in the Haemodoraceae. Note uncertainty as to homology of two-layered exine of Pyrrorhiza modoraceae). Thus, Simpson (1987) proposed that this ““modified” tectate-columellate exine structure in the Pontederiaceae may not be homologous with the more typical tectate-columellate structure (e.g., in the Philydraceae and in many other monocots) and may have been secondarily derived from an exine structure comprised of rod-shaped elements. (Developmental studies are underway by the author to test this hypothesis.) Because both a one-layered exine and a two-layered exine (with a delimiting ‘commissural line") occur in the Pontederiaceae, this outgroup is coded as polymorphic (**?") fo characters #33 and #35, 36. Character #34 is coded as lacking (state 0) a three-layered exine, and character #37 is coded as lacking (state 0) a discontinuous subexterior exine wall. Character #38. Enantiostyly. Enantiostyly is the curvature of the style to either the right or left, thus defining so-called “right-handed” and “left-handed” flowers. Displacement of one or more stamens usually accompanies enantiostyly, which results in an asymmetric flower. aemo- doraceae several genera exhibit enantiostyly. All five species of the zygomorphic Wachendorfia have enantiostylous flowers, in which the arcuate style is deflected to one side of the flower and the median anterior stamen is deflected to the other side (Fig. 51). Ornduff € Dulberger (1978) have hypothe- sized that such stylar enantiomorphy functions to promote pollination between flowers of different morphs, in effect promoting outcrossing. Flowers of Schiekia have a similar enantiostyly in which the style and median anterior stamen are oppositely displaced (Fig. 47); the two latero-posterior sta- mens ike those of Wachendorfia) are consid- erably smaller. Both Schiekia and Wachendorfia have what are presumed to be nectar guides near n the e line at arrows); X17,700.— 105. Pyrrorhiza neblinae (two-layered exine; note amorphous, discontinuous inner exine wA de x19,400.— 106. Schiekia orinocensis (three-layered exine; note amorphous and discontinuous middle layer), s an 21, [oA — 107. Orthothylax glaberrimus (tectate-columellate exine with lamellae inner to foot-layer); x 38,600. — nochoria vaginalis (tectate- columellate exine with ill-defined ит pet x 24,300. 18,500.— exine composed of laterally appressed and basally fused rod-shaped ele — 109. Pontederia ec no Ег gardneri (one-layered emen; x39, 000.—111. Zosterella dubia (two-layered exine, inner layer papillate; note commissural line at arrows); x1 754 Annals of the Missouri Botanical Garden the base of the posterior tepals. Thus, it is probable that in both of these genera (which have zygo- morphic, horizontal flowers) an insect visitor would alight in a consistent orientation, forwardly directed toward the ovary base where nectar would collect. Pollen initially transferred to one side of the insect's body would, in subsequent visitations, more likely contact the stigma of a flower of opposite hand- edness (see Ornduff & Dulberger, 1978, re. Wach- endorfia). Pyrrorhiza, the only other genus of the family with zygomorphic flowers (not caused by longitudinal splitting of the perianth; see Perianth splitting), does not possess enantiostyly. The style of Pyrrorhiza is relatively straight and is positioned directly above the single anterior stamen (Fig. 44). Styles of four other family genera, Barberetta (Fig. 71), Dilatris (Fig. 72), Lachnanthes (Fig. 6), and Xiphidium (Fig. 75), also are strongly curved to one side of the flower. In Haemodorum the styles of adjacent flowers in a flower pair are each slightly incurved, forming mirror image pairs. These styles are only slightly curved and are not strongly displaced to one side (Fig. 74); however, they are probably homologous with the strongly displaced stylar curvature in the other six genera and are so coded. In contrast to Schiekia and Wachendorfia, however, the above genera have actinomorphic and erect (not zygomorphic an horizontal) flowers without any type of bilaterally symmetric nectar guides. It is, therefore, likely that an insect visitor to flowers of any of these four genera would be positioned inconsistently with re- gard to the deflected style. Enantiostyly in these taxa may be adaptive in simply decreasing the chance of self-pollination by physically displacing the stigma from anthers. Among the outgroup taxa, enantiostyly occurs in all species of the Philydraceae. In fact, the style is displaced to one side of the (sole) anterior stamen, similar to enantiostyly in the Haemodoraceae. In the Pontederiaceae only Heteranthera and Mono- choria have **weakly enantiostylous flowers" (Eck- enwalder & Barrett, 1986); all other Pontederia- ceae lack enantiostyly. The Pontederiaceae are thus coded as polymorphic (**?””) for this character. Character #39. Ovary position. of the Haemodoraceae have an inferior ovary. Cer- tain species of Conostylis have a mostly inferior Nine genera ovary; these are coded as having inferior ovaries. The other five family genera clearly have superior ovaries. Within the angiosperms as a whole, an inferior ovary is generally considered to be a derived fea- ture (Bessey, 1915; Sporne, 1975), evolving pos- sibly because of selective pressure by floral her- bivores or pollinators (Grant, 1950) or in response to an adaptive advantage caused by increased pro- tection of seeds or increased energy allocation to developing ovules (Stebbins, 1974). (See, however, Eyde & Tseng, 1969, regarding a possible case of derivation of a superior ovary from an ancestral inferior ovary.) All members of the two designated outgroups in this study have superior ovaries and are so code Character 440. Septal nectaries. Within the Haemodoraceae all genera possess septal nectaries, with the exception of Xiphidium. In genera with inferior ovaries, the septal nectaries consist of slit- like channels, lined with a single epithelial layer, which traverse the septa at the upper part of the ovary and release nectar at the ovary apex near the base of the style. Among family members with inferior ovaries, Dilatris and Phlebocarya are ex- ceptional in having extremely minute septal nec- taries, consisting of a very short chamber located at the extreme ovary apex. Of the genera with superior ovaries, Barberetta, Pyrrorhiza, Schiek- ia, and Wachendorfia have very small septal nec- taries, arising and opening up at the ovary base. In the case of Schiekia and Wachendorfia, nectar secretions presumably arise from these septal nec- taries and collect in the two lateral “apertures” characteristic of these two genera. The complete absence of septal nectaries in Xiphidium (whic has a superior ovary) may be correlated with the fact that it is visited by pollen-feeding, not nectar- feeding, bees via an anther vibrational (“‘buzz’’) mechanism (Buchmann, 1980). Thus Xiphidium may represent a specialization in which the selec- tive pressure for nectar secretion was eliminated, establishing pollen as the primary visitor attractant. In view of the reduced septal nectaries in most superior-ovary taxa, it is not unreasonable to en- vision a loss of these structures in Xiphidium, particularly in light of its pollination mechanism. mong the outgroup taxa, septal nectaries are absent from all four genera of the Philydraceae. Septal nectaries are present in at least Eichhornia and Pontederia of the Pontederiaceae but are ab- sent in Heteranthera of that family (present study). The septal nectaries of the Haemodoraceae are very likely homologous with those in the Ponte- deriaceae and probably with those of numerous other monocots. However, because of their vari- ability, the Pontederiaceae are coded as polymor- phic (**?”’) for this character. Character #41. Fertile carpels. All species of the Haemodoraceae have three connate carpels. Volume 77, Number 4 1990 Simpson 755 Phylogeny and Classification of Haemodoraceae In the ee iq vini etica however, the seeds of two carpels y the аш carpel is fertile at maturity. aia aborting carpels are not found in any other mem- bers of the Haemodoraceae. (Barberetta is also unique in the family in having the style displaced laterally relative to the ovary apex; this displace- ment is certainly correlated with the development of only one carpel containing one ovule.) Among the outgroup taxa, abortive carpels are absent in the Philydraceae. Abortive carpels are found only in Pontederia and Reussia of the Pon- tederiaceae, being absent in all other genera of the family. Thus, the Pontederiaceae are coded as poly- morphic (**?””) for this character. Character #42. Locule number. All mem- bers of the Haemodoraceae have three locules, with the sole exception of Phlebocarya. Ovaries of Phlebocarya possess three basal, epitropous ovules with rudimentary or only partially protruding lat- eral septa and no apical septa. Thus, the ovary of Phlebocarya is technically uniloculate. A three-loculate ovary is found in all members of the Pontederiaceae, and in all Philydraceae with the exception of Philydrum. Here most of the apical volume of the ovary is unilocular, with pro- truding, bibrachiate parietal placentae bearing nu- merous ovules. Thus the Philydraceae are coded as polymorphic (**?””) for this character. Character #43. Placental sclereids. Dis- tinctive sclereids occur in the ovary placental tis- sues in six family genera: Anigozanthos (Figs. 113, 114), Blancoa, Conostylis, Macropidia, Tribon- anthes (Fig. 115), and Phlebocarya (Fig. 116). These placental sclereids consist of isodiametric to slightly elongate (generally 3—4 times longer than road) cells with moderately thick secondary cell walls. Characteristic of these sclereids are the pres- ence of numerous spherical, nodulelike structures attached to the inner surface of the secondary cell wall (Figs. 114-116, 119, 120); both walls and nodules stain densely with safranin and are appar- ently lignified. The placental sclereids are distrib- uted somewhat randomly, either singly or in small clusters, throughout the nonvascular tissue of pla- centae and styles and occasionally the ovary wall. Placental sclereids were not observed in any organ of the other eight family genera: Barberetta, Di- latris, Haemodorum, Lachnanthes, Pyrrorhiza, Schiekia, Wachendorfia, and Xiphidium These characteristic placental sclereids are pres- ent in all four genera of the Philydraceae (illus- trated by Philydrum in Figs. 117—119). These sclereids also occur in at least Pontederia of the Pontederiaceae (Fig. 120) but are absent in Het- eranthera reniformis of that family. Despite the observed polymorphism, it seems very likely that the sclereids of the Pontederiaceae are homologous to those of the Haemodoraceae in view of their anatomical and positional similarity. Because the independent evolution of sclereids in the Ponte- deriaceae seems quite unlikely, this family is coded as having placental sclereids present. Of the above genera having placental sclereids, four also possess tannin cells in the ground tissue of ovary placentae: Helmholtzia, Philydrum (Fig. 117), Phlebocarya (Fig. 116), and Pontederia (Fig. 120); the remaining five genera possess only placental sclereids. These placental tannin cells are usually found in small clusters in the ground tissue of ovary placentae. They consist of generally iso- diametric (occasionally slightly elongate) cells with thin, unlignified cell walls. The entire cell lumen of these cells is filled with an ergastic substance that stains very densely with safranin; this is char- acteristic of tanniniferous compounds (Johansen, 1940), though no specific chemical tests were made. These placental tannin cells are very similar to and very likely homologous with the perianth tannin cells discussed above (see Perianth tannin cells). Characters 144—47. Ovule morphology and number. The morphology and number (per car- pel) of ovules in the Haemodoraceae are quite variable (see Figs. 121-139). Virtually all ovules in the family have a narrow micropylar “neck” arising from the ovule body. The shape of the ovule body itself, however, can be somewhat globose, as in Dilatris (Fig. 122), Haemodorum (Fig. 124), and Lachnanthes (Fig. 123), or variably elongate and often flask-shaped. Placenta morphology, ovule position, and ovule number also vary. In most taxa the Mina are RpoHopeus t in position, with the hlebo- н 15 йе only genus of the family with epitro- pous ovules; these are arranged, one per carpel, in a ring, arising from the base of the ovary (Fig. 137). Lachnanthes (Fig. 123) and Schiekia (Fig. 126) are similar in having 5-7 and (3)4 ovules, respectively. These ovules are pleurotropously po- sitioned (i.e., with the micropyle directed sidewise, towards the central flower axis) on a placenta that protrudes into the locule. This placenta is quite enlarged and peltiform in Lachnanthes, having marginally disposed ovules (Fig. 123). Curiously, however, Pyrrorhiza (Fig. 125) and Haemodorum (Fig. 124), which possess two hypotropous ovules per carpel, have an enlarged placenta similar to that of Lachnanthes and Schiekia. In addition, an 756 Annals of the Missouri Botanical Garden RES 113-120. Placental tissue anatomy in the зол ишер and outgroups. 113, 114. Anigozanthos i. —113. Elongate sclereid (scl) with nE cell ET x 347.— 114. Close-up of sclereid, showing nodulelike structures (arrow) appressed to inner cell w hae —115. Tri кшз variabilis. Note lignified sclereid pu with nodulelike ы (arrow); х85 6.—1 Bag elon ciliata. Note sclereid (scl) with nodules н апд tannin cell (їс); x856. 117-119. Philydrum o 117. Sclereid (scl) and tannin cell (tc); х 443. —118. Elongate sclereids with nodulelike structures (arrow); x 437.— 119. Close- up of sclereid (scl), showing nodules (arrow); х 949, —120. Pontederia cordata. Note tannin cells (tc) and sclereid (scl) with nodules (arrow); x 887. Volume 77, Number 4 1990 Simpso 757 scd and Classification of Haemodoraceae FIGURES 121-144. 122. Dilatris corymbosa; x19.— 125. Pyrrorhiza neblinae; x19. с. Schiekia orinocensis; x19. 128. oe coeruleum; «з — 129, 130. оо flavidus; x9.5 x9.5 air xd (133). — 134, 135. Conostylis aurea; 419 (135). a Macropidia fuliginosa; x19.— 38, 19.— 132, 133. Blancoa canescens; zanthos rubra; x9.5 (134), x Ovule ge б in the Haemodoraceae and cra — 121. Barberetta aurea; x19.— . Lachnanthes caroliniana; x19.— . Haemodorum spicatum; x19.— Wachendoría thyrsiflora; x19.— 129), x19 (130). —131. Anigo dt ~ Phlebocarya ciliata; x19.— Tribonanthes variabilis; x9.5 (138), x19 (139). — 140, 141. m zia acorifolia; X19 (140), x9.5 (141).— 142, 143. Heteranthera reniformis; x9.5 (142), x19 (143). — 144. Pontederia cordata; x19. enlarged placenta consisting of a ringlike mass of tissue surrounding the base of the ovule is present in Barberetta (Fig. 121), Dilatris (Fig. 122), and Wachendorfia (Fig. 127). (In Dilatris this ring of tissue expands during seed development, forming a false aril; see De Vos, 1956.) It seems likely that the enlarged placentae of all seven of these genera are homologous; however, it is not immediately evident which morphology (i.e., the ringlike pla- centa surrounding a single ovule or the peltiform placenta bearing several marginal ovules) is most ancestral. Among other OTUs, Anigozanthos rubra (Fig. 131) has two hypotropous ovules per carpel, and Macropidia (Fig. 136) has a single ovule per car- pel. In both of these taxa, a distinct ringlike or enlarged placenta like that found in other Hae- modoraceae is lacking; however, the inner ovary wall surrounding the ovule(s) is somewhat swollen and may be homologous to the enlarged placenta 758 Annals of the Missouri Botanical Garden BARBERETTA WACHENDORFIA ———> м —_—> Y SCHIEKIA LACHNANTHES ; PYRRORHIZA ANIGOZANTHOS DILATRIS MACROPIDIA OR 9 BLANCOA TRIBONANTHES PHLEBOCARYA —. Ej ANIGOZANTHOS CONOSTYLIS ТҮҮЛҮ Ficu 45. Hypothesized intergradation series of ovule types in the Haemodoraceae. Note uncertainty of R evolutionary direction of single, epitropous ovule in PAlebocarya. discussed above. Finally, numerous (up to 40) ovules pel, is questionably linked to taxa with one hypo- per carpel are found in Anigozanthos flavidus — tropous ovule/carpel. (Figs. 129, 130), Blancoa (Figs. 132, 133), Cono- In order to code the character “оуше mor- stylis (Figs. 134, 135), Tribonanthes (Figs. 138, phology and number" numerically without biasing 139), and Xiphidium (Fig. 128). Although the the input, this feature was divided into the following placentae in these taxa may be somewhat expanded four characters. (e.g., certain species of Conostylis), it appears unlikely that the greater ovule number in these Characters #44, 45: Ovule position: PLEU- taxa is directly related evolutionarily to that in ROTROPOUS -[444]^ HYPOTROPOUS Schiekia and Lachnanthes. =[#45]— EPITROPOUS. Note that the epitro- The possible evolutionary changes in ovule num- — pously positioned ovules of Phlebocarya are evolu- ber, ovule morphology, and placenta morphology tionarily linked to the hypotropous condition (Fig. in the Haemodoraceae are rather complex. An 145). Correspondingly, it seems quite unlikely that unambiguous intergradation series is not evident. the specialized pleurotropous ovules arising from a However, the morphocline presented in Figure 145 — peltiform placenta are evolutionarily intermediate is presented as the most likely a priori hypothesis. to a hypotropous and epitropous condition (Fig. Note that Lachnanthes and Schiekia, having 4- 145). 7 pleurotropous ovules/carpel, are coded either as derived from two-ovulate taxa or as intermediate Character #46: Ovule number one per carpel vs. between two-ovulate taxa and numerous-ovulate — two-numerous per carpel. This character allows taxa. Phlebocarya, with one epitropous ovule/car- for a single evolutionary step between those taxa Volume 77, Number 4 1990 Simpson 759 Phylogeny and Classification of Haemodoraceae with one ovule and those with two or more (see Figs. 145, 165C). Character #47: Ovule number one or two per carpel vs. numerous per carpel. Lachnanthes and Schiekia, having (respectively) 4 and 5-7 ovules per carpel are coded as “Х” so as to avoid bias for evolutionary linkage either directly with two- ovulate taxa or intermediate between two- and nu- merous-ovulate taxa (Fig. 165C). Figure 165C il- lustrates the character coding for ovule position and number (characters 444-47). Note that the number of steps between adjacent ovule groupings is one. Among the outgroup families, all members of the Philydraceae have numerous ovules per carpel with a hypotropous to pleurotropous position (il- lustrated for Helmholtzia in Figs. 140, 141). How- ever, in view of the distinctive placental morphol- ogy of those Haemodoraceae with pleurotropous ovules (i.e., positioned along the margin of a pro- truding placenta), the pleurotropous ovule position in the Philydraceae is not coded as being homol- ogous with that in the Haemodoraceae. In the Pontederiaceae all genera have nume ally hypotropous ovules (illustrated for Heteranthe- ra in Figs. 142, 143), with the exception of Pon- tederia (Fig. 144) and Reussia, which have one epitropous ovule per ovary (contained in the single fertile carpel). Thus, the Pontederiaceae are coded as polymorphic (**?””) with respect to characters #45-47 (but not, of course, with respect to char- acter #44). rous, gener- Characters #48-50. Seed morphology. Seed morphology in the Haemodoraceae is quite vari- able. Dilatris, Haemodorum, and Lachnanthes have glabrous, discoid seeds (generall proximally and convex distally) which are peltately attached (i.e., the funiculus is positioned centrally on the proximal face). Various degrees of a mar- ginal wing occur on the seeds of these taxa. Seeds of Dilatris (Fig. 146) are comparatively large (cer- tainly correlated with having only one ovule per carpel) and have a narrow encircling wing. modorum seeds (two per carpel) are slightly seller and have a prominent marginal wing (Fig. 147). Seeds of Lachnanthes (5—7 per carpel) have only a slight marginal wing (Fig. 148), which develops from the growth and “buckling”” of the outer in- tegument (Simpson, 1988). Seeds of Pyrrorhiza (Figs. 149, 150) resemble those of the above three genera in being rather discoid and peltately at- tached but differ in having a dense tomentum o mostly marginal trichomes. Wachendorfia seeds concave range from globose (Fig. 154) to ovoid (Fig. 155) but are consistently pubescent to tomentose throughout. The individual trichomes of the seeds of Wachendorfia are quite similar to those of Pyr- rorhiza; thus the seed morphology of Pyrrorhiza seems intermediate between that of Wachendorfia and the three genera with discoid glabrous seeds. Seeds of Xiphidium (Fig. 151) and Schiekia (Fig. 152) are globose and tuberculate. Barberetta (Fig. 153) has ovoid, glabrous seeds, which are very similar in size and shape to some species of Wach- endorfia. Phlebocarya (Fig. 156) and Tribonan- thes (Fig. 157) have glabrous and somewhat glo- bose seeds. Seeds of Anigozanthos (Fig. 158), Blancoa, and Conostylis (Fig. 159) are quite sim- ilar to one another, being ellipsoid and ridged lon- gitudinally. М pidia (Fig. 160) has similar seeds to these three genera, but the ridges are much shallower and the seeds are significantly larger. The hypothesized intergradation series for seed morphology in the Haemodoraceae is seen in Fig- ure 163. Note that the direct evolutionary linkage between taxa with discoid, marginally winged seeds and the somewhat discoid, marginally tomentose seeds of Pyrrorhiza is questionable; developmental and anatomical studies are needed for confirmation. Because of the length of the morphocline of Figure 163 and the possibility of biasing the data if the groupings are coded sequentially, the character "seed morphology" was subdivided into the follow- ing discrete characters: Character #48: Seed shape ovoid-globose or ellip- soid (and longitudinally ridged) vs. discoid. This allows for a single evolutionary step between taxa having the distinctive discoid seed morphology and all other taxa. Pyrrorhiza is coded as “X” for this character because of the uncertainty of homology between its somewhat flattened seeds and the dis- coid seeds of Dilatris, Haemodorum, and Lach- nanthes. Character #49: Seed shape ellipsoid vs. discoid, flattened, or ovoid-globose. This allows for a single evolutionary step between taxa having ellipsoid (and longitudinally ridged) seeds and all other seed types. Character #50: Seed vestiture glabrous vs. pu- bescent or marginally winged. Taxa having discoid seeds with marginal wings are coded with those having tomentose seeds because of the presumed homology between the marginal wing and the tri- chomes. Figure 165D illustrates the character cod- ing for seed morphology (characters #48—50). Among the outgroups, seeds of all members of 760 Annals of the Missouri Botanical Garden FicunES 146-162. (below) and distal (above) sid x3.8.— 147. caroliniana, proximal (left) and gen аш sides; placenta with two attached seeds; 150 side; x 7.9. Wachendorfia seeds (side view). — 154. ciliata; x 7.9. —157. Tribonanthes variabilis; x 7.9.— —152. Schiekia orinocensis, distal side; x 7.9.— W. paniculata; x3. " —155. W. t 158. Anigozanthos flavidus; x7.9.—159 160 ! Seed ae in the Haemodoraceae and outgroups.— 146. Dilatris viscosa, proximal Haemodorum spicatum, distal side; x3.8.— x3.8. 14 Proximal side ps seed; x3.8.— achnanthes 9-150. d ue —149. Face view of - Xiphidium coeruleum, distal 53. Barberetta aurea, side view; x 7.9. 154-155. thyrsiflora; x 3.8.—156. Phlebocarya Conostylis sp.; x7.9. x7.9.— 160. Macropidia fuliginosa, x7.9.—161. Eichhornia sp.; x 7.9.—162. Philydrum ae the Pontederiaceae (see Fig. 161) are consistently small and ovoid with longitudinal ridges. Species of the Philydraceae (see Fig. 162) have twisted, yriform-terete seeds with longitudinal rows of epi- dermal cells. Thus the seeds of both outgroups, particularly the Pontederiaceae, resemble Anigo- zanthos, Blancoa, Conostylis, and Macropidia in having longitudinal ridges but differ somewhat in shape. Because of the similarity, seed HL in the Philydraceae and Pontederiaceae are c as homologous with the ellipsoid, longitudinally ridged seeds in the Haemodoraceae. Character #51. Haploid chromosome num- ber. aploid chromosome numbers for the Hae- modoraceae and outgroups are listed in Table 6. The chromosome number of Xiphidium coeruleum is here described for the first time as n — 19 (Fig. 164; this number confirmed by P. Goldblatt of the Missouri Botanical Garden, pers. comm.). Note that chromosome numbers of Pyrrorhiza and Schiekia are unknown. Of the genera in the Haemodoraceae, only Conostylis has a variable number. However, most species of that genus have a count of n = 8, and the exemplar OTUs are either n = 5 or n = 8. One possible coding for chromosome number is the arrangement of taxa in classes via a linear mor- phocline, such as: n = 5-8 e n= 15 n = 19- 2] ^ n= 24. (Chromosome numbers ranging from 9 to 8 in the OTUs are lumped into one class because of the difficulty of assessing evolutionary direction in presumed minor aneuploidy events.) Since nothing in the Haemodoraceae is known re- garding karyological details, it is not unlikely that the evolution of chromosome number may have occurred differently from the simple numerical se- quence listed above. For example, the fact that Lachnanthes (n = 24) has a chromosome number Volume 77, Number 4 1990 Simpson 761 Phylogeny and Classification of Haemodoraceae LACHNANTHES BLANCOA CONOSTYLIS MACROPIDIA ANIGOZANTHOS FIGURE 163. exactly 3 times that of the most common number (n = 8) may be evidence of a hexaploid derivation from an ancestral n = 8; alternatively, a count of = 24 may have arisen from an allopolyploidy event. The n = 15 count of some taxa may have been derived via a tetraploidy event, e.g., from an ancestral n = 7 or n = 8, followed by aneuploidy. A count of n = 19 could have arisen directly from a number of n = 5-8 or from an immediate ances- tor having either n = 15 or n = 24. After considerable experimentation with this pip it was a that the best coding is simply: л = 5-8 ө п = 15, 19-21, or 24. The os series Genes the higher haploid numbers is ambiguous, and more elaborate and complex coding schemes result in the same topol- ogy in the cladistic analysis. Chromosome counts in the outgroups vary con- siderably (Table 6) and are consequently coded as polymorphic (**?””). The only haploid chromosome number found in both outgroups and in the Hae- modoraceae is a count of n = 8. It seems most likely that the ancestral haploid chromosome num- ber for the Haemodoraceae is near n = 8, a hy- pothesis to be tested with the cladistic analysis. Character #52. Phenalenones absent vs. pres- ent. his character is included to establish the PYRRORHIZA XIPHIDIUM SCHIEKIA BARBERETTA PHLEBOCARYA TRIBONANTHES Hypothesized intergradation series of seed types in the Haemodoraceae. monophyly of the Haemodoraceae relative to the two outgroups (see Monophylesis of the Haemo- doraceae). It is very likely that all species of the aemodoraceae, if investigated, would be found to have these compounds. As a reasonable compro- mise, however, genera of the Haemodoraceae not yet chemically investigated are coded as unknown (“2”). If any species of a genus has been found to have the compounds, all species of that genus are coded as possessing them. Both outgroup families are coded as lacking phenalenones. Character #53. Tapetum glandular vs. amoe- boid. This character is included to consider the presence of an amoeboid tapetum in the Haemo- doraceae and Pontederiaceae as a possible syn- apomorphy (see Outgroup Taxa, Table 5). As above, genera of the Haemodoraceae not yet investigated for this feature are coded as unknown (**?”) any species of a genus has been found to have an amoeboid tapetum, all species of that genus are coded as having it. Character #54. Pollen exine wall structure tetacte-columellate vs. non-tectate-columel- late. his feature is included to denote the evi- dence discussed earlier that the Pontederiaceae are the sister group to the Haemodoraceae (see Out- 762 Annals of the Missouri Botanical Garden TABLE 6. Chromosome numbers in the Haemodoraceae and related families. Citation Taxon 2n = n Tribe Haemodoreae Hilliard & Burtt (1971); Ratter & Milne (1973); Barberetta aurea Harv. 15 Ornduff (1979a Dilatris pilansii Barker ca. 19-21 Ornduff (1979a) Haemodorum sp. 8 G. Keighery (pers. comm.) Lachnanthes caroliniana (Lam.) 24 Ornduff (1979a) Dandy Lanaria lanata (L.) Dur. € 36 Ornduff (1979a) Schinz Lophiola aurea Ker-Gawler 21 Ornduff (1979a) Phlebocarya ciliata R. Br. 7 G. Keighery 691 (PERTH) filifolia К. Muell. 7 S. Hopper 840 (PERTH) Wachendo paniculata L. 15 Ornduff (1979a) thyrsiflora L. ca. 15 Ornduff (1979a) Xiphidium coeruleum Aubl. 19 Simpson (present Tribe Conostylideae Anigozanthos ancoa canescens Lindl. Conostylis P. Goldblatt (pers. comm.) Green (19 61) Stenar (1927); Green (1961) 1) oooooc0oc Qa n Ф la) 5 = © a mM Green (1961) aculeata R. Br. (7 subspecies) 8 Green (1961); Hopper (1978) androstemma F. Muell. 5 Green (1961) aurea Lindl. 5 Green (1961) bealiana F. Muell. 8 Green (1961) breviscapa R. Br. 4 Green (1961) candicans Endl. 8 Green (1961) caricina Lindl. 7 Green (1961) filifolia F. Muell. 8 Green (1961) juncea 8 Green (1961) phathynenthe Diels 8 Green (1961) seorsiflora F. Muell. 8 Green (1961) serrulata 8 Green (1961) setigera R. Br 14 Green (1961) setosa Lindl. 7 Green (1961) stylidioides F. Muell. 8 Green (1961) stylidioides F. Muell. 16 Hopper (1978) vaginata Endl. 8 Green (1961) Macropidia = (НооК.) Огисе 6 Green (1961) Tribonanthes 7 G. Keighery (pers. comm.) Apostasiaceae Е counts reported) Hypoxidaceae Curculigo orchioides Gaertn. 36 Sharma & Ghosh (1954); Sharma & Chaudhuri (1964); Mitra (1966) orchioides Gaertn. 18 Raghavan (1957); Sharma & Bhattacharyya (1960) Volume 77, Number 4 Simpson 763 1990 Phylogeny and Classification of Haemodoraceae TABLE 6. Continued. Taxon 2n = n= Citation recurvata Dryand. 18 Sharma & Ghosh (1954); Sharma & Chaudhuri (1964) Hypoxis acuminata 18, 20 Wilsenach & Papenfus (1967) aurea Lour 54 Mehra € Sachdeva (1971, 1976) decumbens L 42 Naranjo (1975) filiformis T Wilsenach & Papenfus (1967) longifolia Baker T2 Wilsenach (1 967) multiceps Buchinge 36 Wilsenach (1967) nitida ca. 40-42 Wilsenach (1967) pusilla Hook.f. 28 Beuzenberg & Hair (1963) rooperi 96 ca. 43-58 Wilsenach & Papenfus (1967); Wilsenach (1967) rooperi Moore 76,114 18 Fernandez & Neves (1962) stellipilis (Ker.) 16 Fernandez & Neves (1962) zeyheri Baker (?) 32 Fernandez & Neves (1962) Rhodohypoxis baueri Nel. (9 cultivars) 18, 12 Saito (1975) Philydraceae Helmholtzia acorifolia F. М 34 Briggs (1966) glaberrima Toe A ) Caruel 34 Briggs (1966) Orthothylax glaberrimus 16 Hamann (1966) Philydru langinosum Banks & Soland. 16 8 Hamann (1966) Briggs (1966) Pontederiaceae Eichhornia crassipes (Mart.) Solms 36 Briggs (1966) crassipes Solms 32 Krishnappa (1971) crassipes Solms 30, 32, 58 16 Banerjee (1974) speciosa Kunth 8 Sarkar et al. (1975) Monochoria hastifolia Presl. 28, 34-84 14, 40,42 Banerjee (1974) korsakowii Rgl. 52 Sokolovskaya (1966) vaginalis 28 Hsu (1967 aginalis 26, 52, 72, 74, 80 30, 40 Sharma (1970) vaginalis Presl ex Kunth 52 Krishnappa (1971) vaginalis Presl 80 40 Sharma & Sarkar (1967-1968) vaginalis Pax. var. plumbaginia Solms-Laubch. 52 Sharma & Sarkar (1967-1968) vaginalis Presl var. plumbaginea Solms-Laubch. 52 Banerjee (1974) var. plumbaginea Solms-Laubch. 52 Sharma (1970) vaginalis (L.) Presl 80 40 Sharma (1970 vaginalis (L.) Presl 52 Mehra & Pandita (1978) Pontederia rdata L. 8 Ornduff (1969); Lowden (1973) parviflora Alex. 8 Lowden (1973) rotundifolia L.f. 16 Lowden (1973) sagitta Presl 8 Lowden (1973) Tecophilaeaceae Cyanastrum cordifolium L. 22 Sato (1942) 764 Annals of the Missouri Botanical Garden TABLE 6. Continued. Taxon 2n= n= Citation cordifolium L. 12 Nietsch (1941) Cyanella abla L.f. Ornduff (1979b) hyacinthoides L. 12, 14, 24 Ornduff (1979b) lutea L.f. var. lutea 8,12, 24 Ornduff (1979b) lutea L.f. var. rosea Bak. 12 Ornduff (1979b) orchidiformis Jacq 12 Ornduff (1979b) Odontostomum hartwegii Torr 10, 10 + f Cave (1970) Tecophilaea 12 LaCour (1956) cyano-crocus Velloziaceae (no counts reported) Walleriaceae (Walleria spp.) 24 Goldblatt (1989) group Taxa; Simpson, 1987). Thus, all Pontede- riaceae are coded similarly to all Haemodoraceae as having a non-tectate-columellate architecture. Note that this feature was not taken into account with respect to the early characters (#33-37) "exine wall structure," as the Philydraceae were coded as (**?””) for number of exine layers (char- acters #33, 34) Character #55. Leaves unifacial vs. bifa- cial. All Haemodoraceae and all Philydraceae have unifacial leaves, whereas all Pontederiaceae have bifacial leaves. This character is included to take into account what may represent a shared derived feature between the Haemodoraceae and Philydraceae. CLADISTIC ANALYSIS METHODS Cladistic analyses were implemented using PAUP (Phylogenetic Analysis Using Parsimony), version 2.4 (Swofford, 1983), on an IBM-AT compatible microcomputer. The most parsimonious trees were constructed using the “branch and bound” algo- rithm from the data matrix of Table 8. Wagner parsimony was used throughout. Coding is sum- marized in Table 7. (See Conclusions for results of recent analyses using unordered character cod- ing. In the initial analysis all characters were as- signed equal weight. In a subsequent analysis char- acters that could originally have been treated as states of a single character (e.g., characters #4— 6, all dealing with inflorescence type) were ““scaled,” weighted as a proportional fraction of their original weight of ** 1" (characters #4, 5, and 6 were each given a weight of 4). This takes into account (often unintentional) a priori weighting of a feature be- cause it can be subdivided into two or more cor- related binary characters. The following characters were fractionally weighted in this second analysis: #4-6, 7-13, 15, 16, 21, 22, 23-25, 27, 28, 29, 30, 33-37, 44-47, 48-50, and 51-53. In a third cladistic analysis, only those char- acters for which the polarity could be determined at the outgroup node, using the method of Mad- dison et al. (1984), were included. Thus, the fol- lowing eleven characters, for which polarity could not be determined at the outgroup node, were omitted from this analysis: characters #4, 17, 23, 24, 25, 33, 35, 36, 53, 54, and 55. RESULTS AND DISCUSSION For the complete data set of Table 8, there are two equally most parsimonious cladograms (Figs. 166, ). The consistency index of the global analysis, including both outgroups, is 55/88 = 0.625. (If the 10 characters which are nonhomo- plasious and autapomorphic for OTUs of the Hae- modoraceae are deleted, the consistency index is 45/78 = 0.577.) When “fractional weighting” was performed (see Methods), a single most par- simonious cladogram was computed, equivalent to that of Figure 166. Thus, it could be argued that coding several characters in a multistate transfor- mation series results in little change in cladistic relationships. When the characters for which po- larity could not be determined were omitted from the data set, the same two cladograms of Figures 166 and 167 were obtained. This confirmation of results is important because, with some data sets, inclusion versus omission of nonpolarized charac- ters may yield significantly different cladistic re- Volume 77, Number 4 1990 Simpso 765 it and Classification of Haemodoraceae FIGURE 164. point to overlapping, but different, bivalents. x1,750 (left), x 2,770 (right). lationships. (Further comparison of alternative cladograms to Figures 166 and 167 will be dis- cussed below. As discussed earlier, the two outgroups were treated as OTUs equivalent in treatment to the intrafamilial OTUs of the Haemodoraceae. Because the outgroup relationships were not coded as re- solved, this cladistic analysis confirms the Hae- modoraceae as monophyletic; i.e., all ОТО» of the ingroup are more closely related to one another than to either outgroup. The overall tree was rooted at the Philydraceae clade, based on assumptions of polarity of characters #52-55 (Table 8; see Fig. 176). The alternative methodology of assign- ing a hypothetical ancestor and determining the states of that ancestor via the **2-step" algorithm of Maddison et al. (1984) would have yielded iden- tical results within the Haemodoraceae. However, one possible advantage of this *“*1-step” analysis (sensu Maddison et al.), performed without spec- ifying outgroup relationships, is that it provides a test both for the monophyly of the ingroup and for the relationship of the ingroup to the outgroup (see Interfamilial Relationships). Synapomorphies for the Haemodoraceae (Fig. 166) are: bifurcate cymes (character 45); inferior ovary position (character #39); discoid, flattened, or ovoid-globose, i.e., not ellipsoid and ridged, seeds (character #49); and presence of phenalenone compounds (character #52). Of these, only the Anther microsporocyte squash of Xiphidium c coeruleum at metaphase I. Note 19 bivalents. Arrows presence of phenalenone compounds was hypoth- esized a priori as synapomorphic for the family (see Monophylesis of the Haemodoraceae). Corrobo- ration here confirms its valid use as a synapomor- phy for the family. (The indicated occurrence of bifurcate cymes, inferior ovary position, and nonel- lipsoid, ridged seeds as synapomorphies for the aemodoraceae is discussed below The two equally parsimonious topologies bend only in the relationship of Dilatris, port as the sister taxon of Lachnanthes ао (Fig. per or as the sister taxon of the monophyletic group containing SC-PY-XI-BA-WA (Fig. 167). In either topology it is interesting that the initial bifurcation defines two monophyletic groups: designated here tribe Haemodoreae and Conostylideae. These roughly correspond to the traditionally defined tribes Haemodoreae and Conostylideae (e.g., Bentham & Hooker, 1883; Hutchinson, 1973; Melchior, 1964; Geerinck, 1969a) or the subfamilies Haemodoro- ideae and Conostyloideae (Cronquist, 1981; Dahl- gren & Clifford, 1982). Thus, the present analysis confirms the taxonomic ““integrity”” pa tribes Con- ostylideae and Haemodoreae and for their being monophyletic taxa. энне the evidenre many differences in significant patristic distance between these two major clades of the Haemodor- aceae, I conform with historical priority and recent consensus of use in retaining the tribal rather than the subfamilial rank. I see no valid reason for 766 Annals of the Missouri Botanical Garden E 7. Character listing and coding. Originally multistate characters were recoded as two or more binary characters. — — m — — — — 1. N Ww EN en ~J ec © . Inflorescence type 2 О=с . Trichomes pilate О = absent . Trichomes, if tapering, uniseriate or multiseriate ? = nontap g . Trichomes, if tapering, branched . Perianth symmetry Reddish root and stem coloration — absent 1 = present . Stem structural type 19, = rhizome = corm . Plicate leaves 20. — absent 1 = present . Inflorescence type 21. O = cyme absent 1 = cyme simple, bifurcate, or trifurcate N e absent or simple e bifurcate or bifurcate and trifurcate ym ‚ Inflorescence type 0 = cyme absent, simple, or bifurcate 23. 1 = cyme bifurcate and trifurcate N EN - present ? — homology uncertain . Trichomes, if pilate, with basal rosette O — absent 25. 1 = present ? ot pilate . Trichomes tapering N С = absent 1 = present . Trichomes, if tapering, unicellular or multicellular O = unicellular 1 = multicellular ? = nontaperin N - © = uniseriate 1 = multiseriate N © erin O = absent (unbranche 1 = present (branched) 30. = i ? pering . Trichomes, if tapering, with basal rosette O = absent 31. 1 = present ? = nontapering . Perianth apertures 32. 0 = absent 1 = present . Perianth tube 33. bsent 1 = present (short or long) be . Perianth tu 34. О = absent or short 1 = long w en ) — actinomorphic 1 = zygomorphic (without perianth splitting) 18. , ына к = топ 28. Perianth splitting О = absent = present Perianth aestivation O = imbricate 1 = valvate Perianth tannin cells bsent 1 = present Stamen (fertile) number = six r e . Stamen (fertile) number 0 = six or three stamens + two latero-anterior staminodes 1 = three (without staminodes) or one Anther dimorphism : — anthers equal — anthers eni i e dimorphis О = anthers ш. or one large + two small anthers 1 = one large + two caducous anthers or one anther + two staminodes Anther dimorphism O = anthers equal or one large + two small or caducous anthers 1 = one anther + two staminodes . Stamen connective appendages = absent = present sulcate i = por Pollen are О = monosulcate or 2-3-porate — oligoforaminate . Pollen sculpturing 1 = verrucate or rugulate Pollen sculpturing О = foveolate or verrucate 1 = rugulate Pollen apertural border = absent = present iat mid proximal verrucae — abse | = pre Number s exine wall layers О = one-layered 1 = two- or three-layered Number of exine wall layers . Exine wall, if two-layered, with papillate inner elements О = absent Volume 77, Number 4 1990 Simpso 767 Pb and Classification of Haemodoraceae TABLE 7. Continued. 1 — present ? — exine one- or three-layered 36. Exine wall, if two-layered, with only papillate inner elements 1 = present ? = exine one- or three-layered . Subexterior exine wall discontinuous O = absent w -l resent 38. Enantiostyly О = absent = present 39. Ovary position O = superior 1 = inferior 40. Septal nectaries = absent 1 = present . Fertile carpels O = three D uA 1 = one . Locule number О = three 1 = one . Placental sclereids aN N EN w l= hypotropous or epitropous . Ovule position = pleurotropous or hypotropous l = epitropous 46. Ovule number > л O = one рег carpel 1 = two-numerous per carpel 47. Ovule number = one or two per carpel 1 = numerous per carpel = four-seven per carpel 48. Seed shape 0 = ovoid-globose or ellipsoid (& longitudinally idged) 1 = discoid ? = flattened (Pyrrorhiza) 49. Seed shape 0 = ellipsoid (& longitudinally ridged) 1 = discoid, flattened, or ovoid-globose 50. Seed vestiture O = glabrous 1 = pubescent or marginally winged 51 . Haploid chromosome num 0 — 5-8 1 = 15, 19-21, or 24 ? — unknown or polymorphic 52. Presence of phenalenone compounds О = absent л Ww 3. Tapetum type = secretory = amoeboid ? = уп 54. Exine structure O = tecate-columellate 1 = non-tectate-columellate . Leaf type O = bifacial 1 = unifacial en л promoting a change in rank that conveys no change in cladistic relationships. he present circumscription of the tribes differs from past treatments primarily in the removal of Hagenbachia, Lanaria, Lophiola, and Pauridia from the family and in the transfer of Phlebocarya from its usual placement in the tribe Haemodoreae to the Conostylideae. Phlebocarya has traditionally been placed in the tribe Haemodoreae because of its imbricate as opposed to valvate tepals. However, an imbricate perianth is clearly plesiomorphic for members of the family; its occurrence cannot be used to unite taxa in a phylogenetic classification. Phlebocarya is united with the other Conostylideae by the common occurrence of derived features (see below). Synapomorphies shared by the six genera of the Conostyloideae are: loss of pilate trichomes (char- acter #7, although this may be synapomorphic only for members of the tribe other than Phlebo- carya; see Fig. 166), possession of multiseriate trichomes (character #11, which has an alterna- tive state change possibility; see below), branched trichomes (character #12), six fertile stamens (character #21), six or three stamens + two latero- anterior staminodes (character 422), porate apertures (character 427), rugulate pollen wall sculpturing (character #30), and absence of enant- iostyly (character #38). Character #11 exhibits a reversal in the clade to Tribonanthes (Fig. 166); an equally parsimonious alternative (but highly un- likely in the author's view) is the independent evo- lution of multiseriate trichomes in the clades to Phlebocarya and to the COau-MA monophyletic group. Of the other indicated synapomorphies, it seems extremely likely that the palynological fea- 768 Annals of the Missouri Botanical Garden TABLE 8. Character x Taxon matrix for the Haemodoraceae. A ‘*?” indicates uncertain ancestry, absent data г “X” coding. (See Table 7 and text for discussion of taxa and character coding.) PHIL = Philydraceae, PONT = 166. “Жол Other taxa abbreviations are listed in Figure 1 Characters: 00000 00001 11111 11112 22222 22223 33333 33334 44444 44445 55555 Taxa 12345 67890 12345 67890 12345 67890 12345 67890 12345 67890 12345 PHIL 07000 01011 00000 07001 11??? 000?? 00??? ??100 17110 11000 ?0001 PONT 00010 01011 0000 ??001 ????? 00010 00?0? ?0?0? ?Ol1? ??000 ?0110 ANfI 00011 00?11 11001 10110 00000 01011 00101 10011 00110 11000 01111 ANru 00011 002711 11001 10110 00000 01011 00101 10011 00110 10000 01111 BA 00100 0110? ???00 00000 11000 00010 11100 00101 10010 00010 1??11 BL 00011 00211 11001 10010 00000 01011 00101 10011 00110 11000 01?11 СОап 00011 00211 11001 10010 00000 01011 00101 10011 00110 11000 0??11 COau 00011 00?11 11001 00010 00000 01011 00101 10011 00110 11000 0??11 CObe 00011 00?11 11001 10010 00000 01011 00101 10011 00110 11000 0??11 DI 10011 11111 00100 00000 11100 00010 10100 00111 00010 00111 1?111 HA 11011 0100? ???00 00001 11100 00010 0000? ?0111 00010 10111 01?11 LA 10011 1?111 00000 00000 11000 00010 0000? ?0111 00000 17111 11111 MA 00011 00?11 11001 10110 00000 01011 00101 10011 00110 00000 0??11 РН 00011 0??11 11000 00001 00000 01011 00101 00011 01111 00010 0??11 РҮ 11011 01110 00100 01000 11111 00010 00100 01001 00010 10?11 ???11 5С 00011 01110 00110 01000 10110 00000 0011? ?1101 00000 17011 ???ll TR 01011 00?11 01001 00011 00000 11111 00101 10011 00110 11010 0??11 WA 11110 01110 00110 01000 11000 00010 11100 00101 00010 00011 11111 XI 10010 0110? ???00 00000 11100 00010 00100 00100 00010 11011 11111 tures are indeed synapomorphic, as supported by pollen investigations of all taxa considered closely related to the Haemodora Within the tribe байыш, Phlebocarya is the most basal taxon (Fig. 166). Characters shared by Tribonanthes and the remaining four genera (CO, BL, AN, & MA) are a short perianth tube (character #15), valvate perianth aestivation (character # 19), inner exine layer composed solely of papillate elements (character #36), and nu- merous ovules per carpel (character #47). The last-mentioned feature (447) requires a conver- gence (clade to XI) and a reversal (clade to ANru & MA). The placements of Phlebocarya and Tri- bonanthes are questionable. They are similar to one another in chromosome number (n = 7; see below), but no coded features argue for their being sister taxa. Phlebocarya shares two similarities with CO, BL, AN, & MA that Tribonanthes does not: presence of 2—3-porate pollen grains (char- acter #27) and of multiseriate, dendritic trichomes pollen grains, however, is undoubtedly morphic for the tribe (Fig. 166), and ТИБ Кез is presumed to have evolved this feature in common with other members of the tri Conostylis, Blancoa, Anigozanthos, and Mac- ropidia are members of a monophyletic group that is defined by two homoplasious synapomorphies: absence of perianth tannin cells (character #20 C) and presence of ellipsoid, longitudinally ridged seeds (character #49 Б). All four of these genera have protruding hemispheric aperture walls (not found in Phlebocarya or Tribonanthes), which may represent a unique evolutionary change and thus constitute further evidence for the monophyly of these four genera. Conostylis and Anigozanthos are paraphyletic in the present analysis (Fig. 166). In Conostylis two exemplar species, COan & CObe, share more recent common ancestry with BL, AN, and MA than with COau as shown by the common presence of an elongate perianth tube (character #16), a derived feature. Anigozanthos and Mac- ropidia clearly form a monophyletic group, as supported by their common possession of a unique longitudinal perianth splitting (character #18). However, Anigozanthos is portrayed as paraphy- letic because MA and ANru share a derived fea- ture: reduction of ovule number from numerous to one or two per carpel (character #47 R), a feature not shared with ANfl, which has numerous ovules per carpel The present study tends to support Geerinck’s (1969a, b, 1970), treatment of Macropidia as a species of Anigozanthos (A. fuliginosus) and Blancoa as a species of Conostylis (C. canescens). Volume 77, Number 4 1990 Simpson Phylogeny and Classification of Haemodoraceae 769 110XXXX <> 1110001 <= 1111001 e 000X —— 1000 <= 1010 X111000 1 ' |. 0X11110 | 100XXXX ien ' 1 1110 С 1 A XX11110 <> 0X11010 00XXO0 11XX1 <> 10001 ——» 10000 ' 111 —— X11 <» 011 10100 t à D 000 <= 010 B 10110 FIGURE 165. “X”-character state coding of din ie Haemodoraceae. Sequence of numbers and Xs represent character states of related characters for a given taxon. The “X” symbol represents an equivocal character state coding (co s a “?” in Tables 7 and 8). Arrows represen evolutionary transitions in the morphocline. Numbers A. of character state changes are indicated beside arrows — В. Pollen exine wall structure (characters #33-3 pe of Figure 28. 112.—C vule la ору (characters #44 47), reflecting morphocline of Figue 145.—D. Seed morphology Е #48-50), reflecting morphocline of Figure 163 Following Geerinck’s treatment and accepting the results here, Anigozanthos s. ampl. would then be monophyletic. On the other hand, Conostylis, as currently defined, is still likely paraphyletic, even when merged with Blancoa. Much more detaile phylogenetic studies are needed to resolve the phy- logenetic relationships of Anigozanthos and Cono- stylis, particularly in the recognition of genera segregated from Conostylis: Blancoa, Greenia, and Androstemma (see Keighery, 1981). Hopper & Campbell (1977), in study of seed morphology of Anigozanthos and Macropidia, attained results markedly different from those portrayed in Figure 166. Their phylogenetic tree suggests that Ani- gozanthos is a monophyletic group, and that the lineages to both Anigozanthos and Macropidia were derived separately from an ancestral taxon most closely related to the extant Conostylis brevi- scapa, the only species of Conostylis with distinct tepals. It is my view that Anigozanthos and Mac- ropidia are likely most closely related to the long- tubular species of Conostylis, namely С. bealiana, С. androstemma, and/or С. (Blancoa) canescens. Clearly, however, a thorough character analysis and phylogenetic study of all species of this complex of taxa is needed. Characters that unite the genera of the Hae- modoreae are: reddish coloration in roots and root- stock (character 4 1), absence of placental sclereids (character #43), discoid seed shape (character H48, which has an alternative state change pos- sibility; see below), and pubescent or marginally winged seed (character #50). Of these, only char- acter #43 is nonhomoplasious and exhibits no re- versals. Character #48 has an equally parsimo- nious alternative of two convergent events (in the clades to HA and to DI-LA). Additionally, note that characters #23 & #35 may be synapomorphic for either of the tribes Haemodoreae or Conosty- lideae, the uncertainty resulting from the fact that the AP of these characters at the ingroup node ., the common ancestor of the Haemodoraceae) ME not be determined Within the tribe Haemodoreae, Haemodorum is the most basal (earliest diverging) lineage. The seven other genera of the tribe are united by four synapomorphies (see Fig. 166). Two of these, pres- ence of pilate trichomes with a basal rosette of epidermal cells (character #8) and an increase in chromosome number from n = 5-8 to n = l5- 24 (character #51), are nonhomoplasious in both equally parsimonious topologies and seem good evi- dence for the recognition of this subgroup. Absence (loss) of perianth tannin cells (character #20 C) occurs independently as a derived feature in the lineage to CO, BL, AN, and MA of the tribe Cono- 770 Annals of the Missouri Botanical Garden Tribe Haemodoreae Tribe Conostylideae WA BA XI PY SC DI LA HA PH TR COau COan CObe BL ANfI ANru MA 166. One of two equally most parsimonious cladograms of the Haemodoraceae derived from the data a '**" (characters 9, 11, 13, and one reversal; only one of these alternatives (the most likely in the author's view) is displayed in the cladogram. State changes indicated with a and 48) have two equally parsimonious alternatives, " (c haracters 7 and 35) may occur with equal probability at the internode immediately above the displayed position. tgroup node; — Conostylis bealiana, DI — — Macropidia fuliginosa, PH = Phlebocarya spp., Tribonanthes spp., WA = Wachendorfia thyrsiflora, XI = stylideae. The occurrence of tapering trichomes having a basal rosette of epidermal cells (character #13) either exhibits a reversal in the clade to LA (Fig. 166) or is synapomorphic only for DI, PY- SC, XI, BA, and WA (Fig. 167) As mentioned above, the phylogenetic relation- ships of Dilatris and Lachnanthes are portrayed in two equally parsimonious topologies. In Figure 166 DI and LA are sister taxa, united by a single synapomorphy: the occurrence of bifurcate and trifurcate helicoid cyme units (character #6). Note that the presence of tapering trichomes with a basal rosette (character #13) is seen as an evolutionary event in the lineage preceding the DI-LA clade, and that a reversal of this feature (#13 R in Fig. 166) occurs along the clade to LA alone. The alternative and equally parsimonious possibility of independent evolution of such trichomes in the lineages to both DI and to WA, BA, XI, PY, & Y = Pyrrorhiza neblinae, SC = АМЕ = Anigozanthos flavidus, ANru = Anigozanthos rufus, BA = u = Blancoa canescens, COau = Conostylis aurea, COan = Conostylis androstemma, CObe Xiphidium coeruleum. C = с, К = Reversal. SC seems highly improbable. In Figure 167 the lineage terminating in Dilatris is depicted as arising after the lineage giving rise to Lachnanthes. Note that Dilatris is here united with WA, BA, XI, PY, SC by character #13 (derivation of tapering trichomes with a basal rosette, requiring a single character state change and not two as in Fig. 166). In this scheme, changes in characters #6 (cyme bifurcate and trifurcate) and #33 (exine wall one- layered) may occur equally parsimoniously either as a pair of convergences (illustrated in Fig. 167) or as one unique event and a reversal (not illus- trated). Wachendorfia, Barberetta, Xiphidium, Pyr- rorhiza, and Schiekia are members of the next monophyletic unit of the tribe. Three derived fea- tures support this grouping as seen in Figure 166: derivation of unicellular, tapering trichomes (char- acter #10), a reversal to a superior ovary (char- Volume 77, Number 4 1990 Simpso 771 P and Classification of Haemodoraceae acter #39 R), and a reversal to an ovoid-globose seed shape (character #48 R*). Pyrrorhiza and Schiekia compose a monophyletic group by virtue of three synapomorphies. One of these, presence of a zygomorphic perianth (character #17 C), oc- curs independently in the lineage to Wachendorfia; however, this may be questionable, as will be dis- cussed. The other two synapomorphies, anther di- morphism with one large and two cad (character #24) and a discontinuous inner sub- exterior exine layer (character #37), furnish ex- cellent evidence for the close phylogenetic rela- tionship of these taxa. Wachendorfia and Barberetta are a monophyletic assemblage as evi- denced by five evolutionary steps (Fig. 166). Only two of these— presence of plicate leaves (character ucous anthers #3) and presence of large proximal verrucae on the pollen grains (character #32)—are nonhomo- plasious, but these alone seem sufficient to warrant the monophyly of this group. Xiphidium is united with Wachendorfia and Barberetta by a single (rather weak) synapomorphy: occurrence of a sim- ple cyme unit (character #5 В), a reversal from the occurrence of bifurcate cyme units, indicated as synapomorphic for the Haemodoraceae as a whole Character convergences and reversals. Certain characters exhibiting reversals or convergences in the cladogram of Figures 166 and 167 are worth discussing. The change in ovary position from su- perior to inferior (character #39) occurs before the initial bifurcation into two tribes. Thus, the inferior ovary found in DI, LA, HA, and all Cono- stylideae is interpreted as synapomorphic for the family as a whole (i.e., present in the common ancestor of the family but not in either immediate outgroup). А reversal of ovary position (character H39 R) occurs along the lineage to WA, BA, XI, PY, and SC. This is problematic in that the evo- lution of an inferior ovary is generally viewed as being irreversible and is often well correlated with other characters in general taxonomic studies. Such an assumption of irreversibility would require an additional three or four evolutionary steps in the most parsimonious cladograms of Figures 166 & 167. (See Alternative Cladograms and Figure 174 for consideration of the unique, irreversible evo- lution of an inferior ovary and of the independent evolution of an inferior ovary in two independent clades.) Convergence in exine structural pattern is seen for Lachnanthes and Haemodorum. Even though both of these taxa have one-layered exine walls, the cladogram of Figure 166 supports the hypoth- Tribe Haemodoreae FIGURE 167. grams of the Haemodoraceae (only tribe shown; tribe Conostylideae каа. to that of Second of two most parsimonious clado- к g. 166). equally parsimonious alternatives, either two convergent events or one apomorphy and one reversal; only the two convergent events (most likely in the author’s view) are displayed in the cladogram. Abbreviations as in Figure 166. esis that evolution of this exine wall type occurred independently in the lineages leading to the two taxa (character #33 С); as discussed above, the cladogram of Figure 167 permits an equally par- simonious alternative. The occurrence in Hae- modorum of an occasional inner exine layer (see Simpson, 1983) may lend support to the lack of homology between Lachnanthes and Haemodo- rum in this feature and thus the convergent evo- lution of a one-layered exine Convergence in perianth symmetry (character #17 С) is seen in the lineages leading to WA and to PY-SC, arguing that a zygomorphic perianth evolved independently in these lineages. Similarly, the distinctive perianth apertures present in Wach- endorfia and in Schiekia are portrayed as having been derived independently. But in view of the complexity of these perianth apertures, this pos- sibility seems extremely unlikely. Because the monophyletic groupings of WA-BA and PY-SC are well supported by other characters (discussed above), it seems more likely that a reversal of perianth symmetry (zygomorphic to actinomor- phic) occurred in the lineages to XI and to BA, 772 Annals of the Missouri Botanical Garden A | | : = > ` ” WA BA XI PY SC DI LA Fs TR COau COan CObe BL АМ ANru MA / | REDUCED PILATE ы | 1088 ud. CELLS | ON T ERING) | LOSS TAPERING UMICELLULAR, TAPERINA W/BASAL ROSET BASAL ROSETTE CELLS FIGURE 168. evolutionary changes in trichome anatomy are indicated. and that a loss of perianth apertures occurred in the lineages to XI, BA, and PY. (See Alternative Cladograms, below, for a consideration of this pos- А convergence in stamen number (character #22 C) occurs in the lineage to Schiekia. This convergence supports the idea that the staminodia in the outer whorl in Schiekia are not direct ves- tiges of ancestral stamens but rather are de novo floral modifications. The loss of perianth tannin cells (character #20) occurs independently within each family tribe. The significance of this cell type is not known. Further anatomical studies or a chemical investigation of the precise contents of these tannin cells might be valuable. Trichome evolution. The hypothesized homologies of trichome types in the Haemodoraceae can be assessed by a comparison of trichome anatomy with the branching patterns of the two most parsimo- nious cladograms (Fig. 168). First, it is most par- simonious to hypothesize that the common ancestor of the family (at the ingroup node) had pilate tri- chomes and uniseriate, tapering trichomes (which occur in members of both outgroup families). Sec- N UNICELL. (P. ciliata) Loss MULTISER, MULTISERIATE, DENDRITIC TAPERING PILATE Superposition of illustrations of trichome anatomy on cladogram of Figures 166 and 167. Possible ond, the cladistic analysis supports the hypothesis of homology between the pilate trichomes of Hae- modorum and those of the remaining Haemodo- reae. The evolution of vestiture in Haemodorum involved the loss of ancestral tapering trichomes. Third, the distinctive trichome basal rosette evolved de novo in a separate lineage to that terminating in Haemodorum; the basal rosette cells may be correlated with further modification of the pilate trichome to one with a glandular terminal cell. The occurrence in Lachnanthes of long, uniseriate, ta- y the /oss of that basal rosette; further evolutionary modifications in Lachnanthes in- volved the reduction of pilate trichomes to unicel- lular trichomes (having the basal rosette, however). The unicellular tapering trichomes (in PY, SC, and WA) evolved by reduction from a multicellular tapering trichome, and the presence of only pilate trichomes (in XI and BA) evolved by the indepen- dent loss of the unicellular type. In the tribe Con- ostylideae the most parsimonious explanation for the uniseriate, basally branched trichomes in Tri- bonanthes is reduction from the highly branched, dendritic type; the unicellular trichomes of Phle- ocarya ciliata are portrayed as having arisen de Volume 77, Number 4 1990 Simpso 773 eae) and Classification of Haemodoraceae BELEIRET WA BA XI 5С О! LA HA PH 4 Li 1 OVULE/CARPEL, * оша PLACENTA RING-LIKE anpeL, / | hd Pieunovnorous 6-7 OVULES/CARPEL, PLEUROTROPOUS ae QD TR COau COan CObe BL 1 OVULE/ CARPEL 2 OVULES/CARPEL 1 OVULE/CARPEL, EPITROPOUS 2 OVULES/CARPEL, HYPOTROPOUS FIGURE 169. Superposition of illustrations of ovule morphology on cladogram of Figures 166 and 167. Possible ted. evolutionary changes in ovule number, shape, and position are indicate novo (although it is possible they could be reduc- tions from an ancestral pilate trichome). Further studies of trichome anatomy in members of the Haemodoraceae may prove useful in firming or refuting the above hypothesized homol- ogies. Of particular value would be ultrastructural studies (e.g., details of cell wall structure or cy- toplasmic contents) or developmental studies (e.g., cell division patterns in trichome ontogeny). con- Ovule evolution. The evolution of ovule number and morphology in the Haemodoraceae is some- what complicated, as seen in Figure 169. The most parsimonious explanation, following the morpho- cline of Figure 145, is that the ancestral condition for the family is two hypotropous ovules per carpel. An evolutionary step (to numerous ovules per car- pel) occurs in the lineages to XI and to the TR- CO-BL-AN-MA clade. А reversal in ovule number from numerous to two per carpel occurs in the lineage to ANru. Reductions to one ovule/carpel occur independently in the clades to WA-BA, DI, PH, and MA. The epitropous ovule position in Phlebocarya occurs nowhere else in the family and is portrayed as an autapomorphy for the genus. The pleurotropous ovule position and increase in ovule number in Lachnanthes and Schiekia are most parsimoniously explained as being derived independently by convergence, which may seem doubtful in view of the distinctiveness of this mor- phology. The cladogram shows that a hypothesis of common ancestry for this feature in the two genera would necessitate two extra state changes. In fact, the rather thickened placental tissue of Pyrrorhiza and the ring of placental tissue in Di- latris may both be homologous with the thickened, peltiform placenta of, e.g., Lachnanthes. Devel- opmental studies would be extremely informative in assessing the significance of placental morphol- ogy and of ovule number, shape, and position. Seed evolution. As seen in Figure 170, the coded ancestral condition of seed morphology (elliptic, longitudinally ridged seeds) is not supported, as changes from this morphology are indicated in each tribal clade. A discoid, marginally winged seed mor- phology is portrayed as synapomorphic for the tribe Haemodoreae; presence of seed coat pubescence is synapomorphic for WA, BA, XI, PY, and SC (although this was not coded separately from the the SC-PY-XI-BA-WA clade and that of the Cono- stylideae. The flattened, marginally tomentose seeds of Pyrrorhiza are portrayed as having arisen de 774 Annals of the Missouri Botanical Garden Воо Ө РҮ 5С О! LA HA РА 4 FLATTENEO / | o» е WA BA XI | | GLABROUS М. ES FicunE 170. Q u PH T 09 6 @ @ 6 R COau COan CObe BL ANru MA ELLIPSOID, LONGITUDINALLY RIDGED or DISCOID, B o WINGED GLOBOSE-OvOIO Superposition of illustrations of seed morphology on cladogram of Figures 166 and 167. Possible h evolutionary changes in seed shape and vestiture are indicated. Note that other equally parsimonious evolutionary events may be possible. novo, but an equally parsimonious explanation (not illustrated in Fig. 170) is that these flattened seeds are evolutionarily intermediate between the discoid, marginally winged seed and the globose-ovoid, pu- bescent seed type. In the tribe Conostylideae the ellipsoid, longitudinally ridged seed morphology is synapomorphic for CO, BL, AN, and MA. There is, therefore, no evidence that this seed type is plesiomorphic for either the family or the tribe Conostylideae, even though the seed morphology of both outgroups resemble it (see Character Anal- ysis). It is the globose, glabrous seed type of PAle- bocarya and Tribonanthes that is either synapo- morphic for the tribe Conostylideae or possibly ancestral for the whole family. Future studies of seed anatomy and development may prove useful in assessing the homologies of seed evolution in the family. Chromosome evolution. The cladogram of Figure 171 can be used to assess evolutionary changes in chromosome number in the Haemodoraceae. The two most basal taxa of each tribe have very similar chromosome numbers. Perhaps the most likely pos- sibility is that the chromosome number for the common ancestor of the family was n — 8, whic agrees with п = 8 as one of the more common chromosome counts among the outgroups; how- ever, a base number of n — 7 1 monious. The common chromosome number of n = 7 for PH and TR is compatible with their close proximity on the cladogram (although not aiding 7 is equally parsi- in resolving their interrelationship) and can be ex- plained possibly as an aneuploidy event from an ancestral n — 8 condition. Species of Conostylis have a variety of chromosome numbers, n — being the most common. Figure 171 shows that the COan clade might be resolved from those of CObe and BL based on a common chromosome number (n — 5) with COau. However, the present study is much too incomplete to resolve interre- lationships among Conostylis species; detailed studies of all species of this genus are needed to assess its karyological history better. А common chromosome number of n — 6 in AN and MA supports the very close relationship of these two genera. Figure 171 portrays an evolutionary step from to either n = 15 orn = in the lineage that includes LA, DI, SC, PY, XI, BA, and WA. This event (a synapomorphy for this lineage in the cladistic analysis) could have arisen via tetraploidy or hexaploidy from an n = 8 ancestor. However, it is apparent that chromosome number evolution Volume 77, Number 4 Simps 775 1990 on and Classification of Haemodoraceae 15 15 19 ? ? 19-21 24 8 7 7 5 5 8 8 6 ? 6 WA BA XI PY sc DI LA HA PH TR COau COan CObe BL АМ ANru MA FIGURE 171. Superposition of haploid chromosome evolutionary changes in chromosome number are indicated. Note that other equally parsimonious evolutionary events may be possible. in the tribe Haemodoreae is still virtually unre- solved; numerous scenarios of evolutionary change are possible (Fig. 171). Although chromosome number is a valuable taxonomic character, there are problems due to ambiguity of character coding (see Character Analysis) and lack of data. Future karyotypic studies and determination of chromo- some numbers for Pyrrorhiza and Schiekia should aid greatly in testing the hypotheses of Figure 171 and in refining phylogenetic relationships, partic- ularly in the tribe Haemodoreae. Biogeography. The cladograms of Figures 166 & 167 can be used to assess the biogeographic history of the Haemodoraceae. It should be stressed that biogeographic data, being extrinsic data, were not included in the original data set. As seen in Figure 172, all members of the tribe Conostylideae are restricted to southwestern Australia. Haemodo- rum, the most basal genus of the tribe Haemo- doreae, is the only other Australasian member of the family, being distributed in western, northern, and eastern Australia, Tasmania, and parts of New Guinea. This distributional pattern can be explained most simply by a single vicariance event: the split- ting of a continuous ancestral population via the separation of (or establishment of an effective re- productive barrier between) Australia—Antarctica from the remainder of Gondwanaland. That portion remaining on Australia- Antarctica eventually gave number on cladogram of Figures 166 and 167. Possible rise to Haemodorum; the ancestral stock remaining on South America—Africa eventually diverged to give rise to the seven other genera of the tribe (Fig. ). The present distributions of Pyrro- rhiza, Schiekia, Xiphidium, Wachendorfia, and Barberetta could be explained by a single vicari- ance event: the continental separation of South America and Africa at the point of divergence of the lineages to XI and to WA-BA. As is evident from Figure 172, the phylogenetic relationship of Dilatris is problematic with respect to vicariance biogeography. If Dilatris is accepted as the sister taxon of Lachnanthes (as in Fig. 166), then it could be hypothesized that the splitting of the lin- eages to Dilatris and Lachnanthes was via vicar- iance: the separation of an ancestral population (with subsequent divergence) by the splitting of South America from Africa. Lachnanthes could then have attained its present distribution via dis- persal or vicariance from South America to North America. This possibility would necessitate an in- dependent (but contemporaneous) vicariance event at the point of divergence of the lineages to XI and to WA-BA. If Dilatris is accepted as being the sister taxon to WA, BA, XI, PY, and SC (as in Fig. 167), then its present distribution is most simply explained as long-distance dispersal (Fig. 72). The fact that Dilatris and Wachendorfia are sympatric over much of their range may be evidence that they attained a common range by 776 Annals of the Missouri Botanical Garden S Am., Aust., S Afr. SE Afr. С Am. S Am. S Am. S Afr. М Am. М. Guin. WA BA XI PY SC DI LA HA AFR. GONDWANA AUSTRALIA FIGURE 172. SW Aust. „ы т lp a TR COau COan CObe BL АМІ ANru MA Superposition of geographic ranges on cladogram of Figures 166 and 167. Arrows represent major biogeographic changes, including possible vicariance events (double arrows) and dispersals (single arrows). separate biogeographic processes (Croizat et al., 1974). Alternative cladograms. In view of the incom- patibility of certain character changes in the clado- grams of Figures 166 € 167, a few alternative cladistic relationships are worth consideration. One major possibility is that an inferior ovary evolved uniquely and irreversibly in the Haemodoraceae; the resultant cladogram is seen in Figure 173. This topology requires 92 total character state changes (C.I. = 0.598), a length of four state changes greater than the most parsimonious solution of Fig- ures 166 & 167. Note that in Figure 173 the traditionally defined tribe Haemodoreae is now paraphyletic, with Dilatris, Haemodorum, and Lachnanthes more closely related to the six genera of the Conostylideae (the sole synapomorphy being an inferior ovary, character #39) than to other traditionally classified Haemodoreae. This alter- native cladogram contains additional problems of conflicting character state changes. These include introduced homoplasy in presence/absence of pi- late trichomes with basal rosette cells (character #8), placental sclereids (character #43), seed ves- titure (character #50), and chromosome numbe (character #51). It seems particularly unlikely, for example, that there would have been a secondary decrease in chromosome number (to л = 8 in the lineage to Haemodorum) as well as the loss (for r the Haemodoraceae as a whole) and reacquisition in the lineage to the Conostylideae) of placental sclereid cells. However, in view of the generally accepted irreversibility of ovary position, the clado- gram of Figure 173 is a possible alternative to the more parsimonious cladograms of Figures 166 & 167. Detailed investigation of ovary development and floral vasculature might prove extremely valu- able in assessing the uniqueness and irreversibility of the evolution of ovary position in the Haemo- doraceae. Another alternative portrays the independent evolution of an inferior ovary in the Conostylideae and in one clade of the Haemodoreae (Fig. 174A; length = 90; C.I. = 0.611). Such a topology would link DI, LA, and HA as members of a monophyletic group (as in Fig. 173) within the tribe Haemodo- reae; synapomorphies for these three genera are: exine one-layered (character #33, necessitating a reversal in lineage to DI, although two convergent events is equally parsimonious), inferior ovary (character #39, convergent in lineage to all mem- bers of tribe Conostylideae, although a unique event plus a reversal is equally parsimonious), and discoid seed shape (character #48). Of these, the con- vergent evolution of an inferior ovary and discoid seeds seem most likely to be synapomorphies for the three genera. The cladogram of Figure 174A is, of course, less parsimonious (two steps greater) than that of Figure 166 or 167 and necessitates Volume 77, Number 4 Simpson 777 1990 Phylogeny and Classification of Haemodoraceae WA BA XI PY SH TR COau COan CObe BL ANfI ANru MA ФС 51 FIGURE 173. а cladogram (to Figs. 166, 167): most parsimonious cladogram in which an inferior ovary po evolves u ed are listed; all other state P cba “e have two equally and irreversibly. Only those vg eae below the es are identical to Figures 166 a Minim s alternatives, either two conv divergence of PY-SC and of PH 167. hospes state changes indicated with a orgent e or one apomorphy and one reversal; only one of these alternatives (the most likely in the author's view) is аса: in the cladogram. additional homoplasy for some characters, includ- ing: trichome anatomy (characters #8 R and 413 С), perianth tannin cells (character #20 C, an additional convergence portrayed), and chromo- some number (character #5 A third alternative cladogram is presented in Fig. 174B. This possibility differs from the most parsimonious topology of Figures 166 € 167 in placing the lineage to XI basal to that of PY-SC (i.e., in portraying WA-BA and PY-SC as sister groups). Figure 174B requires only a single extra step (length = 89; C.I. = 0.618) compared with the cladograms of Figures 166 & 167. In fact, the phylogenetic relationship of Xiphidium as por- trayed by Figures 166 € 167 could be viewed as questionable. This most parsimonious explanation, that Xiphidium is most closely related to the WA- BA subgroup, is supported by only one synapo- morphy: presence of simple cyme units (character H5). The rationale for perhaps preferring the cladogram of Figure 174B is that it is more par- simonious than the cladogram of Figures 166 & 167 if it is assumed that zygomorphic perianth symmetry (character 417) and perianth apertures (character #13) are homologous when present (i.e., that they arose by common evolutionary origin, not by convergence). If, for example, the common possession of a zygomorphic perianth is synapo- morphic for WA, PY, and SC (as seen in Fig. 174B), then one fewer reversal for this character occurs in the cladogram of Figure 174B (vs. those of Figures 166 & 167). In addition, if perianth apertures arose in WA and SC by a single evolu- tionary event (at “+” sign in Fig. 174B), then one fewer reversal (i.e., loss of perianth apertures) is required in Figure 174B than is required in Figures 166, 167. It is the author's view that each of these are very likely possibilities, particularly with regard to the distinctive perianth apertures of Wachendorfia and Schiekia. Thus, the cladogram of Figure 174B is to be preferred over that of Figures 166 & 167 (at least with respect to these five genera of the Haemodoreae) even though it is overall less parsimonious by one step. This rea- soning is, in effect, a type of a posteriori Ricoh of characters. It is similar to utilizing ““Dollo”” char- acter parsimony, which allows no convergences and minimizes any subsequent character state re- versals. A fourth possible alternative cladogram (length = 90; C.I. = 0.611) is that of Figure 175A, which portrays DI as the sister group of WA-BA, the three taxa united by two synapomorphies: a pollen apertural border (character #31) and a single ovule/carpel (character #46 C). However, this topology is difficult to explain in view of a host of 778 Annals of the Missouri Botanical Garden B b- FIGURE 174. Alternative cladograms (to Figs. 166, 167). Only those characters that differ in distribution from Figure 166 are listed. Character state changes indicated with a “*” have two equally po турак айы either two convergent events or one apomorphy and o reversal; only one of these Aena (ihe. most likely s view) is displayed in the cladogram. — A. wn) portraying W'achendofia - Barberetta and Pyrrorhiza -Schiekia as sister groups other characters and necessitates additional ho- position (character #39), and seed shape (char- acter . Homoplasy in ovary position, tri- chome anatomy, and seed morphology is particu- larly difficult to reconcile in view of similarities of Dilatris with LA and HA. Thus, the cladogram of Fig. 175A cannot, in the author's opinion, be readi- ly supported over that of Figures 166 and 167. B FIGURE 175. Alternative cladograms (to Figs. 166, 167). Only those characters that differ in distribution from Figure 166 are listed. Character state changes indicated with a “*” have two equally Dm alternatives, either two convergent events or one apomorphy and one reversal; only one of these alternatives (the most likely icio b Dilatris as the sister group to Wache endorfia- Barber tril с shown) portraying Sehiekia as the sister group of Wachendorfia-Barberet A final alternative cladogram (length = 92; C.I. = 0.598) is portrayed in Figure 175B. This alter- native differs from that of Figures 166 & 167 in removing Schiekia as the sister group of Pyrrorhi- za and placing it as the sister group of WA-BA. Figure 175B has the advantage of treating the distinctive perianth apertures of Wachendorfia and Schiekia as nonparallel features, necessitating a loss only in the clade to Barberetta. However, the evidence for uniting Schiekia and Pyrrorhiza, Volume 77, Number 4 1990 Simpso 779 pta and Classification of Haemodoraceae Y PHILYDRACEAE X Z LAMELLAE INNER OOT-LAYER 1 STAMEN/FLOWER HAEMODORACEAE FUSION OF POSTERIOR TEPALS PONTEDERIACEAE А НАВІТ ARYLPHENALENONES BIFACIAL, LIGULATE LEAVES P ak NON-TECTATE-COLUMELLATE EXINE STRUCTURE VERRUCATE EXINE WALL SCULPTURING AMOEBOID TAPETUM UNIFACIAL LEAVES Po LEAVES o PLACENTAL SCLEREIDS Li TAPETUM Cladogram portraying hypothesized relationships among the Haemodoraceae, Pontederiaceae, and FIGURE 176. Philydraceae. Major evolutionary events are portrayed. Note possible clades to “X, discussion. based on exine ultrastructure (character #37) and on stamen/staminode morphology (character #24 seems, to the author, more convincing. The clado- gram of Figure 174B is accepted as more likely than that of Figure 175B. Of the above alternative, but less parsimonious, cladograms, that of Figure 175A is most compat- ible with a vicariance explanation of present dis- tributional ranges. This cladogram would require only two separate vicariance events correlated with the splitting of Gondwanaland into Australia, Af- rica, and South America. The distribution of Lach- nanthes could, as above, be explained via dispersal or a vicariance event between South and North America. However, as discussed, accep ptance of this cladogram would necessitate several лау һо- moplasious events. Thus, it must be concluded that, despite the better biogeographic “fit” of the alter- native cladogram of Figure 175A, the data support those of Figures 166 & 167 (or, perhaps, of Figure 174B) better. — Interfamilial relationships. Although a strict cla- distic analysis of the Haemodoraceae and all pos- sible outgroup families in the complex is beyond the scope of this paper, a major premise of the " “Y” and “Z.” (See text for present analysis is that the families most closely related to the Haemodoraceae are the Philydraceae and Pontederiaceae (see Outgroup Taxa). The cladogram of Figure 176 illustrates major hypoth- esized evolutionary changes among lineages leading to the Philydraceae, Pontederiaceae, and Hae- modoraceae, including those treated as characters (#53-55) in the cladistic analysis of the Haemo- doraceae. One or more autapomorphies are shown for each terminal clade, evidence that the three families are monophyletic. The cladogram was root- ed at the Philydraceae under the assumption that the unique exine structure of the Haemodoraceae and Pontederiaceae (found in no other considered outgroup family) constitutes a very reliable syn- apomorphy and unites the latter two families as sister taxa. Їп addition to the evolution of unifacial leaves (character #55 in the cladistic analysis), a possible synapomorphy linking the three families is the presence of placental sclereids and perianth tannin cells. An amoeboid tapetum (character #53) and a non-tectate-columellate, verrucate exinous pollen wall (character #54) constitute synapo- morphies for the Haemodoraceae and Pontederi- aceae. It is hypothesized that evolution of the Pon- tederiaceae involved a major adaptive shift and Annals of the Missouri Botanical Garden ш < ш ч 3 y + ш z о < I & Б & [2 LI 33 w ш 3 $ 8 бё 9 $ + * x ч ш о t [3 « ш о ш ш ш g < < S | 2 d ч N 5 о x о 9 « © з > > a = ш S CN o y t Q 3 a X x Ы E ї ш & 3 ч 34 9 -38 a q LS 25 19 31 a Sz 21 16" 36 +27 3 à 17 1 13 15" 2 3 15 9 2 м 9 8 5 1 37 6 L E 177. Cladogram of the Bromeliiflorae, re- pied from Dahlgren & Rasmussen (1983). radiation to an aquatic environment with subse- quent changes in anatomy, such as the develop- ment of spongy aerenchyma in some taxa. This shift to a free-floating or emergent growth form may be correlated with an evolutionary change from unifacial to bifacial leaf morphology. This hypothesis is supported by the fact that the Pon- tederiaceae typically have inverted vascular bun- dles analogous to a “radial” (unifacial) origin (Arber, 1925, and references therein). The grouping of the Philydraceae, Pontederia- ceae, and Haemodoraceae as a monophyletic group (Fig. 176) agrees well with the delimitation of the Pontederiinae, sensu Walker (1986). Figure 176 shows many similarities as well with the eclectic cladistic portrayal of the Bromeliiflorae, Dahlgren & Rasmussen (1983; see Fig. 177). Dahlgren & Rasmussen portrayed the Typhales (Sparganiaceae and Typhaceae) in a tritomy with the Haemodoraceae and Pontederiaceae; the Phily- draceae are basal to this clade (Fig. 177). I propose the existence of synapomorphies, derived primarily from palynological data, linking the Haemodora- ceae and Pontederiaceae as sister taxa. Dahlgren & Rasmussen linked the Typhales to the Ponte- deriaceae and Haemodoraceae via a single syn- apomorphy: presence of an amoeboid tapetum. Be- cause investigated members of Typha and Sparganium have a typical tectate-columellate ar- chitecture, it is possible that the lineage to the Typhales could be placed in Figure 176 at “Х,” sensu i.e., before the evolution of the palynological spe- cializations. In addition, at least one species of the Typhales, Sparganium eurycarpum, possesses placental sclereids similar to those found in the Haemodoraceae, Philydraceae, and Pontederia- ceae. However, in view of the lack of additional supporting evidence, the linkage of the Typhales to the Haemodoraceae—Pontederiaceae seems quite speculative. Based on these few characters, it is equally parsimonious to hypothesize the cladistic position of the Typhales at “Y” in Figure 176 (assuming the Typhales to belong within the Bro- meliiflorae to begin with). This possibility requires the independent origin of an amoeboid tapetum but no reversal to a bifacial leaf. It is apparent that the numerous specializations possessed by the Ty- phales have obscured their affinities. The relation- ship of the Typhales to other monocotyledons con- tinues to be an intriguing systematic problem (see Dahlgren & Clifford, 1982). In contrast to Dahlgren & Rasmussen (1983), Walker (1986) placed the Zingiberiflorae (Zingi- berales) as the sister taxon of his Pontederiiflorae (= Pontederiidae; including the Haemodoraceae, Philydraceae, and Pontederiaceae), equivalent to position “Ү” in Figure 176. This would fit the distribution of characters in Figure 176 fairly well, as the Zingiberales have mostly distichous leaves possibly an ancestral character) and at least one species, Strelitzia reginae, has placental sclereids similar to those in the Haemodoraceae, Philydra- ceae, and Pontederiaceae (Simpson, unpublished). However, the Zingiberales possess an amoeboid (not secretory) tapetum, requiring either an inde- pendent evolutionary event in the Zingiberales or the separate evolution of a secretory tapetum in the lineage to the Philydraceae. One other possi- bility is worthy of consideration: that the Zingi- berales are the sister group of the Haemodoraceae— Pontederiaceae (position **Z" in Fig. 176). This cladistic hypothesis would necessitate the indepen- dent evolution of a bifacial leaf morphology. How- ever, it is intriguing that all members of the Zin- giberales have a thin, modified exine wall, often consisting of scattered deposits atop a thick, cel- lulosic/pectic intine (Kress et al., 1978). In ad- dition, ultrastructural developmental studies by the author (Simpson, 1989) indicate that early exine deposition in Xiphidium (Haemodoraceae) is strik- ingly similar to that occurring in Heliconia of the Zingiberales (see Stone et al., 1979). In both taxa sporopollenin is deposited on one to several ex- truded “‘white lines," defining an inner and outer exine layer. Thus, the two-layered nature of the exine in most Haemodoraceae may be structurally ~ Volume 77, Number 4 1990 Simpso 781 dad and Classification of Haemodoraceae homologous with that found in at least one member of the Zingiberiflorae. This developmental evidence would argue for the placement of the Zingiberales at “Z” in Figure 176. Further studies of pollen wall development, especially in other families of the Bromeliiflorae, sensu Dahlgren & Rasmussen (1983) may prove extremely useful in confirming the distinctiveness of this developmental pattern in these taxa CONCLUSIONS It is hoped that the detailed description of the characteristics of the Haemodoraceae and of the rationale for character coding will serve as a basis for future criticism and refinement of the phylo- genetic relationships presented. The occurrence of incompatibilities between several characters makes difficult any reasonable certainty of phylogenetic relationships among some genera. The interrela- tionships among the tribe Haemodoreae pose a particular problem. Aside from certain groupings (such as the sister-group relationships of Pyrrorhi- za and Schiekia and of Wachendorfia and Bar- beretta), several other possibilities having a mini- mum of additional evolutionary steps are evident. Recent analyses using alternative coding schemes (emphasizing unordered coding) and new data ob- tained since this paper went to press have yielded three additional most parsimonious trees for the tribe Haemodoreae; each of these topologies in- includes Dilatris, Haemodorum, and Lachnanthes as a monophyletic clade most closely related to Pyrrorhiza-Schiekia. These possibilities will be considered in light of future research (see below). The major phylogenetic relationships a the Con- ostylideae appear at this time to be r firm, with the possible exception of the еа of Phlebocarya and Tribonanthes. As was em- phasized, critical analysis of many more species of Conostylis and Anigozanthos is needed before their interrelationships can be understood. The included studies suggest, however, that Conostylis and An- igozanthos, as usually circumscribed, are very likely not monophyletic groups. Merging of Blan- coa with Conostylis and Macropidia with Ат- gozanthos may be warranted in a strict phyloge- netic classification. Consideration of the Haemodoraceae and im- mediate allies are pivotal in analyzing the validity of the Bromeliiflorae sensu Dahlgren & Rasmussen. In fact, a major discordance in the classification of monocots proposed by Walker (1986) versus that of Dahlgren and coworkers centers on the recognition of the Bromeliiflorae and their rela- tionships to the Zingiberales and Commelinidae. Future investigations of the interfamilial relation- ships of the Haemodoraceae should prove quite intriguing in this regard. he present study underscores the need for ad- ditional research, particularly in karyology, ultra- structure, and development. My current project on morphometric analysis of ovule and seed de- velopment in the complex might prove particularly intriguing in tracing discrete evolutionary events. A better understanding of biogeographic history and ecology could provide insight into the possible adaptive significance of these events. Finally, the relatively new techniques of DNA restriction site analysis and sequencing could provide very im- portant data in validating or refuting proposed re- lationships. This study serves well to exemplify that the problems and uncertainties typically evident in phylogenetic analyses may lead to future research that will further clarify relationships and present new insights into plant evolution. LITERATURE CITED ADAMSON, R. S. & Т. М. SALTER. 1950. Flora of the Cape oo Juta & Co., Cape Town and Johan esbur pa I D. 1980. A re-evaluation of the Melan- P ges (Liliaceae) using numerical analysis. In: С. D. Brickell, D. F. Cutler & M. Gregory T Potaleid Monocotyledons. Linn. Soc. Symp. 8: 65-82. 1985. a = affinity with the Ш- iaceae. Taxon 34: ARBER, A. 1925. ates ons, a крон St = ер Univ. Press, Cambridge, Мазза chus AYENSU, E. ©. & J. J. SKVARLA. Fine structure of Velloziaceae pollen. Bull. Torrey Bot. Club 101: 250-266. BaILton, Н. 1894. Amaryllidacées. Hist. Pl. 13: 1- 85. BANERJEE, M. 1974. Cytological studies on some bers of the family Pontederiaceae. Cytologia 39: 483- 49]. н, I. & Н. С. GANGULEE. 1937. Spermatogenesis n Eichhornia crassipes. J. Indian Bot. Soc. 16 289-295. HALDAR. 1942. A contribution to the morphology and cytology of Monochoria hastaefo- lia. Proc. Indian Acad. Sci. B 16: 91-106. Hooker. 1883. Genera Planta- Bessey, C. Е 1915. flowering plants. Ann. Missouri Bot. Gar The phylogenetic taxonomy of d. 2: 109- BEUZENBERC, E. J. & J. B. Hair. 1963. каше o chromosome atlas of the New Zealand flora. 5. Miscellaneous families. New Zealand J. Bot. 1: e Briccs, B. G. 1966. Chromosome numbers of some Australian monocotyledons. Contr. New South Wales -34. Natl. Herb. 4: 24 782 Annals of the Missouri Botanical Garden BRIGHIGNA, L., К. C. FiorDI & М. R. PALANDRI. 1981. Ultrastructure investigations on the two-nucleate d len grain of Tillandsia caput: -medusae Morr. (Bro meliaceae). Amer. J. Bo : 10 Brown, В. 1810. Prodromus Florae бе Hollandiae et adas von Diemen. Engelmann (Cramer) a nd Wheldon & Wesley, Weinheim. [1960 Facsimile, Hafner Publ. Co., New Yor BUCHMANN, S. L 0. Preliminary anthecological ob- servations on Xiphidium caeruleum Aubl. (Mono- cotyledoneae: Haemodoraceae) in Panes J. Kansas Entomol. Soc. 53: 685-699. 2. Sporogenesis and oe in е шаа hartwegii. Phytomorphology 2: 2 | . Chromosomes of California Liliaceae. Univ. Calif. Publ. Bot. 57: 1-58. COOKE, & J. M. EDWARDS. occurring phenalenones and related compounds. Fortschr. Chem. Org. Naturst. 40: 158-190. & W. SecaL. 1955. Colouring matters of seii plants IV. Haemocorin: a unique glycoside Tum corymbosum Vahl. Austral. J. 1981. Naturally Chem. 8. ,B.L. i SON & W.SEGAL. 1958. Colouring matters of ры ж plants VI. Haemocorin: the structure of the aglycone. Austral. J. Chem. 11: 930. CroizaT, L., С. J. NELSON & D. E. Rosen. 1974. Cen ters of m and related concepts. Syst. Zool. 23: ee A. 1981. An Integrated System of Classi- n of Flowering Plants. Columbia Univ. Press, rom | баш н R. 980. A revised system of classification the angiosperms. Bot. J. Linn. Soc. 80: 91-124. & Н. Т. CLIFFORD. 1982. The Monocotyle- dons, a Comparative Study. Volume 2 in Botanical Systematics: An Occa Ed Series of Monographs. Academic Press, New York. . М. RASMUSSEN. 1983. Monocotyledon evolution: characters and pL estimation. In: M. K. Hecht, B. Wallace & G. T. Prance (edi- tors), d Biology 16: 255- 305. ‚ CLI & Р. F. Yeo. 1985. The Fai ol the Monocotyledons: Structure, Evolu- n, and Taxonomy. Springer-Verlag, New Yor jen. E 1872. The Origin of Species by Means of Natural Selection, 6th edition. Down, Beckenham, Kent. Mentor edition, 1958. New American Library, k New Yor DELLERT, R. 33. Zur systematischen Stellung von Wachendorfia. Oesterr. Bot. Z. 82: 335-345 De Vos, M. P. The development of the eis and seed in the Hypoxideae. 1. lanthe. J. S. Bot. 14: 159-169. 949. The development of the ovule and the seed in the iode 2. The Lea Pauridia and Forbesia. J. S ot. 15: 1950. Die Ontwikkeling v van qe Saadknop en Sa ad by Cyanella S M ra "n Gefal von Polyem- bryonie. S. Afr. J. Sci. 46: 220-226. 6. = on the ura and rela- tionships of South African genera of the Haemodora- ceae: veo Vidi and Wachendorfia Burm. J. S Afr. Bot. -63. v з the embryology and relationships of the South African genera of the Haemodoraceae. In: Recent Advances in Botany 1: 194-698. . Studies on the embryology and rela- tionships of South African genera of the Haemodora- т. Bot. 29: 79-90. "1979. angaroo paws and related | species Austral. Plants 10: 199-237. "Рохе, J. A. & M. J. DoNocHUE. 1986. Seed plant loge and the origin of angiosperms: an exper- mental cladistic approach. Bot. sath 52: 321-431. бен LDER, J. E. & S. С. Н. ВАК 1986. РҺу- к systematics of пуа Syst. Bot. 11: 373-391. EDWARDS, J. M . CHURCHILL & Ea Weiss. 1970. A chemical ‘contribution to the taxonomic status of Lophiola americana. йек енш 9: 1563- 15 ENDLICHER, I. L. 1836-1840. Genera Plantarum Se- cundum Ordines Naturales OS Apud Fr. Beck Universitatis Bibliopolam, Vienna E, В. Н. & С. С. TSENG. EYDE, 1969. Flower of Tetra- plasandra gymnocarpa, hypogyny with epigynous ancestry. Science 166: 506-508. AEGRI, K . VAN DER PUL 1966. The Principles of Pollination Ecology. Pergamon Press, Oxford. FERNANDES, A. & J. B. Neves. 1962. Sur la caryologie de quelques monocotylédones africaines. Compt. Rend. de la IVe Réunion Pléniére de l'Assoc. pour l'Etude Taxonomique de la Flora d'Afrique Tropicale: 439- 463. Fries, T. C. E. 1919. Der Samenbau bei Cyanastrum. Svensk Bot. Tidskr. 13: 295-304. ioci 5 Considerations taxonomiques au Sn 1 cts) Bull Soc. Roy. Bot. Belgique 101: 265- 278 Génera des Haemodoraceae et des . 1969a. A Bull. Jard. Bot. Natl. Belgique 39: ———. 1969b. Le genre Conostylis R. Br. (Hae- modoraceae a ke ua Bull. Jard. Bot. Natl. Bel- gique 39: 167- 1970. a du genre Anigozanthos Labill. (Hasciedoratess d'Australie). Bull. Jard. Bot. Natl. Belgique 40: 261-276. GorpBLaTT, P. 1989. Chromosome number in Walleria (Ticophilaeaceae). Ann. Missouri Bot. Gard. 76: 925- GRANT, V. 1950. The protection of x ovules in flow- ering plants. Evolution 4: 179-20 GREEN, J. W. 1 The genus (odd R. Br. I. Leaf UA Proc. Linn. Soc. New South Wales 84: 194-20 1961 The genus Constylis. 11. Taxonomy. Proc. Linn. b New South Wales 85: 334-373. BENE. А. Beitrage zur Entwicklungsge- schichte der Тассасееп, Bot. Not. 1921: 189-268. n U. 1 mbryolagische, morphologisch- natomische und systematische Untersuchungen an Phivdraceen. Willdenowia 4: 1-178. Harris, P. J. & R. D. HARTLEY. 1980. Phenolic con- stituents of the cell srs of monocotyledons. Bio- chem. Syst. Ecol. 8: -160. HENNIG, W. 1966. Phylogenetic Systematics. Univ. Illinois Press, Urbana, Illinois HILLIARD, О. M. & B. L. Burtt. Notes on some 1971. Volume 77, Number 4 1990 Simpson 783 Phylogeny and Classification of Haemodoraceae plants Africa chiefly Vu Natal: II. Notes Roy e» Gard. Edinburgh 31: 1-33. Horts, 8, D. 19 Variation 23 pipi hybridiza- tion in the Conostylis aculeata species group near Dawesville, Western Australia: poni J. Bot 25: 395-411 1978. Консо поїеѕ апа пем їаха іп the Conostylis Err group (Haemodoraceae). Nuytsia 2: 254-2 A. H. aie 1978. Assortative polli- nation by red wattlebirds in a hybrid population of Anigozanthos Labill. (Haemodoraceae). Austral. J Bot. 26: rl 50. ———— & N. A. CAMPBELL. 1977. A multivariate morphometric study of species relationships in kan- garoo paws (Anigozanthos Labill. and Macropidia umm. ex Harv.: Haemodoraceae). Austral. J. Bot. 25: 523-544 Hsu, C. C. А > Pidinings chromosome studies on the vascular plants of Taiwan (1). Taiwania 13: 117- 130 Huser, Н. 1969. Die Samenmerkmale und Verwandt- idi чанын ue е Mitt. Bot. Staats- samml. Miinchen 8: -9 a нен J. The Families of Flowering Plants, lst edition, Volume 2. Monocotyledons. Oxford Univ. Press, m on 59. The Families of Flowering Plants, 2nd Ae е: 2. Monocotyledons. Oxford Univ. Press, London. . 19 3 The Families of Flowering Plants, 3rd edition. Oxford Univ. Press, London. JOHANSEN, D. A. 1940. Plant Microtechnique. McGraw- KEIcHERY, G. J. 1981. Pollination and the generic status of Blancoa canescens Lindl. (Haemodora- ceae). Flora 171: 521-52 KORNFELD, J. M . M. EDWARDS. 1972. An inves- tigation of the photodynamic pigments in extracts of Lachnanthes tinctoria. Biochim. Biophys. Acta 286: 88 Kress, W. J., D. E. Stone & S. C. SELLERS. т Ultrastructure м exine-less pollen: + (Heli- coniaceae). Amer. J. Bot. 65: 1064- KnisHNAPPA, D. C. "197. Cytological Pudet in some aquatic angiosperms. Proc. Indian Acad. Sci. 73: 9-185 La Cour, L. К. 1956. Chromosome count of Tecophilaea cyano-crocus. In C. D. Darlington & A. P. Wylie (editors), Chromosome Atlas of Flowering Plants. Macmillan, London. LiNDLEY, J. 1830. An Introduction to the Natural Sys- m of Botany. Longman, Rees, Orme, Brown and Green, London. LowbEN, R. M. 1973. Revision of the genus Pontederia L. Rhodora 75: 426-487. MADDISON, W. Р J. DONOGHUE & D. R. MADDISON. 1984. Outgroup analysis and parsimony. Syst. Zool. -103. MAGUIRE, B. & J. J. WurDAck. 1957. The botany of . & Т. К. PANDITA. 1978. In: IOPB chro- mosome number reports LXI. Taxon 27: 375-392. . К. SACHDEVA. 1971. In: IOPB chromo- some reports XXXIII. Taxon 20: 609-614 & —— ——. 1976. Cytological observations on some west Himalayan monocots. IV. Cytologia 41: Мао H. 1964. Haemodoraceae. In Engler’s Syl- der Pflanzenfamilien, 12th edition, Volume 2. ebruder Borntraeger, Berlin. MILLSPAUCH, C. F. 1 . American Medicinal Plants. крз Press of Sherman & Co., Philadelphia, Penn- sylvan MITRA, K. "1966. Karyological a in Curculigo orchioides Gaertn. Sci. Cult. -202. NARANJO, С. А. 1975 Cotta а іп Нурохіѕ L. (Hypoxidaceae). » enm of H. decumbems L. Phyton 33: T N. v. В. JosH1, S. KRISHNAN, M. V. Panse & M. N. WAMBURKAR. 1968. Antibacterial noe rties of perinaphthenone derivatives. Curren ia) 37: 288. NiETSCH, Н. 1941. Zur systematischen Stellung von 90: 31-52. 1977. Atlas of Airborne Pollen Grains and Spores in Northern urope. Naturoch Kultur, Stockholm — R. 1969. The breeding system ot Pontederia m т Bull. Torrey Bot. Club 93: 407-416. a. Chromosome sip i and relation- ships of « certain African and Am Haemodoraceae. Ann. Missouri Bot. ‘Card. 66: 577- 580 Chromosome numbers in Cyanella (Tecophilaeaceae). Ann. Missouri Bot. Gard. 66: 581- 583. & R. DULBERGER. 1978. Floral жен and the reproductive system of Wachendorfia panic- ulata. New Phytol. 80: 427-434. : iia with fleshy bil n Australian survey -215 in J. S. Pate А. ]. McComb (editors), The Biology of Australian а Univ. W. ustral. Press, Ne Pax, F. 1888 aemodoraceae, Deia In: A. Engler 8 K. Prantl, Die P ichen Pflanzenfami- lien. aa) 92-96, 122- edis e. In: А. Engler & К. Prati. зэ Netürlicben Phanzenfamilien. 15a: 386- 391. & К. HOFFMANN. 1930. Amaryllidaceae. In: A. Engler & K. Prantl, Die Natürlichen Pflanzen- familien. 15a: 391-430. PLATNICK, N. I. 1976. Are monotypic genera possible? Syst. Zool. 25: 198-199. RAGHAVAN, R. S. Chromosome numbers in Indian medicinal plants. Proc. Indian Acad. Sci. 45: 294- 298. RATTER, J. A. & C. MILNE. 1973. Some angiosperm chromosome numbers. Notes Roy. Bot. Gard. Edin- burgh 32: 429-438 RAVENNA, Р. 1977. Neotropical species threatened and endangered by human activity in Iridaceae, Amarylli- daceae, and allied bulbous families. Jn С. T. Prance . Elias (editors), Extinction is Forever. New York. The genera of Haemodoraceae in the Southeastern United States. J. Arnold Arbor. 97: 205-216. SAITO, K. 1975. Studi th f polyploidy 784 Annals of the Missouri Botanical Garden and its contributions to the flower plant breeding. II. On the nature and sterility in the spontaneous triploid cultivars of Rhodohypoxis baueri Nel. Jap. J. Breed. 25: 355-362 SARKAR, А. K., №. Datta & U. CHATTERJEE. 1975. In: IOPB chromosome number reports XLVIII. Taxon 24: 367-372. Sato, D. Karyotype alteration and ро of Liliaceae and allied families. Jap. J. Bot Т SCHNARF, K. 1931. Vergleichende Embryologie der An- giospermen. Gebruder Borntraeger, Berlin. SCHULZE, R. 1893. Beitrage zur vergleichenden Anato- ili n, Haemodoraceen, Hypoxidaceen, und Vellosiacnn. Bot Jahrb. Syst. 17: 295-394. RH 22. Die "Teilung des vegetativen Pollenkerns bei [red crassipes Ber. Deutsch. Bot. Ges. 40: 60-63. ScHWENK, E. 1962. Tumor action of some quinoid "AIC in the Cheekpouch Test. Arzneim. Forsch. 12: 1143. SHARMA, А. К. . Annual report, 1967-68. Res. Bull. Univ. Calcutta (Cytogenetics Lab.) 2: 1-50. & М. К. BHATTACHARYYA Ап investi- gation on the scope of a number of pretreatment chemicals for chromosome studies in different groups of plants. Jap. J. Bot. 17: 152-162. ——&M 1964. Cytological studies aid in assessing the status of dee evieria, (эы and Curculigo. Nucleus 7: 43-58. . GHOSH. 1954, Further ie on the cytology of the family “Amaryllidaceae” and its eg p E interpretation of its phylogeny. Genet. Iber. 6: E oe (editors). 1967-1968. Chro- mosome number reports of plants. /n Annual Report, Cytogenetics Laboratory, Dept. of Botany, Univ. of Calcutta. Simpson, М. С. 1981. Embryological development of Lachnanthes caroliniana and Lophiola aurea (Hae- modoraceae) and its o significance. Ab stracts, Botanical Society of America. Miscellaneous Series. Publication 160: A.I.B.S. meetings, Bloomington, Indian 1983. Pollen ‘ultrastructure of the Haemo- Arcana and its taxonomic significance. Grana 22: 3. 985a. Pollen ultrastructure of the Philydra- ceae. ips 24: 23-31. 985b. Pollen ultrastructure of the Tecophi- vss Grana 24: 77-92. Pollen d of the Pontede- mology with the Hae- “| 26. Embryological development of Lach nanthes caroliniana (Haemodoraceae). Amer. J. Bor 75: 1394-1408. 1989. Pollen wall development of Xiphidium coeruleum (Haemodoraceae) and its systematic im- plications. Ann. Bot. ches 64: 257-269. . Haemodoraceae. Tecophilaeaceae. Jn R. Dahl- gren & P. Goldblatt (editors), Families and Gen nera of Flowering Plants, Volume 2. Monocotyledons (in press) & W. C. Dickison. 1981. Comparative anat- omy of La eee and Lophiola (Haemodora- 5-11 6. Geografischskoe raprostra- nenie poliploidnykh vidov rasteniy. (Issledovanie flor y Primorskogo AA .) Vestn. Leningradsk Univ., Ser. Biol. 3: 92-10 SPORNE, K. R. s The Morphology of Angiosperms. St. Martin's Press, New York. a G.L. 1974. Flowering Plants, Evolution Above e Species Level. Belknap Press of Harvard Univ. setts. 1925. Embryologische Studien. 1. Zur Em- bryologie eniger Columniferen. 5 Die Embryologie der Amaryllidaceen. Diss. Uppsa ——. . Zur н РЕ der Gat- nr Anigozanthos Labill. Bot. Not. 1927: 104- 1938. Die systematische Stellung der Gattung Xiphidium. Svensk Bot. Tidskr. 32: 274 STONE, D. E., S. C. SELLERS & W. J. Kress. 1979. Ontogeny a exineless pollen in a : rir nie . Missouri Bot. Gard. 66: SWOFFORD, D. L 1983. Phylogenetic ы using parsimony (PAUP). Illinois Natural History Survey, Champaign, Illinois. [Computer program TAKHTAJAN, А Outline of the classification of flowering pd ши Bot. Rev. (Lan caster) 46: THISELTON- Dre, Y. T 96-1897. Flora ke toca modoraceae M Liliaceae. L. Reeve & Co., Covent t Саг n. WALKER, J. W. 1986. oe ie ат of ie лү с : ‚А. 75. The ida of angiosper p palynology. Ann. Missouri Bot. Gard. 62: 664- лаа R 1967. Toa a on some Hypoxis: 1. Somatic chrom and meiosis in some Hypoxis species. J. S. “Afr. Bat 33: 75-84. & 1967. Cytological obser- me Hypoxis species. J. S. A TANADA: M. S. 1983a. On "p taxonomic status of Lophiola aurea Ker-Gawler. Rhodora 85: 73-81. 983b. Comparative morphology of monocot pollen and evolutionary trends of apertures and wall structures. Bot. Rev. (Lancaster) 49: 331-379 A CONTRIBUTION TO THE ORCHID POLLINIA! Michael S. Zavada? ABSTRACT Pollen grains of 30 orchid species from the Spiranthoideae and V dinis (including the Vandeae) were studied ultrastructurally. This study complements the published palynological literature e Apostasioideae, „р Spiranthoideae, and Orchidoideae. The orchid pollen thus far studied exhibits a p range of variation in pollen un only the most peripheral grains in the pollinium is accompanied d some po shed icd in e features observed of the intine. Аз in other monocots, endexine has not been unequivocally аккош ny in orchi The Orchidaceae comprise about 1,000 genera and 15,000- 20,000 species. Dressler (1981, 1983) has separated the orchids into five subfamilies: Apostasioideae, Cypripedioideae, Spiranthoideae, Orchidoideae, and Epidendroideae (including the Vandeae). There is general agreement among taxonomists that the Apostasioideae are the least specialized (primitive) subfamily of the Orchidaceae (Dressler Dodson, 1960; Dressler, 1981, 1983), with two genera and about 16 species, found primarily in tropical Asia. Pollen morphology and ultrastructure have been investigated (Schill, 1978; Newton & Williams, 1978) and the pollen shows features of many other monocots (Zavada, 1983). The grains of the Apostasioideae are usually shed in monads and are monosulcate. Exine sculpturing is reticu- late. Pollen wall structure is tectate-columellate with a thin footlayer and no endexine. The Cypripedioideae, primarily a tropical group, have four genera and about 130 species. This subfamily likewise exhibits many primitive features (Ackerman & Williams, 1980, 1981). Pollen is generally monosulcate or ulcerate (i.e., with an ill- defined pore). Exine sculpturing can be reticulate or verrucate to scabrate. The pollen wall infra- structure, as revealed by SEM, appears to be col- umellate (Newton & Williams, 1978; Schill & Pfeif- fer, 1977). Representative taxa of the rather large subfam- ily Spiranthoideae have been investigated palyno- logically to some extent (Balogh, 1979; Williams & Broome, 1976). Pollen grains are shed in loose pollinia and are generally inaperturate. Exine sculpturing is reticulate (especially in the Spiran- thinae), and wall structure varies depending on the position of the pollen grain in the pollinium: pe- ripheral grains are usually tectate-columellate with a thick footlayer; those in the interior of the pol- linium usually lack the tectum, with only the col- umellae and footlayer being prese Pollen of the Spiranthoideae and Tpidendvotioas have been little studied using transmission electron microscopy (however, see examples in Chardard, 1958, 1969; Heslop-Harrison, 1968; Dulieu 1973; Caspers & Caspers, 1976; Balogh, 1982; Balogh & Mann, 1982; Hesse & Burns-Balogh, 1984; Wolter & Schill, 1986). The present contribution describes the pollen wall structure of 30 taxa from these subfamilies (only one taxon from the Spiran- thoideae METHODS Pollinia were removed from living material and placed immediately in cacodylate HCl-buffered glu- taraldehyde-formaldehyde, followed by fixation in osmium tetroxide, dehydration in an ethanol series, and embedding in Dow Epoxy Resin 334. Pollinia were sectioned on an LKB-1 ultramicrotome, se- ! I thank Tim Reeves and Eric Christenson for providing the live material used in this study. з Department of Biology, The University of Southwestern Louisiana, Lafayette, PO 70506, U.S.A. ANN. Missouni Вот. Garb. 77: 785-801. 1990. 786 Annals of the Missouri Botanical Garden quentially poststained in uranyl acetate and lead citrate for 15 minutes, and viewed with a Philips -300 transmission electron microscope. Vouchers for all live specimens have been deposited in the herbaria CONN or MRD (Appendix 1). The classification used follows Dressler (1981, 1983). DESCRIPTIVE PALYNOLOG Y SPIRANTHOIDEAE Erythrodeae-Spiranthinae Sacoila lanceolata (Aubl.) Garay. Pollen grains occur in loosely packed pollinia and are inaper- turate. Exine sculpturing is reticulate and pollen wall structure varies depending on the position of the pollen grain in the pollinium. Pollen grains peripheral in the pollinium are tectate-columellate with a relatively thick footlayer (Fig. 1). Pollen in the interior of the pollinium usually lack a tectum; however, the columellae and footlayer are present (Fig. 2). The footlayer is underlain by a thin fibrillar intine (Fig. 2) EPIDENDROIDEAE Arethuseae- Bletiinae Acanthephippium sylhetense Lindl. and Bletilla striata (Thunb.) Reichb. f. In both species pollen occurs in tightly packed pollinia. Pollen of Bletilla striata appears multiaperturate in the peripheral region of the pollinium, the exine having a number of thin areas (Fig. 3). Pollen wall structure is tec- tate-columellate (Fig. 4). The columellae are thin and rest on a unilamellar footlayer that is underlain by a thick multilayered intine (Fig. 4). Pollen in the interior of the pollinium lack an exine and are surrounded by intine; however, there appear to be small sporopolleninous granules embedded in the outer portion of the intine (Fig. 5). FicunES 1-8. , 2. Pollen of the Spiranthoideae. — 1. showing tectate- ш wall structure with the well- Wan footlayer (FL), x interior in the pollinium showing columellae, footlayer, and t — Sacoila lanceolata, Ley peripheral in the pollinium 0.—2. S. lanceolata pollen in fibrillar intine. The tectum appears to be absent, x 6,500. 3-8. Pollen of the Epidendroideae.—3. Bletilla striata pollen peripheral in the pollinium showing tectate- columellate wall, the thin footlayer, and the thin areas of the exine (arrowheads) that may function as apertures, x 6,500. — 4. High magnifica x 3,450.— T. intine. No a wall is present, x 8,200. FIGURES 9-16. Pollen of the Epidendroideae. —9. Co the thick tectum and granular infrastructural layer interior in the pollinium showing m yered fibrillar i a x 200.— 12. High magnification o M. palu "sutures" between individual pollen Note suture between in the кү showing atectate wall, x 7,000. — wall, FIGURES 17-24. showing oval columellae- like structures = rest C. skinneri sho (arrowheads), x 12,0 x 6,500. — 21. Pollen peripheral in in that rests directly on the thick i in the pollinium of E. anceps showing d. pollen x4,150.— 23 the D of Epide Not that rests directly o ntine. Exine is absent, ad, compare to the surface features of M. paludosa, F ~ = = = 3 = & = + yal о. Ф =) ge % =] gz) SP c “ч e “Яя "d = O ч e Е 5 n =E © 5 =} О = Ld qc © Lad = E © 3 e tion of B. striata pollen peripheral in the pollinium showing wall structure, х 8,200. — 6. A AD 4. sylhetense pollen most interior in the pollinium showing the fibrillar oelogyne nitida pollen peripheral in the pollinium showing e intine, X6,500.— . М. paludosa pollen al . М. unifolia, pollen peripheral in the pollinium showing atectate Pollen of the Ерійепӣгоійеае. — 17. Pollen и іп the pollinium of Cattleya skinneri on a thin intine, х15,125. . Po llen interior in the polliniu m of n from the interior of the pollinium of E. cochleata showing fibrillar intine; often sporopolleninous granules (arrows) are embedded in this wall layer at oe juncture of three o ndrum ance e е of the tectum бойо, х11,200.— 22. Pollen interior L. autumnalis showing exineless pollen with a apo ov Note cytoplasmic connection between adjacent pollen grains (arrow) and minute granules (arrowheads), х 3,5 Volume 77, Number 4 Zavada 787 1990 Pollen Wall Ultrastructure of Orchid Pollinia 4 PA A н 788 Annals of the Missouri Botanical Garden Zavada 789 Volume 77, Number 4 1990 Pollen Wall Ultrastructure of Orchid Pollinia 790 Annals of the Missouri Botanical Garden FIGURES 25-30. Pollen of the Epidendroideae. — 25. Pollen igi dels in the pollinium of Masdevallia caudata showing thick atectate exine that is underlain by a thin intine. Note the microchannels that traverse the lower portion of the exine (small arrowheads) and the "suture" betwe RC pollen grains (large arrowhead), х 5,300.— 26. Pollen interior in the pollinium e e caudata ШОО. б е а pollen grains that are surrounded by a thin intine. The dense nonacetolysis resistant layer between pollen grains may be sie of the intine, x10 peripheral in the pollinium of Dryad ella edwallii ан atectate exine with thin bilayered intine (small arrowhead). Note suture between adjacent pollen grains. When the tectum of adjacent pollen grains is closely appressed to one another the exine may prevent dessication of the exineless interior pollen grains. As the pollinium expands these sutures may open permitting the emergence of the pollen tubes (also see Figs. x 16, 25, 29), x6,500.— 28. Pollen interior in the gcns of D. edwallii showing multilayered intine, X 10,000. — 29. Pollen peripheral in the pollinium a ee janetiae showing atectate exine. Note the suture ест adjacent pollen grains он 30. Pollen i interior in the pollinium of P. janetiae рези the presence of only the intine, х 10,500 oS c c | to N "Y e. o 5 Volume 77, Number 4 1990 Zavada Pollen Wall Ultrastructure of Orchid Pollinia Acanthephippium sylhetense pollen grains are inaperturate. Exine sculpturing is psilate. The wa structure of pollen peripheral in the pollinium is tectate with oval columellae-like elements in the infrastructure (Fig. 6). The closely spaced colu- mellae-like elements are often thick branching rods to oval and granular-like structures that support the tectum (Fig. 6). The bases of the columellae rest directly on a thin fibrillar intine. In the interior of the pollinium the tectum is generally absent, being comprised of short, stout columellae or irregularly shaped granules (Fig. 7). The exine is entirely absent from the innermost pollen in the pollinia; each pollen grain is surrounded by a fibrillar intine Coelogyneae—Coelogyninae Coelogyne nitida Lindl. Pollen occurs in tightly packed pollinia and appears inaperturate. Pollen peripheral in the pollinium have a thick tectum, which is underlain by a massive granular layer (Fig. 9). The granular layer rests directly on a thick bilayered intine (Fig. 9). Pollen interior in the pollinium lack an exine, and each pollen grain is surrounded by a bilayered intine (Fig. 1 Malaxideae Malaxis latifolia J. E. Sm., М. paludosa Sw., and M. unifolia Michx. Pollen occurs in small, tightly packed pollinia and are inaperturate (Figs. 11, 12). The outer wall of pollen peripheral in the pollinium is thick and homogeneous (atectate) (Figs. 13, 15, 16). Each pollen grain peripheral in the pollinium has its own wall (Figs. 12, 13); however, the walls are closely appressed to one another, often giving the appearance that the entire pollinium is enclosed by a single sporopolleninous wall. Pollen interior to these outermost pollen grains appear to lack exine in M. latifolia (Fig. 13) and are surrounded only by intine; however, a thin homogeneous spo- ropolleninous wall surrounds each microspore in the two North American species (Figs. 15, 16). All three species investigated are identical with regard to wall ultrastructure of pollen most peripheral in the pollinium. The cytoplasm in М. latifolia has what appears to be condensed rough endoplasmic reticulum (Fig. 14), a condition often found in resting cells in animals, but a rare occurrence in plants. Epidendreae—Laeliinae Cattleya skinneri Batem., Encyclia cochleata (L.) emee, Epidendrum anceps Jacq., and Laelia autumnalis Lindl. Pollen in all four taxa occur in tightly packed pollinia. Pollen grains peripheral in the pollinium have a thick tectum underlain by large, stout columellae or oval granules that rest directly on the intine (Figs. 17, 19, 21). Pollen more interior in the pollinium are exineless and surrounded only by an intine (Figs. 18, 20, 22). However, at the junctures of two or more pollen grains, sporopolleninous granules are evident (Cat- tleya, Fig. 18 and Encyclia, Fig. 20). In Epiden- drum (Fig. 22) minute sporopolleninous granules are embedded in the intine between adjacent pollen grains. Laelia autumnalis deviates from the above taxa by the occurrence of the tectate-granular wall. Pollen grains peripheral in the pollinium have a thick tectum underlain by a granular layer (Fig. 23). The tectum peters out on the lateral faces of the peripheral pollen grains (Fig. 23). Pollen in the interior of the palcos dr are similar to Loses белү in having granules embed ded in the outer уер ‘of the intine (Fig. 24). Epidendreae—Pleurothallidinae Pollen of all four taxa investigated in this sub- tribe occur in tightly packed pollinia and are inaper- turate. Masdevallia caudata Lindl. Pollen grains periph- eral in the pollinia have a thick homogeneous outer layer (Fig. 25). The inner portion of the exine is also homogeneous but is traversed by minute chan- nels (Fig. 25). The inner surface of the exine is endo-rugulate and rests directly on a thin intine (Fig. 25). Pollen in the interior of the pollinium are exineless, surrounded only by the intine (Fig. Dryadella edwallii (Cogn.) Luer. The wall struc- ture of pollen most peripheral ju in к the рокид consists of a thick | 27). This wall layer rests ey on a thin ed Pollen grains in the interior of the pollinium are exineless and are surrounded by a multilayered intine (Fig. 28). Pleurothallis janetiae Luer and Restrepia striata will be treated together. Pollen grains peripheral in the pollinium have a thick homogeneous tectum (atectate) that rests directly on the intine (Figs. 29, 31). However, the exine becomes rugulate (the rugulae are large granular structures) on the lateral walls of the outer pollen grains (Fig. 29, 32). Pollen interior in the pollinium lack exine and are sur- 792 Annals of the Missouri Botanical Garden Volume 77, Number 4 1990 Zavada Pollen Wall Ultrastructure of Orchid Pollinia rounded only by a multilayered intine (Figs. 30, 33). Epidendreae—Dendrobiinae Dendrobium nobile Lindl. Pollen grains occur in tight pollinia and are inaperturate. Grains periph- eral in the pollinium have a thick tectum underlain by a granular layer (Fig. 34). The tectum peters out toward the contact areas of adjacent pollen grains, where the granular layer is only present. The granular layer rests on a fibrillar intine (Fig. 34). Pollen interior in the pollinium are exineless and surrounded by a bilayered intine (Fig. 35). D. aphyllum (Roxb.) Fischer. Pollen grains oc- cur in loosely packed pollinia and are inaperturate. Pollen in both the peripheral and interior regions of the pollinium are exineless (Figs. 36, 37); how- ever, pollen peripheral in the pollinium have a massive multilayered intine (Fig. 36). Pollen in the interior of the pollinium have a thin multilayered intine (Fig. 37). Epidendreae-Bulbophyllinae Bulbophyllum imbricatum Lindl. Pollen grains oc- cur in tightly packed pollinia and are inaperturate. Pollen peripheral in the pollinium have a tectate- granular wall (Fig. 38), the granular layer resting directly on a thin intine. Pollen grains interior in the pollinium are exineless, surrounded only by intine (Fig. Vandeae-Sarcanthinae Schoenorchis fragrans (Par. & Reichb. f.) Seid. & Smit. Pollen grains occur in tightly packed pol- linia and are inaperturate. Pollen peripheral in the pollinium are surrounded by a homogeneous atec- tate exine (Fig. 39). Pollen interior in the pollinium is exineless, and each pollen grain is surrounded by a thin bilayered intine (Fig. 39). Ascocentrum ampullaceum (Roxb.) Schltr., Cleis- ostoma racemiferum (Lindl.) Garay, Haraella re- trocalla (Hayata) Kudo Luisia teres (Thunb.). Bl. var. botanensis (Fuk.) T. P. Lin, Phalaenopsis lueddemanniana Reichb. f., Thrixspermum saru- watarii (Hayata) Schltr. Pollen grains of all the species investigated occur in tightly packed pollin- ia. Pollen peripheral in the pollinia have a thick homogeneous, atectate wall with endo-rugulations (Ascocentrum, Fig. 46; Phalaenopsis, Fig. 40). Pollen of Luisia, Thrixspermum, Cleisostoma, and Haraella have tectate-granular exines (Figs. 42, 44, 48, 50). The exine rests on a thin (Thrixsper- mum, Fig. 42) or thick (Phalaenopsis, Fig. 40; Luisia, Fig. 45; Ascocentrum, Fig. 47; Cleisosto- ma, Fig. 49) multilayered intine. Pollen interior in the pollinium are exineless and are surrounded by a thick, often elaborated intine (e.g., Thrixsper- mum) (Figs. 41, 43, 45, 47, 49, 51) Vandeae- Zygopetalinae Cochleanthes discolor (Lindl.) Schultes & Garay. Pollen grains occur in tightly packed pollinia. Pol- len grains peripheral in the pollinium have a tec- tate-granular exine (Fig. 52), the granular layer resting directly on a bilayered intine (Fig. 52). Pollen grains interior in the pollinium lack exine and are surrounded only by the bilayered intine (Fig. 53). Vandeae- Bifrenariinae Rudolfiella aurantiaca (Lindl. Hoehn. Pollen grains occur in tightly packed pollinia. Pollen pe- ripheral in the pollinium have a tectate-granular exine, the granular layer resting directly on the intine (Fig. 54). The exine is perforated with minute channels (Fig. 54). Pollen in the interior of the pollinium are exineless, surrounded by a thin intine (Fig. 55); however, sporopolleninous granules oc- casionally are embedded in the thin intine. Vandeae—Ornithocephalinae Zygostates alleniana Kraenzl. Pollen grains occur in tightly packed pollinia. Pollen peripheral in the pollinium have a tectate-granular wall structure (Fig. 56); the granular layer rests directly on a FIGURES 31-37. showing atectate exine, X 2 grains peripheral in the pollinium, x 16,000. Pollen of the Epidendroideae.—31. Pollen peripheral in the pollinium of Restrepia striata 0,500.— 32. R. striata pollen showing sporopolleninous granules between adjacent pollen —33. R. striata pollen interior in the pollinium showing the thin intine, 0.—34. Pollen peripheral in the pollinium of Dendrobium nobile showing tectum and the dense granular layer that rests directly on a fibrillar intine, x15,1 bilayered intine, x 6,550. fibrillar intine, x 5,30 intine, x 3,450 5.—35. Pollen interior in the pollinium of D. nobile showing — 36. Pollen peripheral i in the pollinium of D. aphyllum showing the presence of only the 0.— 37. Pollen interior in the pollinium of D. aphyllum showing loosely packed pollen and the 794 Annals of the Missouri Botanical Garden FIGURES 38-43. Pollen of the Epi dendroideae.— 38. Pollen 4 ев imbricatum showing ошег grains with tectate- granular wall and exineless interior pollen grains, x 6,5 39. Pollen of Schoenorchis fragrans showing the atectate e that is cc uou ound the entire pollinium ie exineless interior pollen grains, x 3,500.— 40. Pollen peripheral in the pollinium of Phalaenopsis lueddemanniana showing atectate exine with endo-rugulations (arrowhead), х 20,500.— 41. Exineless pollen interior in the оаа of Р. lueddemanniana showing multilayered intine, d o 4.2. нше шыла уш in the pa of Thri rmum saruwat tar showing the exine that is underlain by a nty lar А , X6,500.— 43. Exineless s interior in the тэе of Т. saruwatarii showing каин intine, x28, 500 Volume 77, Number 4 Zavada 1990 Pollen Wall Ultrastructure of Orchid Pollinia Ficures 44-49. Pollen of the Epidendroideae.—44. Pollen peripheral in the pollinium of Luisia teres var botanensis showing thick tectum and the thin granular infrastructural layer that rests directly on the bilayered intine, x 6,500.— 45. Pollen interior in the pollinium of L. teres var. есе showing multilayered intine. Note the small | : 5 Pollen : : granules embedded in the dense intinous layer (arrowhead), x4 .—46. Pollen peripheral in the pollinium of Ascocentrum ampullaceum showing atectate exine with endo-rugulations ж te the thinning of = exine; these thin areas may function as apertures (arrow), X11, .— 47. Exineless pollen interior in the pollinium of 4. ampullaceum showing multilayered intine, x 10,000. — 48. Pollen of Cleisostoma racemiferum showing ce ollen- inous wall between adjacent pollen grains, x 7,100.—49. Exineless Eg inferior in the pollinium of C. racemiferum, x3 , Annals of the 796 Missouri Botanical Garden Volume 77, Number 4 1990 Zavada Pollen Wall Ultrastructure of Orchid Pollinia bilayered intine (Fig. 56). In the tectum are urn- shaped structures of unknown function (this feature is not known to occur in any other pollen, gym- nosperm, or angiosperm) (Fig. 56). Pollen grains in the interior of the pollinium are exineless and are surrounded by a bilayered intine (Fig. 57). Cymbidieae Cymbidium aloifolium (L.) Sw. Pollen grains occur in tightly packed pollinia. Pollen peripheral in the pollinium have a tectate-granular wall structure (Fig. 58), the granular layer resting directly on the intine (Fig. 58). The tectum is traversed by minute channels (Fig. 58). Pollen interior in the pollinium are loosely attached and exineless; the intine is the only wall layer present (Fig. 59). However, spo- ropolleninous granules are occasionally embedded in the intine (Fig. 59). Cymbidieae-Oncidiinae Trichophilia marginata Henfr. Pollen grains oc- cur in tightly packed pollinia. Pollen peripheral in the pollinium have a tectate-granular wall structure (Fig. 60). The granular layer rests on a thin intine (Fig. 60). Pollen interior in the pollinium are ex- ineless and surrounded by a multilayered intine (Fig. 61); however, sporopolleninous granules often occur in the contact areas between adjacent pollen grains (Fig. 61). DISCUSSION Pollen grains from each of the five orchid subfamilies were studied ultrastructurally. Al- though the number of species studied is insubstan- tial in relation to the size of the family, the available data, when considered in the context of other mor- phological features, provide insight into the evo- lutionary trends of aperture and wall structure of the Orchidaceae. Burns-Balogh (1983) has proposed evolutionary trends for exine structure of orchid pollen. She considered pollen of the Apostasioideae most prim- itive. This is based on the pollen characters asso- ciated with unspecialized reproductive organs of the subfamily. The pollen wall of Apostasia is tectate-perforate-columellate with a well-developed footlayer. Burns-Balogh (1983) recognized two ma- jor lines diverging from the Apostasioideae. One ine culminates in the tectate-imperforate-granular wall of the Cypripedioideae via a tectate-imperfor- ate-columellate wall. The other line progresses to the tectate-perforate-columellate type, eventually giving rise to tectate-granular wall types of the Orchidoideae and Epidendroideae with the loss of the footlayer. In all of these types Burns-Balogh (1983) recognized the common occurrence of en- dexine. Endexine does not usually occur in mono- cots (Zavada, 1983), and this wall layer has not been unequivocally demonstrated in orchids: the footlayer or infrastructural layer in all cases rests directly on a fibrillar intine. Although Burns-Bal- ogh's (1983) treatment of evolutionary trends of wall structure in orchids may be reasonable, the small sample size does not permit placing the large subfamilies in an evolutionary sequence based on pollen alone. In addition, some workers have sug- gested that the large subfamily Epidendroideae may be polyphyletic (Dressler, 1983). The data in the present contribution will be treated in terms of major evolutionary trends of pollen wall structure within the Orchidaceae (not necessarily represent- ing cladistic relationships). This will be subject to revision and refinement as data accumulate on pollen ultrastructure in this fam ree major trends are ETIA in the Orchi- daceae. The first and most obvious is toward ad- herence of pollen grains into polyads (pollinia). This trend is best developed in the Orchidaceae and does not characterize many other monocot groups. However, polyads occur in the dicots (e.g., Ascle- piadaceae, Fabaceae). In dicots with polyads, there is observed a pollen wall ultrastructure comparable to that of the orchids, i.e., a shift from the tectate- columellate wall structure in the taxa considered least specialized to tectate-granular or atectate in the taxa considered derived (Dicko-Zafimahova & — FIGURES 50- 55. crochannels, arrowheads), g e, and thin intine, x 10,000. Pollen of the Epidendroideae. — 50. Pollen bases in the pollinium of Haraella retrocalla к infrastructure 50 x сл — t M. 3 Ф © Ф n y Ro. o 5 5 e 8 d © іа! Б dolfiella aurantiaca showin — 55. Exineless pollen interior in the pollinium of R. aurantiaca showing bilayered шше. x 10,000. 798 Annals of the Missouri Botanical Garden ^ y ~ е * * '"IGURES 56-61. Pollen of the Epidendroideae. — 56. Pollen peripheral in the pollinium of Zygostates alleniana showing tectum (T) with the unusual urn-shaped structures of unknown function (arrowheads), granular infrastructure (GR), and intine (IN), x11,500.—57. Exineless pollen interior in the pollinium of Z. alleniana showing intine, Volume 77, Number 4 1990 Zavada Pollen Wall Ultrastructure of Orchid Pollinia Audran, 1981). This raises the possibility that pol- len wall structure in taxa with polyads may be more related to the spatial configuration of the pollen and development of the sporopolleninous wall around the pollinia than a reflection of the evolutionary level attained by a particular taxon. Within the Orchidaceae the Apostasioideae have monads, but polyads and/or mealy pollinia occur in the Cypripedioideae and Spiranthoideae. The Orchidoideae and Epidendroideae have tightly packed, waxy pollinia (Dressler, 1981, 1983). The second trend is toward a well-defined sulcus in the Apostasioideae, an ill-defined ulcerate pore in some Cypripedioideae and Spiranthoideae, and the inaperturate condition in the Orchidoideae and Epidendroideae (Zavada, 1983). The monosulcate condition is found in the least specialized subfamily, Apostasiodese: In the more derived subfamilies (Cy ) the ulcerate con- dition i is known, МЕ іп the most derived taxa (Ep- idendroideae) the inaperturate condition is preva- lent. These trends in apertures parallel those observed in the other monocot groups, and in gen- eral, are characteristic of pollen of the Alismatideae and Zingiberidae (monosulcate to inaperturate condition). The Commelinideae exhibit a parallel trend in the reduction of the sulcus to an ill-define pore, but here culminate in the development of a well-defined graminoid pore. The inaperturate con- dition is not observed in this group (Zavada, 1983). The third major trend concerns exine structure and stratification. The occurrence of the tectate- columellate wall structure and monosulcate aper- ture in the Apostasioideae is typical of numerous monocot families including those generally consid- ered primitive (Zavada, 1983 and references there- in). There derived subfamilies that exhibit primitive wall struc- ture type (tectate-columellate; Cypripedioideae, Spiranthoideae, Orchidoideae, and a few Epiden- droideae). However, the primitive wall structure type often occurs with the more derived apertural are a number of species in the more types and pollen units. The combination of prim- itive and derived features in the few taxa investi- gated does not reveal the phylogenetic relationships among the Apostasioideae and these other subfam- ilies. In addition, the derived wall structure types (tectate-granular) also occur in the derived taxa (e.g., Orchidoideae and Epidendroideae). The Vandeae consistently exhibit the tectate- granular and atectate wall types, no footlayer, in- aperturate pollen, and compact pollinia, suggesting this is the most palynologically derived group of the family The cdi parallel many of the palynological evolutionary trends proposed for the other mono- cots (Zavada, 1983). However, a few features are unique to the monocots and the Orchidaceae in general. The first is loss of the sulcus via the ul- cerate-porate condition. This is a major monocot trend, but is relatively rare in dicots. This trend, as mentioned above, occurs in the Alismatideae, Zingiberidae, and in part, the Commelinideae. The second is the occurrence of the tectate-colu- mellate wall structure in monocots considered prim- itive and the least specialized orchid subfamily, Apostasioideae. The more advanced taxa have the tectate-granular, atectate, or exineless pollen. This is also a major monocot trend and is restricted to only a few dicot groups. Although some derived dicots have a granular or atectate wall structure, these wall structure types are thought to be prim- itive in some dicots (Walker, 1974). Concomitant with the loss of the tectum, infrastructure, and footlayer, there is elaboration of the intine in many monocots including the Orchidaceae. The occur- rence of the elaborated intine is often found in monocot taxa that have specialized pollination mechanisms (e.g., Cannaceae, Kress tone, 1982). However, the morphological diversity and significance of the intine to pollination mechanisms and reproductive biology need to be investigated further. Further pollen studies of orchids will undoubt- edly bring many surprises. lt is significant that orchids exhibit a wide range of pollen types that occur in the most primitive to the most derived monocots. Further detailed palynological studies of orchids and closely related families will be impor- tant to our understanding of pollen evolution, not only in the Orchidaceae, but in the monocots in general. x 5,300.— 58. Peripheral pollen in the pollinium of Cymbidium aloifolium showing microchannelled tectum (arrow- heads), granular infrastructure, and thi ck intine, x5,300.— 59. Pollen interior in the showing intine with minute sporopolleninous granules (arrowhead), x 3,450.— 60. of Trichophila marginata showing thick tectum and massive granular infrastructure that pollinium of C. aloifolium Pollen peripheral in the pollinium rests on a thin intine, x 6,500.— 61. Pollen interior in the pollinium of 7. marginata showing pales granular layer (arrow) and dense intine, X11,5 800 Annals of the Missouri Botanical Garden LITERATURE CITED ACKERMAN, J. N. H. WiLLiams. 1980. Pollen morpho э of the tribe Neottieae and its impact on the classification of the Orchidaceae. Grana 19: 7- 18. & 1981. Pollen morphology of the С EM (Orchidaceae: Diurideae) and related subtribes. Amer. J. Bot. 68: 1392-1402. BALOGH, P. en Pollen a of the tribe Cra- nichideae Endlicher subtribe Spiranthinae Bentham ras Orquidea 7: 242-260. 1982. Pollinarium morphology of Mexican Orchidaceae. I. Subtribe Laeliinae. Oides 327- 342. M.J. Mann. 1982. Column and pollinarium а of Rhizanthella gardneri Rogers. Instit. Physical Science and Technology, Univ. Maryland, Newsltr. Issue 10: 4-5. Burns-BALOGH, P. the exine in Ochidacesa, Amer. J. 1312 A theory on the evolution of Bot. 70: 1304- Caspers, N. & L. Caspers. 1976. Zur Oberflaechen- skulpturierung der Pollinien Mediterraner Orchis- und Ophyrs-arten. Pollen & Spores 18: 203-215. CHARDARD, R. L'ultrastructure des grains de pollen FOrchidacées. Rev. Cytol. et Biol. Veg. 19: i ks Aspects infrastructuraux de la matu- ration des grains de pollen de quelques Orchidacées. Rev. Cytol. et Biol. Veg. 32: 67-100. CRONQUIST, уа 9 An Integrated Ao of Flowering Plants. Columbia Univ. Press, New York. picó ан ул, L. D. & J. C. AUDRAN. 1981. Étude nique de la pollinie de Calotropis procera е) Сгапа 20: 81 DRESSLER, К. L. The Orchids, Natural History and Classification. Harvard Univ. Press, Cambridge, e ts. 983. Classification of the r and their б origin. Telopea 2: 413-4 & С. SON. 1960. aa and Bot. phylogeny of the Orchidaceae. Ann. Missouri Gard. 47: 25-68. Ошко, D. 1973. Etude morphologique de la surface pollinique de Ponthieva maculata Lindl. Orchida- ceae en microscopie électronique a balayage. Adan- sonia 13: 229-234 HEsLoP-HARRISON, J. the formation of the orchids. J. Cell Sci. 3: 457- 466. Hesse, M. & P. BuRNS-BALOGH. 1984. Pollen and pol- linarium morphology of Habernaria (Orchidaceae). Pollen & Spores 26: 385-400. Kress, W. J. D. E. STONE. 1982. Nature of the sporoderm in monocotyledons, with special reference to the pollen grains of Canna and Heliconia. Grana 21: 129-148 NEWTON, С. D. & М. Н. WiLLIAMs. 1978. Pollen mor- phology of the са and the шш deae (Orchidaceae). Selbyana 2: 169-1 SCHILL, R. 1978. Palynologische "Un: tersuc AER zur systematischen Stellung der Apostasiaceae. Bot. Jahrb. Syst. 99: 353-362 & 1968. а ройеп mitosis generat ive W. P Untersuchungen a Orchideenpollinien unter besonderer Berucksichti- gung ihrer Feinskulpturen. Pollen & Spores 19: 5- 118. WALKER, J. W. 1974. Evolution of exine structure in the pollen of primitive angiosperms. Amer. J. Bot. 61: 891-902. WiLLIAMS, N. H. & C. R. BRooME. 1976. Scanning electron microscope е i orchid pollen. Amer. Orchid Soc. Bull 45: 699- WOLTER, M. & К. ScHILL. HAA С MP von Pollen, Massulae, und Pollinien bei den ra rd Tropische und subtropische dp 56: 1-93. ZAVA ps M. 1 ee morphology of nocot pollen and ‘evolutionary trends of apertures and wall structure. Bot. Rev. (Lancaster) 49: 331- 379. APPENDIX I. Specimens investigated in this study. Acanthephippium sylhetense Lindl., Christenson 381 CONN, India Ascocentrum ampullaceum (Roxb.) Schltr., Christenson 4 ONN, Thailand Bletilla striata (Thunb.) Reichb. f., Christenson 374 C , Japan Bulbophyllum imbricatum Lindl., Christenson 426 С , Africa Cattleya skinneri Batem., Christenson 393 CONN, Costa Rica Cleisostoma racemiferum ( (Lindl.) Garay, Christenson 365 CO ndia Coc Diles discolor (Lindl.) Schultes € Garay, Chris- tenson 436 CONN, Costa Ric Coelogyne nitida Lindl., Christenson 1017 CONN, India Cymbidium aloifolium (L.) Sw., Christenson 384 CONN, ex hort. Dendrobium aphyllum (Roxb.) Fischer, Christenson 837 CONN, India Dendrobium nobile Lindl., Dryadella edwallii (Cogn. ) М, Brazil Encyclia cochleata (L.) Lemee, Christenson 1016 CONN, ex hort. Epidendrum anceps Jacq., Christenson 383 CONN, Flor- ida Haraella Sea (Hayata) Kudo, Christenson 1022 CONN, Tai Christenson 387 CONN, India Luer, Christenson 1015 = autumnalis ‘Lindl. Christenson 366 CONN, Mex- Luisia teres (Thunb.) BI. var. botanensis (Fuk.) T. P. in, Christenson 433 CONN, Malaxis latifolia Sm., Christenson 400 CONN, ex hort. Malaxis paludosa Sw., Reeves, 18 Aug. 1984 MRD, Minnesota ни: саш Michx. Reeves, ѕіп., 19 Aug. 1984 МКР, ota Masdevallia caudata Lindl., Christenson 379 CONN, Colombia Phalaenopsis lueddemanniana Reichb. f., Christenson CONN, Philippines Pleurothallis janetiae Luer, Christenson 1014 CONN, osta Ruca Restrepia striata Rolfe, Christenson 1018 CONN, ex ort. Rudolfiella aurantiaca (Lindl.) Hoehne, Christenson 386 CONN, Brazil Volume 77, Number 4 1990 Zavada Pollen Wall Ultrastructure of Orchid Pollinia 801 Sacoila lanceolata (Aubl.) Garay, Christenson 1013 ONN, Florida Schoenorchis fragrans (Par. & Reichb. f.) Seid. & Smit., Christenson 1019 C ONN, India ) Schltr., Christenson 1021 CONN, Taiwan - Tricophilia marginata Henfr., Christenson 380 CONN, Guatemala Zygostates alleniana Kraenzl., Christenson 1020 CONN, Brazil THE CORYPHOID PALMS: PATTERNS OF VARIATION AND EVOLUTION! John Dransfield,? I. К. Ferguson,? and Natalie W. UAL ABSTRACT Within members of the palm subfamily Coryphoideae are to be found a larger number of unspecialized character Pa l mae; these in clude primitive leaf forms, flowers, and pollen ast, shows specialization. о id subfamilies and with putative early monocotyledonous pollen from the Of the three cupy an intermediate position; and arly Cretace such as the great variation in the position of the splits and the ola are bl and possible trends of leaf evolution proposed. The position of the palms within the monocotyledons is considered. A new classification of the palms has recently been completed. An outline of the classification was published by Dransfield € Uhl (1986) in order to validate the names of new suprageneric categories. The basis for the classification is discussed in detail in Genera Palmarum: A Classification of Palms Based on the Work of Н. E. Moore Jr. (Uhl & Dransfield, 1987). In the new treatment, the palms are divided into six subfamilies: Coryphoideae, Ca- lamoideae, Nypoideae, Ceroxyloideae, Arecoideae, and Phytelephantoideae. The most important char- acters used in the separation of the subfamilies are the nature of the leaf (whether palmate, pinnate, or bipinnate, and whether induplicate or redupli- cate), the inflorescence and its bracts, the arrange- ment of flowers on the rachillae, floral structure, and the nature of the fruit. The development of the new classification was greatly influenced by a detailed survey of characters (Moore & Uhl, 1982), where the authors have summarized the major trends of evolution in the palms and suggested directions of change in character states. The greatest concentration of character states believed to be unspecialized (Moore & Uhl, 1982) is found in the Coryphoideae. These include the least specialized inflorescences, the least complex flower groupings, and the only apocarpic palms apart from Vypa (Nypoideae). Moore & Uhl (1973) identified the least specialized extant palms as the genera of the apocarpic Trithrinax alliance (Moore, 1973), those genera making up subtribe Thrinaci- пае (Coryphoideae: Corypheae) (Dransfield & Uhl, 1986). Moore & Uhl (1973) pointed out that palms retain many characters considered primitive for the monocotyledons as a whole, but emphasize that they do not suggest linear derivation of all mono- cotyledons from palms. The present paper discusses two aspects of the evolution of palms, pollen, and the leaf, by refer- ence to their manifestation in subfamily Cory- phoideae. As circumscribed by Dransfield € Uhl (1986), the Coryphoideae are distinguished by the following characters. The leaves are palmate or costapal- mate, rarely entire, induplicate, rarely reduplicate (then with the flowers apocarpous), or mixed in- duplicate-reduplicate, or pinnate (then the leaflets with entire tips). The flowers may be solitary or clustered; they are never arranged in triads of one central pistillate flower and two lateral staminate flowers. The combinations of these characters allow separation from other subfamilies, but their dif- fuseness and exceptions indicate the great vari- ability in the subfamily, which includes about 400 ieee in 39 genera arranged in three tribes: Cor- pheae, Phoeniceae, and Borasseae. The least spe- са tribe, Corypheae (31 genera), is distin- guished by the palmate leaf; by genera being ' We are grateful to Miss Caroline Batchelor (Coventry occ Polytechnic) с сап ув out much of the preparation and electron up the pollen plates, and we ? Herbarium, Roy n microscopy of the pollen of the Corypheae a (Sandwich Course) мн үр. at the Royal Botanic Gardens, Kew. xs also bbs her ше help in oeniceae dur ear undergraduate making ank Andrew McRobb of the Photographic Unit di for printing the micrographs. nic бис Kew, Richmond, Surrey, TW9 3AE, U 1 Bota * L. H. Bailey ra Cornell University, Ithaca, New York 14853, U.S. E ANN. MISSOURI Bor. GARD. 77: 802-815. 1990. Volume 77, Number 4 1990 Dransfield et al. 803 Variation and Evolution in Coryphoid Palms hermaphrodite, polygamous, or very rarely strictly dioecious; by the not or only slightly dimorphic flowers; and by the rachillae lacking pits. Tribe Phoeniceae (1 genus), intermediate in specializa- tion, is distinguished by apocarpic flowers and a pinnate leaf with basal leaflets modified as spines. Tribe Borasseae (7 genera), the most specialized of the coryphoid tribes, is distinguished by being strictly dioecious, by the staminate and sometimes also the pistillate flowers borne in deep pits on the rachillae, and by usually strongly dimorphic flow- ers. The Corypheae are divided into four subtribes on the basis of the degree of fusion of the carpels. Subtribe Thrinacinae contains 14 apocarpous gen- era. Subtribe Livistoninae, with twelve genera, has gynoecia of three carpels, free at the base but joined throughout their styles. In subtribe Cory- phinae, with four genera, the carpels are basally fused with free or connate styles but often with separate stylar canals. In subtribe Sabalinae, the single genus Sabal has carpels joined throughout, the stylar region with a single stylar canal. POLLEN OF CORYPHOIDEAE Although the pollen morphology of the Palmae has been surveyed by various workers over the last 20 years (Punt & Wessels Boer, 1966a, b; Thani- kaimoni, 1970a, b; Sowunmi, 1972; Kedves, 1981), virtually no electron microscopy of the pol- len has been attempted. In fact, the paucity of published electron micrographs of palm pollen is remarkable and commented on by Zavada (1983). Recently the present authors have started a pro- gram at Kew to study palm pollen using light and electron microscopy, and some of the results of these studies have been published (Ferguson et al., 1983, 1987; Frederiksen et al., 1985; Ferguson, 1986; Mendis et al., 1987; Harley, 1989). The pollen morphology of the Coryphoideae has been studied quite extensively with light micros- copy (see Thanikaimoni, 1970a; Sowunmi, 1972). Very recently the pollen of tribe Borasseae has been studied in detail with electron microscopy (see 1987). The pollen grains of tribes Corypheae and Ferguson et al., Phoeniceae are generally monosulcate, elliptic to subcircular in polar view, small to medium in size with L = 20-52 um, 1 = 18-45 um, and h = 12- 34 um (the terminology L, 1, h follows Thanikai- moni, 1970a). The apertures are elliptic or rarely subcircular, and very rarely trichotomosulcate, more or less equal in length to the longest axis, and covered with a very thin smooth membrane. The genus Chamaerops has pontoperculate ap- ertures. The tectum is reticulate perforate, or fove- um thick; some species of Pritchardia have thicker walls up to 3 um, while Colpothrinax has pollen with a very thick exine of 3-5 um. In Corypheae and Phoeni- ceae the exine stratification is simple with a well developed tectum (which is often relatively thick), columellate layer, and foot layer. The columellae olate. The exine is usually 1- may be arranged in single, double, or many rows supporting the tectum. This character seems to vary within pollen of the same species (Thanikai- moni, 1970a) and does not appear to have any taxonomic significance. In the tribe Borasseae, four genera have pollen very similar to that of the two preceding tribes, but very large pollen size and monoporate apertures occur. Also there is a very coarse reticulate tectum and supratectal gemmate ornamentation present. TRIBE CORYPHEAE THRINACINAE. Although there is relatively little variation in the pollen morphology of the genera in subtribe Thrinacinae, a number of small dis- tinctive features are present. The tectum of some genera, including Coccothrinax (Figs. 1, 2) and Maxburretia, is perforate or occasionally foveo- late. Rhapis, Schippia, Thrinax (Figs. 3, 4), Trithrinax, and Zombia have similar but finer or smaller tectal structures. Chamaerops has very remarkable pontoperculate pollen (Figs. 6, 7) and Cryosophila sometimes has pollen with trichoto- mosulcate apertures. In Chelyocarpus and Cry- osophila (Figs. 10, 11) the tectum on the nonaper- tural side is reticulate, while it is perforate on the apertural side or even complete adjacent to the aperture margins. Trachycarpus has pollen with a reticulate or coarsely reticulate tectum with fine granules in the lumina. In this genus also the re- ticulum is finer on the apertural side than on the nonapertural side. Fine granules are present in the lumina of the reticulate tectum of Guihaia and Cryosophila. LIVISTONINAE. The pollen morphology of many of the genera in this subtribe is again of a very generalized monocotyledonous type, and there is a uniformly perforate tectum. This type of pollen is found in Acoelorraphe, Pritchardiopsis, Serenoa, Brahea, and Johannesteijsmannia (Figs. 15—18). There is a range of variation from a finely perforate tectum to reticulate with small granules in the lumina in Livistona; Licuala has similar but even greater variation in tectal structure, ranging from finely and sparsely perforate in L. glabra Griff., 804 Annals of the Missouri Botanical Garden for example, to reticulate with isodiametric lumina having dense coarse granules on the nonapertural side in L. sp. aff. peltata Roxb. (Figs. 8, 9) and with often rather rugulate-perforate aperture mar- gins. Pritchardia (Figs. 21, 26) has a sparsely or densely perforate tectum but larger pollen (L = 5-50 um) than is usual in the subtribe, and the exine is thicker (2-3 um). Pollen of Washingtonia shows a range in tectal perforation with some ten- dency for smaller lumina on the apertural side and along margins of the aperture in comparison with those on the nonapertural side. The ornamentation of the tectum is somewhat striate-rugulate in Cop- ernicia with the muri in more or less parallel rows. Colpothrinax has the most distinctive pollen in the subtribe, with differentiation between the tectum of the nonapertural and apertural sides (Figs. 19, 20). Fine granules are present in the lumina. The exine is very thick on the nonapertural side (3—5 um), where the tectum is thick (Fig. 25). CORYPHINAE. The pollen of Corypha somewhat resembles that of Colpothrinax but the exine is thinner (2-3 um). Nannorrhops (Figs. 30, 31) and Chuniophoenix (Figs. 34, 36) have pollen similar to that throughout the subfamily. There are very small differences in tectal structure between the species of the two genera. Kerriodoxa has the most distinctive pollen or- namentation in the subfamily. The muri are dis- continuous and form a loosely arranged reticulate pattern on the nonapertural side (Figs. 28, Supratectal granular processes occur on the muri (Figs. 28, 29, 32, 33), and there are fine dense granules in the lumina. ae AE. The pollen of the genus Sabal (Figs. -46) is of the same general type common ш the subfamily. The size is generally larger than average (L = ca. 40 um) and the exine thicker (2-3 um). TRIBE PHOENICEAE The pollen morphology is generally similar to that described in the Corypheae (Figs. 37-42). The tectum is usually distinct on the margins of the aperture from that on the nonapertural side (Figs. 38, 39) — ,2. oo argentata (Jacq.) L. Н. Bailey (Small & Carter s.n.).—1. Whole pollen grain y finely granular tectum, TEMG URES 1-9. showing apertural E SEMG (scanning electron micrograph) x2, side, SEMG x8,00 ] ] side, SEMG ,90 4 oe ral exine section showi о aP x 20,000.— 5. x 20,0 7. Chamaerops humilis L.—6. ( SEMG E 500. — 7. (Brum show apertural thinnings on either side of operc . Nonapertural ipn e TEMG x15,000. in lumina, SEMG x ulum, T FIGURES iin x ot. Gar 11. Whole me grain showing ape Note th exine section, T x (tectum) on nonapertural side, SEMG x 8,000. E. Moore (Dransfield 916). — 16. Whole pollen e difference in exine surface between the apertural and a и sides. TEMG x 2,300. — ,000. — 14. е папа (Kunth) лл aa field 8 non whole pollen grain: showing i a SEMG x3,3 grain exine sit ee aperture at top, TEMG Coccothrinax ique Неба Wright 3220). Nonapertural exine section, TEMG oris mitt & Ernst 5936) hole ese grain exine section with aperture at the top; arrowheads 2,600 8, 9. Licuala sp. aff. peltata (Kerr 11726).— — 9, Exine surface (tectum) on nonapertural side, showing granules 10-13. Cryosophila warscewiczii (Н. A. Wendl.) Н. Н. Bartl. (cultivated at Herrenhausen Whole pollen grain showing nonapertural side with coarsely perforate tectum, SEMG x1,550.— ertural side with a finely perforate tectum in comparison with 10, SEMG x 2,500.— 12. Whole pollen grain exine section with aperture at top covered by ery thin aperture vendo (arrowhead). 3. Nonapertural Bl. ex Salom. (Langlassé E exine surface . J. perakensis J. Dransf. (Drans- 16-18. J. altifrons (Reichb. f. & Zoll.) H. x 2,000.— 17. fi d exine section, TEMG x 20,000. — 18. Exine surface (tectum) on nonapertural side, SEMG x8,000. FIGURES 19-27. grain showing E side, SEM ,650.— 25. Whole pollen grain exine section Sis "Барси at to 19, 20, 25. үш wrightii Griseb. & Н. A. Wendl. (Wright 3964). — 19. Whole pollen 20. Whole pollen grain sowing apertural side, SEMG х 1,650. — between apert ural sides (arrowheads С "TEMG x grain exine E Mes aperture at top; n heads), TEM section, EE put nb. erences in tectu — 27. Wa E. vob (Linden) H. ж Wendl. (Wright s.n.), whole pollen grain exine ор, . Pritchardia minor Becc. (Cranwell et al. 3103), whole pollen grain m between apertural and nonapertural sides (arrow- Volume 77, Number 4 1990 Dransfield et al. Variation and Evolution in Coryphoid Palms 805 JM ig д Pe a, at P de p s А 9 4 т) - + araga е p 806 Annals of the Missouri Botanical Garden 807 Dransfield et al. Volume 77, Number 4 1990 Variation and Evolution in Coryphoid Palms 808 Annals of the Missouri Botanical Garden k 33 36 FIGURES 28-36. 28, 29, 32, 33. Kerriodoxa elegans J. Dransf. (Bhoonab s.n.).—28. Whole pollen grain showing nonapertural side, SEMG х 2,400. — 29. Exine surface on aperture margin, SEMG x8,000.— 32. Whole pollen grain section with aperture at top, TEMG x 2,000. — 33. Nonapertural exine section, granularlike supratectal processes indicated with arrowhead, TEMG х 20,000. — 30. Nannorrhops ritchiana (Griff.) Aitch. (cultivated in Volume 77, Number 4 1990 Dransfield et al. 809 Variation and Evolution in Coryphoid Palms TRIBE BORASSEAE There is more variation in the pollen morphology of the genera of this tribe than in the two preceding (Ferguson et al., 1987). The genera Latania, Lo- doicea, Medemia, and Bismarckia have pollen of the general type found throughout the subfamily. Borassus and Hyphaene have pollen with supra- tectal gemmate processes. Borassodendron has very large circular pollen (L = ca. 73 um, | = ca. 2 um) with very thick walls (4-6 um). One species, B. machadonis (Ridley) Becc., has porate apertures Within the subfamily there are few pronounced trends in the pollen morphology, yet tribe Boras- seae tends to have more specialized pollen with larger pollen size, porate apertures, and supratectal structures. Within the other two tribes a slight tendency for increase in pollen size occurs, as does an in- crease in reticulum size together with the presence of granules in the lumina. These characters may be interpreted as being indicative of a low level of specialization. Likewise, the increase in thickness of the pollen walls and the differences in tectal ornamentation between the apertural and nonaper- tural sides are perhaps specialized. COMPARISON OF CORYPHOIDEAE POLLEN WITH THAT OF OTHER SUBFAMILIES The pollen morphology of Coryphoideae is fairly uniform, and only Kerriodoxa and some genera in tribe Borasseae do not conform to a very gener- alized type. The significance of the pontoperculate pollen of Chamaerops (which is paralleled in the small genus /riartella in subfamily Arecoideae) and the distinctive ornamentation in the pollen of Ker- riodoxa is unclea Uniformity of den morphology in the Cory- phoideae contrasts markedly with that found in the other subfamilies. The monosulcate pollen type with a perforate or reticulate tectum occurs throughout the entire family (Harley, 1990). However, there is a huge range of variation in ornamentation, apertures, and in exine stratification. For example, in the Calamoideae there is intectate gemmate and spinose pollen in Salacca, Daemonorops, and Kor- thalsia as well as tectate psilate or sparsely per- forate tectate pollen in the two former genera (Fer- guson, 1986). Extended sulcate, dicolpate, and diporate pollen occurs in Calamus, Salacca, Dae- monorops, and Korthalsia (Thanikaimoni, 1970a; Frederiksen et al., 1985; Ferguson, 1986). Su- pratectal spines and supratectal gemmae occur in Retispatha and Calamus. Mauritia, Mauritiella, and Lepidocaryum have spinose pollen, the spines being characteristically sunk into the foot layer (Sowunmi, 1972; Ferguson, 1986). Nypa has tectate spinose pollen with an ex- tended sulcus (Thanikaimoni, 1970a; Ferguson, 1986). In the Ceroxyloideae, Louvelia has monoporate pollen (Ferguson et al., 1988), whereas Ravenea has tectate spinose pollen (Ferguson, 1986). Tri- chotomosulcate pollen occurs in Chamaedorea, al- though resembling bond the generalized type found in the Coryphoidea The pollen " the Arecoideae is probably the most varied, although there are many genera with the generalized monosulcate type (Harley, 1990); a very great range of types occurs within the genus Pinanga alone (Ferguson et al., 1983). Areca has triporate, porate, extended sulcate, and simple monosulcate pollen (Ferguson & Dransfield, un- published). A mixed granular and columellar in- terstitium occurs in the monoporate Areca caliso Becc. (Ferguson & Dransfield, unpublished). Sclerosperma also has triporate pollen (Thanikai- moni, 1970a). Trichotomosulcate grains occur in, for example, Pinanga, Elaeis, Bactris, Astrocar- yum, and Acrocomia (Thanikaimoni, 1970a; So- wunmi, 1972; Ferguson, 1986; Ferguson & Dransfield, unpublished). Intectate gemmate pollen is present in Arenga, Caryota, and Dictyocaryum (Ferguson, 1986). Socratea, Catoblastus, and Wettinia have large spines interspersed with dense granular spinules (Ferguson, 1986). many species in the genus have pollen COMPARISONS WITH THE FossiL RECORD Comparison of the pollen morphology of the Coryphoideae with putative fossil monocotyledon- Indi a —Saharanpur), whole pollen grain showing пире шо es pis x 2,000.—31. ag ae umbraculifera L. 4, 36. Chu 2,200. 32, 33. See above. 3 niophoenix — (Thwaites 2336), hainanensis Burret (Whitmore 3152). below. — 36. Nonapertural exine section, TEMG x30,000.— —34. Whole ойе grain ied apertural side, SEMC x 2,200.— 35. 3 See 5. Nannorrops ritchiana (Griff.) Aitch. (Radcliffe- Smith 547 1), nonapertural exine section, TEMG x 20,000. Annals of the Missouri Botanical Garden FIGURES 37-46. 37, 38. Phoenix sp. (Kerr 6872). — 37. Whole pollen grain showing nonapertural side, SEMG x 3,000. — 38. Exine surface on aperture margin, SEMG х8,000. 39-41. Phoenix paludosa Roxb. (Schmidt 362).—39. Whole pollen grain section with aperture at top, some contents remaining after acetolysis, TEMG um 700.— 40. Nonapertural exine section, TEMG x36,000.—41. Exine surface (tect x3, SEMG x8,000. ) on nonapertural side, 42. Phoenix dactylifera L. (Guiaro 2314), exine surface (tectum) on nonapertural side, SEMG Volume 77, Number 4 1990 Dransfield et al. 811 Variation and Evolution in Coryphoid Palms ous pollen suggests that there are very close sim- ilarities between those of the Liliacidites—Reti- monocolpites type pollen and the pollen of subfamily Coryphoideae. Pollen types, some remarkably sim- ilar to those of Coryphoideae, are described by Doyle (1973) and Walker & Walker (1985, 1986) from the Lower Cretaceous Potomac Group of North America. The more coarsely reticulate Clavati- pollenites grains are not so readily matched among the Coryphoideae. The differences in tectum be- tween the apertural and nonapertural sides in ex- tant palm pollen might prove to be of value in relating the pollen of Coryphoideae to fossil pollen types. It is perhaps noteworthy that the trichoto- mosulcate apertures occurring in early fossil de- posits are relatively rare in Coryphoideae but wide- spread in other subfamilies of the palms. Likewise, the very coarsely reticulate ornamentation like that in early fossils (Retimonocolpites) occurs in the Arecoideae (Venga and Gronophyllum, for ex- ample) but is absent from the Coryphoideae. There are also faint similarities between the pollen of Kerriodoxa and some Clavatipollenites types with ridged or granular structures on the muri. It can be postulated that the pollen morphology of the Coryphoideae is unspecialized. This view is supported by other morphological characters and from comparison with the fossil record where sim- ilar types have been shown to occur in the Lower Cretaceous (Walker & Walker, 1985, 1986). The pollen morphology of the other subfamilies is much more variable and can be regarded as being more specialized, and appears much later in the fossil record, extending from the Maestrichtian, but is much more frequently found in Eocene and Mio- cene deposits (Muller, 1980, 1981). Identification of Coryphoideae pollen in the fossil record seems likely to be very difficult because it is so very similar to the relatively commonly oc- curring and widespread monocotyledonous gener- alized type. However, we stress that, although it may be impossible to equate such early pollen types with palms positively, it may be equally impossible to rule palms out. THE LEAF OF CORYPHOIDEAE The plicate and usually split leaf is the most distinctive organ of the palm family and more than any other structure links palms together. Such plicate leaves are scattered in other monocotyle- donous families, but outside of the palms splitting of the plicate leaf is found only in some members of the Cyclanthaceae and in Curculigo seychel- lensis Bojer (Hypoxidaceae). The structure of the leaves of Cyclanthaceae has been the subject of recent investigations by Wilder (1976, 1981). In Curculigo seychellensis, the leaf blade is borne on a spiny petiole and is deeply bifid down the midline at maturity, thus presenting a remarkably palmlike appearance (Dransfield, pers. obs.); this is the only member of the genus to possess a split leaf. Palm leaves, as is well known, may be palmate, costapalmate, pinnate, bipinnate, or entire and pin- nately or palmately ribbed. The blade is always plicate in bud and the folds are usually prominent in the mature leaf, although occasionally very in- distinct, as in some species of Chamaedorea. The origin and development of the plications, for long a mystery and the source of much speculation, has recently been elucidated by Kaplan et al. (1982). In some previous classifications of the family (e.g., Satake, 1962; Saakov, 1954), the nature of the splitting, whether along the adaxial or abaxial folds, has been considered to be of fundamental importance. In the most recent classification (Dransfield & Uhl, 1986; Uhl & Dransfield, 1987), although the position of the splits is largely con- sistent with division of the family based on other characters, there are some exceptions of great in- terest, which are almost all to be found within the Coryphoideae. Indeed subfamily Coryphoideae has a wider ага т MAS form = any other of the d & Uhl, 1986). Yet this great range has pole not been fully appre- ciated in the past, and contrasts markedly with the six palm general uniformity in pollen morphology. Most members of Coryphoideae have palmate or costapalmate leaves with blades partially divided along the adaxial folds into induplicate segments. Costapalmate leaves differ from strictly palmate leaves in the presence of a costa, an extension of the petiole into the blade, representing the true midrib of the whole leaf. Entire undivided leaves are quite common in juvenile stages of many cor- yphoid genera but at maturity occur only in Jo- hannesteijsmannia and a few species of Licuala, e.g., L. grandis Н. A. Wendl. and L. orbicularis Becc. Pinnate leaves are found only in Phoenix, the only genus in tribe Phoeniceae. P 000.—43. Sabal mexicana Mart. (Palmer 193), whole pollen grain on nonapertural side, SEMG x1,650.— 44. Sabal yapa Wright ex Becc. (Gentle 1156), exine surface (tectum) on nonapertural side, SEMG x8,000. 45, p is е ранено Lodd. ex Schult. (Fredholm 5390). — 45. Whole pollen grain on nonapertural side, SEMG 46. Nonapertural exine section, TEMG x15,000. 812 Annals of the Missouri Botanical Garden Until recently the division of the leaf blade in Coryphoideae was considered to be consistent] into induplicate segments. It is now known that several types of splits occur in the subfamily. We can distinguish splits along the adaxial folds, along the abaxial folds, and between-fold splits. Splits may be shallow or deep, or may even reach the insertion of the blade on the petiole or costa. Preliminary investigations of the development of the leaves suggest that the various types of splitting occur at different times during development and that the types of splitting are not equivalent (Dransfield, 1 . The diversity of leaf form within the subfamily is caused by combinations of the several types of split. Elsewhere in the family, there is diversity of leaf form but, except in Arecoideae (where adaxial splits occur), the range of splitting mechanisms appears not to occur. Still much work needs to be done on the different forms of pinnate leaves. The simplest form of blade in the subfamily is exemplified by Chamaerops humilis L. and many other genera and species; in this type the blade is regularly divided along the adaxial ribs to about half the blade radius into single-fold, induplicate segments, which in turn are divided shallowly along the abaxial rib. Splitting of the blade occurs rela- tively late in the development of the leaf and is completed by expansion of the sword leaf from the apical bud. Splitting seems to be intimately asso- ciated with the mechanical forces of the expanding leaf. The leaf of Trachycarpus fortunei (W. J. Hook.) H. A. Wendl. is only slightly different: the divisions of the blade are unequal and the resulting segments are of differing lengths. Most costapal- mate leaves have blades divided rather regularly into single fold segments of similar length, but in some taxa, such as all members of the genus Phol- idocarpus and Livistona saribus (Lour.) Merr. ex Chev., the blade is divided by a few very deep adaxial splits into many fold segments, which are in turn divided along adaxial folds by shorter splits into single fold segments. The timing of the deeper splits in relation to the shallow splits has not been investigated. Many members of tribes Corypheae and Boras- seae have rather strongly costapalmate leaves. In some costapalmate leaves, such as those of Liv- istona decipiens Becc. and L. loriphylla Becc., the splits in the distal part of the leaf may nearly reach the costa; the distal part of the leaf resembles the leaf of Phoenix and suggests perhaps how the leaf of Phoenix may have evolved. However, leaf development in Phoenix is strikingly different (see below) Splitting along the abaxial folds occurs in a few genera. In Guihaia the blade is divided to about V$—34 the radius into neat reduplicate segments (Dransfield et al., 1985). This clearly reduplicate palmate-leaved genus is closely related by inflo- rescence and flower structure to the induplicate palmate-leaved Maxburretia. Guihaia has the only leaf of its type in the subfamily. Elsewhere in the palms, strictly reduplicate palmate leaves are known only in three calamoid genera, Mauritia, Mauri- tiella, and Lepidocaryum. In other coryphoid gen- era displaying abaxial splits in the leaves, the abax- ial splits appear to be superimposed on a basically induplicate leaf. In most species of Cryosophila, Chelyocarpus, and Itaya the leaf blade is divided into induplicate segments, but the whole blade is bisected right through to the insertion by a deep split along the central abaxial fold, i.e., along the true midrib of the leaf. The timing of this split has not been examined, but in Sabal, some species of which also display this central split, the split occurs early in the development of the leaf, several plas- tochrons before the adaxial splits occur (Uhl, un- published). This deep central split is paralleled in many pinnate-leaved palms, in some members of the Cyclanthaceae, and in Curculigo seychellen- sis. In Licuala, apart from the species with entire leaves, the blade is regularly divided right to the insertion along the abaxial ribs. The segments thus produced are usually broadly wedge-shaped and composed of several folds. In most species the individual segments are lobed along the apical mar- gin by short splits of varying depth, longer splits occurring on the adaxial folds, shorter on the abax- ial folds. This short lobing thus appears to corre- spond with the induplicate segments in other cor- yphoid palm leaves. Furthermore, the deep abaxial splits that divide the leaf into segments occur very early in the development of the leaf, usually by the third plastochron (Dransfield, 1970). This sug- gests that the splitting mechanism may be different, and is superimposed upon a basically induplicately split leaf. In one species, L. bidentata Becc., the segments are sometimes composed of only one fold, and the blade superficially resembles that of the three reduplicately palmate calamoid genera men- tioned above. We may regard the entire leaves of L. orbicularis and L. grandis, which, although not being deeply split, still retain the shallow marginal induplicate lobing, as being intermediate stages in the evolution of the more typical Licuala leaf. The striking entire leaf of Johannesteijsmannia, once suggested to represent a primitive leaf form (Cor- ner, 1966), develops in a manner similar to that Volume 77, Number 4 1990 Dransfield et al. Variation and Evolution in Coryphoid Palms of the entire-leaved species of Licuala and pos- sesses the shallow induplicate lobing of the apical margins (Dransfield, 1970). Another unusual type of splitting occurs in Rhapis and Rhapidophyllum. In these two gen- era, the major leaf splits occur between the folds rather than along the folds. The folds do not reach the insertion of the blade, although they may very nearly do so. These inter-fold splits usually divide the blade into segments composed of several folds; the apical margins of the segments are shallowly induplicately lobed as in Licuala. The inter-fold splits occur much earlier in the development of the leaf than the shallow adaxial splits of the margins. Phoenix is the only pinnate-leaved genus in the subfamily. The leaflets are induplicate and the leaf is exceptional in its development. A broad expanse of tissue, called the **haut," on the adaxial surface of the developing leaf develops from interdigitation and proliferation of epidermal cells at the adaxial folds (Periasamy, 1967). As the sword leaf expands the haut disintegrates. There is no known parallel to the haut elsewhere in the family and its mor- phology and development deserve further investi- gation. The adaxial splits in the leaf of Phoenix divide the blade seemingly to the rachis, although close examination shows a thin band of lamina tissue along the rachis connecting the bases of the leaflets. Some generalizations can be made about the dif- ferent types of splits in the coryphoid leaf. Adaxial splits never reach the insertion of the blade and usually seem to occur relatively late in leaf devel- opment. We regard this type of splitting mecha- nism to be the simplest, and it seems to be inti- mately related to the mechanical forces imposed on the expanding leaf. Almost always, abaxial splits extend to the insertion on the costa or petiole. An exception is the leaf of Guihaia, in which the abaxial splits only reach about 142-% of the radius. Where the development of the abaxial splits has been investigated, they occur much earlier in leaf development than the adaxial splits, long before the mechanical forces of leaf expansion. Adaxial splits appear to be the generalized state in the Coryphoideae; abaxial splits, such as those in Licu- ala, appear to be a superimposed secondary de- velopment, a specialization. However, the deep, central, abaxial split found in /taya, Chelyocarpus, and some species of Cryosophila, paralleled in Cyclanthaceae and Curculigo seychellensis, may be the simplest way in which a plicate leaf can split, and could be ancestral in the family. nlike Corner (1966), who suggested that the complex bipinnate leaf of Caryota represented the primitive leaf form in the family, we regard the much simpler, simply plicate leaf, divided by in- complete splits through mechanical forces of the expanding blade, to be least specialized. The fore- runners of the palm leaf, we propose, were undi- vided and plicate. From this relatively simple mod- el, the complex leaves of modern palms have evolved by elaboration of the midrib of the leaf into a costa or rachis, the development of different splitting mechanisms, the development of secondary pli- cations, and the development of complex patterns of necrosis at the margins to give the highly char- acteristic praemorse margins of several calamoid, ceroxyloid, and arecoid palms. The simplest palm eaves are, we believe, to be found among the Coryphoideae, whether palmate or costapalmate. The earliest fossil leaves that can definitely be assigned to the palms are palmate and costapalmate forms (Daghlian, 1981). However, this may rep- resent differential preservation or the greater ease of identifying fragments of palmate leaves than pinnate leaves (Read & Hickey, 1972 A feature of the palmate leaf in need of further study is the hastula. Hastulae are triangular flanges occurring at the base of the blade, usually on the adaxial surface only, occasionally also on the abax- ial surface. They are present in all coryphoid pal- mate leaves except in Chuniophoenix and Nan- norrhops (Corypheae) and Lodoicea and Medemia (Borasseae). In Johannesteijsmannia the devel- oping leaf bears a well defined hastula, which dis- integrates just before the sword leaf emerges, leav- ing almost no vestige in the mature leaf. The reduplicate palmate leaves of the calamoid palms Mauritia, Mauritiella, and Lepidocaryum do not display clear hastulae. Hastulae are usually small, rarely more than 1 cm long, but in some Cuban members of Copernicia (Corypheae) they are spiny margined and greatly enlarged, sometimes over 50 ^. rigida Britt. & The adaptive significance of these structures is not known. They could be of mechanical signifi- cance or perhaps direct rainwater away from the apical bud. There is no doubt that the hastula directs rainwater away from the petiole and hence the palm apex (Dransfield, pers. obs. on Livistona rotundifolia (Lam.) Mart. in the wild), and Medem- ia and Nannorrhops, lacking hastulae, are plants of low-rainfall areas, but so are the species of Copernicia with the largest hastulae. A. K. Irvine (pers. comm.) has recently drawn our attention to the presence of a small flange of tissue on the adaxial surface of the leaf rachis of the pinnate-leaved Oraniopsis appendiculata (F. 814 Annals of the Missouri Botanical Garden М. Bailey) J. Dransf., A. К. Irv. & N. Uhl (Ceroxy- loideae: Ceroxyleae); it bears some resemblance to a hastula (Uhl & Dransfield, 1987). Subsequently we have found similar flanges in members of the closely related Ceroxylon and the much more dis- tantly related pinnate-leaved Polyandrococos pec- tinata Barb. r. and Cocos nucifera L. (Are- coideae: Cocoeae). It is tempting to suppose that this flange is homologous with a hastula, but we may only speculate on its nature until develop- mental work can be carried out. DISCUSSION Within the palm family can be found many of the features regarded as apomorphic for and help- ing to define the monocotyledons: sympodial habit, leaves with sheathing bases and parallel venation, and floral parts composed of three. Some features of palms are interpreted as plesiomorphic in mono- cotyledons; for example, in the apocarpous palms of the Coryphoideae (Thrinacinae and Phoeniceae) the carpels are conduplicate and follicular, and have open ventral sutures, features regarded as plesiomorphic in the Angiosperms as a whole. Pol- len of the least specialized modern group of palms is, for the most part, as demonstrated above, of a generalized type common in other monocotyledon- ous families and considered to be plesiomorphic. The most distinctive feature of the palms, the leaf, seems clearly to have been derived from an un- divided but plicate form, a leaf type found scattered among the monocotyledons. The plicate leaf divid- ed by clearly defined splits is perhaps the only reliable apomorphy for the whole family. In the past, the palms have usually been asso- ciated with the Pandanaceae and Cyclanthaceae because of superficial similarities of habit and leaf structure respectively. Yet little, if anything, links the families in inflorescence or flower structure. Recently, in fact, the distinctness of the three fam- ilies has been recognized. Thorne (1983) and Dahl- gren et al. (1985) placed palms in a separate evo- lutionary line. With no synapomorphies clearly established to link the palms with other families, the palms remain isolated taxonomically. Perhaps current cladistic studies will clarify this (Dransfield & Uhl, in prep.). Some authors, such as Cronquist (1981) and Muller (1984), have regarded palms as a secondary radiation from a monocotyledonous stock. Muller (1984) indeed stated that the palms did not evolve until the late Cretaceous, citing as support the appearance of supposed /Vypa pollen in the Maes- trichtian. Although the present authors do not dis- pute that the earliest definite fossils do not occur until the late Cretaceous, the fact that they are recognizable as palms is due to specialized features. As demonstrated here, many of the extant palms regarded as being least specialized have pollen grains of a plesiomorphic nature, indistinguishable as yet from many other monocotyledons and in fact very similar to the early Cretaceous grains illustrated by Walker & Walker (1985, 1986). It is suggested that palms probably arose before the late Creta- ceous. Palms display many evolutionary trends; while possessing characters regarded as plesiomorphic in the monocotyledons (and this seems not to be widely appreciated), they also display many specializa- tions. It appears to the authors that palms retain characters of a very early monocotyledonous stock from which more specialized palms and, perhaps, some other monocotyledonous families may have evolved. LITERATURE CITED Corner, E. J. H . The Natural History of Palms. Weidenfeld 8 Nicolson, London CRONQUIST, А. An Integrated System of Classi- fication of Flowering Plants. Columbia Univ. Press, New Yor DAGHLIAN, C. Р. A review of the fossil record of monocotyledons. Bot. Rev. (Lancaster) 47: 517- 555. DAHLGREN, R. M. T., Н. T. CLIFFORD & P. F. Yeo. 1985. The Families of Monocotyledons. Springer- Verlag, Berlin. DovLE, J. A. 1973. Fossil evidence on early эңе of the monocotyledona. Quart. Rev. Biol. 48: 13. 4 DRANSFIELD, J. 1970. Studies in the Malayan palms Eugeissona and Johannesteijsmannia. Ph.D. Dis- sertation. Univ. of Cambridge, Cambridge & N. 1986. An outline of a classi- fication of palms. Principes 30: 3-11. E . Wer. 1985. Guihaia, a new coryphoid s genus ‘fom China and Vietnam. Prin- 3- s 29: ни І. К. 1086. Observations on the variation ш pollen morphology of Palmae and its significance. Canad. J. Bot. 64: 3079-3090. , J. DRANSFIELD & I. FLAWN. A review of the pollen morphology and systematics of the genera Ravenea and Louvelia (Ceroxyleae: Ceroxy- loideae: Palmae). J. Palynology (Lucknow) 23-24: "Ta. HavarD & J. DRANsFIELD. 1987. The pollen morphology of the 2e у (Palmae: Coryphoideae). Kew Bull. 42: , J. DRANSFIELD, F. C. Pac “л С. Томка almae: Arecoideae). FREDERIKSEN, N. O., QV Wiccins, I. K. FERGU RGUSON, J. DRANSFIELD & С. M. AGER. 1985. Distribution, Volume 77, Number 4 90 Dransfield et al. Variation and Evolution in Coryphoid Palms palaeoecology, d and botanical affin- ity of the Eocene pollen genus Diporoconia, n. gen. Palynology (Dallas) 9: 37 60. HarLeY, M. M. 19 Occurrence of simple, tectate, . 1982. The айный of plication inception in palm leaves; problem and = morphology. Canad. J. Bot. 60: 2999- Kepves, M. 1981. " Morphological investigation of re- cent Palmae pollen grains. Acta Bot. Acad. Sci. Hung. 26: 339-373. MENDIS, N. M., I. К. FERGUSON & J. DRANSFIELD. 1987. The pollen morphology of the subtribe Oncosper- B matinae (Palmae: Arecoideae: Areceae). Kew Bull. 42: 47-63 MoonE, H. E. 1973. The major groups of | and their distribution. Gentes Herb. 11: 27- N. Unt. 1973 he c РРА Weir evolution and comparative biology. VI. Palms Quart. = ah Biol. “48: 414- B 19 Major Paci of evolution in palms. Bot. Rev. т 48: 1-69. MULLER, J. 1980. Reflections on fossil palm pollen. Proc. IV Intern. Palynol. Conf. Lucknow (1976- 1977) 1: 568-578. Вга! Sahni Institute of Pa- pu Lucknow, India 1981. Fossil pollen records ы extant angio- hermas. Bot. Rev. (Lancaster) 47: 1984. Significance of fossil e for angio- sperm history. Ann. Missouri Bot. Gard. 71: 419 PERIASAMY, K. 1967. Morphological and Hb cal studies of palms. III. Growth patterns of the lea of Caryota and Phoenix after T initiation of pli cations. соо 16: 0. Punr, W. & J. С. WEsseLs BOER. cal DUM in cocoid Suns Acta Bot. Neerl. A palynologi- 15: 255- = a ————. 1966b. A palynological study in geonomoid sais. Acta Bot. Neerl. 15: 266-275. READ, R. W. & L. J. Hickey. 1972. A revised clas- d o fossil palm and palmlike leaves. Taxon 21: 129- SAAKOV, 5 С а Palms and bees Culture in the SSR. Acad. Sci. U.S.S.R w & Leningrad. SATAKE, T. 1962. A new system of hé саш of Palmae. Hikobia 3: 112-133. жин М. А. 1972. Pollen morphology of the Pal- and its bearing on taxonomy. Rev. Palaeobot. alynol. 13: 1-80. THANIKAIMONI, С. 1970a. Les Palmiers: е еї systematique. Trav. Sect. Sci. Tech. Inst. Franc i pro 11: 70b. Pollen morphology, classification and бата of Palmae. Adansonia ser. : 34 THORNE, R. F. 1983. Proposed new realignments in the angiosperms. : 85-117. UHL, E W.& Genera Palma- m: À Classification of Palms Based on the Work of жын E. Moore, Jr. Allen Press, Lawrence, Kan- Warken, E Ы & А.С̧. WALKER. 1985. ulteasiructure of L and pad Braces of flowering plants. reta pear 521. Bot. Gard. 71: 464-5 & 1986. Ultrastructure of Early etaceous angiosperm pollen and its ev be a] implications. Pp. 203-217 in S. Blackm Ferguson (editors), Pollen and Spores: s and Function. Academic Press, London. WILDER, G. J. 197 Structure and development of leaves in Carludovica palmata (Cyclanthaceae) with reference to other Cyclanthaceae and Palmae. Amer. J. Bot. 63: 1237-1256. 1 Morphology of adult leaves in the С кафа асеае ел л Bot. Gaz. (Crawfords. ville) 142: ZAVADA, monocot pollen and wall structures. Bot. 9 . Comparative morphology of and evolutionary trends in apertures Rev. (Lancaster) 49: 331- POLLINATION AND THE Paul Alan Сох? EVOLUTION OF BREEDING SYSTEMS IN PANDANACEAE! ABSTRACT The Pandanaceae include three genera: Pandanus with 700 species and a large range, Freycinetia with 200 species ч a smaller range, and Sararanga with t two species and a very limited range. Using Cyclanthaceae, Araceae, and А superimposed upon the consensus tree derived from this analysis. Vertebrate pollination and dioecism appear to be ancestral in Pandanaceae, with anemophily in Pandanus and entomophily in Sararanga to be independently derived conditions. Only a few changes in inflorescence morphology were necessary to facilitate these profound changes in pollination syndromes. Dioecism, while perhaps adaptive for vertebrate-pollinated Pandanaceae such as Freycinetia, in Sararanga and may lead to its eventual extinction. The interplay of pollination syndromes and breeding systems appears to have strongly influenced range expansion and speciation in the Pandanaceae. Within the last decade or so, great interest has been directed toward the evolution of breeding systems in seed plants, particularly the evolution of different sexual systems such as monoecism, dioecism, and gynodioecism. Traditionally these systems have been considered to be disparate and distinct, providing the basis for various classifica- tion schemes such as the sexual system described by Linnaeus (Linnaeus, 1754). However, by con- sidering a plant to be a metapopulation of modules (White, 1979), the recognition of nested modular levels in plants permits analysis of the different sexual systems at the modular level. Using this modular approach, all sexual systems can be con- sidered merely as combinations of monomorphic or dimorphic sexual strategies (Cox, 1988) at dif- ferent modular levels. Thus, for example, mon- oecism and dioecism can both be considered asex- ually dimorphic strategies occurring at the level of the shoot and the individual, respectively. Using the techniques of evolutionary game theory (May- nard Smith, 1982) as developed for sex allocation theory (Charnov, 1982) these different sexual strategies can be examined for evolutionary sta- bility in any particular population. By definition, an evolutionary stable strategy (ESS) is one such that if all members of a population adopted it, then no individual with a different strategy could invade the population through the forces of natural selec- tion (Maynard Smith, 1982). The strength of the combination of game theory and a modular approach is that the effects of var- ious ecological factors on the evolutionary stability of different breeding systems can mined quantitatively. For example, vertebrate pollination was found to maintain dioecism as the evolution- arily stable breeding system in Freycinetia reineck- ei in Samoa (Cox, 1982). However, a weakness of this approach, and indeed one shared by all eco- logical approaches to evolution, is that ecological plausibility does not necessarily imply evolutionary causality. The questions of real interest in evolutionary ecology do not merely address the factors that currently maintain a particular trait (such as dioe- cism), but rather seek the factors that favored the evolution of the trait. However, such determina- tions cannot be made effectively by an ecological analysis without due consideration of the phylo- genetic relationships involved. Indeed, ecological ! I thank Greg Anderson, Herbert Baker, Irene Baker, Kamal Bawa, David Cox, Michael Donoghue, Ed Guerrant, Chris Humphries, Bruce Knox, Pauline Ladiges, Harold St. John, Ben Stone, Barry Tomlinson, Merlin Tuttle, Natalie Uhl, and Jim e ety Presidential Young Investigator Award BSR.845 2000 2D ker for useful discus sions and criticisms = ч work at various але This rch in Sci ce, the epartment of Botany and Range Science, Brigham Young University, Provo, Utah 84602, U.S.A. ANN. MISSOURI BOT. GARD. 77: 816-840. 1990. Volume 77, Number 4 1990 Cox 817 Breeding Systems in Pandanaceae FIGURE 1. approaches to evolutionary questions may produce serious errors in interpretation if phylogenies are ignored. For example, an ecological analysis of the child- hood diseases of nineteenth- and twentieth-century European royalty, using the incidence of disease in commoners as a control, would indicate a sig- nificant correlation between hemophilia and access to silver drinking vessels, perhaps suggesting silver to be a causal factor in hemophilia. However, this "silver chalice fallacy" can be avoided by exam- ining the genealogy of the group, which clearly indicates that hemophilia among European royalty is inherited from a single autosomal mutation in Queen Victoria. Cognizance of phylogenetic constraints can also give an indication of characters that are unlikely to have evolved in a particular lineage. Andro- dioecism is very unlikely to have evolved in any of the ancestors of, for example, Scaevola taccada (Goodeniaceae), since phylogenetic analysis shows that all of Scaevola and indeed nearly all of the Goodeniaceae have anthers that dehisce and wither in bud, with pollen being presented by the pen- dulous, indusiate stigma. Thus, it would be very difficult for a male to disperse its pollen in the absence of a morphologically complete gynoecium. pitfalls inherent in plausibility arguments based solely on ecological considerations, it is advanta- geous to determine first the phylogenetic history of the group. Once a phylogenetic hypothesis, in this case represented as a cladogram, is determined with some degree of confidence, various trends of evolution of the breeding system within the group can be analyzed within this systematic context. Note, though, that the resultant analysis should be == fey onsite — Pandanus A Sararanga Geographical distribution of the Pandanaceae. considered only as a poten, since its validity ultimately rests upon the p phy which may change as more ibfocnátion botones avail- able. SYSTEMATIC RELATIONSHIPS IN PANDANACEAE The Pandanaceae, the sole representatives of the Pandanales, are arboreous or scandent mono- cotyledons confined to the Old World tropics and subtropics (Fig. 1). The family consists of three genera. Pandanus is made up of approximately 700 species of trees ranging from the Society Is- lands to west Africa on an east-west axis and from Australia to the foothills of the Himalayas on a north-south axis (St. John, 1965; Stone, 1976b). Freycinetia is a genus of about 180-200 species of lianas ranging from the Society Islands to Sri Lanka on an east-west axis, and from New Zealand to the Himalayas on a north-south axis. Sararanga comprises two arboreous species: 5. philippinensis from the Philippines, of which few recent collec- tions exist; and S. sinuosa, which is found along the edge of the Tethys geosyncline in the Solomons, Admiralties, and North Irian Jaya. The Pandanaceae are very distinctive and dis- junct monocotyledons and likely are a group of great antiquity: Pandanus pollen has been de- scribed from the late Cretaceous and early Tertiary (Jarzen, 1978; Muller, 1981) while at least one megafossil from the early Eocene has been attrib- uted to the family (Nambudiri & Tidwell, 1978; Bande & Awarthi, 1986). Systematic studies of the Pandanaceae have been made by several investigators including Warburg (1900), Martelli (1934), Merrill & Perry (1939), and most recently St. John (1960, and elsewhere), 818 Annals of the Missouri Botanical Garden Stone (1968a, and elsewhere), and Huynh (1980). The revision of Pandanus by St. John (1960, and elsewhere) is now in its 47th part, and the mono- graph of Freycinetia by Stone (1967, and else- where) will soon be completed. However, even though Pandanaceae appear to represent a mono- phyletic group, the relationship of Pandanaceae to other families of monocotyledons remains poorly understood. Conjecture regarding the systematic relationships of the family ranges from the view that Pandanaceae are very primitive monocotyle- dons to the view that they are very advanced. Meeuse (1965, 1966) suggested the Pandanaceae are relict representatives of the fossil order Pent- oxylales, while Engler & Gilg (1924) regarded it as the most primitive monocotyledonous family with affinities to the Typhales (a view shared by Rendle, 1904, and Takhtajan, 1980). However, Stone (1972a) and Cronquist (1968) suggested affinities with Aracaceae and Cyclanthaceae, a view which has been supported recently through phenetic anal- ysis by Dahlgren & Clifford (1982) and in a cla- distic revision by Dahlgren & Rasmussen (1983). However, the precise relationship of Pandanaceae to other families of monocotyledons remains un- determined as even its sister group has yet to be demonstrated convincingly. Relationships among the genera of Pandanaceae are similarly unclear even though these three gen- era are quite distinct; as Stone (1972a) pointed Pandanus species out, there are no “‘freycinetoid”’ "pandanoid" Freycinetia species. Stone and no suggested that Freycinetia is the most ized" genus but “ other two genera than the latter two are from each other” (Stone, 1972a, p. 40). Stone believed Sara- ranga possesses the greatest number of primitive characters. In a similar vein, Dahlgren et al. (1985) ee special- somewhat more distant from the argued that “Sararanga is more ancestral in its floral construction than either of the other genera." Stone’s (1972a, 1982a) view of the relationships among the three genera, with Sararanga and Pan- danus being more closely related to each other than either is to Freycinetia, is supported by my preliminary cladistic analysis of generic relation- ships within the family. Three major difficulties were encountered in this analysis at the generic level. First and foremost among these is lack of a clear sister group to Pandanaceae, although Ara- ceae, Arecaceae, or Cyclanthaceae are all potential outgroups (Harling, 1958; Dahlgren & Clifford, 1982; Dahlgren & Rasmussen, 1983; Dahlgren et al., 1985). Second, the taxonomic status of these potential outgroups is unclear. If, for example, th Cyclanthaceae prove not to be monophyletic, their use as a single outgroup in a cladistic analysis would be invalid. Third, the distinctiveness of the family makes it difficult to find a large number of char- acters that vary at the generic level within the Pandanaceae but remain constant at the familial level in the outgroups. This would not be a problem if in-group analysis had been done within these other families, but such analyses are far beyond the scope of this paper. To overcome these problems, the potential out- groups Araceae, Arecaceae, and Cyclanthaceae were represented by single genera. These gene Rhaphidophora (Araceae), Balaka (Arecaceae), and Asplundia (Cyclanthaceae)—were chosen ra— solely on the basis of my familiarity with them in the field, although in the case of Asplundia my observations were checked against Harling's (1958) monograph of the Cyclanthaceae. Fourteen differ- ent character states were scored for these outgroup genera and the three genera of the Pandanaceae. Using this data matrix (Table 1), the most parsi- monious cladogram for the Pandanaceae was de- termined for each outgroup. All three of these trees, however, were found to have the same general topology (Figs. 2-4), which indicates that Pan- danus and Sararanga are more closely related to each other than either is to Freycinetia. This result is believed to be relatively robust since the general topology remains unchanged even when the entire Cyclanthaceae, Araceae, and Arecaceae are used instead of their representative genera as outgroups, although the character number is reduced. From these three different phylogenetic trees, a consen- sus tree (Fig. 5) can be drawn that shows features common to the other trees. Аз a caveat it should ohn, eneral remain poorly collected and insufficiently studied, and hence this prelimi- nary analysis of generic relationships is subject to revision as data accumulate. Monophyly of the genera, for example, should be rigorously dem- onstrated rather than assumed as I have done here. The current analysis is intended as a method of generating hypotheses concerning evolution of plant breeding systems and as a means of pointing out areas where more information is required. In the consensus tree (Fig. 5) Pandanaceae are characterized by three synapomorphies, namely spiny leaf margins, dioecism, and tristichous phyl- lotaxy (characters 12, 13, and 14, respectively). A fourth character, clustering of ovaries into syn- carps, is a probable synapomorphy for the family but was not used in this analysis since many ap- parent reversals occur, such as sections Acrostig- Volume 77, Number 4 1990 OX 819 Breeding Systems in Pandanaceae TABLE 1. Character states in Pandanaceae. Num- rey- Pan- Sara- Raphi- As- bers Character transition (old to new) cinetia danus ranga Balaka dophora plundia 1. Numerous ovules to solitary ovule 0 1 1 0 0 0 2. Parietal placentation to other placentation type 0 1 1 1 0 0 3. Seeds unprotected by seedcoat to thick, or endo- 0 1 0 0 0 0 ic 4. Spikes arranged in pseudoumbellate involucre to 1 0 0 0 0 0 spikes otherwise dispose 9. Inflorescence bracts not fleshy to inflorescence 1 0 0 0 0 0 fleshy 6. Inflorescence spicate to inflorescence paniculate 0 0 1 1 0 0 T. Primary phyllotaxis maintained to secondary tor- 0 1 1 0 0 0 sion causing pseudospiral leaf arrangement 8. Phyllotaxis same in juvenile and adult to phyllo- 0 0 1 0 0 0 taxis changes with maturity 9. Leaves with sheathing base or auricle to leaves 0 1 1 0 0 0 without sheathing base or auricle 10. Aerial roots present to aerial roots not present 0 0 1 1 0 0 11. Scandent habit to arboreous habit 0 1 1 1 0 0 12. No spines on leaf margins to spines on leaf mar- 1 1 1 0 0 0 gins 13. Breeding system various to breeding system dioe- 1 1 1 0 0 0 cious 14. Phyllotaxis various to phyllotaxis tristichou 1 1 1 0 0 0 15. Тера! primordia present to tepal к ub 0 1 1 0 0 0 sent ma, Jeanneretia, Bryantia, Curviflora, Maysops, and Cauliflora of Pandanus, which have single- celled drupes that are not connate into phalanges (Stone, 1972b). The belief that these cases illus- trate reversals from ancestral forms with compound ovaries must remain an assumption until a clado- gram for internal relations of genera and subgenera within Pandanaceae is developed. Major characters of systematic importance with- in the family include ovule number (character 1), placentation (character 2), and secondary torsion in the phyllotactic arrangement (character 7). Pan- danus and Sararanga are characterized by solitary ovules, absence of tepal primordia during floral organogenesis, and leaves without sheaths or au- ricles at the base. However, Pandanus freycine- tioides and Р. parvus exhibit rare or occasional multiple ovules (though few—only two or three) h & Stone, 1975). The evidence for tepal primordia is based on only a few examples out of some 800 species. In Pandanus and Sararanga juvenile (i.e., prereproductive) axes, the initial tris- tichous arrangement of the leaves is obscured by secondary torsion of the axis, creating a pseudospi- ral leaf arrangement (Figs. 13c, 15c). The sister group to Pandanus and Sararanga, Freycinetia, is characterized by numerous ovules, the presence of tepal primordia during floral organogenesis (Fig. 8b), no secondary torsion in the tristichous phyl- lotaxis, and auricles at the leaf bases. The system- atic usefulness of the last-mentioned character at an intergeneric level may be subject to revision, however, as studies of leaf organogenesis in all three genera of Pandanaceae and in the probable out- roups are necessary to test whether auricles and leaf sheaths have a similar ontogeny and are indeed homologous. Such an ontogenetic analysis is im- portant since the leaf bases in Pandanus sect. Jeanneretia and sect. Dauphinensia have auri- clelike sheaths almost as distinct as in Freycinetia, though they are not caduc Two autapomorphies aes 4, 5) charac- terize Freycinetia in all three trees: fleshy bracts (Fig. 6c) and telescoping of the inflorescence, which results in a pseudoumbellate involucre. The role of both features in pollination will be discussed later. There is an additional possible autapormophy, viz. epidermal-cell papillae on filaments of stamens (Stone, 1971). A single autapormophy, i.e., a hard endocarp (character 3), characterizes Pan- danus in the consensus tree. Sararanga is characterized by a single autapo- seeds protected by morphy common in the consensus tree: an onto- 820 Annals of the Missouri Botanical Garden Pandanus Rhaphidophora Freycinetia Sararanga FIGURE 2. Most iaa cladogram of the Pan- danaceae using the characters able 1 and using Rha- peores ere as the мөн genetic change in phyllotaxis (character 8). Since this change has not previously been noted it merits brief discussion here. n Sararanga sinuosa Hemsl. populations in Guadalcanal (Fig. 7), Tulagi Island (Fig. 8), and along the Siota Passage dividing big and little Ngela islands, I noticed a distinct difference in phyllotaxis in axes produced before and after flowering. The seedling and juvenile axis (Figs. 9, 10, 11) is always tristichous, in this respect resembling the phyllo- taxis of Pandanus and Freycinetia juvenile and adult axes. As this orthotropic axis develops, a secondary torsion results in a pseudospiral leaf arrangement similar to the adult and juvenile axes of Pandanus. The tristichous axis of Sararanga continues growth without branching to a height of approxi- mately 6-10 m when it is terminated by an inflo- rescence (Figs. 7c, 10c, 10d). Three buds in the Balaka Pandanus Sarajanga Freycinetia 10 FIGURE 3. Most parsimonious cladogram of the Pan- danaceae using the characters in Table 1 and using Ba- laka (Arecaceae) as the outgroup Asplundia Pandanus Sarajanga Freycinetia Most parsimonious cladogram of the Pan- d sing the characters in Table 1 and using As- plundia (Cyclanthaceae) as the outgroup. axils of leaves beneath this inflorescence continue growth, but these axes and all subsequent axes appear to be tetrastichous, with their leaves in four rather than three ranks, and lack any secondary torsion of the axis (Figs. 10, 11). The precise nature of the phyllotactic arrangement in these secondary axes can be determined by measuring the length of very young leaves using a method developed by Don Kaplan (pers. comm.). For this purpose, buds of primary and secondary axes of Sararanga sinuosa Hemsl. collected in Tulagi and Guadalcanal were dissected from the axes in the field and then pickled in FAA (Fig. 11). These buds were subsequently sectioned serially. Each section was examined for the first or last appearances of a leaf. Thus the length of each leaf could be found by multiplying the number of sections it occurs in by the section thickness (in this case 5 um). This FIGURE 5. Consensus cladogram of the Pandanaceae using cladograms shown in Figures 2-4. Only ed polarities common to all three cladograms are show Volume 77, Number 4 1990 Cox Breeding Systems in Pandanaceae FIGURE 6. nidae) in Samoa.—c. Pollination by Conservation International). d. Pollen on facial hair with permission from J. L Oxford Univ. Press, Oxford.) analysis revealed that the apparent tetrastichy of mature (postreproductive) Sararanga axes results from a distichous leaf arrangement with each suc- cessive node producing a pair of opposite leaves in a plane orthogonal to the plane of the leaf pair produced by the previous node. This arrangement is obscured by the highly compressed internodes except in the inflorescence axis where the internode length is increased, and the distichous phyllotaxis is apparent. REPRODUCTIVE BIOLOGY IN PANDANACEAE FREYCINETIA Floral biology. Freycinetia inflorescences are borne on hapaxanthic axes, either terminating the major axis of the liana (F. reineckei, F. marginata) with renewal growth occurring from an axillary bud beneath the inflorescence, or terminating ax- illary shoots (F. arborea, F. scandens). In some species the shoots arise on defoliate branches (F. funicularis). The pattern of serial changes in leaf Pollination ER of Freycinetia reineckei Pteropus marianensis (Pteropidae) in .—a. Habit. — b. Pollination by Aplonis artifuscus (Stur Tuttle, Bat Figures used uam photo by ean teropus samoensis (Pteropidae) in amo ovett Doust & L. Lovett Dou (йг) Plant Reproductive Ecology: insti and Patterns. form and size along the Freycinetia axis is similar to the pattern found in other monocotyledons (Kap- lan, 1973), although the reduction in bract size toward the distal end of the axis is strikingly dif- ferent from some Cyclanthaceae (Harling, 1958). The changes in leaf length, basal width, and the occurrence of marginal spines along a vegetative axis are illustrated in Figure 12. There is an in- crease in leaf length from the oldest leaf, the pro- phyll (leaf number 34) to the mature foliage leaves (leaf number 16) and thence a decrease in length to the youngest leaf (number 1), which encloses the meristem. If the axis terminates in an inflo- rescence, there is a distinct decrease in length and number of marginal spines toward the distal end (Fig. 12) with a smooth transition from foliage leaves to the fleshy bracts. Frequently foliage leaves immediately beneath the inflorescence have a patch of bright coloration at their base, which undoubt- edly adds to the attractiveness of the inflorescence to potential pollinators. For an illustrated example of these transitions see Stone (1967). In Freycinetia angustifolia and F. jagorii the inflorescence is clearly racemose. In all or most Volume 77, Number 4 1990 Breeding Systems in Pandanaceae FIGURE 8. Sararanga sinuosa in the Solomon Islands.— a. Pendulous pistillate inflorescence in Tulag п.—с. Juvenile in Tulagi showing безе чеп phyllotaxis. —d. Seedlings in Tulagi showing tristichous ses E — FIGURE 7. Sararanga sinuosa in the Solomon Islands.—a. Habit in Guadalcanal.—b. Root system in Tulagi; note absence of stilt roots. —c. Pistillate infructescence in Guadalcanal. —d. Cephalia in Guadalcanal. 824 Annals of the Missouri Botanical Garden E9. Sararanga sinuosa in Guadalcanal, Solomon Islands. —a. Juvenile. —b. Field dissection of “а” showing ha phyllotaxis.— c. Larger juvenile. —d. Field dissection of “c” showing tristichous phyllotaxis. other species, the inflorescence is a false umbel of spikes with the subtending fleshy bracts telescoped by a reduction of internode length yielding the pseudoumbellate structure. Various colors are found in Freycinetia inflorescences. Some species have white bracts (F. hombronii, F. scandens), while others have colored inflorescences ranging fom pale salmon (F. arborea) to bright reddish orange (F. reineckei, F. marginata) (Cox, 1984; Cox et al., 984; Stone, 1982b). Some species have bright yellow bracts (F. biloba), while others have very dark purple bracts (F. negrosensis). Species with Volume 77, Number 4 1990 Cox 825 Breeding Systems in Pandanaceae Field dissection of Large juvenile. — b. e-up of tristichous inflorescence ; note superficial tetrastichy of се 55 FIGURE 10. Sararanga sinuosa іп Guadalcanal, Solomon Islands. “a” showing tristichous phyllotaxis.—c. Field dissection of зеен ий indiv се showing tristichous base of a на three axillary branches produced beneath inflorescence. —d. Clo base сва branches photographed prior to din stage shown in “с branches. white inflorescences produce a fetid, musky smell while those with colored inflorescences frequently lack a distinctive smell; however, this topic lacks substantive data as yet. No liquid nectar is produced by Freycinetia species. Instead, the bracts themselves function as hexose-rich “‘solid nectar," containing up to 29% by dry weight total sugar (Cox, 1983) with partic- ularly high concentrations of fructose (Cox, 1983, 1984; Cox et al., 1984). The bracts are also rich Annals of the Missouri Botanical Garden FIGURE 11. Sararanga ѕіпиоѕа. —a. Superficially O apex of branch produced subsequent to first werin . Cross section of “a” showing distichous nature of phyllotaxis. —c. Cross section of tristichous apex of prereproductive primary axis showing Шао» ар d Pollen with reticulate exin Volume 77, Number 4 1990 in free amino acids, with up to 9% by dry weight total amino acids (Cox, 1983), and up to 12 dif- ferent amino acids present (Cox, 1984). The number of spikes in an inflorescence usually varies between three and seven, while in F. celebi- ca, the spike is usually solitary. Since each of these spikes is subtended by a fleshy bract, the shortening of internodes in this region of the axis permits the clustering of the spikes and bracts into a single anthecological unit that seems well adapted for vertebrate pollination (Fig. 6). The staminate spikes are often the same color as the distal bracts. The pistillate spikes, however, are usually green, though they may be white or pink in some species. The staminate spikes represent a high-quality reward for pollinators, as they may provide up to 26% dry weight crude protein and up to 24% dry weight lipid (Cox, 1984). Their high lipid content probably results in large part from the lipid-rich pollenkitt that covers the pollen. At the terminus of the axis are several fleshy bracts (Bekonstigungskorper) that do not subtend spikes and that differ from outer bracts by being smaller and cylindrical, and by lacking marginal teeth on their tips. The true flowers are extremely tiny and are congested on the spikes. The absence of perianth members at maturity makes delimitation of each flower very difficult, though their individuality can be discerned by studies of floral organogenesis. Experimental. Developmental stages of Frey- cinetia arborea inflorescences were collected on the islands of Kauai and Hawaii and preserved in FAA. The spikes were bisected longitudinally. Half of each spike was stained in acid fuchsin, de-dif- ferentiated in 75% ethanol, and studied using the epi-illumination techniques of Sattler (1968). The other half was critical-point dried, coated with a silver-gold-palladium amalgam, and studied using the scanning electron microscope techniques of Uhl & Moore (1980) With both techniques, floral units, each of which is subtended by a tiny bract (Fig. 13a), can be distinguished along the inflorescence. The outer whorl usually consists of six perianth primordia, inside of which is the whorl of androecial primordia (indicating six stamens). At a later stage of devel- opment (Fig. 13b), several gynoecial primordia de- velop inside the androecial whorl. In staminate spikes, however, the floral bracts, perianth, and gynoecium soon cease development (Fig. 13c) and only the androecial primordia continue developing, with the aborted gynoecial primordia forming a ring-shaped pistillode (Fig. 13d). Comparable de- velopmental information for pistillate spikes is un- 827 Breeding Systems in Pandanaceae 64 2.00 56 x 175 Y 48 ` 1.50 s [ 40 Y £ Г *; 125. о a ` 5 = a в 2 2 32 4 Y 1003 3 f i Е = i * 24 H ` 075% è Е) y i 16 4 ЕЙ 0.50 F X - аф Е ЕЕ - 0.25 y *. ig $i 5 30 25 20 15 10 5 leaf number FIGURE 12. Leaves of transition along a vegetative axis of Freycinetia reineckei. Dotted line indicates leaf length in ыл solid line indicates basal widt centimeters; + or — in e youngest leaf) were obtained by dissection of the apical bud. available because the appropriate stages have yet to be collected; however, a similar sequence of events probably occurs since the mature gynoecia of many Freycinetia species are surrounded by numerous diminutive staminodia (Fig. 14 Breeding systems. Although Freycinetia and all Pandanaceae have been believed to be strictly dioe- cious (Hutchinson, 1973; Dahlgren et al., 1985), recent fieldwork has indicated that a variety of breeding systems exists in Freycinetia (Cox, 1981; Cox et al., 1984; Poppendieck, 1987). To discuss this diversity and its evolutionary significance, it is useful to consider a Freycinetia individual as a metapopulation (White, 1979) of modules. More precisely, a Freycinetia (or any other) plant can be viewed as having modularity at different levels, thus being composed of a nested set of modular units (Cox, 1988). In all known species of Frey- cinetia a minimum of four levels of modular con- struction can be recognized. For example, entire genetic individuals (Fig. 15a) of F. reineckei may be considered as modules at one level (let us call this level 1), and these modules could theoretically be either sexually monomorphic (all hermaphro- ditic) or dimorphic (pistillate and staminate). А second modular level (level 2) consists of hapax- anthic axes terminated by an inflorescence (Fig. 15b); again, these hapaxanthic axes can be sexually monomorphic or dimorphic. А third modular level Missouri Botanical Garden Annals of the Volume 77, Number 4 1990 829 Breeding Systems in Pandanaceae “ШИЕ. „әм, сү КЛИ Wan UN h 4 DV VA TR S К AA DURS ^. SAM КОШ nit A, ey AMT Aie y FIGURE 14. Gynoecial structures of e reineckei. — А. Mature gynoecium with bilobed stigma, gland ots on ovary neck, and staminodia at ovary base staminodia (level 3), which may be either monomorphic or dimorphic, is the spike together with its subtending bract (Fig. 15c). Within each spike is found the fourth modular level (level 4), the floral unit sub- tended by a tiny bract (Fig. 15d), which could be sexually monomorphic or dimorphic. For a Freycinetia species to be dioecious, i.e., —B. Aspect of gynoecia on surface of spike.— C. Detail of sexually dimorphic at modular level 1, all lower modular levels also must be dimorphic. However, recent field research has revealed some individuals of F. reineckei in Samoa to be hermaphroditic at level 3 by producing hermaphroditic spikes (Cox, 1983), as are some individuals o strobilacea in Indonesia (Cammerloher, 1923; Stone, 1971). — E13. Floral iia d in Freycinetia arborea. —A. Early Mibi age stage of flower with subtending bra bel : (b) perianth primordia (p), a primodia (g).— C. Subs gynoecial primordia into staminode. stage showing Later stage of development showing gynoecial several flowers along spike. (yn noecia, bracts, and р perianth members have all ceased еи with only androecial members continuing growt 830 Annals of the Missouri Botanical Garden MODULAR LEVELS IN FREYCINETIA шэ 0g b. Level 2 : determinate shoots UUUUUUUOCOUUO 00000000000 00000000000 aa LOOQOOOOOOOo c. Level 3: spikes а. Level 4:floral primordia FIGURE 15. The different modular levels in Freycinetia reineckei at which various dimorphic or monomorphic reproductive strategies could be expressed. The floral primordia in (d) occur early during organogenesis; crowding on the mature axis obscures individual floral units, particularly on staminate spikes. Volume 77, Number 4 1990 Cox 831 Breeding Systems in Pandanaceae Some individuals of F. scandens in Australia are hermaphroditic or monomorphic at level 1 by pro- ducing male and female shoots on the same plant (Cox et al., 1984); sexual monomorphism at this level is usually termed **monoecism."' Similar cases of monoecism were described by Stone (Stone, 1972c; Cox, 1981) in individuals of Freycinetia imbricata in Sumatra and Freycinetia negrosensis in the Philippines, which also produce staminate and pistillate shoots on the same plant. Such di- vergences from a dioecious breeding system may be important in island colonization (Baker & Cox, 1984), particularly since monoecious individuals of Freycinetia scandens have been found to be self-compatible (Cox et al., 1984). Bagging and exclusion experiments for several species of Frey- cinetia indicate that apomixis is absent (Cox, 1983, 1984). Pollination biology. Freycinetia inflorescences are adapted to pollination by several kinds of verte- brates such as flying foxes (Burck, 1892; van der Pijl, 1956; Cox, 1982, 1984); smaller bats (Knuth & Loew, 1904; Daniel, 1976); squirrels (Heidi, 1927); crows (Cox, 1983); pigeons (Cox, 1984); honeycreepers (Cox, 1983); and white eyes (Cox, 1983); but not rats (Cox, 1983). The openness and availability of the inflorescence to a wide taxo- nomic variety and size range of pollinators suggest a lack of tightly coupled plant-pollinator coevolu- tion. Indeed, Freycinetia pollinators tend to be frugivores. Biochemical assays of Freycinetia re- ineckei inflorescences and a variety of indigenous Samoan fruits eaten by its major pollinator, Ptero- pus samoensis, revealed striking similarities be- tween hexose/disaccharide ratios and amino acid compositions (Cox, 1984). In contrast to po mnada systems that are believed to have in other plants, no apparent evolutionary response of pol- linators to Freycinetia has been found. There is, however, strong evidence that pollinator behavior has affected the evolution of breeding systems in Freycinetia. The vertebrate pollinators of Freycinetia tend to be destructive. For example, the flying foxes that pollinate Freycinetia reineckei in Samoa and Guam, Pteropus samoensis and P. mariannensis, respectively, eat the inner and outer staminate and pistillate bracts as well as the lipid-rich staminate spikes (Fig. 6c, d). Pistillate spikes, which are green and mucilaginous, are rarely disturbed and set abundant seed when pollinated. Although staminate spikes transmit their genes via pollen on the pol- linators” faces, and pistillate spikes transmit their genes via ovules, hermaphroditic individuals lose a large proportion of their investment in gynoecial structures due to pollinator damage, and they have a low fitness, represented by a concave fitness set (Cox, 1982). Quantitative comparison of the fit- nesses of staminate, pistillate, and hermaphroditic individuals shows vertebrate pollination maintains the dioecious breeding system in Freycinetia rei- neckei (Cox, 1982). Dispersal. The infructescences of Freycinetia are usually brightly colored, frequently red, and are rich in sugars. These attributes make them attrac- tive to various birds and perhaps some flying foxes. Freycinetia reineckei in Samoa is dispersed by Aplonis artifuscus (Sturnidae). Fauna Hawaiien- sis (Perkins, 1902) is filled with references to birds of various genera eating F. arborea fruits. Guppy (1906) believed birds to be the primary dispersers of Freycinetia species and reported finding bird pellets below the inflorescences filled with Frey- cinetia seeds. Experiments on seed germination (Cox, unpublished) suggest adaptation to endozoo- choric dispersal, since F. reineckei seeds require a mild acid treatment for successful germination. Dispersal by flying foxes is undoubtedly important as well. The type specimen (at BISH) of Freyci- netia degeneri (Degener 15128) has written on the label, “fresh roots pounded as roofs to build their grass houses. They do not eat fruit. Flying foxes eat it.” Smith (1979) also believed flying foxes to be important dispersal agents in Fiji. Ter- restrial mammals also probably aid dispersal; Per- kins (1902) reported that rats eat F. arborea in Hawaii, and they are implicated as seed dispersers in New Zealand. A specimen of F. banksii (Mee- bold 18245) has written on the label, “eaten by rats," which Daniel (1973) confirmed. The attrac- tiveness and availability of Freycinetia infructes- cences to a wide variety of vertebrates has perhaps been best stated by Stone din pp. 85-86): "Probably the fruits—technically berries—are dis- persed by birds, bats, and possibly other mammals. Being juicy, they are edible.” PANDANUS Floral biology. Stone (1968b) grouped pistillate inflorescences of Pandanus according to whether the cephalia are solitary or multiple, and whether the fruits are free or connate into syncarps. In species with single cephalia (e.g., P. tectorius sensu latissimo, P. leram, Stone, 1983), numerous char- taceous bracts, usually pale yellow or white, are produced beneath the inflorescence and are in- serted tristichously. In species with multiple ce- phalia (e.g., P. nepalensis, P. spinistigmaticus), 832 Annals of the Missouri Botanical Garden FIGURE 16. the cephalia are borne on racemes, and beneath each cephalium is a single bract. Pandanus species in general do not show the extreme telescoping of the main inflorescence internode that occurs in Freycinetia, although the cephalia themselves re- veal much telescoping of secondary and tertiary axes. Staminate inflorescences of Pandanus usu- ally consist of panicles of staminate spikes, each spike subtended by a large, often pale-colored bract. Floral units homologous to those found in Frey- cinetia arborea have yet to be discovered in Pan- danus, but those of P. androcephalanthos and other species of subg. Martelidendron in Mada- gascar perhaps have a similar construction. Sta- mens are usually fasciate in phalanges, pistillodia infrequently occur (e.g., P. douglasii, Stone, 1968b), but tepals have yet to be identified. Sim- ilarly, staminodia infrequently occur in pistillate Pandanus individuals (e.g., Р. cominsit, Stone, 1968b), but again tepals have yet to be identified. Detailed studies of floral organogenesis in Pan- danus are needed. Pandanus inflorescences, particularly staminate inflorescences (Fig. 16), sometimes have a subtle, staminate inflores- pleasant fragrance. Indeed, cences of cultivated P. odoratissimus are used in Pandanus tectorius. —a. Habit in Maui. — b. Staminate inflorescence in Moorea. India as a base for perfumes, the “ой of Keura" Purseglove, 1972). Neither sex, however, duces nectar, and fleshy bracts similar to those ~, pro- found in Freycinetia are unknown. Breeding systems. A true departure from strict dioecism in Pandanus has yet to be found. Given the widespread occurrence of staminodia and pis- tillodia (Stone, 1968b) in the genus, however, it seems likely that further field studies will reveal aberrations. For example, Stone (1972Ь) found two pistillate individuals of Pandanus cominsii var. Augustus on Buka Island that had drupes sur- rounded by staminodia, some of which had anthers; P. microcarpus produces staminodia with anthers adnate to the exocarp (Vaughn & Wiehe, 1953). Departures from strict dioecism, if they occur, either are extremely rare or are limited to certain h t species; I have yet to find, despite es throughout Polynesia, a Pus шй йш of Pan- danus tectorius that produces both staminate and pistillate inflorescences. Bagging experiments in Kauai and Maui, Ha- waii, and Moorea (French Polynesia) indicate that Volume 77, Number 4 1990 Cox 833 Breeding Systems in Pandanaceae FIGURE 17. near cephalium of breeding system. Pandanus tectorius is facultatively apomictic (Cox, 1985). All bagged inflorescences produced fruit (Fig. 17). Possibility of contamination of the bags was checked with scanning electron microscopy of stigmatic surfaces; no pollen grains were found. Parthenocarpic fruit development was rejected since bagged cephalia produced viable embryos and en- dosperm. To determine whether apomixis was ob- ligate or facultative, the genetic diversity of sam- ples of endosperm from bagged and unbagged cephalia was studied with starch gel electrophoresis (Cox, 1985). Five different loci were examined. Endosperm samples from different syncarps from the same unbagged cephalium had significant iso- zyme diversity, while endosperm samples from syn- carps from the same bagged cephalium were mono- morphic at all loci. Thus the system is facultatively apomictic, with asexual reproduction occurring in the absence of pollination. Campbell (1911) noted migration of somatic nuclei into the embryo sac, which was confirmed by Cheah & Stone (1975), who found supernumerary nucellar nuclei in the embryo sac of Pandanus parvus. Facultative apo- mixis may occur throughout the genus; as early as 1867 Kurz found isolated Pandanus dubius females to produce fruit, while Fagerlind (1940) reported apparent parthenocarpy in Pandanus co- Pandanus tectorius. —a. Pistillate inflorescence. —b. Flow visualization in field showing deceleration .—c. Tristichous phyllotaxis with secondary spiral. —d. Bagged cephalium indicating apomictic nature lumniformis. | have found in the Royal Botanic Gardens, Sydney, a sole pistillate individual of Pan- danus forceps that sets abundant fruit, and in the Royal Botanical Garden, Melbourne, a single pis- tillate individual of Pandanus forsteri that also produces copious fruit; whether this indicates apo- mixis or parthenocarpy remains to be determined. Pollination biology. The pollination biology of nearly all species of Pandanus remains unstudied. I have made an extensive study, however, of the pollination ecology of Pandanus tectorius sensu latissimo in Hawaii and French Polynesia (Cox, 1985). Although insects (e.g., bees) are frequent visitors to staminate inflorescences where they gather pollen, they seldom visit pistillate inflores- cences. Neither type of inflorescence produces nec- tar or any pollinator reward other than pollen. Peroxidase assays indicate an early onset of stig- matic receptivity that extends until well after fruit formation. The copious pollen produced by the pendulous, paniculate staminate inflorescences (Fig. 16b) is devoid of a pollenkitt and blows easily in the wind. However, a pollenkitt occurs in other species; according to Stone it is formed in Pan- danus beccarii and similar species of sect. May- sops. To test for wind pollination, flow patterns 834 Annals of the Missouri Botanical Garden around pistillate inflorescences were examined with a laboratory wind tunnel using the techniques of Niklas (1982), and in the field using a portable smoke-injection apparatus (Cox, 1985). Regardless of inflorescence orientation, flows in the range of 1 m/sec. resulted in back eddies that caused sig- nificant deceleration in the region of the stigmatic surfaces (Fig. 17b). Spiral patterns in the xy and xz planes indicated that pistillate Pandanus tec- torius inflorescences function as highly efficient pollen receivers, being hydrodynamically analo- gous to some filter-feeding marine invertebrates. This anemophilous nature of P. tectorius, together with its system of facultative apomixis, was con- firmed through bagging studies. Dispersal. The primary unit of dispersal on most Pandanus species is the syncarp, frequently termed a phalange.* The phalanges are brightly colored, and the basal part is sweet. In coastal species, such as P. tectorius, the phalanges are buoyant and can frequently be found in beach-drift throughout the Pacific (Gunn & Dennis, 1976). Guppy (1906) and Ridley (1930) considered P. tectorius to be dis- persed primarily by ocean currents, although Lee (1985) reported localized dispersal by crabs. Bird dispersal of Р. tectorius also occurs in Samoa, and flying fox dispersal has been observed in Micronesia (Stone, pers. comm.). Stone (1982a, pers. comm.) reported turtle dis- persal of P. helicopus, a freshwater species, in the Malay Peninsula, and endozoochoric dispersal of inland species is highly likely (Stone, 1982a). Dis- persal by humans of economically useful species has occurred as well. In Hawaii a census of four 0.25-m? plots of P. tectorius phalanges lying on the ground revealed that of those phalanges that produce seedlings, 69% produce two-six seedlings (Cox, 1985). Using data from this survey, a probability analysis re- vealed that the likelihood of any established phal- ange producing at least one male and one female seedling exceeds 55% (Cox, 1985). Further work by Lee (1989) in Moorea has confirmed that the dispersal of a single phalange may establish a sex- ually reproductive population. ' Terminology follows Stone. The levels employed are (1) carpel, (2) diei = (3) cephalium. Older works use syncarp, but this can refer to either level 2 or 3. Each carpel ripens to form a drupe; in species with free or solitary drupes, the cephalium consists of drupes. How- ever, in species such as P. tectorius, each phalange is ormed of several connate drupes; hence the term “‘poly- drupe” is sometimes used. SARARANGA Floral biology. The massive (up to 1.7 m long) pendulous inflorescences of Sararanga are always terminal (Figs. 14c, 15a), with renewal growth occurring through the development of axillary buds beneath the inflorescence. In contrast to Freyci- netia and Pandanus, the pistillate inflorescence is strongly paniculate while the staminate inflores- cence consists of panicles of staminate heads, sim- ilar to that found in Pandanus species. Both types of inflorescences are subtended by small, hard bracts. The nature of the terminal floral unit in Sararanga remains unclear. The green, terminal floral units (Fig. 7d) are composed of numerous fused carpels, each producing a single anatropous ovule beneath a single pointlike stigma. The sinuous arrangements of the fused carpels and their stigmas "strongly resemble the [pattern of] stitching on an American baseball” (Stone, 1961). A small, whitish cuplike structure is produced at the base of the fused carpels. This structure, when analyzed by paper chromatography, proves to be rich in hexose sugars. Although the terminal floral units were termed “receptacula florifera”” by Hemsley (1893), Stapf (1896) called them flowers, a practice fol- lowed by Stone (1961), who considered the ter- minal floral units “‘pedicellate flowers” with the white cupule termed a “perianth.”” North & Willis (1971) described the fertilized terminal floral units as “‘fleshy fruits, consisting of numerous carpels.”’ However, precise homology between the ter- minal floral units of Sararanga and floral struc- tures of other Pandanaceae, as well as other mono- cotyledons, remains obscure. Studies of floral organogenesis previously described revealed little similarity between the terminal floral units of Sara- ranga and the flowers of Freycinetia. Particularly unclear is the nature of the white cupule that subtends the terminal floral unit. This cupule scarcely resembles the perianth of any other mono- cotyledonous flower; it is much more bractlike in appearance. Ultimate resolution of these difficulties requires, in my opinion, study of the ontogenetic development of these structures; as yet, however, I have been unable to obtain the appropriate de- velopmental stages (Fig. 18). Based on current information, the possibility that the terminal floral unit of Sararanga may be homologous to a Pan- danus phalange, or even homologous to an entire Pandanus cephalium or Freycinetia spike, with the cupule representing the subtending bract, can- not be excluded. The fleshy cupule appears to serve the same function, as pollinator rewards, as the eshy bracts of Freycinetia. Analysis of the cupule Volume 77, Number 4 Cox 835 1990 Breeding Systems in Pandanaceae URE 18. Floral organogenesis in Freycinetia and Sararanga. —a. Dissection of reproductive apex of F. reineckei showing young spikes and subtending bracts.—b. Early developmental stage of a staminate F. reineckei spike. — c. Dissection of Sararanga sinuosa reproductive staminate apex from a secondary axis showing early stages of inflorescence development; note decussate nature of phyllotaxis. —d. Stigmatic surfaces of Sararanga sinuosa cephalium. 836 Annals of th Missouri al Garden reveals it to be similar to Freycinetia bracts by being very rich in hexose sugars. Since Sararanga produces no nectar or other pollinator reward, these white, fleshy cupules serve as the sole pollinator reward. They are, however, extremely small in comparison with the hexose-rich bracts of Frey- cinetia or the bracts of Pandanus inflorescences. Breeding systems. | have surveyed populations of Sararanga sinuosa in the Konnga Region of Gua- dalcanal Island, Tulagi Island, and along the Siota Passage on Big and Little Ngela islands and have yet to discover any departures from strict dioecism. Staminodia and pistillodia are unknown in the ge- nus. Sararanga sinuosa does not appear to be apo- mictic, as some isolated trees have very poor or no fruit set, and within any particular inflorescence, numerous terminal floral units fail to develop. An isolated pistillate tree in the Honiara Botanical Gar- dens, Guadalcanal, sets some fruit, although the set is poor compared with wild trees. Because of deforestation, this individual is at least 6 km from any other possible conspecific and 15 km from any other known conspecific. Even this does not, how- ever, provide evidence for apomixis or partheno- carpy, since, as will be discussed below, Sararanga sinuosa appears to be entomophilous. As Kerner von Marilaun (1895, p. 208) reported, a Dra- cunculus creticus (Araceae) planted in the Vienna Botanic Gardens attracted at anthesis a swarm of dung beetles when previous to its opening none could be found anywhere in the Gardens or im- mediate vicinity. The possibility that the massive inflorescences of Sararanga, which rival those of the monocarpic palms in size and productivity, could attract pollinators from several kilometers away cannot be discounted. Pollination biology. The pollen grains of Sara- ranga sinuosa average 13 um in diameter, have reticulate exines (Figs. 15b, 18d), and lack a pol- lenkitt. The anthers are borne in groups, each of which is subtended by a small fleshy bractlike cu- pule nearly identical to that produced beneath the pistillate terminal floral units. As mentioned above, this cupule or bract is rich in hexose sugars. The entire staminate inflorescence has a slightly pep- pery smell; no similar odor can be detected from the pistillate inflorescences. The sessile stigmas of the pistillate floral units, borne in sinuous rows (Fig. ‚ 18d), show peroxidase activity while the unit is small and green. In staminate and pistillate in- florescences no nectar is produced; the sole polli- nator rewards are pollen and the whitish, hexose- rich cupules or bracts that subtend the floral units. Since the inflorescences are pendulous from the tops of these tall (20 m) trees, determination of floral visitors is difficult. I have witnesed small beetlelike flying insects visiting the pistillate inflo- rescences but have not captured any for identifi- cation and analysis. A night I spent in the crown an S. sinuosa tree in Tulagi Island similarly failed to yield any evidence of nocturnal visitors. Wind tunnel analysis of flow patterns around pis- tillate inflorescences indicated no significant eddies or flow patterns produced by the inflorescence or by the terminal floral units that cause flow decelera- tion near the stigmatic surfaces. I therefore believe Sararanga sinuosa to be pollinated solely by flying insects, although further fieldwork in the Solomons to gain more details of the pollination ecology of Sararanga. Dispersal. Guppy (1906, p. 156) believed Sara- ranga sinuosa to be bird-dispersed. Although I have yet to witness dispersal, the bright red (A. D. E. Elmer called them “candy red") terminal floral units, which are rich in disaccharides, have a de- licious taste reminiscent of cherries or strawberries. Given the position high in the trees, they are almost certainly dispersed by birds and bats, although con- firmation of this awaits further fieldwork. is needed BREEDING SYSTEMS, EVOLUTION, AND SYSTEMATICS In principle, the phylogenetic tree derived from systematic information can be used to interpret evolutionary trends in breeding system and polli- nation syndromes in the three genera of Pandana- ceae. For example, because pollination syndromes appear to be relatively constant within each genus (with the exception of possible chiropterophily in some species of Pandanus sect. Maysops), and since each pollination syndrome obviously has an evolutionary history, the genera can be replaced with their respective pollination syndromes as shown in Figure 19. This modified cladogram indicates that within Pandanaceae, wind and insect polli- nation have evolutionary histories more closely re- lated to each other than either does to vertebrate pollination. Only a few details concerning possible trends in pollination biology can be derived from the consensus tree, however, since information from pollination biology was not used to derive the tree. Unfortunately, the pollination biologies of three genera I have used as outgroups, Balaka, Rha- phidophora, and Asplundia, remain unknown. However, to illustrate the possibility for interpret- ing the evolution of breeding systems and polli- nation syndromes, should the sister group of the Volume 77, Number 4 1990 Cox 837 Breeding Systems in Pandanaceae Pandanaceae be convincingly demonstrated, І sub- stitute for Asplundia in Figure 20 the genus Cy- clanthus (Cyclanthaceae) whose pollination biology has been well studied ier 1982). In Costa Rica, Cyclanthus bipartitus is polli- nated primarily by beetles (Beach. 1982). The first night after pistillate anthesis, scarab beetles of the genus Cyclocephala forage on the lipid-rich fleshy adaxial surfaces of the inner bracts. Later, after staminate anthesis, the beetles feed on the pollen. If we put these details on the cladogram (Fig. 19), we find Cyclanthus and Pandanaceae sharing two features of pollination biology as a symple- siomorphy: fleshy bracts and pollen used as polli- nator rewards. А synapomorphy а Pandanaceae, is tristichous phyllotaxis, whic arts a radial symmetry. Of importance to the pollination biology of Pandanus, this phyllotactic pattern positions the bracts in three ranks beneath the inflorescence. Wind tunnel experiments on Pandanus showed such tristichy of the bracts to impart aerodynamic characteristics necessary for pollen capture by the pistillate inflorescence (Cox, 1985). Although highly functional, this feature cannot be considered adaptive in the strict sense since pistillate inflorescences of vertebrate-polli- nated Freycinetia reineckei also produce similar flow patterns when placed in a wind tunnel. Given the cladogram here derived, tristichous phyllotaxis appears as a preadaptation to anemophily that was possessed by some of the ancestral pandans, and proved functional only after changes in the sta- minate inflorescence, such as lengthening of the MI wee allowed pollen to be wind-dispersed. s of interest that a synapomorphy charac- c Sararanga and Pandanus is relatively elongated internode length in the staminate inflo- rescence. This results in a pendulous inflorescence that facilitates dispersion of pollen on the wind. As shown by the wind tunnel experiments with Frey- cinetia, the ancestral tristichy of the family made the involucrate pandanaceous inflorescence ready to function as an aerodynamically efficient pollen receiver. Thus, merely lengthening the staminate inflorescence internode made a jump from verte- brate pollination to wind pollination suddenly pos- sible. With the advent of wind pollination, produc- tion of pollinator rewards became unnecessary. Here the change of fleshy bracts to chartaceous bracts appears as an autapomorphy characterizing Pan- anus. A possible autapomorphy characterizing Frey- cinetia is a lipid-rich pollenkitt that holds pollen to the feathers of birds and to the facial hairs of flying foxes as well as providing a pollinator reward — Freycinetia eetle) (Vert Pandanus Sararanga ebrate E (Wind) (Insect) — chartaceous long staminate inflorescence internod nvolucre tristichous phyllotaxis fleshy bracts FIGURE 19. Consensus tree of pollination syndromes in Pandanaceae with Cyclanthus (Cyclanthaceae) used to determine polarities. ~ (Cox, 1984). A pollenkitt, however, also occurs in Pandanus (in P. beccarii and perhaps other species of sect. Ma s). e characteristics of vertebrate-pollinated pan- danaceous inflorescences, as revealed by extant Freycinetia species, make them unlikely to be pol- linated by insects. For example, the open involucres cause the head of a large vertebrate pollinator, such as a bird or flying fox, to come into direct contact with the spikes as they feed on the bracts. Contact with the spikes would not be made, how- ever, by small insects feeding on the bracts, given the distance between the attractive bracts and the spikes. Thus, even though pollen could be deposited on insects that fed on the staminate spikes, this pollen would not be transferred effectively to pis- tillate spikes, given the lack of nectar or other pollinator rewards offered by the pistillate spikes themselves. A reduction in size of the inflorescence resulting in closer spatial proximity of the pollinator reward to the stigmatic surfaces would be necessary for insect pollination to occur. Something similar appears to have occurred in Sararanga, pending, of course, accurate deter- mination of the morphological identity of the ter- minal floral units. The inflorescence is paniculate and massive, creating a large display for potential insect pollinators. The basal hexose-rich cupule or bract is positioned on the pedunculate terminal floral unit so that the body of any insect icedung on it would come in contact with th The cupule or bract of the staminate terminal үө units is in a similar position, and would thus deposit pollen in a similar manner on the body of any insect feeding on the bract Thus, if the Cyclanthaceae indeed proved to be the sister group to the Pandanaceae, an evolution- 838 Annals of the Missouri Botanical Garden Freycinetia (Vertebrate) Pandanus Sararanga psc (Insect) E 20. Consensus tree of Pandanaceae with pri- mar е are reeding systems superimposed; polarities deter- mined as in Figure 5. ary sequence from vertebrate pollination to a branching point between wind pollination and insect pollination could be envisioned. In terms of breeding system evolution, Cyclan- thus bipartitus is monoecious, with unisexual flow- ers produced in alternating cycles on the congested spadix. Because, as shown in extant hermaphroditic Freycinetia inflorescences (Cox, 1982), vertebrate pollinators feeding on pollen cause significant dam- age to gynoecial structures, the advent of verte- brate pollination would cause a monomorphic (her- maphroditic) population to become vulnerable to invasion by dimorphic (dioecious) mutants (Cox, 1982, 1986). Thus dioecism, which is a synapo- morphy uniting Pandanaceae (Fig. 20), may have evolved in the family in response to ancestral ver- tebrate pollination. Similarly, vertebrate pollination maintains dioecism in Freycinetia populations to- day (Cox, 1982). Dioecism, however, greatly re- duces colonization ability (Baker € Cox, 1984). Thus, the evolution of facultative apomixis in Pan- danus increased its ability to colonize new ара and islands, resulting in a dramatic Dioecism may be maladaptive in мкр to м nization potential in Sararanga because it reduces likelihood of successful pollination, and thus may partially account for its inability to colonize oceanic islands successively. The relationships between breeding systems, pol- lination syndromes, ranges, and numbers of species may be highly significant in the Pandanaceae. For example, the genus Pandanus, with wind polli- nation, facultative apomixis, and water- or animal- dispersed syncarps, has the broadest range and greatest number of species. Pandanus, with its anemophilous pollination system, can colonize any appropriate island regardless of the island’s faunal composition. Since Pandanus is facultatively apo- mictic, successful colonization does not require the establishment of both sexes. Thus a single female propagule may eventually fill an entire island with its apomictic progeny. Through founder effect and genetic drift alone, such facultative apomixis in a plant colonizing disparate islands would be expected to result in a huge range and a massive number of species. Indeed this breeding system may par- tially Ha some of the controversies in Pan- danus tax For example, Stone (1976a, 1982c) and Fosberg (1977) have frequently dis- agreed with St. John (1979a, b) on the identity of Pandanus tectorius. Fosberg, for example, re- duced six binomials erected by St. John from Al- dabran specimens to synonymy with Pandanus tectorius, and Stone reduced 15 Pandanus species erected by St. John from coastal Australian col- lections to synonymy with P. tectorius. As Stone (1982c, p. 135) perceptively suggested, “The ob- vious taxonomic difficulties in certain species-groups of Pandanus such as the ‘P. tectorius” group and also in Pandanus чш Austrokeura Stone, тау be caused by apomixis.” Facultative apomixis and anemophily may also permit Pandanus species to colonize new habitats that would otherwise be unavailable to it if it were linked to a specific type or class of animal polli- nators. Thus, allopatric speciation processes as well as genetic drift may have contributed to the high speciation rate in Pandanus. Freycinetia species, on the other hand, lack facultative apomixis and water dispersal. However, their attractiveness to a wide variety of vertebrate pollinators and dispersers, as well as infrequent “leaky dioecy”” (Baker & Cox, 1984), would assure them a large range and high speciation rate, albeit lower than Pandanus, where the apomixis coupled with abiotic dispersal has likely resulted in specia- tion through genetic drift. Conversely, Sararanga lacks facultative apo- mixis, “leaky dioecy," and is apparently tied to a single class of pollinators of possibly a limited range, which may explain its small range and low number of species. Its distribution along the edge of the Tethys geosyncline and its absence from oceanic islands suggest that the distribution of Sararanga may be entirely due to vicariance processes, i.e., local differentiation resulting in two species. n conclusion, recognition of the phylogenetic relationships of Pandanaceae can help in the for- mulation of various hypotheses concerning breed- ing system evolution, evolution of different polli- nation syndromes, and their effect on range Volume 77, Number 4 Cox 839 1990 Breeding Systems in Pandanaceae extension and speciation rates. In this light, Pan- ————, B. WaLLacE & I. Baker. 1984. Monoecism danaceae do not appear to be primarily anemoph- ilous and diclinous, as Meeuse (1972) suggested, but rather they spring from vertebrate-pollinated ancestors that developed dioecism in response to destructive pollinators. Subsequently, there have been interactive effects between pollination, breed- ing system evolution, speciosity, and range exten- sion, with only a few changes in inflorescence mor- phology, leading to the evolution of strikingly different pollination syndromes. Obviously this view is limited by the accuracy and resolution of the proposed phylogeny. It is hoped that continued interest in tropical monocotyledons will one day yield persuasive evidence for a sister group to the Pandanaceae. Such a discovery would facilitate rigorous testing of the hypotheses proposed here. LITERATURE CITED Baker, Н. С. & Р. A. Cox. 1984. Further thoughts rd. on islands and dioecism. Ann. Missouri Bot. Ga 71 39. BANDE, M. oe 8 Н. алап. 1986. New thoughts on nities of Viracarpon hexarpermum Sahni from the eie Intertrappean beds of India. Stud. Bot. Hung. 19: 13-22. Beetle pollination of P lanthus bipartitus (Cyclanthaceae). Amer. J. Bot. 69: 1074- 1081. Burck, W. 1892. Wandelingen door den ae Tuin te Buitenzorg. Landsdrukkerij, Batavi CAMMERLOHER, H. 23. Over eenige Minder Bekende Lukmiddelen van Bluemen. Trop. Natuur 12: l. CAMPBELL, D. H. 1911. The т of Pandanus. Ann. Bot. — 25: 77 CHARNOV, E. L The ы of Sex Allocation. Princeton i dessus Princeton, New Jerse CHEAH, C. Н. & B. C. Sto 1975. Embryo sac cand microsporangium development in Pandanus (Pan- 8-238. danaceae: new findings in the genus Freycinetia. Biotropica 13: 195-198 . Vertebrate pollination and the main- tenance of dioecism in Freycinetia. Amer. өе 120: 65-80. 1983. Extinction of the Hawaiian avifauna —— in a change of Popp for the ieie, Frey- 'inetia arborea. Oikos 41: 105-199. 19 Chiropterophily and ornithophily in Freycinetia i in Samoa. Pl. Evol. & Syst. 144: 277- 985. Islands and dioecism: insights from the reproductive ecology of Pandanus tectorius in Poly- nesia. Pp. 355-368 in J. White (editor), Studies on Plant Demography: A Festschrift for John L. Harper. Academic Press, London 1988 Monomorphic and dimorphic sexual strategies: a modular approach. Pp. 80-97 in J. Lovett Doust & L. Lovett Doust "oed Plant Re- productive Ecology: Strategies and Patterns. Oxford Univ. Press, Oxford. n the genus Freycinetia. Biotropica 16: 313-314 CRONQUIST, А. 1968. The Evolution and Classification of Flowering Plants. Houghton Mifflin, Boston DAHLGREN, R. M. T. & H. T. eed A Comparative Study. Academic Press, Lon "х ET SMUSSEN. 1983. Monocotyledon Evol. Biol. 16: 255-389 ‚ H. T. CLIFFORD & Р. Е. Yeo. 1985. The Families of Monocotyledons. Springer-Verlag, Berlin. DANIEL, M. 1973. Seasonal diet of the ship rat (Rattus rattus) in the lowland fo ys of New Zealand. Proc New gd Ecol. Soc. 20: 21-30. 91 Feeding " the short- tailed bat (Mys- | evolution. ta y nectar. New Zealand J. Zool. 3: 391- ENGLER, А. & E. Сис. 1924. Syllabus der Pflanzen- familien. Verlag von Grebuder Borntreager, Berlin. FAGERLIND, F. 1940. Stemplebau und ки ош wicklung bei na Pandanazeen. Ann. Jard. Buitenzorg 49: 55-78. ка F. R. Miscellaneous notes on the flora Aldabra and neighbouring islands: V. Pandanus tectorius Parkinson sensu latissimo (Pandanaceae). Kew Bull. 31: 837-840. Gunn, C. R. & J. V. DENNIS. 1976. World Guide to Tropical Drift Seeds and Fruits. Quadrangle, New ork Times Press, New York. Guppy, Н.В. . Observations of a Naturalist in the Pacific between 1896 and 1899. Macmillan, London. HARLING, С. BP aug of the Cyclanthaceae. Acta Horti Mere: 18: 1-428 Observation of some apes in the Dute h East ha, Dansk. Bot. Ark. HemsLeEY, W. B. 1893. On the flora " E P Or Friendly Islands. J. Linn. Soc., Bot. HuTCHINSON, J. 1973. The Families of es Plants, 3rd edition. Clarendon, Oxford. HuvNH, 1980. La morphologie du pollen de Pandañus subg. Vinsonia (Pandanaceae) et sa sig- ош taxonomique. Pollen & Spores 222: 173- E. n M. 1978. Some Maestrichian palynomorphs and their phytogeographical and paleoecological im- plications. Palynolo -38. KAPLAN 973. The monocotyledons: their evolution and comparative biology, III: the problem of leaf morphology and evolution in the monocotyledons. Quart. Rev. Biol. 48: 437-457. KERNER VON MARILAUN, A. 1895. The Natural History of Plants, Volume П. Gersham, London. 1904. К W. Oliver, pica KNUTH, Р. & E. LOE 1904. Handbuch der Blüten- biologie, toris 3. Engelmann, Leipzi Kunz, 5. 67. igre the Indian a and their allies. J. Bot -106, 125-136. LEE, M. A. B. 1985. 2 fie of Pandanus tec- torius by the land crab Cardisoma carnifex. Oikos 9-173 Seed and seedling олю! in Pan- danus tectorius. Biotropica 21: 1754. LINNAEUS, C. Genera ad Pus Quinta. Laurentii Salvii, Homiae. MARTELLI, U. amoan Pandanaceae. Occ. Pap. -24. Bernice Pauahi, Bishop Mus. 10(13): 1 840 Annals of the Missouri Botanical Garden MAYNARD sea J. 1982. Evolution and the Theory s. Cambridge Univ. Press, Cambridge MEEUSE, A. р. J. 1965. Angiosperms— past and pres- ent. Advancing Frontiers Pl. Sci. 11: 170-182. 66. Fundamentals of Phytomorphology. Ronald Pess, New pu k ———. 1972. Palm and pandan pollination: primary ыш. ог een entomophily ? The Botanique Merrit, E. D. & L. M. Perry. 1939. On the brass жес dE of Pandanaceae from New Guinea. J. Ar- 1019 Arbor. 20: 139-186. Mois. J. 1981. Fossil em records of extant an- дин, Bot. Rev. -146. NAMBUDIRI, E. . & W. D. Тїрк. 1978. ы affinities of Viracarpoon Sahni from the deccani bus flora of India. ЕИ 6 Abt. В: 30-43. NIKLAS, K. Sam and airflow ‘say oa around conifer ovulate cones. Science 217: 442- NoRTH, C. A. & A. J. Willis. 1971. Contributions to the anatomy ш Sararanga (Pandanaceae). J. Lin Soc., Bot. 64: 41 ge PERKINS, R. С. auna Hawaiiensis, Volume B Part IV. Ee Cambridge Univ. Press, Cam- idge. aa H. H. 1987. Monoecy and sex-changes in Freycinetia EAE Ann. Missouri Bot. Gard P PURSEGLOVE, J. 1972. ойый Crops: Monocotyledons. А . The Classification of Flowering Plants, 2nd edition, Volume 1. Cambridge Univ. Press, . 1930. Dispersal of Plants Throughout the World. Reeve, London SATTLER, R. 1968. A technique for i of floral development. Canad. J 720-7 SMITH, A. C Flora Vitiensis Nova: k n Flora of Fij. Pacific Tropical Botanical Garden, Lawai, auai. STAPF, O. 1896. On the structure of the female flower and fruit of Sararanga sinuosa Hemsl. (Pandana- 2 iig J. Linn. Soc., Sr. Jonn, H. 1 Revision of the genus Pandanus Strickman, Part 1. Key to the sections. Pacific Sci. 14: 224- Le Distribution mondiale du Genre Pan danus Гети new series, 5(3): 307-308, Plates я : 2g ws of the genus Pandanus Stick- man, Part 42. Pandanus tectorius Parkins. ex and Pandan odoratissimus L.f. Pacific Bu 33: 395-4 А TU Revision of the genus Pandanus Stick- man, Part 43. Pandanus in Tahiti. Pacific Sci. 33: STONE, B. C. The genus Sararanga (Pandana- iid ge 13: 212-224. 967. Materials for a monograph of Frey- cinetia ness uo cue Gard. Bull. Straits Set- tlem uk for a monograph of Frey- cinetia Gaud, IV. Subdivisions of the genus, with fifteen new sections. Blumea 16: 361-372. 9 Morphological studies in Pandana- ceae. I. Staminodia and pistillodia of Pandanus and their hypothetical significance. Phytomorphology 18: 509. 70. Malayan climbing ber р genus Freycinetia in Malaya. Malay Nat. J. 23: 84-91. 1971. Material: for a monography i Frey- cintia Gaud, XII. Fedn. Mus. J. (Kuala Lumpur) N.S. for 1968, 13: 155-165. 72a. A reconsideration of the evolutionary status of the family Pandanaceae and its significance in monocotyledon phylogeny. Quart. Rev. Biol. 47: 34-4 1972b. The genus Pandanus in the Solomon йн ds with notes оп ы regions (Part 1). Ма- каш Sci. (А): 93-132. 2с. үз pe a p A of Frey- cinetia vend Mace ж he Sumatran species. Fed. Mus. J. 15: 1976a. The De itd of the New Heb- rides with an essay on ЭУЕ variation іп Pan- danus tectorius. Kew Bull. 31: 47-70. . 1976b. On the с of Pandanus (Pandanaceae). Compte Rendu du Soc. Biogeogr. Paris, Seance 458: 69- 1982a. New Guinea Pandanaceae: first ap- proach to ecolo Volume 42). W. Junk, The Hague. . 1982b. The uie species of Freycinetia (Pandanaceae). Brunonia 5: 79-94. 1982c. Pandanus к Parkins. in Aus- n. Soc., Bot. 85: ake a conservative view. J. Lin 133-146. 1 Revisio каран Рагї П. Рап- danus subgenus Rykia (De urz (sections Gres- 2 R Pt sis, and Е мо. Fedn. Mus J. e des P. " 1980. Outline of the flowering plants. Bot. Rev. (Lancaster) 46: 225-359. UnL, №. W. & H. E. Moore, Jr Androecial development i in six polyandrous genera representing five + groups of palms. Ann. Bot. (London) 45: 57- classification of VAN DER Pu, L. 1956. Remarks on pollination by bats in the genera Freycinetia, Duabanga, and Haplo- phragma and on rr in general. Acta Bot. Neerl. 5: 13 тш R. & Р. WIEHE. n The genus Pandanus n the Mascarene Islands. J. Linn. Soc., Bot. 55: 1- WARBURG, О. 1900. Pandanaceae. In A. Engler & K. Prantl a Das Pflanzenreich 3, IV. 9: 1-99, Wuite, J. 1979. e plant a a metapopulation. Ann. Rev. Eed. Syst. 10: 109-145. NOTES WEBERBAUERA PERFOR ATA (BRASSICACEAE), A NEW SPECIES FROM PERU Weberbauera Gilg 8 Muschler is a South Amer- ican genus of 16 species distributed along the Andes from northern Peru south into Patagonia. The ge- nus has been recently revised (Al-Shehbaz, 1990), and its boundaries were then expanded only slightly to include three artificially delimited segregates that Schulz (1924) proposed as Alpaminia O. E. Schulz Pelagatia O. E. Schulz, and Stenodraba O. E Schulz. As I delimited the genus in 1990, Weber- bauera is a well-defined, monophyletic genus some- y what related to the South American Englerocharis Muschler (2 species), Onuris Philippi (5 species), and Sarcodraba Gilg 8 Muschler (3 species). We- berbauera perforata Al-Shehbaz was not included in my recent revision of the genus because the collections on which this new species is based were misidentified as Draba L. Weberbauera perforata Al-Shehbaz, sp. nov. TYPE: Peru. Depto. Cuzco: Nevado Auzangate (as Ausangate), [13°48’S—71°14’W], 4,800 m, 17 May 1957, R. Hirsch P1255 (holo- type, GH). Figure 1. a perenna caespitosa ins = basalia ro- ile pen езе petiolata, supern arginem seri- ceis m longa, 2-6 mm lata; aan кз lutea, 2-2.5 mm longa; petala late ок i vel suborbiculata, unguiculata, lutea, 2.5-3 mm longa; siliqua oblonga, 4- 7 1.5-2 mm lata; septum incompletum dim- idium Cespitose, perennial herbs forming dense cush- ions. Caudex simple or few-branched, densely cov- ered with persistent petioles of previous years. Flowering stems slender, decumbent, falling off at end of growing season. Basal leaves petiolate, ro- sulate, spatulate, 1-3 cm long, 2-6 mm wide, obtuse to rounded at apex, entire, densely sericeous on the upper surface with trichomes to 2 mm long, ciliate at margin, glabrous beneath; petioles 6-16 mm long, persistent, stramineous, becoming pa- pery. Cauline leaves linear to narrowly oblanceo- late, pubescent on the upper surface, 4-12 mm long, 0.5-1 mm wide. Inflorescences few-flowered, corymbose racemes, only slightly elongated in fruit. Sepals oblong, yellow, erect, nonsaccate, glabrous, mm long, ca. 1.5 mm wide. Petals broadly spatulate to suborbicular, yellow, clawed, 2.5-3 mm long, 1.6-2 mm wide; claws erect, 1.2-1.8 mm long. Stamens tetradynamous; filaments 1.8- 2.2 mm long; anthers oblong, 0.6-0.8 mm long. Nectar glands confluent, subtending the bases of all filaments. Fruiting pedicels divaricate-ascend- ing, straight, glabrous or sparsely pubescent, 3-5 mm long. Fruits oblong, glabrous, terete, 4-7 mm long, 1.5-2 mm wide; septum with a central per- foration; gynophore 0.3-0.4 mm long; style 0.2- 3 mm long; stigma entire. Seeds oblong, ca. 1 mm long; cotyledons incumbent. Additional specimen examined. PERU. DEPTO. CUZCO: Ocangate, [13?38'S- 71?24"W], 4,700 m, 11 May 1957, R. Hirsch P1215 (GH). Weberbauera perforata is readily distinguished from all other species of Weberbauera in having yellow flowers, incomplete septa, and densely se- riceous upper surface of basal leaves. The other species of Weberbauera have white flowers, com- plete septa, and differently pubescent basal leaves. I am most grateful to George K. Rogers for his critical review of the manuscript. LITERATURE CITED AL-SHEHBAZ, I. . A revision of pony Вака), J. Arnold Arbor. 71: 221- ScHuLz, O. E. ruciferae— (e a e. or i m ue Phlanzenreich IV 105(Heft 86): 1- —Ihsan A. Al-Shehbaz, Missouri Botanical Gar- den, Р.О. Box 299, St. Louis, Missouri 63166, U.S.A ANN. Missour! Bor. GARD. 77: 841-842. 1990. 842 Annals of the Missouri Botanical Garden 4.1. ATA “ae [ NE rr es. Ен ae | FIGURE 1. Weberbauera perforata. — а. Plant.—b. Basal leaf, upper surface. — c. Cauline leaf, upper surface. — d. Sepal. —e. Petal. —f. Stamen.—g. Receptacle with nectar glands. —h. Fruit.—i. Replum and septum. Scales a — l cm, b-i = 1 mm. Drawings a-c, h, i, from holotype; d-g from paratype. BRAYOPSIS GAMOSEPALA (BRASSICACEAE), A REMARKABLE NEW SPECIES WITH GAMOSEPALOUS CALYX During the preparation of a monograph on the South American species of Draba L., a new Bo- livian species of Brayopsis Gilg & Muschler was discovered. The species, hereafter B. gamosepala Al-Shehbaz, is unique among all of the New World members of the mustard family (Brassicaceae) in having typically united instead of distinct sepals. True and apparent gamosepaly in the Brassi- caceae have been previously reported. Hedge (in Hedge & Rechinger, 1968) described Sisymbrium gamosepalum Hedge from Afghanistan, but the species has connivent instead of truly connate se- pals. Haussknecht (1897) established the genus Gamosepalum Hausskn. as distinct from Alyssum L. solely on his assumption that Gamosepalum has a gamosepalous calyx. Schulz (1936) retained Gamosepalum as a distinct genus, but as Dudley (1964) showed, the calyx in this group consists of distinct sepals that falsely appear to be connate because of interlocking idumentum. The only other gamosepalous species of the Brassicaceae was de- scribed by Pampanini (1926) as Desideria mirabi- lis Pamp. The species, which Jafri (1973) justifi- ably transferred to Christolea Camb., to Karakorum, Pakistan. Evidently, gamosepalous calyces evolved independently in Brayopsis and Christolea. is endemic Brayopsis gamosepala Al-Shehbaz, sp. nov. TYPE: Bolivia: La Paz, Bautista Naavedra, “Kalkrippen etwa 4 km 5 K der Strasse nach Escoma, Felsspalten, Geroll, y cm Exp. W; Neigung 15-45 Grad," 4,460 m, 10 Jan. 1980, J. Krach 7640 (holotype, MO; isotypes, GH, M). Figure 1. Herba perenna caespitosa scaposa nana; folia basalia rosulata densissime conferta, E e vel oblonga, 6- 8(-15) mm longa, (1.2-)2-3 mm lata, supra et ad mar- ginem longe villosa; calyx ela (3-)3.5-4.5(- 5) mm longus, sepala connata; petala lineares, 6-7 mm longa, 12-1 4C 1.7) mm lata; ovarium pilis appressum dense obtectum; siliquae 7-11 mm longae; styli 1.6-2 mm longi; semina 1 х 0.7 mm Cespitose, scapose, small, perennial herbs form- ing dense cushions. Caudex usually branched, cov- ered with persistent leaf bases of previous years, the branches terminated in rosettes. Basal leaves petiolate, rosulate, 6-8(-15) mm long, (1.2-)2-3 mm wide; blades elliptic or sometimes oblong to ovate, entire, obtuse to subacute, the upper surface and margin sparsely to densely villous with simple trichomes to 2.5 mm long, the lower surface gla- brous. Peduncles 1-flowered, sparsely pubescent, 6-10(-15) mm long in fruit. Calyx campanulate, gamosepalous, (3-)3.5-4.5(-5) mm long, sparsely pubescent with short trichomes; teeth broadly tri- angular, 1-1.8 mm long. Petals linear, 6-7 mm long, 1.2-1.4(-1.7) mm wide, slightly narrowed to base. Stamens tetradynamous, erect, 4-5.5 mm long; anthers narrowly oblong, 1.4-1.8 mm long. Nectar gland ringlike, low, subtending the bases of all filaments. Ovary sessile, densely pubescent with linear, densely pubescent, 7-11 m mm wide; septum complete; style slender, glabrous, 1.6-2 mm long; stigma capitate, entire. Seeds ovate, subbiseriate, ca. 1 х 0.7 mm Brayopsis gamosepala is most closely related to B. alpaminae Gilg € Muschler, a Peruvian species which the former resembles greatly in leaf pubescence, ovary and fruit shape and pubescence, and style length. Brayopsis gamosepala is distin- guished readily by its connate sepals that form a campanulate calyx and larger, only slightly bise- riate seeds. In contrast, B. alpaminae has a po- lysepalous calyx and smaller (0.6-0.8 x 0.4-0.5 mm), typically biseriate seeds. The sepals in the latter species are often connivent, and during fruit maturity they, as in B. gamosepala, separate from the receptacle as a unit. Although the development of a gamosepalous calyx in the Brassicaceae is now known in two species of unrelated genera, the feature alone does not justify the establishment of a new genus. In fact, had it not been for the gamosepalous calyx, Brayopsis gamosepala could have been easily mistaken for B. alpaminae. Brayopsis (six spe- cies) is а well-defined, monophyletic genus most closely related to Eudema Humb. & Bonpl. The taxonomy and the delimitation of the two genera ANN. Missouni Bor. GARD. 77: 843-844. 1990. 844 Annals Missouri Pod Garden ur de ү "| Үү! : D | | | ie y үр bl | f fi FIGURE 1. ys каш gamosepala. —a. Plant.—b. Leaf. —c. Flower.—d. Petal. —e. Stamen.—f. Pistil. Scales = 4 mm; = 1 mm. Drawn from the ernie’ have been recently treated (Al-Shehbaz, 1989, 1990) I am grateful to Hannes Hertel, the director of Botanische Staatssammlung, Munchen, for per- mitting the deposition of a few rosettes of the type collection at MO and GH. I am also thankful to George K. Rogers for his critical review of the manuscript. LITERATURE CITED AL-SHEHBAZ, I. A. 1989. The South American genera Brayopsis and Englerocharis (Brassicaceae). Nord. J. Bot. 8: 619-625. 1990. Generic limits and taxonomy of Bray ч in Eudema (Brassicaceae). J. Arnold. Arbor. dde 09. DupLEv, Т. R. 1964. Synopsis of the genus Alyssum. J. Arnold Arbor. 45: 358-373 Начо, С. 1897. Gamosepalum Hskn. gen. nov. Mitt. Thür. Bot. Ver. 11: 73-76. HEDGE, I. & К. Н. RECHINGER. 1968. Cruciferae. Fl. Iranica 57: 1-372. JAFRI, S. M. H Brassicaceae. In E. Nasir & S. I. Ali (editors), Fl. West Pakistan 55: 1-308. PAMPANINI, R. 26. Desideria mirabilis Pamp., gen. et sp. nov., nuova crucifera anomala de Caracorum (Asia centrale). Bol. Soc. Bot. Ital. 1926: 107-111. ScHuLz, О. E. 1936. Cruciferae. In Н. Harms жез, Die Natiirlichen Pflanzenfamilien, ed. 2. 17B: 2 58. —Ihsan A. Al-Shehbaz, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166, U.S.A A NEW SPECIES OF POLYPODIUM (POLYPODIACEAE) AND TWO NEW SPECIES OF HYPOLEPIS (DENNSTAEDTIACEAE) FROM MESOAMERICA The following new species are here described as a result of my work on the pteridophyte volume for the Flora Mesoamericana project. Polypodium alansmithii R. C. Moran, sp. nov. TYPE: Mexico. Chiapas: 22 km from San Cris- tobal de las Casas on the road to Tenejapa, then right 3 km on the road to Matzam, 2,400 т, Huft et al. 2173 (holotype, MO; isotype, UC not seen). Figure 1. Rhizoma 2-4 mm latum latissime repens, squamis 2- tiolois laminas circa hung b (7-)10-18(-22) x 4.5-15 cm deltatis етеген, vel oblongis pinnatisectis vel pinnatis in parte basale; pinnis 3-6(-8) mm latis integris vel raro crenatis, 8-16 utroque rhachidis latere. Epiphytic, rarely epipetric or terrestrial; rhi- zome 2-4 mm wide, long-creeping, the scales 2- 4 mm, lanceolate, bicolorous, black medially with pale brown margins, clathrate or subclathrate me- dially, appressed, denticulate to erose; petiole ca. equaling the lamina, dark brown, nonalate; lamina (7-)10-18(-22) x 4.5-15 cm, broadly deltate to broadly oblong, pinnatisect, sparsely scaly on both surfaces, the scales usually circular or ovate, oc- casionally with an acicular apex, subentire to den- ticulate; pinnae 2.5-8 x 0.3-0.6(-0.8) cm, pairs 8-16, entire to crenate, the distal ones ascending; veins free, obscure; sori round, not embossed adax- ially. Selected specimens examined. MEXICO. CHIAPAS: 11 i ). GUATEMALA. E of Purulha, 1,500 ud forest of Maria Tecún, Molina R. et al. 30394 (MO). HONDURAS. COPÁN: Quebrada La Hondura, 2 km NE de Santa Rosa ANN. de Copan, 1,000 m, Mejia 43 (MO). FRANCISCO MORAZAN: along road to Parque Nacional La Tigra, 22-25 km NE of Tegucigalpa, 1,850-2,125 m, Croat & Hannon 63996 (MO); above Rosario Mine, San Juancito Mts., 1,800 m Morton 7417 (MO); La Tigra, 1,600 m, Regina A. 32 (MO). INTIBUCÁ; Cerro San Cristóbal, La Esperanza, 2,000 m, Mejía O. 100 (MO). EL SALVADOR. SANTA ANA: Cerro Verde, 1,900 m, Siu s.n. (MO). SONSONATE: Cerro Verde cloud forest, 2,000 m, Seiler 1474 (MO). NICARAGUA. ESTELÍ: Cumbre del Cerro Quiabú, 7 km W de Esteli, 1,600 m, Neill 1225 (MO). MADRIZ: Cerro Pataste, ca. 20 km SW de Ciudad Somoto, 1,700 m, Grijalva 913 (MO). JINOTEGA: along a 3, ca. 1 km NW of La Fundadora entrance, unnamed peak ca. 500 m hwy., 1,450-1,520 m, Pn 20369 (MO); carretera Matagalpa-Jinotega, La Fundadora, 1,400 m, Moreno 1865 (MO). MATAGALPA: El Arenal, 500 m denio el camino de Aranjuez, 1,400 m, Moreno 9590 (МО); Fuente Pura, km 142, carretera Matagalpa-Jinotega, 1,400- 1,450 m, Moreno 16997 (MO). This species is named for Alan R. Smith (UC), who has given me much help and encouragement with the Flora Mesoamericana project. The species occurs from southern Mexico (Guerrero, Oaxaca, and Chiapas) to Nicaragua, from 1,500 to 3,600 m. In the Floras of the region, it has previously been included in P. montigenum, which is closely related but differs by the characteristics given in the following key: m Lamina (12-)22-35(-50) x (4-)14-18 cm, l-pinnate throughout; pinnae 7-15 mm wide, pairs 11-21, the distal ones perpendicular to the rachis or nearly so; petiole 4-34 the length of the lamina; Costa Rica, Panama ........................ olypodium montigenum . Lamina (7-)10-18(-22) x 4.5-15 cm, pin- natisect throughout or 1-pinnate basally; pinnae -6(-8) mm wide, pairs 8-16, the distal ones ascending; petiole ca. equaling the length of the lamina; southern Mexico to Nicaragua Polypodium alansmithü — p Since most of the distinguishing characteristics deal with leaf size and dissection, it could be argued that Polypodium alansmithii is merely a smaller, less-cut version of P. montigenum. The two com- Missouni Вот. GarD. 77: 845-850. 1990. 846 Annals of the Missouri Botanical Garden Wd GI Volume 77, Number 4 1990 Notes 847 pletely differ in range, however, which supports their separation. Actually, P. montigenum is most closely related to Р. plebeium Schldl. & Cham., which occurs from Mexico to Panama. Polypo- dium montigenum differs from P. plebeium only by its appressed, rather than spreading, rhizome scales. Polypodium montigenum is redescribed below with a list of specimens examined because it has been too broadly circumscribed in the past. It is endemic to the mountains of Costa Rica and Pan- ama, where it grows in cloud forests from 1,800 to 3,100 m. Polypodium montigenum Maxon, Publ. Field Mus. Nat. Hist., Bot. Ser. 17: 306. 1938. TYPE: Costa Rica. Heredia: along the cart road from Vara Blanca (between Poás and Barba volcanoes) to La Concordia, 1,600-1,950 m, Maxon & Harvey 8479 (holotype, US; iso- type, NY not seen). Figure 2 Epiphytic on tree trunks or clambering on sur- rounding vegetation; rhizome 3—6 mm wide, long- creeping, the scales 2-5 mm, linear-lanceolate, bicolorous, black or dark brown medially with nar- row whitish margins, not clathrate medially, ap- pressed, fimbriate to erose; petioles 14-34 as long as the lamina, dark brown, nonalate; lamina (12-) 22-35(-50) x (4—)14—18 cm, lanceolate to nar- rowly oblong, 1-pinnate throughout, sparsely scaly abaxially, the scales circular to ovate, sometimes with an acicular apex, ciliate; pinnae (2.5-)7-10 х (0.3-)0.7-1.5 cm, pairs 11-21, entire or cre- nate or (in large leaves) serrate, the distal ones perpendicular to the rachis or nearly so; veins free, obscure or inde visible; sori round, not embossed adaxia Additional specimens examined. COSTA Rica. ALAJUELA: Angel Falls on road to Puerto Viejo, 5 km N of Vara Blanca, Mickel 3575 (NY, UC). CARTAGO-SAN JOSÉ: Interamerican Hwy., vic. of Villa Mills and 1 km y 1976, Croat 35402 La Georgina y Vara Blanca, 1,950 m (NY); Cerro Chompipe, N de San Rafael, 2,000 m, Lems 6402829 (NY); 1 km W of Vara Blanca on the slope of ү SAN JOSÉ: road from Cartago to San Isidro del General (Pan American Hyw., 21-22 km SE of El Empalme, cloud forest, 2,750 m, ж К. Smith & Béliz 1998 (MO, UC); along Interamer- n Hwy., Cerro de la Muerte, 3,100 m, Hennipman et iral 7124 (МО); S of Cartago, ca. 4 km S of El Empalme near La Chonta, 2,500 m, Lellinger 1580 (MO). PANAMA. enel Ns adalupe-Cerro Punta, Finca J. L. Caballero, 1 km (MO), L of in ee ,200 m, rege) 672 Bugaba, Cerro Punta, 8°52'3№, 82°33’E, 2,200 m, van der Werff & Herrera 6273 (MO, UC МУ The labels on several specimens (e.g., Smith & Béliz 1998) state that the rhizome is scandent on surrounding vegetation, not clinging to trees. As ar as I know, this growth habit has not been previously recorded in Polypodium. Other labels, however, state that the plants were growing as epiphytes on tree trunks. Hypolepis ee R. C. Moran, sp PE: Panama. Chiriqui: along trail between А fork of Rio Palo Alto and Cerro Pate Macho, ca. 6 km NE of Boquete, 8°48'N, 82?23.5'W, 1,600-2,000 m, 6 Feb. 1986, A. К. Smith et al. 2361 (holotype, MO; iso- type, UC). Figure 3d-f. Petiolus brunneus vel pallide brunneus spinosus; lam- inae utrinque pubescentes, pilis bacilliformibus plerumque VH idibus e saepe absentibus. Leaves to ca. 1.5 m(?), continuous in growth, erect; petiole brown throughout or stramineous dis- tally, epipetiolar branches present laterally near the base, usually 1-2; both surfaces of the lamina moderately pubescent along and between the veins with minute, mostly appressed, pale reddish, ba- cilliform hairs, lacking catenate hairs; lamina mar- gins eciliate; costae of the penultimate segments not bordered adaxially by perpendicular, decur- rent, herbaceous wings; rachis not spiny, brown to stramineous, straight (not flexuose) basally, gla- brous; veins not ending in shallow emarginations; indusia ca. 0.1 mm wide, eciliate, green like the lamina or slightly scarious, not or only scarcely <— FIGURE 1. 1474 (МО). — с. Honduras, Espinal 55 (М Fertile leaves of a alansmithii. —a. Honduras, Mindence 31 (МО). —Ь. El Salvador, ги —d. Guatemala, Molina R. et al. 30394 (MO). —e. Honduras, Mej O. 100 (MO). —f. Honduras, Clare 203 (MOS —g. Honduras, Crüz P. 73 (MO). 848 Annals of the Missouri Botanical Garden Volume 77, Number 4 Notes 849 1990 15 cm MOS 1 mm 1 mm FIGURE 3. Hypolepis ditrichomatis (a-c) and Н. trichobacilliformis (d-f).—a. Abaxial surface of tertiary segment. — b. Side view of acicular hyaline hairs between veins (left ч and appressed reddish hairs alo ong veins (right). — с. Basal portion of pinna.—d. Abaxial surface of tertiary segment.—e. Pinna bases. — . Bacilliform hairs. a-c, van UC). der Werff & Herrera 6265 (MO). d, Smith et al. 2361 (MO). e, f, van der Werff & cioe 6472 ( — Fic Fertile ч of Polypodium montigenum.—a. Costa Rica, Hennipman et al. 7124 (MO).—b. (Holotype Ж. Rica, Maxon & Harvey 8497 (05). — c. Costa Кіса, panda 1580 (МО). —d. Panama, van der Werff & Herrera 6273 (MO). 850 Annals of the Missouri Botanical Garden reflexed and apparently absent, or fully reflexed but covering only !4-V of the sorus. Additional specimens examined. Costa Rica. ALAJUELA: Volcán Poás, Stork 2507 (UC); is Vol- cán Poás, 2,200 m, Tonduz 10696 (US). CARTAGO: Santa Clara de Cartago, 1,950 т, Maxon & Harvey 8229 (US); Roble, massif de l'Irazu, 2,000 m, Pittier 4173 (US). HEREDIA: Vara Blanca de Sarapiquí, N slope of с Cordillera, between Poás and Barba volcanoes, 1,615 т, Skutch 3600 (US). $АМ JOSÉ: Cerro ое SW slopes along ridge trail from Canaan to summit, ca. 2,500 m, Evans et al. 55 (US); Cerro de la Muerte, 1 km NW of Villa Mills on Interamerican Hwy., Hotel La Georgina, 2,900 т, Lellinger 861 (US); Cerro de la Muerte, km 173 along the Interamerican Hwy., 0 m, Mickel 3300 (US); Las Nubes, 1,500-1,900 m, ri dley 38807 (US); near Finca La Cima, above Los Lotes, М of El Copey, 2,100-2,400 m, Standley 42695a (US), on Pan American Hwy. above La Georgina, slopes of Cerro de la Muerte, 3,000 m, Stone 2015 CARTAGO/SAN JOSÉ: ca. 3 km SE of Trinidad or 42 km SE of San Isidro de Cartago along the Carretera Inter- americana, 2,600 m, Wilbur 27801 (DUKE). PROVINCE NKNOWN: San Geronimo, 1,500 m, Wercklé 581 (US). PANAMA. CHIRIQUÍ: Distrito Bugaba, Cerro Punta, along os to watershed to Bo ocas del Toro, 8%52'N, 82°33'Е, 2,200 m, van der Werff & Herrera 6472 (MO, UC). ed BOCAS DEL TORO: trail along the continental di- vide to ca. 3 km E of Cerro Pate Macho, cloud forest, 2, 000- 3, 000 m, Smith et al. 2442 (UC); trailside in Quebrada Bajo Grande ca. 2 mi. E of Cerro Punta, 2,300 m, Wilbur et al. 13146 (DUKE). — This species grows in cloud forests, wet forests, clearings, and pastures from 1,600 to 2,200 m. It differs from all other Hypolepis in Mesoamerica by having short, cylindrical, bacilliform hairs on the lamina and medial sori that are not or onl slightly protected by reflexed indusia. The density of the bacilliform hairs may vary from sparse to moderate, even on the same leaf. Larger catenate hairs, which occur on the abaxial surface in a other species of Hypolepis in Mesoamerica, are lacking in this species. The indusia of H. trichobacilliformis are not or only slightly reflexed and texturally modified. In those cases where the indusia are fully reflexed, they cover 14-14 of the sorus. Because the sori are naked or only partially covered, some specimens of this species have been named as H. nuda Mett. ex Kuhn, a species endemic to the Andes of Ven- ezuela. Hypolepis nuda differs by its marginal sori that tend be elongated, short-stalked pinnae, and oval or semicircular segments. jara оом С. Могап, 5р. поу. a. Chiriqui: vicinity of Guada- = Еа Río Chiriquí Viejo and the con- tinental divide, 8*53'N, 82°36'W, 2,200 m, Churchill £ de Nevers 5048 (holotype: MO— 3 sheets). Figure 3a-c. etiolus brunneus vel pallide brunneus; lamina се nc pilis erectis acicularibus hyalinis inter venus praesentibus et pilis (fere) adpressis ela pallide rubris solum secus venas praesentibus praeditae; margine ciliatae. Leaves to ca. 2 m, continuous in growth, scan- dent; petiole brown to tan, spiny, sparsely pubes- cent to glabrescent; both surfaces of the lamina sparsely to moderately pubescent along and be- tween the veins with two kinds of hair, the first erect, acicular, hyaline, and found throughout, the second appressed or nearly so, slender, pale red- dish, and found only on the veins and costae; lamina margin eciliate; costae of the penultimate segments not bordered adaxially by a perpendicular herba- stramin- eous, straight or only slightly flexuose basally, sparsely pubescent to glabrescent; veins not ending in shallow emarginations; indusia 0.2-0.3 mm wide, ceous wing; rachis spiny, brown, tan, or scarious, eciliate. Additional specimens examined. Costa Rica. ALAJUELA: slope of Volcán Poás at bridge just below Ha cienda El Tirol, 2,000 m, Mickel 2000 (UC). CARTAGO: Tapanti, ca. 15 km S of Paraiso, 1,150 m, Mickel 2306 (US) along Interamerican Hwy. between Cartago and San Isidro de El General, Scamman 7028 (GH, US, UC). pastured slopes of Volcán Barba, ca. Blanca on the N slopes, 1,750 m, Wi m & Stone 10085 (DUKE, US). san José: 25 km М of San Isidro de El General along the Interamerican Hwy., 1,800 m, Burger & Liesner 7038 (F, СН); slopes of Cordillera ee near La Division N of San Isidro de El Gene dond forest, 2, m, W. г (MO, UC), 6265 (GH, MO, UC, US); Guadalupe, Cerro Punta, finca Maduro, 2,000 m, Caballero 120 5 PMA); Cerro Colorado, along road to copper min Pz e Fe beyond turnoff to Escopeta, 1,390 m, Croat 37306 (MO). This species grows in cloud forests from 1,200 to 2,900 m. It can be distinguished from all other Hypolepis in Mesoamerica by the two kinds of hairs on the lamina (see description). The slender, pale reddish hairs along the axes and veins are similar to those in H. parallelogramma Kunze, a South American species. That species, however, lacks the short, erect, hyaline, acicular hairs found on H. ditrichomatis and has a different cutting of the lamina. — Robbin C. Moran, Missouri Botanical Garden, Р.О. Box 299, St. Louis, Missouri 63166, U.S.A. A NECESSARY NEW COMBINATION IN DUROIA (RUBIACEAE) A new combination in Duroia is required in anticipation of the forthcoming Flora of the Ven- ezuelan Guayana (Steyermark & collaborators, in prep.) due to the earlier publication by Ladbrook (July 1920) of a specific epithet in Coupoui, C. micrantha, Ladbrook’s epithet was treated by Steyermark (1965, 1974) as a synonym of Duroia sprucei Rusby (Dec. 1920). Unfortunately, Rus- by’s name must be replaced by the new combi- nation proposed below. The July 1920 fascicle [ No. 691] of the Journal of Botany, British and For- eign, which included Ladbrook’s article, was date stamped by the Missouri Botanical Garden Library as received “JUL 21 1920.” The date of issue of Rusby’s work, as given on the cover and confirmed elsewhere (Stafleu & Cowan, 1983), is “December 20, 1920.” In the protologue of D. sprucei, Rusby (1920) cited Rusby & Squires 171 and 172 and indicated that they are ““the same as Spruce’s 3624.” There- fore, these three collections are syntypes of the name. Steyermark (1965) declared the “holotype” of D. sprucei to be the NY specimen of Rusby & Squires 172 with isotypes at F and US, and the NY specimen of Rusby & Squires 171 to be a syntype. Steyermark clearly meant to lectotypify D. sprucei because he identified Rusby & Squires 171 as a syntype. Therefore, Rusby & Squires 172 (NY) is the lectotype of D. sprucei, and du- plicates of is collection (Е, MO, US [3 sheets J) are its isolectotype Steyermark labeled the NY specimen of Rusby & Squires 171 as “holotype” of D. sprucei (IDC microfiche NY 972-C-4). Possibly there was an error in preparation of the manuscript or in pub- lication causing the numbers 171 and 172 to be switched. The collection at US contains three du- plicates of Rusby & Squires 172 and no duplicate of 171 (D. Nicolson, pers. comm.). Therefore, in accordance with Steyermark's published state- ments, the numbers are correct as published. Stey- ermark either changed his mind after annotating 171 at NY, perhaps deciding at the last minute to use a specimen from the collection with more du- plicates as the lectotype, or placed his annotation label on the wrong specimen by accident. The conclusions drawn herein do not conflict with Ar- ticle 8 of the Code (Greuter et al., 1988). Duroia micrantha (Ladbr.) Zarucchi & Kirk- bride, comb. nov. Based on Coupoui micran- tha Ladbr., J. Bot. 58(691): 176-177. 1920 [ante 21 July 1920]. TYPE: “Guiana,” Martin s.n. (holotype, BM). Duroia sprucei Rusby, Descr. S. Amer. Pl. 133. 1920 [20 Dec. 1920]. TYPES sera. 1965): Ven- Rusby & оо 172 Pee ype, NY not seen; i rig US-1158548, US- es 171 (lectoparatype, он NY 972-С-4 with annotation as “Holotype””]); Spruce 3624 (lecto- paratypes not seen). — According to Steyermark (1974), the species is widely distributed in northern South America in- cluding the Rio Orinoco drainage of Venezuela, the Hios Negro and Amazonas basins in Brazil, Colom- bia, and Peru, and Rupununi River in Guyana. Herbarium specimens also indicate its occurrence in Surinam. An excellent illustration of the species is given in Steyermark (1974, p. 691, fig. 109). We thank Gerrit Davidse and Dan Nicolson for useful comments on the manuscript. LITERATURE CITED GREUTER, W. ET AL. 1988. и Code of Во- tanical Nomenclature. Regnu 18. LADBROOK, J. 1920. A new а of Coupoui. J. Bot. 58(691): 176-177. Коѕвү, H. Н. 1920. Descriptions of Three Hundred ew Species of South American Plants. Published by the author. STAFLEU, F. A. & R. S. Cow 1983. Taxonomic Literature, 2nd edition, Volume 4. P-Sak. Regnum eg. STEYERMARK, J. A. 1965. Duroia [Rubiaceae]. Pp. 198- l in В. Maguire & collaborators, The Botany of the Guayana rad VI. Mem. New York Bot. c 13(3): 1-285. Duroia. Pp. 679-703 in Fl. de Ven "ace = Rohres, Part 2. Edición Especial del In- stituto Botánico, Caracas —James L. Zarucchi, Missouri Botanical Gar- den, P.O. Box 299, St. Louis, Missouri 63166, U.S.A.; and Joseph H. Kirkbride, Jr., USDA, ARS, Systematic Botany Laboratory, Building 265, BARC-East, Beltsville, Maryland 20705, U.S.A. ANN. Missouni Вот. GARD. 77: 851. 1990. 852 Annals of the Missouri Botanical Garden A NEW CLOUD-FOREST ERYNGIUM FROM CHIAPAS, MEXICO A new polyploid species, Eryngium strotheri, L. Strother s.n. (C-2189) (holotype, UC; iso- is described from the high-montane cloud forest of types to be distributed to principal herbaria). Chiapas, southern Mexico. Figure 1. . . Plantae crassae caulibus florentibus sin ularibus foliosis; Eryngium strotheri Constance & Affolter, sp. folia v T eolata nov. TYPE: Mexico. Chiapas: summit of Cerro parallelo-nervata spinoso-lobata lobis cum spinula acces- Mozotal, E Sierra Madre, 15 Oct. 1980, John soria prominente munitis; reani viridis cymoso- FIGURE 1. Eryngium strotheri. —A. Habit. —B. Detail of basal foliage leaf.—C. Head, with involucre. — D. External view of mericarp showing sepals and fruit squamae. (All from type collection.) ANN. MISSOURI Вот. Garb. 77: 852-853. 1990. Volume 77, Number 4 1990 Notes 853 paniculata; capitula subnumerosa magna viridi-alba hemi- spherico-ovoidea glabra; bracteae involucrales distinctae biseriatae lateraliter imbricatae integrae vel spinoso-den- tatae virides capitulo longiores; fructus cuneato-ovoideus squamatus alatus dorsi superficie nuda Plants perennial, stout, erect, 0.8-1.2 m tall, from a short vertical caudex bearing fascicled fi- brous roots, the flowering stem solitary, leafy; basal leaves rosulate, spreading-ascending, nondisti- chous, + ensiform, linear-lanceolate, 30-40 cm long, 2-5 cm broad (including lobes, 0.5-1 cm without lobes), densely spinose-lobed, the lobes lin- ear to linear-lanceolate, spine-tipped, 1-2 cm long, spreading, mostly with an auxiliary axillary spinule, some as large as the lobes and giving a double- lobed appearance, the blade rigid-attenuate at apex, strongly caniculate, slightly tapering at base into a naked sheath no broader than the blade and 3- 4 cm long, deep green, the venation strictly parallel except for veins entering the marginal lobes; cau- line leaves similar to the basal, lanceolate to ovate, very coarsely and densely spinose-lobed with larger lobes, alternate below, whorled or opposite beneath the terminal inflorescence, subamplexicaul, spread- ing to reflexed; inflorescence paniculately cymose, ending in a large terminal cluster of ca. 5 basically trichotomous flowering branches surrounding and usually exeeding the terminal peduncle, without fertile axillary branches (?), conspicuously brac- teate; heads greenish white, hemispheric-ovoid, 15- 20 cm diam., pedunculate, many-flowered; in- volucral bracts 12-30, linear-lanceolate, 2.5-4.5 cm long, 2-7 mm broad, pungent, definitely in more than one series and imbricated laterally, en- tire or the larger outer ones with 2—5 spinose teeth, green on both surfaces, spreading, exceeding the heads; bractlets lanceolate-subulate, ca. 5 mm long, exceeding the fruit, a coma lacking; sepals broadly ovate, acute, apiculate, 1.5-2 mm long; petals ob- long, ca. 1.5 mm long, the inflexed apex about equaling the limb, entire (?); styles 2-2.5 mm long, longer than the calyx; fruit cuneate ovoid, 4—5 mm long (including calyx lobes), З mm broad, compressed dorsally, provided with a row of flat- tened ovate-lanceolate scales below the calyx, the marginal wings lobed, the dorsal surface naked; chromosome number n = 24; cotyledons oval, 3- 5 mm long. atype. MEXICO. CHIAPAS: steep slope, e dod peo with Quercus, Pinus, Abies, Drimys, Pho- tinia, Clethra, Cornus, and Symplocus on the N & W slopes of the Cerro Mozotal below the microwave tower along the road from Huixtla to El Porvenir and Siltepec, Mpio. Motozintla, 3,000 m, 28 June 1972, D. E. Breed- love 25,829 (DS Clearly a member of sect. Spinescentia, Eryn- gium strotheri appears to combine the conspicuous involucre of E. involucratum Coult. & Rose with the branching inflorescence of E. tzeltal Constance and the foliage of E. guatemalense Hemsley. In- deed, the Strother material, which has been grown at Berkeley for seven years, was obtained in a generous attempt to fulfill the senior author's re- quest for “seed” of E. tzeltal. The species of this group are all polyploid (4—9-ploid) so far as known and also strikingly variable. They may likely be capable of exchanging genes or even genomes at a polyploid level, which could account for the vari- ability and their taxonomic complexity. The intri- cate interrelationships require further study in the field and garden. The illustration was designed and executed by Charlotte Mentges Hannan. LITERATURE CITED Worrr, H. 1913. /n Engler, Das Pflanzenr. 61(4228): 106-271. —Lincoln Constance, University Herbarium & Department of Integrative Biology, and James Affolter, Botanical Garden, University of Cal- ifornia, Berkeley, California 94720, U.S.A. Pres- ent address of J. Affolter: Director of Plantations, Cornell University, Ithaca, New York 14853, U.S.A VALIDATION OF CAESALPINIA SUBGENUS MEZONEURON (DESF.) VIDAL AND NEW COMBINATIONS IN CAESALPINIA FOR TWO SPECIES OF MEZONEURON FROM AFRICA While preparing a manuscript on fossil Caesal- pinia (Mezoneuron) fruits from the Tertiary of North America (Herendeen & Dilcher, in press) it became apparent that new combinations in Caesal- pinia have not been published for two Mezoneuron species from Africa. Mezoneuron Desf. was distin- guished from Caesalpinia on the basis of its in- dehiscent winged fruits (Brenan, 1963; Hattink, 1974). However, Mezoneuron is now included in Caesalpinia due to similarities in floral and veg- etative features (Hattink, 1974; Vidal & Hul Thol, 1976; Polhill & Vidal, 1981). New combinations in Caesalpinia have been published for the Asiatic species of Mezoneuron (Hattink, 1974; Vidal & Hul Thol, 1976). Vidal & Hul Thol (1976) re- garded Mezoneuron as a subgenus of Caesalpinia; however, the subgeneric name was not validly pub- lished by these authors. This paper validates the name С. subg. Mezoneuron (Desf.) Vidal and es- tablishes new combinations in Caesalpinia for two species of Mezoneuron from Africa. Generic limits in the Caesalpinia complex have been difficult to define (Polhill & Vidal, 1981; G. P. Lewis, pers. comm.). Numerous genera were recognized by Britton & Rose (1930) based on carpological differences (Polhill & Vidal, 1981). Recent studies placing more emphasis on floral features have resulted in the abandonment of many of these segregate genera, including Mezoneuron (Gilis & Proctor, 1974; Hattink, 1974; Vidal & Hul Thol, 1976; Polhill & Vidal, 1981). Compar- isons of leaflet epidermal anatomical features in Mezoneuron and 25 American and Asiatic Caesal- pinta species, representing Caesalpinia sens. str. and six segregate genera recognized by Britton & Rose (1930), yield no significant differences be- tween these groups (Herendeen, unpublished). These observations are consistent with those based on comparisons of floral and vegetative features. Subgeneric status for Mezoneuron is justified given ANN. MISSOURI Bor. GARD. 77: the differences between these groups in fruit mor- phology. Although the geographic distribution of C. subg. Mezoneuron is restricted today to Old World trop- ics and subtropics, paleobotanical evidence indi- cates that this group occurred widely across North America during the Tertiary (Herendeen, 1990; Herendeen & Dilcher, in press). The paleobotanical data demonstrate that C. subg. Caesalpinia and C. subg. Mezoneuron were distinct taxa by the Middle Eocene (Herendeen, 1990). Caesalpinia L. subg. Mezoneuron (Desf.) Vidal ex Herendeen & Zarucchi, comb. et stat. nov. Mezoneuron Desf. Mem. Mus. Hist. Nat. 4: 245, 1. 10, 11.(1818) (as Mezonevron). TYPE: M. glabrum Desf. = C. pubescens (Desf.) Hattink. NEW COMBINATIONS Caesalpinia L. C. benthamiana (Baillon) Herendeen & Zaruc- chi, comb. nov. Mezoneuron benthamianum Baillon, Adansonia 6: 196 (1866). TYPE: Af- rica. Senegambia: Heudelot s.n. 1837 (holo- type, P not seen). Specimens examined. AFRICA. GUINEA: J. G. Adams 3382 (MO, P, BARC-fruit). LIBERIA: J. G. Adams 30186 (MO). NIGERIA: R. C. Brown 923 (MO), J. Opayemi s.n., 26 Nov. 1970 (MO). GHANa: 4. A. Enti 1637 (MO). C. angolensis (Welw. ex Oliver) Herendeen & Zarucchi, comb. nov. Mezoneuron angolense Welw. ex Oliver, Fl. Trop. Afr. 2: 261 (1871). TYPE: Africa. Angola: Welwitsch 606 (lec- totype, LISU not seen; isolectotype, BM not seen). ecimens examined. AFRICA. LIBERIA: P. M. Dan inl 112 (MO). TANZANIA & KENYA: O. Flock 552 (MO). 854-855. 1990. Volume 77, Number 4 1990 Notes 855 TANZANIA: Harris et al. s 2621 (MO). UGANDA: Р. . Rwaburindore 896 (MO), P. K. Rwaburindore 1018 MO). ANGOLA: Wi аю 607 (syntype, LISU not seen; isosyntypes, BM, K not seen). Three species of Mezoneuron were recognized by Oliver (1871) from tropical Africa. In addition to the two species discussed above, M. welwit- schianum Oliver was also described. This latter species was transferred to Caesalpinia by Brenan (1963) because the fruit is thickened along the placental suture, not winged as in other Mezoneu- ron species. Fruit morphology suggests that this species is not related to C. subg. Mezoneuron. Based on similarities in vegetative and reproductive morphology, Brenan (1963) suggested a relation- ship between С. welwitschiana and the Asiatic С. tortuosa Roxb. Mezoneuron benthamianum and M. angolense were distinguished by Oliver (1871) on the basis of leaf size, and pinna and leaflet numbers: Pinnae 5-6 pairs, leaflets 10-12 to each coc. . benthamianum Pinnae 8-10 pairs, leaflets 12-18 to each o... M. angolense Based on the specimens studied, it appears that these differences remain valid. In addition, differ- ences between these species in details of leaflet epidermal anatomy further suggest that these are distinct species. Trichomes are more frequent on the abaxial epidermis of C. benthamiana than on C. angolensis, and cutinization of anticlinal walls of epidermal cells is less evident in C. angolensis than in C. benthamiana. The authors thank the curators and staff of various herbaria (BARC, F, MO, US), which have made their collections available for ongoing studies of fossil and extant legumes. Special thanks go to C. R. Gunn and C. P. Lewis for constructive com- ments concerning this paper and the general prob- lem of Caesalpinia delimitation. This work was supported in part by NSF Disssertation Grant BSR 88-00900 to PSH. LITERATURE CITED BRENAN, J. P. M. 1963. Notes on African Caesalpinioi- deae. Kew Bull. 17: 197-214 Britton, N. L. & J. N. Rose. 1930. оо (conclusio). North American Flora 23: 30 GiLLis, W. Т. & С. R. PROCTOR. ao subg. Guilandina in the Bahamas. J. Arnold Arbor. 55: 425-430. HATTINK, T. A. A revision of Malesian Caesal- pinia, including Morir d'a (Leguminosae- Caesal- piniaceae). Reinwardtia 9: 1-69. HERENDEEN, P. S. 1990. Fossil History of the Legu- minosae from the Eocene of Southeastern North America. Ph.D. r Indiana University, Bloomington, India Е D. L. DILCHER. - Caesalpinia subgenus Mezo- n (Leguminosae, с from the Meios of North America. Amer. J. Bot. (in press). OLIVER, D. 1871. Flora of Tropical Africa, Volume Leguminosae to Ficoideae. L. Reeve, L POLHILL, R. M . VIDAL. 1981. Caesalpinieae. Pp. 81-85 iR. M. Polhill & P. H. Raven (editors), Advances in Suspe Systematics. Royal Botanic Gar- dens, Kew VipaL, J. E. € S. Hur THor. 1976. Revision des Caesalpinia asiatiques. Bull. Mus. Hist. Nat. (Paris), ser. 3, 395(Bot. 27): 60-135. — Patrick S. Herendeen, Department of Biology, Indiana University, Bloomington, Indiana 47405, U.S.A.; and James L. Zarucchi, Herbarium, Mis- souri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S.A. CONCEVEIBA AUBLET (EUPHORBIACEAE) NEW TO AFRICA The flora of tropical Africa is fairly well known, and in most areas botanical exploration produces few surprises. One exception is the Atlantic coastal forest of southern Cameroon and western Gabon, which is very rich in species and continues to yield new finds. The flora of this part of Africa has many links with that of South America; a common pattern is for a predominantly neotropical taxon (family, subfamily, tribe, or genus) to have a few repre- sentatives in central Africa (Thorne, 1973). This paper describes an African addition to Con- ceveiba Aublet, a small neotropical genus of about nine species with its center of diversity in the Gui- anas (Jablonski, 1967). The African species is known only from mature forest in the Lope Reserve in the Ogooue valley of central Gabon. The specimens were collected . and B. Reitsma from marke trees during a forest plot enumeration for the New York Botanical Garden. The genus Conceveiba has been broadly defined by Jablonski (1967) to include two segregate gen- era, Conceveibastrum Pax & K. Hoffm. and Ve- concibea Pax & K. Hoffm. He characterized the genus as having a trimerous ovary with three short styles, shortly connate at the base. The male flowers ave numerous stamens, all fertile or with an inner ring of long, plicate staminode The leaves, leaf анааран. pistillate flowers, male inflorescence, male calyx, and individual sta- mens are all typical of the genus. Differences in- clude the fruits, which are slightly lobed instead of trigonous or cylindrical, and the inflorescences, which are arranged monoeciously, while the rest of the genus is reportedly dioecious. Conceveiba africana also has distinctive male flowers. The number of stamens (7—10) is low for the genus, and the lack of an inner ring of staminodes distin- guishes it from Conceveiba s. str. The presence of a pistillode in the male flowers is unique in the genus. In view of the African materials strong overall affinities with Jablonski’s genus, I do not consider the differences great enough to warrant the cre- ation of a new genus. I am therefore including it in Conceveiba s.l. If the narrow view of Conceveiba were adopted, then a new genus to accommodate the African material would probably be needed. The combination of characters found in the new ANN. species is not seen in any one of the three narrow genera. Conceveiba africana D. W. Thomas, sp. nov. TYPE: Gabon. Ogooué-lvindo Province: Lope Reserve, 0°30'5, 11?33'E, Chantier SOFOR- GA inventory, tree # 215, in mature forest, 27 Nov. 1986 (fl, fr), J. M. & B. Reitsma 2618 (holotype, MO; isotypes, NY). Figure 1 res monoeciae ad 19 m altae. Folia coriacea den- tibus glandulosis praedita morphologia respectu variantia: 6 cm longa x 1.8 ad 14 cm lata in petiolis ad 10 cm | longis; maioria elliptica basi acuta; minoria ovata adax ialibus ad juncturam laminae petioloque notata. Inflores- ae, ramis ad 5.5 cm; fk bus praedita. Flos foemineus solitatius bractea lanceola a cum saepe sine ow зы ед sepalis quinque iim зан lanceolatis, 3-4 stigma atibus tribus crassis rubris praedito, зр tunc 8 ad 3. BE mm. Fructus gla- brescentes, 1.1 cm longi, 1.7 cm diam., vix 3-lobati laeves in sicco ик Ye planis praediti i. Medium-sized, much-branched, monoecious tree to 19 m tall, wood white, slash white and brittle, bark pale gray. Leaves coriaceous, 1.8-26 cm long, 1.8-14 cm wide, glabrous except for domatia, thinly pubescent when young; larger leaves alter- nate, elliptic, with an acute to obtuse base, the petiole terete, woody, адан at both ends, to 10 cm long; ed apically, ovate, with a rounded to subcordate ^ue die petiole thickened, 0.5-1.5 cm long; leaf margins thick- ened with glandular teeth; apex of both types acu- minate; secondary nerves 5—14 on a side, depart- ing at 45°-60° from the midrib, with a tuft of hairs in the abaxial axil. Male inflorescences yellowish and ramiflorous, from the axils of fallen leaves, forming pyramidal panicles to 21 х 11 cm, the dry rachis slender and ribbed, the lateral branches often with disc-shaped glands at the base. Cymes very contracted, often 3-flowered; first bract nar- rowly triangular, to 1 mm long, glabrescent; brac- Missouni Вот. GARD. 77: 856-858. 1990. Volume 77, Number 4 1990 Notes 857 CD. 2ст__, ABL_4cem_, Ec O-5em y p,9.25cm , URE l. A-F. Conceveiba africana (Reitsma 2618).— А. Branch, leaves, and infructescence. — B. Male le fl Fi inflorescence. — C. Portion of leaf undersurface. — D. Fruit from above. — E. teoles and bracts of lateral flowers smaller and rounded, pubescent with spreading hairs. Flower buds rounded, apiculate, glabrous, 1 mm diam., the calyx valvate, closed. Male flowers apetalous, the calyx lobed to halfway with 5 triangular, acute sepals; stamens all fertile, 7-10 with long glabrous filaments, straight or slightly folded, anthers short, 4-celled, pubescent pistillode present. Female in- florescence a reddish, simple, terminal, spikelike raceme with basal triangular scales, ribbed when Female flower. — Е. Male flower. dry, shortly pubescent, 8-9 cm long; bracts 2-3 mm long, acute with 2 basal glands; bracteoles resembling bracts, with or without glands. Female flowers borne singly on short stout pedicels, apeta- lous, calyx open in bud, sepals 5, acute, 3-4 mm long, sometimes with a basal gland between two of them; ovary densely appressed-pubescent; stigma a stout column to o 3.5 [nd glabrescent, dehiscent, shallowly 3-lobed, 1.1 cm 858 Annals of the Missouri Botanical Garden high, 1.7 cm diam., the surface smooth with oc- casional flattened protuberances in the dry state, style and sepals persistent, dehiscent, splitting into three 1-seeded portions leaving no column. Seeds about 0.9 cm long, smooth and brown with a con- spicuous triangular hilum Additional ааа examined. GABON. OGOOUE- IVINDO PROVINCE: Lopé Reserve, Chantier LUFTEXFA, 0*30'S, 11°29’ E 30 May 1986 (female fl), J. M. & B. Reitsma 2309 NY. I thank John Dwyer for the Latin description, Mary Bucy for the illustration, and an anonymous reviewer for constructive comments. The study was supported by NSF grant BSR-88600682 to Dun- can Thomas. LITERATURE CITED JABLONSKI, E. 1967. Botany of the Guyana High- lands— Part VII | (Euphorbiaceae) Mem. New York Bot. Gard. 17: 190. THORNE, R. F. 1 23. " Florist лок CEU tropical Africa and tropical America. gers, E. Ayensu . D. Du ен акт Tropical Forest Ecosystems in Africa and South America: А Comparative Review. Smithsonian Press, Washing- ton, D.C. —Duncan W. Thomas, Office of International Research and Development, Oregon State Uni- versity, Corvallis, Oregon 97331-1641, U.S.A. A METHOD FOR COLLECTING DRIED PLANT SPECIMENS FOR DNA AND ISOZYME ANALYSES, AND THE RESULTS OF A FIELD TEST IN XINJIANG, CHINA Dried and pressed plant specimens have been used for scientific study since the Italian Renais- sance (Stafleu, 1971), and this remains the most common method of preserving the “information content” of living plants. Today, the standard her- barium specimen is routinely augmented by the preservation of material for anatomical, cytologi- cal, or chemical analysis (Davis 8 Heywood, 1963). The growing application of molecular data in plant systematics requires methods for preserving material from which isozymes and nucleic acids can be recovered. In most cases, it has been con- sidered necessary to transport plants quickly to the laboratory on ice, or to freeze and store plant tissues in liquid nitrogen until they can be returned to the laboratory. Both methods can be used quite suc- cessfully when the plants studied are nearby and in easily accessible localities. They have also been applied to collect material for molecular analysis in more remote regions (Sytsma & Schaal, 1985; P. Peterson, pers. comm.), but in these cases the logistics and the expense are prohibitive. In addi- tion, the use of ice or liquid nitrogen storage is impractical when making large-scale ““floristic”” col- lections. An alternative to these methods seems most desirable. The recovery of high molecular weight DNA from dried plant material has been demonstrated by Rogers & Bendich (1985) and Doyle & Dickson (1987). Pyle & Adams (1989) recently compared 27 treatments of plant specimens and found that only fresh, frozen, or dried plant tissues provided good yields of quality genomic DNA. They obtained high molecular weight DNA from spinach leaves following desiccation and storage in a desiccator for two months, but degradation was observed when tested at five months. These results prompted ad- ditional tests using both fresh and air-dried spinach placed directly into silica gel in jars. The initial results of this experiment are reported here. The primary purpose of this report is to describe a field test of a simple and inexpensive drying method for preserving plant material for subse- quent nucleic acid and isozyme recovery. This method is appropriate for use in remote regions and can easily complement the routine collection of herbarium specimens. MATERIALS AND METHODS Preparation of plant material, DNA extraction, and gel electrophoresis of genomic DNA from fresh spinach (Spinacia oleracea L.) followed the pro- tocol of Pyle & Adams (1989) with the following modification: Fresh and air-dried (42°C, 36 hr., in a plant press) leaves were placed directly in contact with silica gel, sealed in air-tight plastic bottles, and stored at 37°C for five months. A field test was conducted during a 24-day ex- pedition in the Xinjiang Uygur Autonomous Re- strongly continental climate. Collections were made in the Tarim Basin, Songarian Basin, and Tien Shan Mountains from sea level to 3,800 m. Approxi- mately 2—5 grams of plant tissue were wrapped in tissue paper (to prevent plant tissue fragmentation) and placed in a 125-ml Nalgene bottle prefilled to one-third capacity with anhydrous CaSO,, or Drier- ite (W. A. Hammond Co., Xenia, Ohio). Indicator Drierite was mixed with nonindicator Drierite in a l : 5 proportion. Bottles were labeled, tightly sealed, and stored at ambient temperature. Most plant tissue was completely dry within 24 hours. In suc- culent material a single change of Drierite was required. Upon return to the laboratory, bottles were stored at —20°C. The hot CTAB procedure (Doyle & Dickson, 1987; Doyle & Doyle, 1987) was used to extract high molecular weight DNA. Best results were obtained when the dried plant material was ground in the presence of liquid nitrogen. Precip- itation of DNA with isopropanol was carried out at — 20°C overnight in order to increase yields. Re- striction digests were carried out with the endonu- clease Hind III for two hours at 37°C. Agarose- ANN. Missouri Вот. GARD. 77: 859-863. 1990. 860 Annals of the Missouri Botanical Garden ч OOP 0 м — FIGURE 1. Comparison of spinach DNA from air-dried (lanes 2-6 for months 1-5) and fresh leaves stored on silica gel (lanes 9-13 for months 1-5). Lanes 1 and 8 are DNA from fresh spinach (control). Lanes 7 and 14 are lambda Hind III DNA markers. Size standards from the origin (O) are 23.1, 9.4, and 6.6 kb. The dual-loaded gel was run on agarose, 0.6%, 10 V/cm, 30 minutes. gel electrophoresis and transfer of DNA from aga- rose gels to BioTrace (Gelman) nylon filters were conducted following standard protocols (Palmer, 1986). Preparation of digoxigenin-labeled probes, filter hybridizations, filter decolorization, and probe removal followed the manufacturer’s instructions (Boehringer Mannheim Biochemicals; BMB) as modified by Rieseberg et al. (1990 Chloroplast and nuclear ribosomal DNA varia- tion was detected by hybridizing the filter-bound DNAs to plasmids containing a 9.9 kb Sac I frag- ment of the Lactuca chloroplast (courtesy of Rob- TABLE 1. Таха used for DNA extraction and isozyme analysis. Voucher specimens will be deposited in PEK and RSA. ( Jollections are by Aaron Liston (L.) or James Morefield (M.). Specimens were used fo r successful DNA extraction (D), шапке DNA extraction (X), isozyme analysis for ME, PGI, and TPI (I), and isozyme analysis for all enzymes used in unpublished studies of these genera (A). Collec- tion Use Family Species М. 5157 D Asteraceae Artemesia dracunculus L. L. 827-12 I Asteraceae Senecio krascheninnikovii Schischk. L. 829.5 D Brassicaceae Lepidium latifolium M. 4993 I Brassicac Lepidium latifolium L M. 5042 D бл HE Cerastium beeringianum Cham. & Schlecht. L. 809.4 DI Chenopodiaceae Ceratoides latens (Gmel.) Vaid & Reveal M. 5040 D sulaceae Rhodiola coccinea (Royle) A L. 819-7 D Cyperaceae Carex [sect. Lamprochlaenae (Drejer) Bailey] sp. L. 816-1 DA Fabaceae Astragalus contortuplicatu L. 811-1 DI Fabaceae Glycyrhiza inflata Batal. L. 813-1 I Fabaceae Sphaerophysa salsula (Pall.) DC. L. 819-3 D Gentianaceae Gentiana aquatica L L. 835.22 D A Paeoniaceae aeonia anoma L. 823. D Plantaginaceae Planta L. 808-9 D I Poaceae Aeluropus E ош ) Parl 1. 4812 D Primulaceae Glaux maritima L. L. 819-1 I Ranunculaceae Thalictrum alpinum L L. 818-2 D Rosaceae Potentilla fruticosa L. M. 5067 I osaceae Potentilla aff. pennsylvanica Ledeb. L. 821-2 X I Rosaceae Rosa platyacantha Schrenk L. 821.3 I Rosaceae Rosa sp. L. 822-1 I Rosac Rosa platyacantha x Rosa sp. M. 5200 D Scrophulariaceae Lagotis integrifolia (Willd.) Schischk. Volume 77, Number 4 1990 Notes 861 ert Jansen) and a single 185-255 rDNA repeat from Helianthus argophyllus Torr. & Gray (cour- tesy of Mike Arnold). Sample preparation and enzyme electrophoresis followed the general methodology of Soltis et al. (1983). The tris-HCl grinding buffer-PVP solution described by Soltis et al. (1983) was used for enzyme extraction from dried plant tissue. All en- zymes were resolved on 12% starch gels. The gly- colytic enzymes malic enzyme (ME), phosphoglu- coisomerase (PGI), and triosephosphate isomerase (TPI) were examined. A modification of gel and electrode buffer system 8 (Rieseberg & Soltis, 1987) was used to resolve these enzymes. RESULTS Figure 1 shows genomic DNA from air-dried (lanes 2-6 for months 1—5) and fresh spinach (lanes 9-13 for months 1-5). Lanes 1 and 8 are DNA from fresh spinach (control). Lanes 7 and 14 are Lambda Hind III markers. Air-dried spinach DNA shows little change after four months storage at 37°C. However, the fresh spinach, placed di- rectly in silica gel, sealed and stored at 37°C, always appears to be a little more degraded than air-dried and desiccated spinach. Both treatments appear to be degraded after five months storage. Genomic DNA isolated from 15 species, rep- resenting 14 families, collected in China (Table 1) shows little or no degradation after four to six weeks of storage at ambient temperature (Fig. 2). Ge- nomic DNA could not be successfully isolated from a species of the genus Rosa (Table 1). Figure 3 shows the results of hybridization of these same DNAs to the Helianthus 185—255 rDNA repeat. Successful results were also obtained upon hybrid- ization to the 9.9 kb Sac] Lactuca chloroplast DNA probe (not shown). Activity for the glycolytic enzymes PGI and TPI was obtained for 12 species representing eight fam- ilies (Fig. 4). However, only two of these species (Sphaerophysa salsula and Thalictrum alpinum) had discernable activity for ME. DISCUSSION The preservation of plant samples for DNA and isozyme analysis using only a drying agent and plastic bottles is simple, inexpensive, and of wide potential application. However, a few additional factors should be considered before the method is used in a particular study: (1) Although genomic DNA was successfully iso- lated from 15 of 16 species examined, difficulty 2 4 6 8 1012 14 16 — 0 _FIGURE 2. Undigested DNA from plants collected in dards from the origin (O) are 23.1, 9.4, 6.6, 4.4, 2.3, and 2.0 kb. The gel was run on agarose, 0.7%, 5 V/cm, 12 hours.— Lane 2. Aeluropus littoralis. —Lane 3. Ar- temesia dracunculus. —Lane 4. Astragalus contortu- plicatus. — Lane 5. Carex sp. — Lane 6. Cerastium beer- ingianum. —Lane 7. Ceratoides latens. — Lane : Gentiana aquatica. — Lan a^ ы ти maritima. — 10. Glycyrhiza inflata. — 1. нө peri ia. — Lane 12. Lepidium mecs — Lane 13. Paeonia anomala. — Lane 14. Plantago a кин, 15. Po- tentilla fruticosa. —Lane 16. Rhodiola coccinea. was encountered with the genus Rosa. Difficulties were initially found with material of the genus Brongniartia (Fabaceae) similarly collected in Mexico (O. Dorado, pers. comm.). It should be noted that DNA кп {тот fresh material of due to the presence of polysaccharides, dud alternative extraction pro- cedures may be necessary. (2) The examined species represented diverse growth forms including annuals, herbaceous pe- rennials, succulents, and small shrubs. No trees were included in our study, but preliminary data from the genus Arbutus (Ericaceae) suggests that our technique can be used with woody taxa as well (Liston, unpublished). This technique has also been successfully used in recent collections of Ecuadoran romeliaceae (С. Brown, pers. comm.). (3) Only three glycolytic enzymes were examined in the present study and no attempts were made to optimize the detection conditions for each species. In a preliminary analysis of 15 enzymes in the genera Astragalus and Paeonia, activity compa- 862 Annals of the Missouri Botanical Garden 6 8 10 12 14 16 en oe Le €——— FIGURE 3. Southern blot of DNA from plants collected in Xinjiang, China, digested with Hind III and hybridized to a d 185-255 rDNA repeat from Helianthus argophyllus. Lane 1 is lambda EcoR I, Hind 11, Sal I, and markers. Size standards (origin not ited ion: the top of the gel are 32.7, 23.1, 21.2, 19.4, 15.3, 12.2, 9.4, P 8.3, 7.4, 6.7, 5.8, 5.6, 4.9, 4.4, and 3.5 kb. Lanes 2-16 are as in Figure rable to that of fresh material has been found in In conclusion, we suggest that researchers in- dried material of the Chinese species A. contor- terested in applying this technique to a particular tuplicatus and P. anomala (Liston & Zona, un- taxon test the material beforehand, although our published). Further study is needed to demonstrate — high rate of successful DNA isolation from a wide the usefulness of this technique in actual isozyme range of taxa suggests that it is worthwhile for any studies. collector to obtain samples in this manner in ad- FIGURE 4, Enzyme activity for TPI from plants d in Xinjiang, China, according to the described protocol. — Lane 1. Aeluropus littoralis. —Lane 2. Ceratoides latens. — Lane Glycyrhiza inflata. — Lane 4. Rosa platya- cantha. —Lane 5. Rosa sp.— Lane 6. Rosa К dh x Ros did ne 7. Paeonia anomala. —Lane 8. теш salsula. — Lane 9. Thalictrum alpinum. — Lane 10. бун krascheninnikovii. —Lane 11. Potentilla aff. pennsylvanica. —Lane 12. Lepidium latifolium. Volume 77, Number 4 1990 Notes 863 dition to herbarium specimens. We also recom- mend that DNA extraction and isozyme analysis be carried out as soon as possible after reaching an appropriate laboratory, because studies with spinach indicate the possibility of degradation after several months of storage. Alternatively, samples can be frozen for long-term storage. A. L. and L. H. R. thank Li Bosheng, Guo Ke, Kong Lingsaho, Anver Sadir, and James Morefield for assistance in the field. We are also grateful to Thomas Elias for initiating and making possible the trip to Xinjiang and to Scott Zona for critically reading an earlier version of the manuscript. LITERATURE CITED Davis, Р. Н. & V. H. HEywoop. 1963. Principles of -— Taxonomy. Oliver & Boyd, Edinburgh. DoytE, J. J. & E. E. Dickson. 1987. Preservation of plant samples for DNA restricti d ] anal- ysis. Taxon 36: 715-7 J. L. DoYLe 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem. Bull. 19: 11-15. PALMER, J. D. 1986. pred uis от analysis of chloroplast DNA. Pp. 1 n A. Weissbach H. Weissbach (editors), (чөлү in Enzymology, Volume 118. Academic Press, Orlando, Florida. PyLe, M. M. € R. P. Apams. 1989. In situ ом of DNA in plant specimens. Тахоп 38: RIESEBERG, L & D. E. Sorris.. 1987. eed differentiation between Tolmiea menziesii and Tel- lima grandiflora (Saxifragaceae). Syst. Bot. 12: 154- 161. , 8. BECKSTROM-STERNBERG & К. Doan. 1990. Helianthus annuus ssp. texanus has chloroplast DNA and nuclear ribosomal RNA genes of Helianthus debilis ssp. cucumerifolius. Proc. Natl. Acad. U.S.A. 87: 593-597. ROGERS, S. O. & A. J. BENDICH. 1985. Extraction of DNA from milligram amounts of fresh, herbarium, and mummified plant tissues. Pl. Molec. Biol. 5: 69- Sorts, D. E., C. H. HAUF LER, D. C. DanRow & C. J. GasTony. 1983. Starch gel окы of ferns: a compilation of grinding buffers, gel and electrode buffers, and staining schedules. Amer. Fern J. 73 27. STAFLEU, F. A. 1971. Linnaeus and the Linnaeans. A. Oosthoek, 2 SYTSMA, К. J. & B. A. ScHaaL. 1985. Phylogenetics E the i а skinneri (Gentianaceae) species mplex in Panama utilizing DNA restriction frag- ment analysis. Evolution 39: 594-608. —Aaron Liston and Loren Н. Rieseberg, Rancho Santa Ana Botanic Garden Graduate Program in Botany, 1500 North College Avenue, Clare- mont, California 91711, U.S.A. (Present address of A. Liston: Department of Genetics, жт еы of California, Davis, California 95616, U.S.A.); Robert P. Adams and Nhan Do, CSFAA, | po Box 7373, Baylor University, Waco, Texas 76798, U.S.A.; and Zhu Ge-lin, Institute of Botany, Northwest Normal College, Lanzhou, Gansu, China. BOOK REVIEW Tan, Kit (editor). 1989. Plant Taxonomy, Phy- togeography and Related Subjects: The Davis and Hedge Festschrift. Edinburgh University Press, 22 George Square, Edinburgh EH8 OLF, Scotland. Distributed in North America + 351 pp. Hardbound. Retail price: 47.50 pounds. by Columbia University Press. xxvi This volume, commemorating a seventieth and a sixtieth birthday respectively, is one of those delightful books with which it must be so pleasant to have been associated, whether as editor or au- thor. Peter Davis and lan Hedge are, to quote the forward, **. .. two of the outstanding Edinburgh botanists of the latter half of this century.” The contributors to the festschrift are from 25 institutes in 15 countries, and their contributions highlight current taxonomic problems, especially in the Med- iterranean and southwest Asian regions so beloved by Davis and Hedge. The 26-page preface contains brief biographical sketches, photographs, and extensive lists of the publications of the two celebrated botanists. The actual papers in this volume are a mixed bag and should include something of interest to almost everyone. New taxa of Filago and Cichorium (Compositae), Bupleurum and Stoibrax (Umbel- liferae), Grimmia (Musci), Veronica (Scrophu- lariaceae), Astragalus (Leguminosae- Papilionoi- deae), Iris (Iridaceae), Galium (Rubiaceae), Piptatherum and Stipa (Gramineae), and Peristy- lus (Orchidaceae) are described. There are also revisions of Monochoria and parts of Iris and Olea. The phytogeography papers are representative of the old school of geography; those biologists interested in vicariance biogeography will look in vain. One-seventh of the nearly 5,000 species of vascular plants in Greece are endemic. Arne Strid gives examples of these in relation to phytogeog- raphy and conservation. Another paper discusses floristic links and endemism in the Armenian high- lands. Rupert Barneby discusses some problems and consequences of typification in Mimosa in his usual ANN. MISSOURI Bor. GARD. 77: 864. 1990. eloquent fashion. A second paper by Barneby de- scribes the identification of an object dug from the site of a former English bakery as a mesocarp of a menispermaceous genus that probably arrived on the site as a foreign body in a sack of Brazil nuts. Two recurrent problems also reappear in the volume. The systematic position of Rhabdoden- dron is discussed in light of new and extensive embryological information. Although authors Tobe and Raven would place the genus in the Rosiflorae (but not in Dahlgren's Rosales), the mostly plesio- morphic embryological characters allow the actual placement of the genus to remain enigmatic. Of interest to a wider audience of botanists will be the chapter by Gertrud Dahlgren entitled “The Last Dahlgrenogram, System of Classification of the Di- cotyledons." Published by his widow, this modified bubble diagram represents Rolf Dahlgren's last work on this topic before his fatal traffic accident in February 1987. A list of families and orders, as well as a discussion of the changes between this and the 1980 scheme, is given. Interestingly, the endings '-florae' for superorders have now been changed to *-an Finally, two light but enjoyable papers round out the volume. Plantsmen & Pottery discusses, among other things, Peter Davis's interest in the Scottish pottery known as Wemyss ware, slotting him among former and living biologists with close ceramic ties. The difficulties met on ten botanical expeditions to Iraq, Iran, Afghanistan and Pakistan over a period of 40 years are described in almost 50 pages by K. H. Rechinger. Air-travel is no- ticeable by its absence, as are the political problems and fighting, which now prohibit most travel to Almost 60,000 numbers were col- lected despite the perils and problems encountered. І recommend this account of recent travel in these these areas. areas, which have now largely changed forever. Altogether, this is a most interesting and worth- while compilation of papers, and it belongs on the shelves of most botanical libraries. —P. Mick Rich- ardson, Missouri Botanical Garden, P.O. Box 299, St. Missouri 63166, U.S.A. „OUES, Volume 77, Number 4 1990 Erratum 865 ERRATUM FOR CAMPYLONEURUM NITIDISSIMUM var. ABRUPTUM An error in the comparison between Campy- loneurum nitidissimum var. abruptum (Lindman) León and C. coarctatum (Kunze) Fée gives the exact opposite of the concept intended. The faulty sentence reads: ““It is usually misidentified in her- baria as С. coarctatum (Kunze) Fée, from which it differs by its narrow, long-creeping stem, widely spaced phyllopodia, and leaves less than 40 cm long” (Ann. Missouri Bot. Gard. 77: 212). This should have read: “It is usually misidentified in herbaria as С. coarctatum (Kunze) Fée, which differs by its narrow, long creeping stem... . and leaves more than 40 cm long.” Campyloneurum coarctatum was described by Kunze in Polypodium based on material collected by Poeppig from Peru. The holotype was probably destroyed in Leipzig; however, there is other type material at Paris (Р) and Vienna (W). I have seen both isotypes: at Paris there is a complete specimen with stem and leaf, while the specimen at Vienna (photo, BM) has only the leaf. Kunze applied the name to a plant with narrow (2-3 mm), long- creeping stems, well-spaced phyllopodia, elliptic to ovate-elliptic leaves with narrowly to abruptly cu- neate bases, and undivided primary areoles. This name represents a well-defined species, which oc- curs from Costa Rica (León, Flora Mesoamericana, in press) to Bolivia, usually as an epiphyte. Thus C. coarctatum as here defined belongs to the C. sphenodes complex with long-creeping stems, well- spaced phyllopodia, and mainly undivided primary areoles, a group not closely related to С. nitidis- simum. Campyloneurum nitidissimum var. abruptum was described by Lindman as a variety of Poly- podium repens, based on his own specimens col- lected during the Regnell expedition to Brazil. It is characterized by wide (5—10 mm), short-creeping stems, closely spaced phyllopodia, lanceolate to elliptic-lanceolate leaves, with attenuate to abruptly cuneate bases, and with or without divided primary areoles. Besides, it is only known from Colombia to Bolivia and Brazil, where it is commonly ter- restrial. The misunderstanding concerning the applica- tion of the name C. coarctatum for the species now recognized as С. nitidissimum var. abruptum was based on similarities of the leaf size and shape of base, which are characters of relatively little value. —Blanca León, Museo de Historia Natural, Av. Arenales 1256, Casilla 14-0434, Lima 14, Peru. ANN. Missouni Вот. GARD. 77: 865. 1990. Instructions for Authors, Annals of the Missouri Botanical Garden The Annals publishes (Pom articles in systematic botany and related fields. n English or Spanish. When in Spanish, an English Bude in addition to a Spanish abstract is required. Short papers, especially nomenclatural matters and new taxa, are treated as "Notes," which have their own format as described below. Authors are urged to contact the editor prior to submitting manuscripts. Manuscripts sent already reviewed and re- vised are generally subject to the full normal review pro- cess initiated here. Manuscripts not properly prepared may be returned for revision prior to review. This includes manuscripts with improper abbreviations or with diacrit- ical marks not indicated. Page costs are $80 per page, although charges may be reduced under certain circumstances. Corrections in proof will be billed nonnegotiably to the author ar $2.50 per line. Manuscripts must be typed (output from dot matrix printers must be entirely clear) on one side of nonglossy 8% x 11 inch paper. Leave at least one inch as margin all around, except on the first page. There three inches should be left blank at the top. The text should be double-spaced or triple-spaced throughout, in- cluding abstract, notes, legends, tables, specimen lists, Pru Cited, and Киши. Do not justify the ds there. Italica, boldface, or г other pate typefaces should not be used— they will appear later as communicated to the printer by de editor. Words to be set in italics should be underlined. Accents, umlauts, and other diacritical marks are the responsibility of the author Footnotes should be typed double-spaced as paragraphs each separate additional footnotes. No other footnotes are permitted, except footnotes in tables, and these are best avoided. In papers formatted as "Notes," there are no footnotes: the acknowledgments make up the paragraph preceding the Literature Cited section, and the author's name and address make up the last paragraph of the article. In Notes, the names and addresses of multiple authors are all placed in the same paragraph Long, complex tables are discouraged. Tables not neat and easily understood will be returned for improvement. Leave out vertical lines, and avoid horizontal lines. Lines are added as needed by the editor and/or printer. Type table “он double-spaced as paragraphs at the tops of the t x concise iade should accompany each paper other than **Not Periods a are used after all abbreviations (which should ass direc- e not abbre viated, and cities are usually spelled out, although “St.” is acceptable. Abbreviate periodicals according to B abbreviations are to be capitalized, and that books listed in the Literature Cited section should have their titles spelled out when feasible. Note further that names of taxonomic authorities may be spelled out in major taxo- nomic citations such as the formal citation at the head of a taxonomic treatment. Except by prior arrangement with authors not having access to crucial references, the edito does not provide the service of looking up куне деин п the Literature Cited, periodicals should be listed as follows: author’s last name, initial(s). Year. Full title of article. Journal abbreviated as in BPH. Volume: pages. Do not give parenthetical part numbers after volume numbers unless each part is paginated separately, which is unusual. For more than one author, follow this style: author’s last name, initial(s), second author’s initial(s), last name & third author’s initial(s), last name. Citations of books should appear as follows: author’s last name, ini- tial(s). Year. Full Title (edited by Editor), ч edition, Volume 2. Publisher, Place. сы» of work “in prep.” and similar references t are song discouraged. Except by prior arrangement when ne sary, the editor does not perform library research to fil in missing bibliographic information. A brief Latin diagnosis for each new taxon is preferred to a complete Latin description. Use one paragraph per basionym as follows: Taxon author, literature citation, type citation, e.g., Bauhinia dipetala Hemsl., Diag. Pl. Nov. 1880. Casparia dipe- tala (Hemsl.) Britt. 8 Rose, N. Amer. Fl. 23: 216. 1930. TYPE: Mexico. Veracruz: Valle de Cordóba, Bourgeau 1713 (holotype, K; isotypes, F, GH, P, US). Exclamation points are not use for specimens examined. Indicate unseen types as “n.y.’ Keys should be dichotomous, indented, and consistent with descriptions. Couplets should be numbered as la, lb, 2a, and so forth. Cite specimens as follows: Additional specimens examined (or Selected speci- mens examined or Paratypes). MEXICO. OAXACA: Sierra San Pedro Nolesco, Talea, 3 Feb. 1987 (fl), Jergensen 865 (BM, G, K, US). Dates and reproductive status are optional Send copies of illustrations; do not send originals until requested if the manuscript is accepted. Illustrations will be returned after publication. All illustrative materials should be mounted on stiff cardboard. The maximum size for printed illustrations is 5/4 x 8% inches. Presize illus- trations to fit either column width (25 inches or ca. 68 mm) or full page width (51% inches ог ca. 140 mm). Figures are numbered in Arabic numerals in the order of their citation in the text. Parts of figures are labeled with capital letters. Photographs should be black-and-white abut edges of photographs. Do not use stripping between plate or figure segments. Make sure edges are squared. Maps should include reference to latitude and longitude. Scale bars should appear on illustrations, on photographs, and on maps. Magnifications/reductions indicated in cap- tions are unwelcome to the point of usually not being accepted. Scanning electron micrographs should be free of conspicuous charging. Axes on graphs are all labeled. No explanatory text is to appear in figures. Rather, provide clear captions in sequential paragraphs apart from the other text, one paragraph for each group of figures, and following the style in current issues of the Annals r other aspects of style, consult a recent issue of the Annals, the latest edition of The Chicago Manual of Style, or write the editor. Communication with the editor at all points of the publication process is encouraged. Manuscripts (three copies) and correspondence should be addressed to the Editor, Annals of the Missouri Bo- tanical Garden, P.O. Box 299, St. Louis, Missouri 63166, S.A ANNALS OF THE MISSOURI BOTANICAL GARDEN VOLUME 77 1990 Colophon This volume of the ANNALS of the Missouri Botanical Garden has been set in APS Bodoni. The text is set in 9 point type while the figure legends and literature cited sections are set in 8 point type. The volume has been printed on 604 Warrenflo (No. 1) and 704 Vintage Gloss (Nos. 2, 3, 4). These are acid-free papers designed to have a shelf-life of over 100 years. Vintage Gloss is manufactured by the Potlatch Paper Company. Warrenflo is manufactured by S.D. Warren Paper Co. 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 1990 ISSN 0026-6493 VOLUME 77 ADAMS, ROBERT P. (See Aaron Liston, Loren Н. Rieseberg, Robert P. Adams, Nhan Do € Zhu Ge-lin) AFFOLTER, JAMES M. (See Lincoln Constance & James М. Affolter) ............. AL-SHEHBAZ, [HSAN A. Sisymbrium llatasii and 5. morrisonii (Brassica- ceae), New Species from Coastal Peru AL-SHEHBAZ, IHSAN, A. The South American Eremodraba (Brassicaceae) . AL-SHEHBAZ, IHSAN A. Brayopsis gamosepala (Brassicaceae), a Remark- able New Species with Gamosepalous Calyx AL-SHEHBAZ, IHSAN A. Weberbauera perforata (Brassicaceae), a New Spe- cies from Peru ARBO, María MERCEDES. Turneraceae: Novedades para la Guayana Vene- zolana BALDWIN, BRUCE G., DonaLD W. Күноѕ & Jan DvoRÁák. Chloroplast DNA Evolution and Adaptive Radiation in the Hawaiian Silversword Alliance (Asteraceae-Madiinae) BALDWIN, BRUCE G. (See Donald W. Kyhos, Gerald D. Carr & Bruce С. Baldwin) BARRINGTON, Рау S. (Organizer). Introduction: A Festschrift in Honor of Alice Faber Tryon and Rolla Tryon, Jr. BARRINGTON, Davip S. Hybridization and Allopolyploidy in Central American Polystichum: Cytological and Isozyme Documentation BARRINGTON, Davip S. (See Cathy A. Paris & David S. Barrington) .............. BARRINGTON, Davip S. (See Diana B. Stein & David S. Barrington) ............. BENÍTEZ DE Rojas, CARMEN E. Dos Especies Nuevas del Genero Schwenckia (Solanaceae) de Venezuela BENÍTEZ DE Rojas, CARMEN. (See William С. D'Arcy & Carmen Benitez de Rojas) BERRY, PauL E. (See Jorge V. Crisci & Paul E. Berry) BonM, BRUCE А. (See William J. Crins & Bruce A. Bohm) Bons, LYNN. The Systematics of Solanum section Allophyllum (Solanaceae) BRAMWELL, Davip. Conserving Biodiversity in the Canary Islands .................. BRITO, YSALENY. (See Nelson Ramirez, Celia Gil, Omaira Hokche, Alberto Seres & Ysaleny Brito) CARNEVALI, GERMÁN & IVON RAMÍREZ. New or Noteworthy Orchids for the Venzuelan Flora. VIII. New Species and Combinations from the Ven- ezuelan Guayana CARR, GERALD D. (See Donald W. Kyhos, Gerald D. Carr & Bruce С. Baldwin 1990 949 84 CARR, GERALD D. (See Robert Н. Robichaux, Gerald D. Carr, Matt Liebman & Robert W. Pearcy) CLARK, LYNN G. A New Combination in Chasmanthium (Poaceae) ............... CLARK, LYNN С. & XIMENA LONDOÑO. Three New Andean Species of Au- lonemia (Poaceae-Bambusoideae) Conant, Davip 5. Observations on the Reproductive Biology of Alsophila Species and Hybrids (Cyatheaceae) CONSTANCE, LINCOLN & JAMES М. AFFOLTER. A New Cloud-forest Eryngium from Chiapas, Mexico COOPER-DRIVER, GILLIAN A. Defense Strategies in Bracken, Pteridium aqui- linum (L.) Kuhn Cox, PAUL ALAN. Pollination and the Evolution of Breeding Systems in Pandanaceae CRINS, WILLIAM J. & Bruce A. BoHm. Flavonoid Diversity in Relation to ystematics and Evolution of the Tarweeds Crisci, JorGE V. & PauL E. Berry. A Phylogenetic Reevaluation of the Old World Species of Fuchsia (Onagraceae) Criscl, JORGE V., ELIZABETH A. ZIMMER, PETER C. Носн, GEORGE B. JOHNSON, Curisty Морр & N.S. Pan. Phylogenetic Implications of Ribosomal DNA Restriction Site Variation in the Plant Family Onagraceae ............. D”ArcY, WILLIAM G. & CARMEN BENÍTEZ DE Rojas. Censtrum neblinense (Solanaceae), A New Species from Venezuela DAVIDSE, GERRIT. Notes and New Combinations in Hawaiian Panicum (Po- aceae: Paniceae) Davis, Scorr K. (See Alan R. Templeton, Kerry Shaw, Eric Routman & Scott K. Davis) Do, NHAN. (See Aaron Liston, Loren H. Rieseberg, Robert P. Adams, Nhan Do € Zhu Ge-lin) DRANSFIELD, JOHN, I. К. FERGUSON & NATALIE W. UnL. The Coryphoid Palms: Patterns of Variation and Evolution DvoRÁK, JAN. (See Bruce G. Baldwin, Donald W. Kyhos & Jan Dvorak) . Клік, DONALD A. Integrated Strategies for Conserving Plant Genetic Di- versity FERGUSON, I. K. (See John Dransfield, I. K. Ferguson & Natalie W. Uhl.) GASTONY, GERALD J. Electrophoretic Evidence for Allotetraploidy with Seg- regating Heterozygosity in South African Pellaea rufa A. F. Tryon (Adiantaceae) GE-LIN, ZHU. (See Aaron Liston, Loren Н. Rieseberg, Robert Р. Adams, Nhan Do & Zhu Ge-lin) GEREAU, Roy E. (See Jon C. Lovett & Roy E. Gereau) GIL, CELIA. (See Nelson Ramirez, Celia Gil, Omaira Hokche, Alberto Seres S Ysaleny Brito) 859 217 GIVEN, DaviD. Conserving Botanical Diversity оп a Global Scale ................. GOLDBLATT, PETER. Cytological Variability in the African Genus Lapeirousia (Iridaceae—Ixioideae) GOLDBLATT, PETER. Systematics of Lapeirousia (Iridaceae—Ixioideae) in Tropical Africa GOLDBLATT, PETER. Phylogeny and Classification of Iridaceae ooo... GOLDBLATT, PETER & JOHN C. MANNING. Devia xeromorpha, a New Genus and Species of Iridaceae—Ixioideae from the Cape Province, South Africa GOLDBLATT, PETER & JOHN C. MANNING. Leaf and Corm Tunic Structure in Lapeirousia (Iridaceae—Ixioideae) in Relation to Phylogeny and In- frageneric Classification GRAYUM, MICHAEL Н. Evolution and Phylogeny of the Araceae ooo... HacBERG, Mats. Two New Species of Monotagma (Marantaceae) from the Venezuelan Guayana HAUFLER, CHRISTOPHER H., MICHAEL D. WINDHAM & Thomas A. RANKER. Biosystematic Analysis of the Cystopteris tennesseensis (Dryopteridaceae) Complex HERENDEEN, PATRICK S. & JAMES L. ZARUCCHI. Validation of Caesalpinia subgenus Mezoneuron (Desf.) Vidal and New Combinations in Caesal- pinia for Two Species of Mezoneuron from Africa HICKEY, R. JAMES. Studies of Neotropical Isoëtes L. I. Euphyllum, a New Subgenus HocH. PETER С. (See Jorge V. Crisci, Elizabeth A. Zimmer, Peter C. Hoch, George B. Johnson, Christy Mudd & N. S. Pan) HOKCHE, OMAIRA & NELSON RAMIREZ. Pollination Ecology of Seven Species of Bauhinia L. (Leguminosae: Caesalpinioideae HOKCHE, OMAIRA. (See Nelson Ramirez, Celia Gil, Omaira Hokche, Alberto Seres & Ysaleny Brito) Horsr, BRUCE A. & CAROL A. Topzia. León Croizat's Plant Collections from the Franco-Venezuelan Expedition to the Headwaters of the Rio Orinoco JANSSENS, JAN A. Book Review JOHNSON, Davip М. & Nancy A. MURRAY. New Species of Guatteria (An- nonaceae) from the Guayana Highland JOHNSON, GEORGE B. (See Jorge V. Crisci, Elizabeth A. Zimmer, Peter C. Hoch, George В. Johnson, Christy Mudd & М. S. Pan) JUDZIEWICZ, EMMET J. Steyermarkochloa angustifolia (Sprengel) Judziew- icz, A New Combination (Poaceae—A y KILLEEN, TimoTHY J. The Grasses of Chiquitania, Santa Cruz, Bolivia ....... KIRKBRIDE, JOSEPH H., JR. (See James L. Zarucchi & Joseph H. Kirkbride, Jr.) KosTERMANS, A. J. G. H. Additional Transfers of Asiatic Machilus Sensu Nees, non Desrousseaux, to Persea Miller (Lauraceae) KRAL, ROBERT. Book Review KRAMER, К. U. The American Paradox in the Distribution of Fern Taxa Above the Rank of Species Kress, W. Jonn. The Phylogeny and Classification of the Zingiberales ...... Күноз, DonaLD W., GERALD D. CARR & BRUCE С. BALDWIN. Biodiversity and Cytogenetics of the Tarweeds (Asteraceae: Heliantheae-Madiinae) KvHos, DonaLD W. (See Bruce G. Baldwin, Donald W. Kyhos & Jan vorak) LEÓN, BLANCA. A New Species and New Combination in Campyloneurum С. Presl (Polypodiaceae) LEÓN, BLANCA. Erratum for Campyloneurum nitidissimum var. abruptum LIEBMAN, MATT. (See Robert H. Robichaux, Gerald D. Carr, Matt Liebman & Robert W. Pearcy) Liston, AARON, LOREN Н. RIESEBERG, ROBERT Р. ADAMS, NHAN Do & ZHU GE-LIN. A Method for Collecting Dried Plant Specimens for DNA and Isozyme Analyses and the Results of a Field Test in Xinjiang, China .. LONDOÑO, XIMENA. (See Lynn C. Clark & Ximena Londoño) ........ Lovett, Јом C. & Roy E. GEREAU. Notes on the Floral Morphology and Ecology of Margaritaria discoidea (Euphorbiaceae) at Mufindi, Tan- zania MANNING, JOHN C. (See Peter Goldblatt & John C. Manning) ocio MANNING, JOHN С. (See Peter Goldblatt & John C. Manning) ................... MAXWELL, RicHarD Н. А New Combination in Dioclea Kunth (Fabaceae— Diocleinae) from the Clarification of D. glabra Bentham, Flora Brasili- ensis MAXWELL, RICHARD H. New Таха of Dioclea Kunth (Fabaceae- Diocleinae) from the Venezuelan Guayana MENEZES, NANUZA Luiza DE & João SEMIR. New Considerations Regarding the Corona in the Velloziaceae MICKEL, JoHN T. Two New Species of Elaphoglossum (Elaphoglossaceae) rom Amazonas, Venezuela Moran, Rossin C. Two New Species of Cnemidaria (Cyatheaceae) from anama MORAN, Rossin C. Three New Species of Ferns from Mesoamerica MORAN, Rossin С. А New Species of Polypodium (Polypodiaceae) and Two New Species of Hypolepis (Dennstaedtiaceae) from Mesoamerica ....... Морр, CHRISTY. (See Jorge V. Crisci, Elizabeth A. Zimmer, Peter C. Hoch, George B. Johnson, Christy Mudd & N. S. Pan) Murray, Nancy А. (See David M. Johnson & Nancy A. Murray) ................ Pan, N. S. (See Jorge V. Crisci, Elizabeth A. Zimmer, Peter C. Hoch, George B. Johnson, Christy Mudd & N. S. Pan) Paris, САтнү A. & Davip S. BanRINGTON. William Jackson Hooker and the Generic Classification of Ferns Pearcy, RoBERT W. (See Robert H. Robichaux, Gerald D. Carr, Matt Liebman & Robert W. Pearcy) Ramírez, Ivón. (See Germán Carnevali € Ivon Ramirez) RAMIREZ, NELSON, CELIA GIL, OMAIRA HOKCHE, ALBERTO SERES & YSALENY Brito. Biologia Floral de una Comunidad Arbustiva Tropical en la Guayana Venezolana RAMIREZ, NELSON. (See Omaira Hokche & Nelson Ramirez) RANKER, THoMas A. (See Christopher H. Haufler, Michael D. Windham € Thomas A. Ranker) Raven, PETER Н. (See Elsa М. Zardini & Peter Н. Raven) .. Reyes J., Irma. (See Ramon Riba & Irma Reyes J.) Ripa, Ramon & Irma Reyes J. Pityrogramma calomelanos (L.) Link (Adiantaceae) on Layers of Volcanic Ash in Los Tuxtlas, State of Ve- racruz, Mexico RICHARDSON, P. Mick. Book Review RIESEBERG, LOREN H. (See Aaron Liston, Loren H. Rieseberg, Robert P. Adams, Nhan Do & Zhu Ge-lin) RopicHaux, RoBERT H., GERALD D. CARR, MATT LIEBMAN & ROBERT W. Pearcy. Adaptive Radiation of the Hawaiian Silversword Alliance (Compositae-Madiinae): Ecological, Morphological, and Physiological Diversity Rosson, NORMAN К. B. Two New Species and a New Combination in Vismia (Guttiferae- Hypericoideae) Rocers, GEORGE К. A Century of Scientific Publications at the Missouri Botanical Garden . ROGERS, GEORGE К. Book Review . Коотмам, Eric. (See Alan R. Templeton, Kerry Shaw, Eric Routman & Scott K; Davis): iiu iate irit a nes ScHATZ, С. E. A New Vitex (Verbenaceae) from Madagascar ................ SEMIR, João. (See Nanuza Luisa de Menezes & Joao Semir) ..............oooomm SERES, ALBERTO. (See Nelson Ramirez, Celia Gil, Omaira Hokche, Alberto Seres & Ysaleny Brito) Shaw, Kerry. (See Alan R. Templeton, Kerry Shaw, Eric Routman $ A A Simpson, MICHAEL С. Phylogeny and Classification of the Haemodoraceae SMITH, ALAN R. New Thelypteris (Thelypteridaceae) from Central America SmITH, ALAN R. A New Species and Combination in Thelypteris for Guyana SMITH, ALAN R. Pteridophytes of the Venezuelan Guayana: New Species . SouLé, MICHAEL E. The Real Work of Systematics „i Sousa S., Mario. Adiciones a las Papilionadas de la Flora de Nicaragua y una Nueva Combinación Para Oaxaca, Mexico STEIN, DIANA B. & Davip S. BARRINGTON. Recurring Hybrid Formation in a Population of Polystichum X potteri: Evidence from Chloroplast DNA Comparisons STOLZE, ROBERT G. Observations on Ctenitis (Dryopteridaceae) and Allied Genera in Ámerica STUESSY, Top F. (See Scott D. Sundberg & Tod F. Stuessy) SUNDBERG, Scott D. & Тор F. SrUEssv. A New Species of Trigonospermum (Compositae, Heliantheae) from Central America TAYLOR, CHARLOTTE M. А New Species of Palicourea (Rubiaceae) from Costa Rica TEMPLETON, ALAN R., KERRY SHAW, ERIC ROUTMAN & Scorr K. Davis. The Genetic Consequences of Habitat Fragmentation Thomas, Duncan W. — Conceveiba Aublet (Euphorbiaceae) New to Africa Topzia, CAROL A. (See Bruce A. Holst & Carol A. Todzia) UHL, NATALIE W. (See John Dransfield, I. K. Ferguson & Natalie W. Uhl) WINDHAM, MICHAEL D. (See Christopher Н. Haufler, Michael D. Windham & Thomas A. Ranker WINGFIELD, ROBERT. Psychotria berryi, a New Name for P. davidsae .... WiTTER, MARTHA S. Evolution in the Madiinae: Evidence from Enzyme Electrophoresis Wu, ZHENG-Yr. The Publication of the Flora of China Will Be a Great Contribution to the Scientific Circles of the World ZARDINI, Ё15А M. & PETER H. RAVEN. А New Combination in Ludwigia (Onagraceae) ZARUCCHI, JAMES L. А New Species of Macrolobium (Fabaceae: Caesalpi- nioideae) from Mesoamerica ZARUCCHI, JAMES L. & ЈОЅЕРН Н. KIRKBRIDE, JR. А Necessary New Com- bination in Duroia (Rubiaceae) ZARUCCHI, JAMES L. (See Patrick S. Herendeen & James L. Zarucchi) ...... ZAVADA, MICHAELS. A Contribution to the Study of Pollen Wall Ultrastruc- ture of Orchid Pollinia ZIMMER, ELIZABETH A. (See Jorge V. Crisci, Elizabeth A. Zimmer, Peter C. Hoch, George B. Johnson, Christy Mudd & N. S. Pan) 973 334 314 Volume 77, Number 3, pp. 427-606 of the ANNALS OF THE MISSOURI BOTANICAL GARDEN was 90. published on August 2, 19 Books on Mosses from the Missouri Botanical Garden Glossarium Polyglottum Bryologiae A multilingual glossary for кешк, В. E. Magill, Edito 297 pp. 1990. $12. 00. This polyglot glossary provides aac for nearly 1,200 terms in four languages— English, German, French, and "me In addition, all English terms are translated into Japanese, Latin, and Russian. The English vocabulary forms the core of the work, with each entry defined and cross-referenced to similar entries in the French, German, and Spanish glossaries, which also contain full definitions. The Japanese and Russian lists give translations of the English terms into these languages. The Latin vocabulary is essentially a translation of the English terms and indicates parts of speech and gender for all terms. Cryptogamic Flora of the diari Delta and Adjacent Regions B.-s. (editor). 973 pp. + 28 Заа. "Tex illustrations. 1989. $25.00. In Chinese with Latin nomenclature. A special arrangement with the Shanghai Museum of Natural History makes this Бен available at reduced cost. It treats about 1,100 species of algae, fungi (including lichens), bryophytes, and fe Mosses of the Interior Highlands of North America PIAR edfearn, Jr. viii + 104 pp. 1972, reprinted with iicet sd тен Jone ut йэ: га ry 300 species with family /genus key, generic d ts, occurrence, distribut Contributions to — Bryology, dedicated to Lewis E. Anderson 6 pp. 1985. $12.75. This volume contains articles by about twenty of Anderson's students, colleagues, and friends honoring his ения contributions to our knowledge of the cytology of mosses and of the moss floristics of North America. Musei Austro-Americani 652 + vi pp. 1982. $11. 25. A reprint of Mitten’s 1869 classic, long out-of-print, which treats about 1,700 mosses from the West Indies ei Central and South Ame : = Flora et Southern Africa, Part 1 Мыш, Forin 1 апа 2 Fascicle 1: ху + 291 pp. 1982. $20.00 Sixty-eight genera and 180 species in the families Sphagnaceae through Grimmiaceae. Fasc: icle 2: аа 15 1 рр. 1987. $20.00. оо Белене схе? families Gigasper- maceae пенй бин . Mosses of Guatemala. E. B. Bartram. 442 рр. strate. 1949, et 1972. $10.00. reor сс уш mame eee Fieldiana: B Bor. 25. To place an order, send hil tally abs oO ie сыы E fuk US ааа mdd O Sere ex ш} ңө. rada lr rares et en аи. | each additional book. Orders should be prepaid; a $1.00 fee added s requiring invoices. i элчили мын ci sas ы сеге DS básica, | Р.О. Box ` он ns колы : 86 р MO 63166-0299, USA. Please send the publications circled above ne et SORS 1 Seni invoice (1.00 fee wl be- _ added to total). —MM——— ume E м eM 6 i ЕТ. тта). cts Pues Ant Sct To Cua Wn Noc o e CONTENTS Systematics and Evolution of the Monocotyledons Phylogeny and Classification of Iridaceae Peter Goldblatt 607 Evolution and Phylogeny of the Araceae Michael H. Grayum 628 The Phylogeny and Classification of the Zingiberales W. John Кге A Phylogeny and Classification of the Haemodoraceae Michael G. Simpson ............. 122 A Contribution to the Study of Pollen Wall Ultrastructure of Orchid Pollinia Michael S. Zavada | 785 The Coryphoid Palms: Patterns of Variation and Evolution John Dransfield, L К. Ferguson & Natalie W. Uhl 802 Tollination and the Evolution of Breeding Systems in Pandanaceae Paul Alan Cox 816 NOTES | is Weberbauera perforata counties a New Species from Peru Ihsan A. EREA Al-She hbaz : .841 _ Brayopsis gamosepala ESE a Remarkable New Species with Gamosepalous Se . Calyx Ihsan A. Al-Shehbaz 843 9 т New Species of Polypodium (Polypodiaceae) and Two New aes of Hypolepis Чел 2r . staedtiaceae) from Mesoamerica Robbin C. Moran 845 = E A | Necessary New Combination i in Duroia (Rubiaceae) i mes L Zarwechi & Joseph : JH. Kirkbride, Pan . 851 ; Su ^ New Cloud-forest Eryngium kom Chiapas, Mexico S Lincoln Constance & James M. xx аўце. k Validation of руе SRS o (Desf) Vidal and New Corbii in - Caesalpinia for Two жо of Mezoneuron вам Ania Patrick 5: Herendeen са & James L. Zarucchi . 8 ү Conceveiba Aublet Euphorbiaceae) Мн to Africa ae v. Thomas. же at a 56 ES = Arr. Robert P. poni Man Do & Zhu Geh MALUS Book Review — P. Mick Richardson ....... | | 864 E E uF Erratum for Cenni. nitidissimum | var. pun (Blanca León - ын 225 865. ne jiang, China © Aaron Liston, Loren H. L1 d p 859